UC-NRLF Β ^ DEfi fi2b m v. ' : ; ■• ΥΛ Π ft! » ι. ■- * : • ; - : ■ : ; ( . c s'.-f f .•■; λ !> < ' i ,• s V \ • Γ χ '( - ΛΓ. Ο* i «2 REESE LIBRARY OF THK UNIVERSITY OF CALIFORNIA. Received _. Access ions No. % /J~A*S?^ Shelf No 88Z- Otr ♦SO k \ :. m '/ /// Vv %• <* J -X' W&Pf/. ν* \.**> \\λ uSs *SS» V γ γ , .• \ Υ \ \ «J § W( \ IM γ Γ 1 1 X*V^ vli 4 • -Ml % -ι 1 >ι I I J 7/ / -r £& THE GBEEK VEEB LONDON : PRINTED BY 8POTTISWOODK AND CO., NEW-STREET SQUARE AND PARLIAMENT STREET THE G-BEEK YEEB ITS STRUCTURE AND DEVELOPEMENT By GEOEG CUETIUS PROFESSOR IX THE UNIVERSITY OP LEIPZIG TRANSLATED BY AUGUSTUS S. WILKINS, M.A. PROFESSOR OF LATIN AND COMPARATIVE PHILOLOGY IS THE OWENS COLLEGE, MANCHESTER AND EDWIN B. ENGLAND, M.A. ASSISTANT LECTURER IN CLASSICS IN THE OWENS COLLEGE MANCHESTER f LIBRARY UNIVERSITY OF { CxYLIFORNIA.. LONDON JOHN MURRAY, ALBEMARLE STREET 1880 All rights reserved ysi^ίου, καθηγητού του 4ν 'Αθήναις Β' γυμνασίου, "Εκδοσις ιτίμπτη δλωι μεταρρυθμισμένη, Έν 'Αθήναις 1877, which was kindly sent to me by the author while I was engaged on this second edition.' PREFACE. [7] former pupil at my suggestion. From Hesychius's Lexicon, which I have gone through in the course of my investigations, I have been able, while rigidly excluding all foreign and doubtful matter, to extract many remarkable forms. The rich treasury too con- tained in Lentz's Herodian has been laid under contribution. For all that, such is the astounding wealth of forms which Grreek possesses, that, with the best of intentions, I have fallen far short of absolute completeness, even within the limits here proposed. Still I hope it will now be approximately possible to ascertain the extent to which the phenomena discussed by me were in living use. It is now and then surprising to find, in the course of this enquiry, how forms, which boys at school learn as the proper and regular ones, either have no authority whatever or only occur in some out-of-the-way place, and stand quite alone. 'There is nothing so prejudicial to an insight into the real structure of the Greek verb as the notion, still widely prevalent, that every verb must admit of being conjugated throughout. In reality, not only does each single group of forms make a separate whole, but very often one such group is formed from one and the same stem many centuries earlier than the other, and — leaving the latest stratum of derived verbs out of the question — almost every verb shows us, so to speak, a separate family, with its own family history and a quite individual stamp of character. It may be doubted if there is another language which has developed this tendency towards individuality so far as that of the Greeks. c Next to the formation of a complete collection of the charac- teristic forms from Greek itself I have made a point of com- paring with them whatever forms can be directly compared from the related languages. It is no slight help towards an insight into the origin and ramification of forms of language if we can see clearly how often in two languages, e.g. Grreek and Sanskrit, or even in more than two, precisely the same form has come from the corresponding stem. In the case of the present- tense formations no comprehensive attempt of this kind had ever been made. All that had been done was to point out similarities of formation without taking the trouble to consider the stems in which they appeared. It may surprise many scholars to find how extensive the agreement between the languages is, even in the [8] PEEFACE. case of forms of such comparatively late stamp as the derived verbs. 6 The oftener we are led by investigations of this kind into regions in which the ground is slippery, the greater the importance,. I think, which must be attached to such bare collections of un- doubted facts about which it is hardly possible there should be two opinions. For the etymologist I have undertaken the collection of such facts in my " Principles of Greek Etymology." The present work is intended to provide, in a similar collection of verbal forms, a firm basis for the investigation of their origin» 4)n this head I have only ventured with some reluctance upon the (very difficult questions of the genesis of verbal forms. These questions must be dealt with by analysis and combination — a province quite distinct from that of the comparison of parallel forms. I have expounded elsewhere (" Zur Chronologie der indo- germanischen Sprachforschung," 2nd edition 1873) my views on the origin and developement of the Indo-Grermanic verbal struc- ture. These views, which I still hold, in spite of some amount of opposition, are naturally those on which I proceed in the present work. 2 With regard to the main questions they are the same views which began with Bopp's foundation of our science in the firm structure of his " Comparative Grammar," which were eluci- dated and corrected by Schleicher's systematising, though perhaps now and then too logical condensation, and may be regarded as the universal doctrine of Comparative Philology. No reasonable man will imagine that this structure is satisfactory at every point. It has its weak sides, and it is the strengthening, perfecting and correction of these to which the science must devote itself as it advances. Hard problems meet us, in which we have often to content ourselves with the indication of a greater or smaller degree of probability, and we must not fancy that we can settle every thing once for all. But I confess that the attacks lately made from different quarters on the foundations of this structure seem to me not at all likely to shake them. f The principal works used for the second edition have been, 2 [A statement of these views, revised by Professor Curtius, will be found in the article on the Greek Language in the ' Encyclopedia Britannica,' vol. jri. (ninth edition).] PREFACE. [9 j before all, Delbriick's " Altindisches Verbum " (Halle 1874), Johannes Schmidt "Zur Geschichte des Vocal ismus " Vol. II. , Grust. Meyer u Die mit Nasalen gebildeten Prasensstamme n (Jena 1873).' In the preface to Vol. II., Professor Curtius writes :-^ 6 This second half of my description of the structure of the Greek verb has not led me so often as the first to the ultimate and most difficult questions as to the origin of the earliest Indo- Grermanic verbal forms. I have had to deal rather, though not by any means exclusively, still for the most part with the completion and carrying out of primitive types by the Greeks ; although these can be understood only by bringing out the special charac- teristics of the Greek verb from the common back-ground. Much however that bears upon this has now presented itself to me in a different light from what it did formerly ; and the doctrine of the perfect especially, which in consequence of the peculiar stamp and varied ramification of this tense takes up a very considerable part of this second volume, is stated here in a manner which in many respects is new. For the perfect I have very thankfully availed myself of the researches of old pupils, of which some are collected in the " Philological Discussions published by Or. Curtius's Grammatical Society " (Leipzig 1874), others are printed in the " Studien," while Windisch's description of the Irish perfect, which I have found instructive on many points, has been printed in Kuhn's " Zeitschrift " Vol. XXIII. But many other chapters too, e.g. that on the Verbal Nouns, and specially the doctrine of the Infinitive, and the description of the Sigmatic Aorist, contain views differing from those most generally adopted. I trust they may recommend themselves to the unprejudiced judgment of other investigators. I cannot, I think, be charged with baviug clung obstinately to doctrines which I previously advanced. On the contrary, I believe that I have never refused to accept more recent views and tendencies, so far as they appeared to me at all justified, without however deviating from the fundamental principles which I followed in my first discussion of the Greek Verb thirty years ago.' The translation of the first volume has been executed by Mr. England, that of the second by Mr. Wilkins, but every page has [10] PKEFACE. been carefully revised by us both, and we are jointly responsible for the whole. The very full indexes to the original work were prepared by Dr. Vanicek of Neuhaus : the task of adapting them to the present translation, involving as it did the verification and alteration of more than 5,000 references, has not been a light one: but it is hoped that they will prove of great value in facilitating the use of the book. The numbers in the margin refer to the pages of the second edition of Vol. I. and of the first edition of Vol. II. in the original. It may be convenient to notice that the second edition Of Vol. I. contains eight pages more than the first, while the second edition of Vol. II., in consequence of the insertion of the excursus at p. 33, will probably contain about twenty pages more than the first. Hence e.g. a reference to p. 206 of Vol. I. 1 will -answer to p. 211 Vol. I. 2 (p. 143 of the translation): p. 370 Vol. I.^p. 376 Vol. I. 2 (p. 258 of the translation): p. 100 Vol. II. 2 =p. 84 Vol. II. 1 (p. 329 of the translation). The kindly welcome given to our translation of the ' Principles of Greek Etymology ' leads us to hope for an equally favourable reception for a work which has been universally recognised as a not less important contribution to the cause of a sound and scientific knowledge of the Grreek language. Manchesteb : Christmas, 1879, TABLE OF CONTENTS. CHAPTER I. PAGE INTRODUCTION . 1 CHAPTER II. THE PERSONAL TERMINATIONS . . . 24 . Active . ... 24 First Person Singular .... 24 Second Person Singular 31 Third Person Singular 37 First Person Plural 41 Second Person Plural 44 Third Person Plural 45 Dual 50 II. Middle .... 55 First Person Singular 57 Second Person Singular 59 Third Person Singular 60 First Person Plural 61 Second Person Plural 63 Third Person Plural 64 Dual Forms 67 Excursus on the σθ 68 CHAPTER ΙΠ. THE AUGMENT 72 A) The Syllabic Augment . 76 1) Double Consonants following the Augment 77 2) Syllabic Augment before a Vowel 78 B) The Temporal Augment 87 C) Absence op the Augment 91 D) The Position of the Augment 94 [12] TABLE OF CONTENTS. CHAPTER IV. PAGE PRESENT STEMS WITHOUT A THEMATIC VOWEL . 96 I. Monosyllabic . . 96 II. Stems of Two or Three Syllables . 105 CHAPTER V. AORIST STEMS WITHOUT A THEMATIC VOWEL . . 125 I. Monosyllabic 126 II. Disyllabic 13a CHAPTER VI. THEMATIC PRESENTS FORMED WITHOUT ANY FURTHER STRENGTHENING OF THE STEM . . .13$ CHAPTER VH. STEMS WHICH LENGTHEN THE VOWEL IN THE PRESENT . 150 I. Diphthongal Intensification . . . . . .153 H. MONOPHTHONGAL INTENSIFICATION 156 CHAPTER VHI. THE T- CLASS 16a I. Labial Stems .164 II. Guttural Stems . . 168 HI. Vowel Stems . . .168 CHAPTER IX. THE NASAL CLASS 169 I. Presents in -ν», -νομαι 178 Η. Presents in -α-νω, -α-νο-μαι 180 IIL Presents in -ναω and -wau> * 18S IV. Presents in -νεω, -νεομαι 184 V. Presents in •**, which point to an older »vm . . . .185 CHAPTER X. THE INCHOATIVE CLASS 187 Ι. •σκω ADDED DIRECTLY TO VOWEL ROOTS 192 Η. -σκω added το Consonantal Hoots which have become Vocalic by Metathesis 193 TABLE OF CONTENTS. [13] PAGE ΠΙ. -σκω ADDED TO VOCALIC STEMS OP TWO OR MORE SYLLABLES . .194 IV. -σκω AFFIXED AFTER THE ADDITION OF A SHORT VOWEL . . . 195 V. -σκω ADDED IMMEDIATELY TO CONSONANTAL ROOTS . . . .196 VI. A TRANSFORMED -σκω 197 CHAPTER XI. THE I- CLASS . . . . . .201 I. Presents in -ιω 207 II. Presents showing the Effects left by an earlier -jw . .211 A) Verbs in -λλω from -λ/ω 211 Β) Verbs with Epenthesis of ι . . . ...•"•.• . . . 213 C) Presents in σσ (ττ) 218 D) Presents in ζ 222 APPENDIX TO THE I- CLASS . . . . 229 Denominative Verbal Formation 229 I. Vocalic Division 234 1. Verbs in -a», -ata>, -αζω . . 234 2. Verbs in -οω, -οιω, -οζω 238 3. Verbs in -e», -eta, -*ζω and -ιζω 239 First Excursus. — On the Interchange and Meaning of the Verbs in -αω, -οω, -e« 244 Second Excursus. — On the Inflexion of the Contracted Verbs . . . 246 4. Verbs in -ιω and -ιζω 249 5. Verbs in -υω and -υζω 250 6. Verbs in -et /ω and -ουω 251 II. Consonantal Division 253 1. Derived Verbs in -m 253 2. Derived Verbs in -ρω 255 3. Derived Verbs in -λ» 255 4. Derived Verbs in -σσω (-ττ«) . ...".. . . 256 CHAPTER ΧΠ. THE E- CLASS AND THE RELATED FORMATIONS . 258 1. Presents in -€« with Forms from a shorter Stem in other Tenses . 262 2. Presents without an -e by the side of other Forms with c or tj . . 263 3. Both Formations side by side in the Present 268 4. Ε -Formations in other Tenses than the Present, where the Present- Stem is expanded in some other way 270 APPENDIX TO THE E- CLASS . . • . .273 [14] TABLE OF CONTENTS CHAPTER XIII. PAGR THEMATIC AORISTS 275 I. AORISTS WITHOUT REDUPLICATION 27S II. Aorists with Reduplication 28& CHAPTER XIV. THE MOODS OF THE PRESENT AND SIMPLE AORIST STEM. 296 I. Imperative . . .296 A) Second Singular Active 297 B) Second Person Middle . > . . 304 C) Third Singular, Active and Middle 305 D) Third Plural, Active and Middle 306 E) Dual Forms 310 Π. Conjunctive 311 HI. Optative 324 CHAPTER XV. VERBAL NOUNS OF THE PRESENT AND SIMPLE AORIST STEM 338 I. Infinitives 338 II. Participles 351 CHAPTER XVI. THE PERFECT STEM AND THE FORMS CONSTRUCTED FROM IT 354 I. Reduplication in the Perfect 356 A) With an Initial Consonant 356 B) With an Initial Vowel of the Stem 365 C) Loss of Reduplication 370 D) Position of the Reduplication 373 Π. The Active Perfect 381 A) Personal Terminations of the Indicative 381 B) Formation of the Stem 386 a. Relics of the Primary Formation 386 b. Formation of the Stem by an added Vowel . . . . 388 c. Changes in the Vowel of the St em- Syllable . . . 395 d. Consonantal Changes in the Stem• Syllable . . . 403 e. The Perfect with κ 408 III. The Middle Perfect 416 IV. Moods of the Perfect 422 V. Verbal Nouns of the Perfect 424 TABLE OF CONTENTS. [15] PAQB VI. Tendencies towards Sigmatic Perfect Forms . . . .427 VII. The Pluperfect .... 428 A) Active Pluperfect . . .428- B) Middle Pluperfect 434 VIII. The Future from the Perfect Stem 436- CHAPTER XVII. THE SIGMATIC AORIST 437 A) Relics of a Primitive Formation 445 B) The Ordinary Formation 449 C) Irregularities 460 CHAPTER XVIII. THE FUTURE 467 I. The Sigmatic Future 468 II. The Future without σ 475 A) From stems in λ μν ρ 475 Β) From other Stems 478 C) Other Futures without e 483 III. Moods and Verbal Nouns of the Future . . . . . . . 485- CHAPTER XIX. THE PASSIVE STEMS 488 I. The Passive Stem in -η 491 II. The Passive Stem in -θη 498 θ ELSEWHERE THAN IN THE PASSIVE STEM .' 500 A) Presents in -θω 501 Β) Formations further derived 503 C) Meaning of these stems . . . . . . . . . 504 Similar Formations in other Languages 506 Origin of the Syllable θ-η (0c) in the Passive Aorist . . . 507 CHAPTER XX. THE VERBAL ADJECTIVES . . . . 511 CHAPTER XXI. IRREGULARITIES OF THE VOWEL STEMS IN THE FORMA- TION OF THE PERFECTS, FUTURES, PASSIVE AORISTS AND VERBAL ADJECTIVES 616- [16] TABLE OF CONTENTS. CHAPTER ΧΧΠ. PAGE THE ITER ATI VES 527 CHAPTER XXHI. DESIDERATIVES, INTENSIVES AND FREQUENTATIVES . 633 CHAPTER XXIV. ANOMALIES 538 EXCURSUS. On some Eecent Explanations of the Vocalism in the Thematic Aorist . . 545 INDEXES. A. Gbeek 553 B. Italic 578 C. Sanskrit 581 D. Ibanic 583 E. Teutonic . 584 F. Letto- Slavonic 584 G. Keltic .585 THE GBEEK "VERB. CHAPTER I. INTRODUCTION. The term verb is not always used by grammarians in the same sense. I We hear, on the one hand, of the verb in a sentence, and 'άειίε or ϊθηκε is said to be a verb ; on the other, the same term is used to denote the numerous forms which along with this άειΰε or εθηκε belong to one stem, and we hear of the verb άείΰειν, or the verb τιθέναι. Dionysius Thrax used the word in the former sense when he thus denned the notion of the verb : ρήμα ίστι λεζις κατηγόρημα σημαίνουσα (Bekker, 1 Anecd.' ii. 672). Schoemann, in what he says about the nature of the verb, at p. 16 of his treatise on the Parts of Speech, agrees with this definition. It is in the power of making an assertion, i.e. in the union of a subject with a predicate within one and the same word, that the peculiarity of this, as opposed to other parts of speech, consists. For our present purposes we shall have to distinguish more carefully between these two uses. We do not call αεώε or εθηκε verbs, but verbal forms, and only use the term verb collectively, understanding by a verb a more or less extensively ramifying system of forms, all of which possess the power of making an assertion, and come from one stem, or are at least held together by the bond of a common meaning. Again, all Greek verbs join to form that higher unity, which, as the sum and substance of the whole mass of kindred phenomena, we may call ' The Greek Verb.' If we proceed to consider a single verb or system of verbal forms from a, so to speak, statistical point of view, this system at once falls 2 into two main groups, which have at all times been kept distinct, though the origin and nature of this difference may not have been understood : these are the verb finite, and the verb infinite. It is only the forms of the verb finite which are capable of expressing a complete assertion, or, in other words, of making little sentences which can be conceived as standing alone. In the forms of the verb infinite there is always an incompleteness in the assertion, which needs to be supplemented by a form of the first kind. Infinitives, participles, and verbal adjectives bear the form of nouns, and belong by right of origin and structure in most instances to the class of noun-forms. But since the language as preserved to us makes a distinction of use between these forms and those called noun- forms, in the strict sense of the term, and since they have several charac- teristic distinctions of form in common with the verb proper, they make INTRODUCTION. CH. Τ. an integral part of the verbal system, and ought not to be considered apart from it — a fact not always recognised in Comparative Grammar. Their double nature was well set forth in the old name μετοχή (partici- pium) j the only pity is that this name was confined to a part only of the group. We shall find it best to speak of the whole class as verbal nouns. The elements of meaning which find expression in the Greek verb finite are of six kinds: 1), Person; 2), Number; 3), Relation borne by the action to the subject, the difference, i.e. between Active, Middle, and Passive, so well named by the ancients ϊιάθεσις ; 4), Kind of Time (Zeitart), by which I mean the varieties of the continuous, momentary, and completed action perceivable, i.e. in ελνε, ελνσε, and ελελύκει respec- tively ; 5), Grade of Time (Zeitstufe), or the difference between present, past, and future ; and 6), Modality. The verbal nouns have no means at all for expressing the first element, but they can all express the third and fourth. Of the fifth, the grade of time, they never had any mark, 3 though, in the case of the participle, a peculiar shifting of function has enabled the distinctive mark of the kind to do duty for that of the grade of time ; and besides this, the latest born of the tense-systems, that of the future, has produced verbal nouns of its own. Participles are of course the only verbal nouns which can mark number (2), and, being real adjec- tives, they mark gender and case as well. Even modality (6) is not entirely absent from the meaning of Greek verbal nouns ; the particle av is added to infinitives and participles with a use analogous to that with verb-forms proper, and in this way some modal differences at least find expression outside the verb finite. The array of forms, which this calculation shows the complete verbal system in Greek to possess, is astonishingly large. As it is seldom or never seen in all its force, it will be worth while to pass the long list under review; Considered genetically, the whole of the forms of the Greek verb divide themselves into seven groups, which, as all the forms in a group have a common unchangeable kernel or stem, we refer to seven stems, or, more accurately, tense-stems. In reviewing these in this statistical manner, we will for the time preserve the order given in my * School Grammar,' i.e.: 1), Present-stem; 2), Strong Aorist-stem (Aor. II.); 3), Future-stem; 4), Weak Aorist-stem (Aor. I. act. and middle); 5), Perfect-stem; 6), Strong Passive-stem (Aor. II. pass.); 7), Weak Passive-stem (Aor. I. pass.). Of these seven groups those called strong and weak are seldom both developed in the same verb, so that, with comparatively few exceptions, each verb can actually show only five groups, the active and middle aorist being either strong or weak, and the passive stem likewise. These five groups, however, may be found entire, but for a few gaps, in a great number of verbs, and so far in current use that we may confidently affirm that there is no single form belonging to one of these groups that a Greek, when Attic was at its prime, could not have used if he liked. That our present attempt at reviewing the numerical strength of this store of verb-forms may give us no deceptive phantom results, but a real 4 idea of the number of actually occurring forms distinct in sound and meaning, I shall proceed on the following principles. In the first place, all very rare forms, e.g. the first person dual of the middle tenses, the feminine dual of the participles, the moods of the active perfect, have been left out altogether. Next, all forms which though of different ch ι. NUMBER OF VERBAL FORMS. 3 meaning are phonetically identical, e.g. the nom. and ace. neuter parti- ciples, and even the phonetically identical ace. sing. masc. and nom. and ace. plur. neut. of the active participles (e.g. Χνοντα), and the identical ace. sing, and nom. and ace. s. neut. of the middle (e.g. λυόμενο v), always count for a single form. But where some classes of verbs or certain dialects have varieties of formation, or where the examination of their origin gives us clear evidence that there was a variety on Greek ground, in such cases no later or accidental identification has made us reckon as one what the language occasionally does, or at one time did, regard as distinct. For instance, ϊλνον is reckoned once as 1st pers. sing, and once as 3rd pers. plur., for the Dorians distinguished between εΧυον and εΧύον, and all Greeks between εψην and εφάν or εφασαν. Λύω, it is true, is 1st sing, for both ind. and conj., but as there are in Homer forms in -ω -μι for the conj., there was once a distinction between the two forms. Avr) is in Attic at once 3rd sing. conj. act. and 2nd sing. conj. middle ; but Homer distinguishes between Xvrjai and Χνηαι, and even Attic at one time between Xvy and λύει. The aor. I. inf. act. and the 2nd sing, imp. aor. I. mid. are at any rate occasionally distinguishable by their accent, e.g. παιΰενσαι and παίΰενσαί — reason enough for counting each form separately. On the same principles the 1st sing. fut. act., e.g. Χνσω for λυσίω, has been distinguished from the 1st sing. aor. I. conj. λύσω. On the other hand, it cannot be shown that there ever was a phonetic distinction in Greek itself between Χνετον, Χνεσθον as 2nd dual ind. and imperat., or between Χνετε, Χνεσθε as 2nd pi. of the two moods, and therefore such forms are only counted once. In this way we get the following result. From the present-stem are formed — Pr. Ind. Pr. Conj. Pr. Opt. Pr. Imp. Impf. Act. 7 7 8 4 8 Mid. 7 7 8 4 8 that is, 68 forms belonging to the verb finite. Besides these there are 5 2 infinitives and 2 participles with 19 each, that is, 40 forms of the verb infinite — in all 108. From the future-stem come — Ind. Opt, Act. 7 8 Mid. 7 8 that is, 30 forms of the verb finite, and then there are 2 infinitives and 2 participles with 19 case-forms apiece. Altogether from the future- stem 70 forms. From the aorist-stem, either the strong or the weak, come — Ind. Conj. Opt. Imp. Act. 8 7 8 6 Mid. 8 7 8 6 In addition to these 58 come the 40 forms of the infinitives and parti- ciples — in all 98. Under the perfect-stem we omit entirely the rare conj., opt. and imp. of the active, but not the imperative middle, which is more frequent. We thus get — β 2 4 INTRODUCTION. ch. ι. Perf. Ind. Imper. Pluperf. Fut. Act. 7 8 3 Mid. 7 4 8 Ind. 7, Opt. 8 altogether 49 forms of the verb finite; and. to this have to be added 3 infinitives and 3 participles — in all 109. The passive-stem, strong or weak as the case may be, gives — Acr. Ind. Conj. • Opt. Imp. 8 7 8 6 Fut, Ind. Opt. 7 9 44 which, with the 2 infinitives and the 38 participial forms, give a total of 84. The verbal adjectives, which belong to no tense-stem, produce 38 case- forms. In all, then, we may get from a complete verb — 249 forms of the verb finite, and 258 forms of the verb infinite altogether 507. 6 A glance at the Latin verb is enough to show us how much poorer it is than the Greek. The Latin verbal forms may be referred to two tense-stems, of which the second, that of the perfect, does not extend beyond the active. The present-stem has — Ind. Conj. Imper. Imperf. Ind. Imperf. Conj. Fut. Act. 6 6 5 6 6 6 Mid. 6 6 4 6 6 6 altogether 69 forms of the verb finite, to which must be added 2 infinitives and a participle with 8 different case-forms, and the gerundive with 12 case- forms 1 — that is, 91 forms in all. The perfect- stem has — 6 forms for the indicative perf. 6 „ „ conjunctive perf. 6 „ „ indie, pluperf. 6 „ „ conj. pluperf. 1 form for the fut. perf. for it is only the 1st pers. sing, that is different from the perf. conj. — in all 25 — which with the addition of the inf. act. make 26. Besides these there is the fut. part. act. with its 12 forms, the perf. pass. part, with the same number, and the 2 supines — in all 26. The verb finite reckons altogether 94 forms, the verb infinite 49 — total 143. Everything besides is periphrastic. In Gothic the resources are still more meagre. We can here only compare the verb finite, as the declension of the participles is so much more complicated that their sum cannot be clearly stated. The strong verb in Gothic as in Latin falls into two groups, here called present and past. The present group comprises in the indicative 7 forms, the 1 I have counted the same form only once when it does duty for more than one case, as e.g. legendi for gen. s. an'd nom. pi. ch. ι. NUMBER OF VERBAL FORMS. 5 3rd sing, and the 2nd plur. being identical, in the conjunctive 8, in the imperative only 1, as 3 forms are identical with the corresponding indie, forms ; then there are 6 middle forms. The past has 8 for the indicative and 8 for the conjunctive. The entire sum is therefore 38. The language that comes nearest to Greek in wealth of forms is 7 undoubtedly Sanskrit. Here all the three numbers have their three persons complete, so that each mood and tense shows 9 forms. Of moods and tenses there are 9, as the tenth system of forms, that of the participial future, being periphrastic, cannot be reckoned here. We thus get 81 forms of the verb finite in active, middle, and passive respectively — in all, therefore, 243, as against the 268 of Greek. Then there are the conjunctive forms and several optatives peculiar to the dialect of the Yedas, in which dialect, however, many of the later forms are wanting. Still the verbal system in Indian is on the whole, as Delbruck remarks (' Altindisches Verbum,' p. 15), not very sharply denned, -so that it hardly admits of this kind of calculation. - Owing to the fact that many verbs have alternative forms of the present- stem freely current side by side, the number often mounts up excessively. There are, for instance, from the rt. kar make, according to Delbruck, 336 forms of the present verb finite alone in Yedic Sanskrit. At a later stage again the language has a much smaller store to show than Greek. No doubt it would be the other way if we reckoned the causative, intensive, and desiderative formations as well. These count in Sanskrit grammar for integral parts of the regular verbal inflexion, whereas in Greek, where they are far less numerous, they are separate verbs. Since each of these derived formations goes through all persons in ■ the three voices, and has only the perfect formed by periphrasis, for every group of 81 forms of the primitive verb we get one of 72 — in all, that is, 216 for each derived formation, and 648 for all the three. Add to these the forms of the primitive verb, and there results the gigantic total of 891 genuine verbal forms. Still we should no doubt go wrong if we treated each and all of the forms in this tabulated grammatical system as actually existing. The whole list may be found conveniently arranged in Max Muller's ' Sanskrit Grammar ' (London 1870, p. 245 ff.). This much may, I believe, be positively asserted, that in the number of verbal forms in living use Greek hardly comes behind Sanskrit. This is in part con- • nected with the far finer distinctions of meaning which are to be found 8 in Greek. Without doubt both tense and mood systems are in the latter language more developed and more compact. If after this detailed survey* of the extensive stock of Greek forms we now try to understand how all this wealth originated, the first certainty we can arrive at is that its formation was a process of time. Of this fact we get some few but important indications from the period of the language's history, which may in the narrower sense be called historic, that, namely, which has left us written specimens. One of the most in- genious formations of the Greek verb, the weak passive future, is entirely wanting in Homeric Greek, and of the strong passive future there is but one certain instance, μιγήσεσθαι, which only occurs at Κ 365, for ΰαήσο- μαι has not the right meaning. These forms, therefore, were clearly not made till the time subsequent to that in which the Homeric Epic was in its prime. They were evidently made to supplement the long current passive aorists and on the analogy of the other futures. There had long 6 INTRODUCTION. ch. ι. been by the side of εβη βήναι a βήσομαι, by the side of τλήναι τλήσομαι, by the side of γοήμεναι γοήσεται — why should not μιγήμεναι have μιγή- σομαιΊ and later on, why should not κινηθήναι have κινηθήσομαιΐ espe- cially as these forms gave greater facility for the expression of passivity than was afforded by the middle forms μίζομαι κινήσυμαι. The future optative likewise is unknown to Homeric Greek. For no doubt La Roche is right in altering the completely isolated and not even well attested άλνζοι at ρ 547, and reading ονΰέ κέ τις θάνατον και κήμας αΚΰζ,ει after the analogy of other passages. This late growth is very significant, and teaches us much of the nature of the verb. While the system of cases not only receives no addition whatever in the period known to us by written records, but is actually curtailed, and while very con- siderable losses can be discovered within the limits of Homeric Greek, in the verb the power of putting out new shoots lasted much longer. In the use of the cases then it is remarkable to find older and nicer distinc- tions of meaning often replaced by a less delicate accuracy, and one case assuming the functions of another as well as its own. With the verb, 9 however, the case is the reverse; here we can, so far at least as tenses and moods are concerned, discern here and there the stamp of a greater delicacy and a more thoroughgoing completeness. We find analogy to be the means by which a still living creative force attains its ends, and we may conclude that analogy was also a material element in producing the results of yet earlier times. The impulse to cany through to the end what is once begun, to fill up the gaps in what was at first an isolated group of forms after the pattern of older types, is one which is specially characteristic of the Greek language. Hence it was comparatively late that the marvellous system we see before us reached its full completeness. By the side of this process of completion of the whole we can also trace a few less important innovations as they arise, e.g. the formation of the aspirated perfect, quite unknown to Homeric Greek, the extensive use of the κ in making the active perfect, of which again we see only the beginnings in Homer. Other Greek dialects are of considerable use in many directions in helping us to ascertain what we can of the phonetic relations of an older time before the division into dialects had taken place. But these are all isolated phenomena compared with the mass of forms which are unquestionably as old as Greek itself, and which prove, by the wide extent to which they accord with pheno- mena in related languages, that they were the common inheritance of all or at any rate several of the Indo-Germanic tongues. The task, therefore, which we have to perform, if we are to understand the structure of the Greek verb, can only be done by going back to the relations and conditions of the language in a pre-Greek age. The main parts of the structure were the work, not of Greeks, but of Indo- Germans far away in antiquity. Our investigation therefore, whether we are examining a single phenomenon or constructing a whole out of many, must always be of two kinds — reconstructive and constructive as well. The former is the easier task. Reconstruction has to take the forms of the several languages and conclude from them what the primi- tive Indo-Germanic form was, and to obtain by a systematic combina- 1 tion of such primitive forms a complete image of the structure such as we may conjecture it to have been before the first encroachments of deface- ment and decay. On the side of construction we have to ask with what ch. ι. GROWTH OF THE INDO-GERMANIC VERB. 7 notion was this primitive structure invented — how did it arise ? In so doing we try to transport ourselves in thought to periods which are still more ancient, when the language bears still less direct resemblance to that of later times. A clear perception of this twofold nature of our task is indispensable. There are cases where the two sides are, so to speak, at odds, where it is a question whether the surplus shown by one language or dialect compared with others is of primeval growth,, or an extraneous imitation of some other similar form — a question we shall have to raise, e.g. in the case of the full termination -μι in the 1st sing, optative. In the same way it is sometimes not easy, in the case of a sound by which a form in one language is distinguished from the form equivalent to it in another, to see at once whether this sound has always had a share in marking the significance of the form, and consequently is to be explained construc- tively, or whether it may not have arisen through a later dulling and special, purely phonetic development. In the latter case a reconstruc- tion is necessary before we can arrive at the older sound. Hence, though in theory it may be possible to keep these two sides distinct in the treat- ment of individual cases, it is practically inexpedient. What is of real importance is rather that we should never lose sight of either. Still, since the examination of the details of the Greek verbal structure cannot fail to be a distraction to the due consideration of the whole, and as at the same time it is of great importance that we should view the whole collectively, it will be expedient to summarise here, by way of introduc- tion, the most essential points of what seems to me ascertainable about the gradual origin of that verbal system which we may regard as already complete before the separation of the Indo-Germanic languages. To this may well be added a short examination of such objections as have been raised against some of the main points in this collective view, and a short estimate of the interpretations — some of them diametrically opposed to each other — which have been suggested by its opponents in its stead. To begin then with the positive part of these considerations ; of this much we may be sure, in the first place, that the Indo-Germanic verb,. so far 1 1 as we can by reconstruction arrive at its fundamental outlines, no more came into being all at once than did the Greek. It did not begin by being a ready-made system of all kinds of form, each with its clearly denned function assigned to it at its birth. Every attempt to conceive of the verb as a definite entity, after the fashion of the philosophising gram- marians of earlier times, or to show how it needs must follow this pattern and no other, is a mistake. This huge system of verbal forms, perhaps the most marvellous creation of the language- making mind of man, is a stratified formation. The science of language has long devoted its atten- tion to the right discrimination between these various strata of forms lying one above the other, of which the younger always presupposes and is qualified by the older. I have discussed these problems before in my treatise, ' Zur Chronologie der indogermanischen Sprachforschung,' 2nd edit. Leipz. 1873, but I must here repeat my main points. All formal structure in the languages of our stock consists essentially in the union of two elements, one with meaning and the other without ; that is, to adopt the usual phraseology, in the union of verbal roots with pronominal stems. Of this union two kinds are possible. Either it is attributive, that is, the pronominal stem is added to the more significant root with the same force with which at a later stage of the language an 8 INTRODUCTION. ch. ι. adjective or pronoun is said by grammarians to be joined attributively to a substantive : that is, ag-a (Gk. άγ-ό(-ς)), ag-man (Lat. ag-men) is like b άνίφ, or ούτος 6 ανηρ. The added pronoun has here no other force than that of pointing, like a local adverb ■ there, 7 to the notion expressed in the root, and bringing it into prominence just as an article might. This kind of union is the main source from which arise the formative suffixes and some of the case suffixes, especially those of the nominative• and accusative. The other kind of union is the predicative, the essence of which is that the added pronominal stem does duty as subject, and consequently turns the significant stem to which it is added into the predicate. In the clear separation of the predicative con- nexion from the attributive, while in their origin the two were hardly 12 distinguishable, lies the chef d'oeuvre of the Indo-Germanic formal struc- ture. By the more detailed arguments of the treatise above mentioned I believe I have shown that the predicative connexion was probably the one that was developed first in this stock of languages. Now herein lies the germ of the verb. When once a root like da was united to a prono- minal stem like ta in such a way that this combination da-ta meant that man giver, or he giver, and nothing else, a verbal form had been made, and when presently corresponding forms were made for the other persons too, the primitive forms being da-ma, da-tva, there existed a set of such forms, a small paradigm, with the consciousness of their inter-connexion as a necessary consequence. And as men's minds were already awake to the necessity of avoiding confusion, and care was taken to keep these forms distinct in sound from the attributive compounds, the verbs as a separate part of speech now existed once for all. The further steps taken before the end of this primitive period, — the expression of the plural by the union of several pronominal elements, and the expression of the middle voice by a different combination of the same, the prefixing of yet another pronominal stem, the augment as it is called, by which the grade of past time was marked off clearly from the grade of present, — all these we shall see more clearly when we come to the examination of the details. What we have to do here is rather to get a bird's-eye view of the process of development as a whole. This first stratum of verbal forms thus given in outline comprehends only such forms as occur, say, in the present indicative and imperfect of the Greek verb ψημ'ι. There is only one verb-stem here, and that as yet quite a simple one. Next to the indicative apparently the imperative was formed, as we shall see further on, and its characteristic mark* lies, as in the indicative, only in the personal terminations. But of any other mark of distinction of mood or of the kind of time there is not a trace. The capital gain of the first verbal period is essentially this, that there were now two sets of clearly stamped personal terminations for active and middle, and an augment. These possessions were lasting, and the i 3 distinctions thus struck out were made use of in all subsequently formed strata. The augment, where it was wanted, at the beginning, and the personal terminations at the end of the word made, as it were, the firm framework for all new productions whatever within the verb finite. Tt is at this point that the language appears to have made use, very early in its progress, of a means which it employs in the most varied ways for the emphasising of a syllable, namely, repetition or reduplication. Instead of the simple stem, e.g. da, there might appear within the same ch. ι. GEOWTH OF THE INDO-GEEMANIC VEEB. 9 verbal framework the reduplicated stem, e.g. dada, and so instead of da-ta dada-ta ; and, as this happened right through all the forms, there arose a twofold series : da-ma, da-tva, da-ta, &c, and dada-ma, dada-tva, dada- ta. These two series could hardly fail to be distinguished in meaning. In many cases the difference was this, that the first series was employed to denote momentary action, the second with its fuller forms to denote continuous. Here then we have the first materials for marking what I have called the kind of time. , Any further formative power therefore had hardly room to exert itself either at the beginning or end of the word, but had to confine itself exclusively to the interior. If we would understand other expansions, we must remember always that the verbal stem forms the predicate to the shifting subjects denoted by the terminations. Now this predicate may in a certain sense be compared to the later noun-stems, although quite at the beginning — that is, before the creation of verbal forms — the distinction between noun and verb did not exist. No doubt every rendering of primitive Indo-Germanic in language of a later development can be only approximative ; for there is in the essence of this oldest mode of expression an indistinctness which must of necessity give way to a greater distinctness in the case of a language which has been actually handed down by tradition. But if we are conscious that we are only very imperfectly reproducing the real meaning of those primitive formations, we may perhaps render da-ma by give I, da-ta by give he, and conjecture that the predicative syllable gradually acquired a force which was not very different from that of the afterwards clearly distinguished participle or nomen agentis, e.g. giving, giver. The idea 14 that a copula is needed here appears completely untenable even when viewed from the position of a later development of the language ; for sentences like ουκ άγαθον πολυκοφανίη have not been wanting in any period, and no doubt they were for a long time the only kind of predica- tions in use. 2 The next expansion of verbal forms then was brought about, we conclude, by way of the more elaborate specialisation of the forms and functions of the predicate. The means used was exactly the same as in the case of the noun-stems just mentioned, i.e. expansion of the stem. As noun-stems, even in periods of the language which are known to us, appear now without any suffix, now with several, so the verb-stem can be used without addition as well as with the addition of a suffix. The commonest and shortest suffix is the vowel a. s Instead of attaching the personal terminations directly to the rt. ag the noun- stem aga is formed from it, and this aga is then connected, e.g. with the sign of the third pers. sing, ta, later ti (aga-ti=ayei, agit), in the same way in which later the sign of the nominative case is added to the same stem attributively (aga-s=ayo-g) l An imaginable 1st pers. plur. ag-mas would bear to the actually deducible aga-mas=άγoμεc, agimus, the same relation as that borne by the Lat. noun-stem ag-men to an agi-men 2 [Cp. Eoby's Latin Grammar, ii. p. xxii.] 3 Fick's attempt to dispute the existence of the suffix a (Beitr. z. K. der Indo- Germ. Sprachen, vol. i. p. 1 ff.) seems to me unsuccessful. Nor can I see what gain is expected to result from dividing, e.g. *hha-ra instead of Mar -α, since it makes both syllables quite unintelligible. There is nothing to prove the priority of the verbal form. It would be just as lawful to deny that na and nu are nominal suffixes. 10 INTRODUCTION. ch. ι. which the analogy of regi-men will readily suggest. These α -stems so outgrew the older stratum in numbers, as time went on, that they de- cidedly formed the rule and turned the first stratum into a group of more or less anomalous exceptions. The original property of forming stems possessed by this a served to give to the stem still more of the character of a noun, and thus to mark the action denoted by it as a continuous, lasting one. This 15 explains the fact that this a, represented in Greek by ε or o, and lengthened in the 1st sing, to ω, belongs especially to the present-stem — that is, to that group of forms intended to express the action in its extent and duration. By the side of this a appear two more expansions of stem which a comparison of the related languages shows to be primitive, i.e. the syllables na and nu, about which little else can be said than that they are used to make other stems beside verb-stems. The syllables in the middle of σκίΰ-να-μεν, ορ-νυ-μεν are compared to the stem-forming elements in υπ-νο-ς (=Skt. svap-na-s, Lat. som-nu-s for 80])-nu-s), in the Skt., Goth., and Lith. su-nu-s son, in the Skt. dhrshnu-s bold, with which we may directly connect dhrshno-mi I am bold (rt. dharsh=GL• βαρσ). It is hard to see any peculiarity in these expansive syllables distinguishing them from the vowel a. Nor is it easy to find any further points of analogy between special forms of present-stems and noun-stems of a similar grade of formation. After these forms had established themselves, nominal and verbal stem-formation went each their own way. The intrusion of these stem-forming syllables into the framework of the verb can only be explained by supposing that at the time of its occurrence the forms had not yet completely set, so to speak, and that there still existed a sense that the terminations were the sub- jects and the stem the predicate. Of marks of case or number these noun-forms show not the faintest trace, and hence we conclude that the inflexion of the noun arose later. It is only in the period of stem- formation that the verb and the noun have anything in common. In this both are alike. But as soon as the noun-forms turned themselves by fresh suffixes, and especially by case-terminations, to polysyllabic formations, they became wholly unfit to be made straight into verbs. Reduplication is an internal, and the attachment of suffixes an external expansion of the stem. But the two methods may be com- bined. The stem that has been expanded externally may be inwardly strengthened as well. We find reduplication and lengthening of the stem- vowel side by side with the attachment of a suffix, especially of an a, and the latter, i.e. lengthening or intensification, becomes an important 16 distinction between different tense-stems. When a distinction arises between a stem bhuga and bhauga, lipa and laipa, we have again a twofold series of forms, and to the old binary stem-formation (the simple and the reduplicated) is added a new means of distinguishing continuous action (ψενγειν, λείπειν) from momentary (φυγείν, λιπεϊν). Meanwhile reduplication, sometimes in conjunction with the suffixed a, sometimes without it, furnishes the means of expressing the more intense, the com- pleted action, and thus when specially developed becomes the source of the perfect tense. All forms characterised by the expansion of the stem by a suffix we may distinguish from primitive forms under the name thematic. But in Greek grammar it is advisable to restrict the term thematic to those forms which show the vowels ε (η, ει) and ο (ω, ου) in regular interchange, or in other words, which belong to what has always ch. ι. GROWTH OF THE INDO-GERMANIC VERB. 11 been called the conjugation in Ω. This same interchange of vowels may be seen in the conjunctive throughout, and this is enough to show that this mood is a product of the period we have just been describing. But since a portion of the so-called verbs in Ml follow, as we saw, in their present stems in -va and -vv the same principle of formation, the term thematic is found inadequate. We shall find it more correct to call this class, as opposed to the primitive or radical stratum, the secondary, or — in so far as we here use the word stem in the sense of the already moulded and modified stem — the stem-stratum. Besides these, however, there is yet a third group of verbal forms, the analysis of which shows fresh elements in addition to those common to all verbal forms alike. The σ in ε-λνσα and λύσω, and the θ in έλνθην do not belong to the root, nor can they be compared with the expansive stem-suffixes used in the verb after the analogy of noun-stems. No noun-stems show anything corresponding to these elements. The source from which noun-suffixes are drawn is pronominal stems, but with these the syllables in question have little or nothing in common. Their origin must therefore be sought elsewhere. Bopp in his time recognised in them auxiliary verbs, and accordingly regarded the verbal forms so originated as compounds. This last expression, now in universal vise in comparative grammar, must be taken, it is true, in a limited and special 1 7 sense, since in the fullest sense of the word all verbal forms are com- pounds. But whereas in ψη-μι, ΰείκ-νυ-μεν, Χείττο-μεν we have a single verbal stem in connexion with one or more pronominal stems, there are in έ-λύ-σα-μεν, έ-\ν-θη-ν at least two verbal stems, and we can thus call the latter compound verbal forms with the same propriety as λογογράψο-ς, Χνσί-πονος can be called compound nominal forms. Still, to denote them more exactly, we shall find it better to use the more significant expression auxiliary forms, or auxiliary stratum. This third stratum of necessity presupposes the other two; for if there were no verbs there could be no auxiliaries to use. Auxiliaries are nothing but verbs which have lost their full meaning. It is the rule in language that the full, the significant, and the lifelike precedes that which is empty, inexpressive, and lifeless ; and every verb that has degenerated into a shadowy auxiliary must have first enjoyed full powers and an independent life of its own. Of this the auxiliary verbs in eveiy language afford the clearest proof in their etymology. In periods of which the language has come down to us verbs which originally had most clearly defined meanings, such as stand {stare, Fr. ete=status), remain, become (Germ, werden, orig. turn), to be bound (Germ, sollen), have, dwell (Goth, wisan [Eng. was\ Skt. rt. vas, dwell), have become mere auxiliaries, and are sometimes nothing more than a copula. The Indo-Germanic tongue must have possessed at least one verb that had degenerated into a copula before the separation of the languages, i.e. as- mi I am. It had, however, other verbs as well, most likely, which already had such small specific force that they could be used to express an action by conjunction with another stem, there being no doubt a shorter way of saying the same thing by the use of one stem only. We can, how- ever, distinguish clearly the first and second strata in the inflexion of the auxiliary-forms. The aorist ελνβην corresponds to εβην, that is, the auxiliary element here is primary or radical in its inflexion. "Ε-λι>-σα, on the other hand, like the simple εα for εσα —eram, shows an expanded stem having the ο added to its root. There must have long existed an 12 INTRODUCTION. ch. ι. εσα, or rather, as the whole formation belongs to the Indo-Germanic 18 period, an asa, before έλυσα arose. "We see from this that the third stratum presupposes both the first and the second. The rt. ές, moreover, must, before entering into this combination, have passed from the full meaning breathe, live, which it is highly probable, it first possessed, to the empty one which makes us give to e? vat the name of verb substan- tive, or even simple copula. There must in fact have already existed a verb substantive, inflected according to the rule of the first or second stratum, before the forms of the third stratum arose. If these hypotheses are granted, there is nothing extraordinary in our theory. Every form of the verb finite is a little sentence. Up to this time speakers had been content with sentences without a copula, in which the connexion between subject and predicate was expressed by the mere juxtaposition of the two, and now it was extremely natural to follow the analogy of sentences where the copula stood separate, and have a copula expressed inside the verb itself. While tdov, i.e. a-da-nt, . translated into the language of later times, would be turn dantes, ϊΰο-σαν, i.e. a-da-sa-nt, would be turn dantes erant. Two points only must be presupposed : first, a kind of fluidity about the verbal forms, in con- sequence of which the sense of the origin of the predicative syllables and their analogy to noun-stems had not yet been lost ; and secondly, an absence of marks of case and number at the time that the combination took place. An uninflected dik, uniting in itself the meanings showing and show, could easily combine with a following asmi I am, to make dik-asmi, shortened dik-smi ; so, too, asa-mi, the later by-form of as-mi, could combine with the same nominal stem to make dih-asami, shortened dik-sami, from which was formed the past tense a-dih-sam, i.e. Skt. ddiksham=ehi^a. : Besides the rt. as we find two other verbal roots used in the same way : the rt. dha do, and the rt. ja go. Later periods give us in- structive instances of the periphrastic use of these two roots. Inasmuch as every verb expresses an action, every verbal form can be replaced by the periphrasis of an abstract noun and the verb ' do.' The infinitive occurs oftenest in this connexion ; cp. e.g. the German er thut kommen, the English how do you do ? The rt. ja, on the other hand, is exactly 19 adapted to express circumstance, inasmuch as go is equivalent to go about, versari in aliqua re. Standing separately it has this force in the Lat. infitias ire, and the German spazieren gehen (to go a walking). Inasmuch, again, as there is in going the idea of motion towards a goal, go — cp. the Fr. je vais /aire — can acquire the meaning strive, pursue, and thus become the source of marks of mood and of the future tense. Finally, it can be used for the passive, as we have it used in venum ire, as the opposite of venum dare. For go is an intransitive verb, and as such stands in a kind of opposition to doing, and the notion ' to get into such and such a plight,' supplies a link between it and the expression of passivity. Since two or more auxiliary elements of this kind can be combined, there arises the possibility of a large number of forms which partly serve to supply the deficiencies of the older strata, especially in cases where phonetic difficulties have arisen, partly offer an opportunity of conveying various meanings which the means at hand are quite or partially unable to express. In the course of time this third stratum outgrew the two earlier ones. ch. ι. OBJECTIONS TO THE AGGLUTINATIVE THEORY. 13 Of course the origin of the auxiliary elements was soon lost to view, just as in the second stratum the sense of the way in which the stems had been expanded had soon disappeared. But these syllables, beginning as they did with a consonant, and capable of symmetrical adjustment to all kinds of stems, met 'the wants of what were relatively late periods. There was not so much force wanted for their articulation nor so great a nicety of distinction required as there was for the production of the more delicate and finer formations of a previous age, which, now that they had themselves provided a pattern for the younger generation of formations, became more and more antiquated, though fortunately they were, too numerous ever to become entirely obsolete. This short sketch of the gradual genesis of the Greek verbal forms is on the whole in accordance with the views which since Bopp's time have obtained among comparative grammarians, and which have only been modified in single points here and there by further investigations, among which those of Schleicher may be named as the most conclusive and com- prehensive. It is scarcely surprising that in so difficult problems there 20 have been differences of opinion on certain points. Still, since the ap- pearance of Bopp's ' Conjugations-system ' the main outlines have received general recognition. Such independent enquirers as Pott, Benfey, and Schleicher have been here completely at one with Bopp. Jacob Grimm, who is repeatedly spoken of by the below-mentioned opponent of the received theory as opposed to Bopp, expresses himself (' D. Gr.' i. 1051 ff.) as essentially of the same opinion. Those who treat the philosophical side of language take the same view. For this it is enough to refer to "W. v. Humboldt's treatise 'Ueber das Entstehen der grammatischen Formen ' (< Ges. Schr.' iii. pp. 290, 297), and to Steinthal's ' Charakteristik der hauptsachlichsten Typen des Sprachbaues ' (p. 285 ff.). As we shall see later on, even before Bopp, Buttmann was of the same view with respect to one of the main points, i.e. the origin of personal terminations from suffixed pronominal stems. This really remarkable unanimity has been met by a very decided opposition from two sources : first from Westphal in his * Philosophisch-historische Grammatik der deutschen Sprache,' and later in his ' Methodische Grammatik der griechischen Sprache,' in both of which books the view we have adopted is called ' die Bopp'sche Agglutinationstheorie,' and most emphatically denounced as erroneous. The second attack has been made by H. Merguet, who in his book ' Die Entwickelung der lateinischen Formenbildung ' (Berl. 1870) makes radical objections to several of the main points. 4 A scientific oppo- sition to widespread views is itself a useful stimulus and may lead to greater 21 certainty if it can be shown to be ill-grounded. I think therefore that it is worth while to make a brief examination of these objections, and for 4 Merguet has since given repeated expression to his views, but, as far as I can see, without going more deeply into the question, or subjecting the views he combats to a thorough examination. I may refer specially to his latest work, TJeber dan Einfluss der Analogie und Differ enzimng avf die Gestaltung der Sprach- formen, Konigsberg, 1876 -Similar doubts have been expressed, though with more reserve, by Bergaigne in the Memoir 'es de la Societe de linguistique, vol. iii., who partly follows Alfred Ludwig {Der Infinitiv im Veda ; Agglutination oder Adap- tation'). A. H. Sayce, who, in his Principles of Comparative PMMogy (2nd ed. Lond. 1875), opposes Bopp's theory in many important points, still, at p. 294, accepts its explanation of the personal terminations. 14 INTRODUCTION. ch. ι. a moment to look the new theories of our opponents straight in the face. In so doing we must treat separately the origin of the personal termina- tion and the construction of compound verbal forms. As far then as the personal terminations are concerned, Westphal acknowledges the phonetic similarity between them and the stems of the personal pronouns, but he adopts the view advanced, though not very positively, by Karl Ferd. Becker, according to which the personal ter- minations, and, as Westphal holds, the middle ones, came first, and the personal pronouns were afterwards formed from them. The positive part of his view we shall have to examine later on, but first of all we must enquire into its negative side. What are the grounds then which decide Westphal to abandon a theory that is so widespread and, as it seems to me, so well considered? It almost looks as if he thought the name ' Agglutinationstheorie ' enough in itself to arouse a feeling of abhorrence in every thoughtful mind, as in fact the expression ' Bopp's Agglutinationstheorie ' — for Bopp's name others with equal capriciousness put Schleicher's — has subsequently been used here and there in a like contemptuous sense. The reasons casually adduced by Westphal are mightily meagre. In spite of repeated perusal I have only been able to discover three definite objections. The first rests on the difference between the termination of the 1st sing. act. mi or m and the nomina- tive of the first personal pronoun. ' Those,' he says ('Philos. Gr.' 129), ' who take the view contrary to mine and explain the termination of the first person in the verb by supposing the attachment of a word which already had its own meaning of /, are forced into a grave self-contradic- tion, for the stem ma to which they have recourse has no meaning but me, to me, and never that of /.' This objection is not hard, I think, to disable. It seems to me that the difference between the nominative and the oblique cases is one which language took cognisance of, not when stems were formed, but after inflexion had begun. No one ever said that a nomi- 22 native ma was the source of the termination mi, but a stem ma, which, like every other stem, possessed the faculty of producing various cases in a period subsequent, as I think I have shown, to that of the origin of verbal forms. That a stem should in itself be adapted only for a certain set of cases and not for others seems to me as inconceivable logically as that a verb-stem should be adapted only for certain persons, moods, or tenses. All these things are merely accidents affecting the substance of the stem after it has taken shape, not before. There is nothing of the kind to be seen in the pronominal stem tva for the second or ta for the third person. If then in the language of later times the stem ma forms no nominative, it must be held in so far defective : it must have left off forming a nomi- native. We find something similar in the case of the stem ta. This stem developes no nom. sing. masc. and fern, ta-s ta as an independent pronoun, but the nom. plur. ta-i, tds is enough to show us that there is no conceivable obstruction producing this defect, and compound forms like αυ-τό-ς and is-te prove conclusively that there is no such thing as the creation of stems for oblique cases alone. Westphal, it is true, regards the assumption that the stem ma may once have had the power of denoting the subject as an hypothesis that we have no right to make. 5 But how is it possible to discuss the first estab* 9 How little scruple Westphal has to assume even for Greek forms not sup- ch. ι. . OBJECTIONS TO THE AGGLUTINATIVE THEOEY. 15 lishment of linguistic forms which undoubtedly took shape in very early times, if we do not use hypotheses 1 Does Westphal then make no hypotheses when he assumes a language without personal pronouns, assumes personal terminations to have arisen from ' essentially meaning- less ' vowels and consonants, taken quite at will and presumed to ' occur naturally ' to the primitive Indo-Germans 1 I think these such violent and improbable hypotheses that by their side the assumption that ma was defective seems perfectly innocent. Why, where are we to look for 23 a language without personal pronouns 1 How are we ever to conceive of a verbal structure so elaborate, with the most accurate means of denoting the I, thou, we, &c, if the language was not at the same time able to ex- press the corresponding persons when standing by themselves, able, how- ever imperfectly, to express ' to him ' or * him,' ' to thee ' or ' thee ' some- how or other ? On Westphal's hypothesis this must have been impossible until this process in the verb was completed. And how is it, if the per- sonal terminations really did fall like drops from the body of the middle voice, or like ripe apples from its branches, that notwithstanding there is so very little likeness between e.g. the plural of the middle terminations and that of the personal pronouns 1 He is obliged to admit in his ' Greek Grammar,' i. p. 391 ff., that even after applying all the 'euphonic' sounds, 'fulcra,' &c, which he has at his beck, he finds the stem of the second person plural ' completely unintelligible.' But if it is necessary, before we can explain the production of the independent pronouns in the plural, to find other tendencies at work than those which are discernible in the personal terminations of the verb, the whole of Westphal's hypo- thesis falls to the ground. A second objection deals with the relation of the secondary to the primary terminations. Westphal will not allow us any right to derive the former from the latter by loss of sound. In the preterite, he says, we never find mi, si, ti, nti, and are not justified in assuming it to have had these forms once. But here, too, the received theory is supported by analogies which are beyond doubt, and which even Westphal cannot reject. If the poetical εσσί had not been preserved we should not have a single Greek second person singular with the full termination j in all other cases the ι has disappeared. In Latin there is no mi, si, ti. The i has been completely lost, with the exception of a single trace in the Carmen Saliare. In the first person plural it is only Vedic Sanskrit in its -masi, which there occurs more often than -mas, and the Zend -mahi, which have kept the i, which we must undoubtedly assume for the original Indo-Germanic tongue. In the perfect active in Sanskrit the personal termination of the 1st and 3rd sing, has disappeared, the primary ending 24 of the 3rd plur. (us by the side of anti) is considerably abbreviated. In short, the rejection of final vowels, especially by polysyllabic forms, is among the best-established facts of the history of language, and since it is quite impossible to understand the verbal system without some recon- struction, there is no excessive boldness in presupposing similar processes to have happened in the earliest period of the genesis and first estab- ported by any authority may be seen from what he says at p. 75 of vol. ii. of his Gk. Gr. : ' We must assume that there was at an earlier stage of the Greek language not only a \eyere say ye, &c., but also Keyov I should like to say, and λ4*γομ€ν we want to say.' 16 . INTRODUCTION. .ch. ι. lishment of these forms when we have such clear analogies to guide us. The moderate assumption of such losses, even for so early times, is justified by the fact that all inflexion not only allows but necessitates some degree of weakening of the constructive elements added to the body of the word. A third argument, on which our opponent lays stress, deals with the η of the 3rd pers. plur. (nti, nt, ' Gk. Gram.' p. 79). He holds that ' it is impossible to discern a mark of the third person in each of the two elements η and t so as to give probability to what analogy would show to be the primary meaning,' i.e. he and he. We shall see below, however, that the pronominal stem an provides us with a satisfactory explana- tion, and this was recognised long since by Schleicher, though to this the author of the ' Methodical Grammar ' did not choose to pay any attention. I have not been able to discover any other objections to the received theory. It would rather seem that this scholar, who many years since showed himself, by his valuable investigation of the laws of final letters in Gothic, to be an acute enquirer, but who has paid little atten- tion to the literature of linguistic science since that time, has been really driven, by a line of argument that does not touch the Indo-Germanic languages at all, to represent the construction of Indo-Germanic speech as different to what all previous enquirers have thought it. In the preface to his * Philosophisch-historische Grammatik,* p. xii, he says : 1 There is no self-evident ground for the assumption that all phenomena of the oldest and most primitive store of Indo-Germanic and Semitic inflexions must necessarily have arisen by agglutination, and admit of 25 absolutely no other explanation or analysis.' With respect to the possi- bility of inflexions of a different origin, he appeals especially to Arabic, saying that we have here * a class of inflexions of the noblest and oldest kind, and here not even an attempt can be made to refer the inflexional endings a, i, u, an, in, un (for this triplet of pure vowels is the basis of the later terminations which were dulled by e and o) to pronominal or significant roots.' But we must not be too sure of this. It is main- tained, e.g. by Dillmann, a scholar of some note (' Aethiopische Gramm.' p. 254), in spite of Westphal's veto, that the a of the ace, by the side of which there occurs in ^Ethiopian ha as well, is a primitive ' impersonal demonstrative particle, meaning here, there, identical with the Hebrew Πτ- of direction.' Besides, these elements belong to the formation of cases, others adduced by Westphal to that of moods ; and so even if they could not be shown to have arisen from the adhesion of originally independent stems, this would prove nothing about the personal termina- tions. That these arose in Semitic from pronominal stems seems generally admitted (cp. e.g. Gesenius, ' Hebr. Gr.' (21st ed.),p. 80 ; Dill- mann, * Aethiop. Gr.' p. 161), and is with respect to many of the ter- minations so evident as hardly to admit of a doubt, especially as the Semitic terminations share with the independent pronoun the power of marking gender, a power which no other verbs possess. This last fact proves clearly that here, as shown by Schleicher, ' Ueber Nomen und Verbum ' (Abh. d. k. Sachs. Ges. d. Wissensch. hist.-pliilosoph. Abth. iv. p. 514 ff.), the distinction between noun and verb has not yet been quite clearly drawn. This theory of agglutination which Westphal attacks is supported ch. ι. . OBJECTIONS TO THE AGGLUTINATIVE THEORY. 17 by an almost incalculable number of facts, and takes account throughout of tangible magnitudes. It is a fact that in a large number of languages the personal terminations are absolutely identical with the possessive suffixes. Compare, e.g. the Magyar (Schleicher, ut supra, 527) — vart-am I have waited hal-atn my fish vart-ad thou hast waited for him hal-ctd thy fish vart-a he has waited for him hal-a his fish vart-unk we have waited. hal-unk our fish. Who can doubt here that in both cases the terminations were in them- 26 selves nothing but expressions of the different persons 1 I and my, thou and thy, are here completely identical, so that we are justified in trans- lating the verbal forms as viewed by the Magyar language by my having waited, thy having waited, &c. ; and it will hardly be supposed that this agglutination took place in the verb sooner than in the noun. Exactly the same process can be discerned in many other languages, as may be conveniently seen in Schleicher (ut supra), e.g. Ostjakish (p. 535). pane-m I laid ime-m my wife pane -η thou laidest ime-n thy wife pane-t he laid. ime-t his wife. Jakutish : byst-ym I cut bas-ym my head byst-yu thou cuttest bas-yu thy head byst-a he cut. bas-a his head. If, then,"Westphal bases his view on the very imperfectly demonstrated impossibility of explaining all inflexion to have arisen from the accretion of separate formations, the opposite view is supported by numerous actual instances of the growth of personal terminations out of pronominal stems. Add to this that, in later periods of languages whose stock of sounds has been much reduced and thus made often undistinguishable from each other, personal pronouns are a second time used with verbal forms to denote the grammatical subject, now of course not as stems but as outworn cases : Igive,je donne, &c, and it will appear that the origin claimed for these terminations — however difficult it may be to explain some individual instances — is really as probable a one as we can ever expect to find in the case of problems which deal with the earliest periods of the life of language. It has, moreover, the support of the grand idea which is so truly in harmony with the researches of natural science, that of the continuity of all linguistic formation. The higher stages of language are not separated from the lower by an impassable gulf, but only by a greater nicety of elaboration to which certain races have never attained. Monosyllabic speech, imperfect combination (agglu- tination), perfect combination (inflexion), these are the three main 27 stages, the third of which, if I am not mistaken, is being every day found to be more like the second. These self-consistent, clear, and simple opinions, which more or less explicitly form the basis of the whole mass of modern linguistic science, will, I venture to think, find no difficulty in holding their own against the diametrically opposite view expounded by Westphal, especially in his t philosophifcch-historische Grammatik/ e.g. p. 94 ff. It does not fall c 18 " INTKODUCTION. ch. ι. within our province to examine his general considerations, which are so suggestive of the natural philosophy of earlier times. To many they will be unintelligible, as I confess they are to me. Westphal regards the linguistic structure of the Indo-Germanic stock as ' an architectural work of art, endowed with endless magnificence and lavish grandeur.' ' The logico-constructive categories followed by the Indo-Germans in the for- mation of their language are the same categories which have sway in the Cosmos, in the macrocosm and in the microcosm alike ; the same that underlie sidereal life and the various forms of telluric existence, whether inorganic, vegetable, or animal.' I readily admit, and perhaps more readers than one would do the same, that I find theories of such a vast sweep- brought no nearer to my comprehension by the following somewhat extraordinary comparison. Westphal goes on to say, * Our primitive Indo-Germanic ancestors followed these categories with the same perfect unconsciousness as when they snatched at the first food to nourish their bodies, or when the first Indo-German man embraced for the first time the first Indo-German woman, who, though he did not know it yet, was to produce him a man like himself.' From this ' dialec- tic of celestial intelligence ' we are at length conducted to the world, with which we are directly concerned, the world of sounds, forms, and linguistic expression. But here on the threshold we are met by assertions for which no support is even attempted—* a is the vowel which came nearest to his (the Indo-German's) organs.' Since "Westphal himself admits that the primitive Indo-German had i and u at command as well, it is quite im- 28 possible to see by what rule he measures the nearness to the Indo-German of these different vowels. What is meant by i coming nearer "? If it means ' being more easy to pronounce,' the history of language and physi- ology both give the assertion a flat contradiction. The vowel a demands a greater tension of the organs of speech than i or u, and hence, as is well known, a tends everywhere, as languages go on, to become more like i or u. And yet it is upon this undefined notion of ' coming nearer,' which surprises us as we pass from the macrocosm to the origin of lan- guage, that all Westphal's subsequent system rests. He confidently applies the same notion of approximation to the consonants as well : * The nasal is the consonant that comes nearest to the organs of speech, the dental mute and the sibilant are more remote, 6 hence in the inflex- ional system the former is the representative of what comes nearer to the speaker among the dialectical series of definite conceptions, the dental mute or the dental sibilant, which takes its place, the expression of some- thing more remote.' Here, as we see, this ambiguous conception is turned to practical account, by being made to serve as an explanation of the personal terminations m, s, t. I have thought it not superfluous to add these samples of the positive side of Westphal's teaching, though I confess that while reading these theorems I have at times doubted whether the author was in earnest, or only wanted to tiy how much nonsense superficial readers could be made to accept. I will leave each reader to take his choice between the much- abused * agglutination ' and this new philosophy of the nearest. Another point of importance in the representation cursorily given • Ok. Or. p. 80 : ' Of consonants those that come nearest are the nasal and the mute which is interchangeable with the dental sibilant.' ch. ι. OBJECTIONS TO THE AGGLUTINATIVE THEORY. 19 above of the way in which verbal inflexion arose is the question of com- pound tenses. It is universally admitted that composition, a source of word-making from which the Indo-Germans have gained so much and various help for the noun, is to be found at work in the verb as well. Who could fail to recognise even in Latin forms like pot-ero, Gothic like sdki-dedum (we did seek), the presence of two verbal stems, the second of 29 which takes a position of subserviency to the first 1 But Westphal (' Philos. Gr.' 107) looks on the whole phenomenon as a comparatively late one. His view is that compound verbal forms are uniformly combinations of inflected noun-forms with inflected verbal forms, as is the case, e.g. in the Skt. periphrastic perfect of the verbs of the 10th class, e.g. tiorajari Jcakdra, properly - I made theft ' for ' I stole,' or Jcorajam dsa, Uorajam babhuva, properly ' I was theft.' As infinitives again are universally held to be petrified case-forms, Westphal is content if he can find an infinitive in the first half of such a verbal compound. A compound therefore like the French fut. aimer-ai, properly - I have to love,' he finds no stumbling- block. On the other hand, he denies that an uninflected or bare verbal- stem can be compounded with an inflected verbal form, which is the assumption made, e.g. by Bopp and others in order to explain the Skt. ά -dik-sha-m = Greek ϊ-ΙειΙα. Here Westphal and Merguet are quite at one, with this exception, that Merguet goes farther than his predecessor in his unqualified objection to the received theory. The difference between us here is by no means so fundamental as that discussed above. It is an actual fact that many verbal compounds are of the kind allowed by both scholars, and hence the question must be asked in each case, whether or not there can be found in the verbal-stem a noun-stem capable of inflexion. This question we shall not fail to investigate below when we Come to the forms concerned. We may, however, notice two points by the way. Great mistakes are often made by those* who look for inflected noun-forms or infinitives in the interior of verbal compounds. For instance, while Westphal (p. Ill) asserts that ' before this old perfect too of the verb to do (soki-da, &c.) there must have been an infinitive originally,' he makes not the faintest attempt to establish this by the investigation of the Teutonic languages. What is the good of this - must ' if he leaves the question in such an imperfect state 1 Again, in spite of his unwillingness to recognise bare stems in verbal compounds, Westphal admits on the same page that in the Latin forms eram, erim (legeram, legerim) - it certainly looks as if the auxiliary 30 form in question had been added to the simple perfect-stem, though these combinations are too obscured to admit of a clear insight into their genesis.' In such a case we may be sure of so much at least, that on his own showing there are still some obscurities left in Westphal's theory. Merguet expresses himself more strongly still. At p. 199 of the above-mentioned work he passes a final judgment on all such formations in the following words : - We must not forget that stems with no inflex- ion can only be assumed to have existed as independent words in a period antecedent to the appearance of inflexion, and must have ceased to exist as such as soon as inflexion arose. Now the auxiliary verb assumed to be the second component appears in an inflected form, and therefore presup- poses the existence of inflexion. So that we should have to suppose two words to be here combined, of which the former could only have existed before inflexion began, while the latter owed its existence to inflexion c 2 20 INTRODUCTION. ch. ι. itself. These two words consequently could not both have been in use at once, and the supposition of their combination involves a contradic- tion.' Linguistic science would indeed be in an evil plight if it had been maintaining for half a century doctrines which a couple of sen- tences could so completely upset. It is a pity that Merguet did not make himself a little better acquainted with the views he attacks before writing these words. He talks throughout as if what he calls inflexion had burst upon the world all at once like some natural phenomenon, revolu- tionising all the previous order of things, and introducing in fact just the inflexions of verbs and nouns which are to be found in the school-books. But all linguistic enquiry, as I have repeatedly pointed out, assumes forms to have arisen gradually and in strata. Where inflexion was of so gradual a growth, there is no absurdity at all in supposing that by the side of and in composition with inflected forms there appeared formations belonging to a previous stage of development. In my treatise * Zur Chronologie,' to which he occasionally refers, I have endeavoured to prove that inflexion in the nouns did not take place till some time after the three main stages had been reached in the inflex- ion of the verb. If this was so, there were, e.g. no case-forms of the noun- 31 stem dik for a long time after the production of a verbal form asmi or asami I am ; that is, there was a bare stem then in use. And why should it be thought impossible that these two forms should have come together with a small change into dik-sami, and that this dik-sami should make a preterite a-diksa-rn as dadd-mi made a-dada-m Ί Merguet himself (p. 64) is obliged to admit that the vocative is an uninflected stem-form. There is here nothing like the anachrouism or ' self-contradiction ■ which Merguet imagines he has found. Again, what are we to say to com- pounds like πνρ-φόρο-ς, πάν(τ)-σο(()ο-ς,λογο-ποω-ς 1 In all these cases we most unmistakeably find uninflected uniting with inflected stems to make organic wholes. Or are we to suppose that in all these cases case-termi- nations have been lost 3 Who would venture to try and establish that] In answer to objections Merguet has published a second work, ' Die Ableitung der Verbalendungen aus Hilfsverben ' &c. (Berlin, 1871). At p. 33 of this work he is already on the road to the discovery that if we want to understand the nature of compounded stems., we must transport ourselves to that period in which words ' had the form of bare stems.' Without doubt even at that early time types had been produced of every kind of composition, and among others of the composition of predicative stems with the auxiliary verbal forms which I suppose to have been already developed. We are, in fact, brought back constantly from all directions to the fundamental truth, that in all linguistic life we find older strata side by side with younger, cropping up here and there, and reaching over from an earlier into a later period. What Merguet goes on to say in his first work about the improbability that auxiliary verbs were older than others is still less to the point. No one ever said they were : it is universally held, on the contrary, that auxiliaries are weakened verbs of independent meaning. Forms then with auxiliaries in them do certainly presuppose older strata of verbal forms, but there is nothing that forbids us to suppose that later, after a number of verbs 32 which originally had a full meaning had become auxiliaries in separate use, the attempt was made to use them in compounds as well, and that ch. ι. OBJECTIONS TO THE AGGLUTINATIVE THEOEY. 21 too when they were bare stems, just as they were used in later times after their stems had been expanded and even provided with case-inflex- ions. In my treatise ' Zur Chronologic,' e.g. p. 55 f. (2nd edit.), I have called attention to all this, and pointed out how vast are the periods which the consideration of all these strata one upon another reveals to us. And in fact I cannot see how what I have there said — and Merguet (nowhere examines more closely — is in the very least refuted by the con- tradiction he says he has discovered. Still less successful are the attempts made to find another explanation for the forms in question. Westphal, inasmuch as he is unable to explain the whole mass of forms by the aid of the elements which he regards as primitive, assumes a twofold series of adjuncts by which what he takes to have been the primitive formations were expanded. To the first series of adjuncts he assigns meanings, e.g. to the i of the term, mi, which — although by his theory i is a ' more remote ' vowel — comes nearest, i.e. is the right one to express present time, to the s of the aorist, to which, for some unknown reason, he ascribes an intensive force, and to the a which he says occasionally denotes the plural. To this list must be added, if I understand Westphal rightly, those expansives to which he gives the name ' fulcra,' e.g. the syllables as and jus in the pronominal stems as-ma, jus-ma, and perhaps too some of what he calls ' strengthenings,' or ' secondary adjuncts,' e.g. the κ in σκ. All these sounds and syllables, of which he nowhere gives any explanation, can according to his theory be introduced, even after the primary structure of the language has been established, as a kind of second instalment or reserve force from the divine ίκμαγείον of forms, at the beginning, middle, or end of words. A second series of subsequently introduced adjuncts are, on the contrary, held to be purely phonetic. It contains ' euphonic,' ' purely euphonic,' ' purely phonetic ' vowels, and consonants as well, the latter, where they appear between vowels, being entitled ' dividing consonants,' the exact counterpart, that is, of the well-known and still favourite * connecting vowels.' In this latter point Westphal and Merguet are at one. Both credit the ' fuga hiatus,' that is, the disinclination to let two vowels come into direct contact, with the production of a goodly row of 33 consonants which Merguet supposes (p. 205) to have arisen ' out of what was originally a very indistinct aspiration.' To criticise these views, which Merguet advances less positively in his second work, is not here our object. There are only two points to which I wish to call attention. The first is, that by such assumptions we are really transported bodily back to the old grammarians' point of view, which it was hoped was left behind for ever. It is notorious that their exploded etymologies were indebted for their existence to the freest possible use of the πλεονασμός of single sounds and whole syllables. What are these ' fulcra,' &c. but the naively admitted pleonasms of the ancients 1 If fulcra, strengthen- ings, purely phonetic adjuncts are to be allowed everywhere, it will be hard to set bounds to the most extravagant caprice. The second point is the marvellous inconsistency with which language would be chargeable, if it had really produced forms in the way supposed. On this head we shall confine ourselves exclusively to Westphal. According to this scholar, the Greeks had a considerable troop of sounds at command for 22 INTRODUCTION. ch. τ. the purpose of preventing vowels from clashing ; e.g. τ (' Gk. Gr.' i. p. 117), which is used, among other things, to form the 'locative' ε-τ-ι from the stem ε," r, introduced in the perfect active ' to avoid a hiatus,' ν in W-r -ος, τί -v-L (ii. 409), δ in έλπί-Ε-ος (i. 254, 266), θ in κόρν-θ-ος (p. 274), στ in θεμι-στ-ος (p. 254), ι in οίκο-ι-ο for ο'ίκο-ο (p. 145). In many of these forms there is not the slightest warrant for such assump- tions ; to and vo are in no way such combinations as were avoided (cp. ■στνγιος, δάκρνυν, εΧυομεν). Another question that arises elsewhere is, why were these auxiliary troops put to so little use % It would have been perfectly easy to prevent vowels from clashing at all. How is it, then, that we get such strange forms as yertoc, γενεί, γένεα, λιληίεαι, in all of which an σ had to be rejected before the clashing was possible? How is it to be explained that the same language which, as Westphal himself allows, often throws out j or ι between two vowels, e.g. (ii. 132) in ΰακρν-ω for the presupposable ΰακρυ-ιω, and in all the verbs in πω, οω, 34 εω, has in other cases actually introduced the sound j afresh Ί On the one hand, as no one denies, μισθο-ιω becomes μισθό-ω ; on the other, as Westphal says, ο'ίκυ-ο becomes οίκο-ι-ο ! And elsewhere the case is the same. The β in άρ-β-μό-ς (i. 184) is said to be 'euphonic,' whereas the same group of sounds was found so objectionable by the Greeks in πέπειθ-μαι that they changed, it to πέπεισ-μαι. Are we to think the Greek ear varied so waywardly that in one century it wanted to have a θ before an μ, and in another wanted to get rid of it 1 It must not be forgotten in this controversy, that assumptions like these of Westphal's have in individual instances found support even among the representatives of comparative grammar. I once thought myself, with Westphal e.g., that the κ in the perfect was euphonic, though I have long given up the idea. A more connected and thorough examina- tion of the matter in hand has led me to the conclusion that the admission of purely phonetic elements in the life of language is only to be made with the greatest circumspection. Language is penetrated everywhere with significance. Its forms can at first have contained nothing but what served for the instinctive expression of an idea or conception. It is true that sounds have a life of their own which is in many respects independent. Their changes can be established for each language accord- ing to its own prevailing analogies. It must, e.g. be allowed that out of the vocalic element contained in every continuous sound a vowel may in some circumstances arise, and that, conversely, a special articulation may give a vowel a nasal pronunciation, and that this nasal echo may develope into an independent nasal. Similar phonetic growths of no meaning are to be allowed to but a Very limited extent (e.g. άν-ΰ-ρ-ός), and in all such cases the germ of the sound is in reality an outgrowth from the sounds already there. But nothing can justify us in going farther than this and admitting the possibility of the generatio aequivoca of a sound out of nothing at all. In these principles I believe I am at one with the great majority of my fellow enquirers. Our object is to explain the phenomena of 7 We should have just as much right to call r euphonic, and thus explain &•ν•6 to be from the stem ά (perhaps originally ά-ιτ-οτ), and so -arrive at an ablative. ch. ι. OBJECTIONS TO THE AGGLUTINATIVE THEORY. 23 language according to the phonetic laws and tendencies of individual languages, and to refer them in all cases to elements which had once a meaning, and which were created in early periods of the linguistic life. We are certainly far from our goal as yet, but it is of the first importance 35 to be sure of out• way ; and so it appeared to me fitting that at the outset of these investigations I should make an express avowal of these main fundamental principles. /library UNIVERSITY OF ' CALIFORNIA 24 THE PERSONAL TERMINATIONS. CHAPTER IL THE PERSONAL TERMINATIONS. ■ I. ACTIVE. Buttmann, in his time, expressed himself as follows on the subject of the origin of the personal terminations (Ansf. Gr. i. 2 496, note) : ' All these syllabic endings were, as the analogy of oriental languages clearly proves, marks of the three persons, and originated in attached pronouns ; they lost, however, their original force as the language took shape and opened the door to the smoothings and roundings which find their way into lan- guage owing to the desire for rapid and euphonious speech.' This judgment has been in every way established and confirmed by comparative philo- logy. ■ It must, no doubt, be admitted that even those who are sure that the principle of this explanation is the right one find great difficulty at many points, especially in the dual, the plural, and the whole of the middle. But we may assume, I think, to begin with, that in this primitive region of the genesis of forms weakenings and dullings of primary forms were more frequent than elsewhere, and that as a result of this we cannot always arrive at certainty as to what the origin was. Our purpose de- mands that we should mainly try and find what relation the given Greek forms bear to the primary forms which may with more or less probability be presupposed. In so doing we may most certainly start with the assumption that the forms of the Greek personal terminations which are fullest, in so far as they can be reconciled with those of the related 36 languages, are also relatively the oldest, while those which are less sub- stantial have been weakened from them. This truth, which has been established by the researches of Bopp (Vergl. Gr. § 434 if.) and Schleicher (Comp. § 269 ff.), has not been upset by Friedrich Miiller's opposition. The last-named scholar (Sitzungsberichte der k. Akad. Philol. Hist. CI. vol. xxxiv., and again vol. lxvi. Oct.) was for starting not from mi, si, ti, but from m, s, t, and regarding the i, strange as it may seem, as a mark of the present tense. The difficulties encountered by such assumptions have already been pointed out by Misteli, Ztschr. xv. 287, and I have defended the received view myself in my ' Studien ' (vol. iv. p. 211 ff.). We hold fast to this then, that, to begin with the singular, those personal terminations in it are the oldest which approach most nearly to the personal pronoun-stems ma, tva, ta. First Person Singular. According to Schleicher the termination of the 1st sing, has split up into two essentially distinct forms, the original ma having on one side lost the to and become a, on the other weakened the a to i and taken the form first of mi and then of simple m (Gk. i'), the former (a) appearing in the ch. π. FIRST PERSON SINGULAR, ACTIVE. 25 perfect, the latter (mi or m) in all the other tenses. In Schleicher's view then the α in γέγο y-a=Skt. gagdn-a is essentially distinct from that in the aor. εΊειξα ; in the former it stands for ma, in the latter it is part of the tense- stem, after which, as is shown by the Skt. d-diksha-m, the personal termination disappeared only on Greek soil. If this could be proved, the 1st pers. perf. would be remarkable at once for its great antiquity and its strange transformation. As far as its antiquity goes, which is shown in the preservation of the primitive a, we have a parallel to the 1st sing, in the tha of the 2nd, which we cannot but derive straight from the pro- nominal stem tva; and herein evidently lay the main ground for Schleicher's view. "We should actually be able to ascend to a still older system of vowels, inasmuch as Kuhn (' Ztschr.' xv. 405) adduces from the Yedas forms like (ja-grdbhd (I grasped), bi-bhdjd (I feared), and 37 from Zend ta-tashd (I framed ; cp. Justi's Lexicon) j and Justi re- cognises the form -td in Zend for the 2nd pers. as well (v&ig-ta=olada). There is, however, a difference between these two persons as well as a resemblance. In the tha or ta, Gk» θα, of the 2nd sing., the consonant has been carefully preserved j and where it clashes with the final con- sonant of the rt. the difficulty is adjusted : Skt. vet-tha for ved-tha=oJn-6a for οϊΰ-θα, but the consonant in the termination never disappears to make way for that in the root. On looking further we find the same process occur- ring throughout. Everywhere in Greek the consonant of the termination is the determining and the persistent one. Perspicuity demanded that this should be so, for since the personal terminations are. distinguished from each other essentially by their initial consonants only, if the opposite process had taken precedence, it would really have led to the annihila- tion of the newly created personal terminations. And Schleicher holds, in fact, that the Skt. 3rd sing. perf. veda came from ved-ta, and thus coincides with what was originally ved-ma in the 1st sing. It is true that for this expulsion of the consonant of the term, we have one certain in- stance, of which, of course, Schleicher was thinking when he formed his views. The 1st sing, of the middle loses its m throughout in Skt. and Zend, as does the 3rd sing, of the middle perfect its t, and the result is that both persons of this tense in these languages coincide. But since Greek has in this instance preserved its μαι and rat everywhere without exception, and since in Gothic, too, -da survived all along, we have no right to assign this strange mutilation in Skt. and Zend to a period ante- cedent to the separation of the European languages from the common stock, but ought rather on every account to assign it to no earlier period than that in which Indians and Iranians made a distinct smaller group. Con- sequently the above-mentioned analogy does not hold when we look more closely at it. It might prove something for oriental languages, but not for Greek forms ; but since the Skt. {jagdna can hardly be separated from the Gk. yiyova, we must give up the idea of establishing for the oriental languages, that the a at the end of the 1st sing, is a curtailed 38 ma. For these reasons I hold to the view that the perfects, in the 1st as in the 3rd sing., lost their termination not before but after the a, which we shall afterwards see still stronger reasons for considering to belong to the stem, and that this loss took place at no very early period. I refer yiyova to γεγονα-μι, and $a$dna to gag ana-mi, and believe we ought to regard the agreement between the two languages here as of the 26 THE PERSONAL TERMINATIONS. ch. ii. same kind as that between the 1st sing, in ω and the Zend presents in «, e.g. jieregd (I ask)=Skt. prJcUhd-mi. In the case of these latter we can hardly doubt that, as will be shown more at length directly, in the time before the separation of the languages, the mi was still intact in the present, and hence that the special correspondence between Greek and Zend here is the result of later development taking place independently in the two languages, and due to kindred causes. And this is just the conclusion to which we must come about the above-mentioned perfect forms. 1 In support of this view we can adduce an isolated but well attested perfect form, the ^Eolic F οίΰη-μι = F ο~ώα (Ahrens, Aeol. 136). Why should we not think that here the μι is just as much a relic of earlier times as in other cases? This Foίlη-μL brings us to an Indo- Germanic vaida-mi. The vowel before the termination will have to be dis- cussed later. Here we have only to do with the termination itself. If our conclusions are correct, they prove that there is no evidence of any form for the 1st pers. sing, older than -mi. We now turn to the considera- tion of the -μι retained in Greek, then of its loss and its transformation. 1) μι in the 1st sing. pres. ind. of primitive verbs. The number of the verbs which get their name from the fact that they retain their old termination is not a very great one. We shall learn 39 more about them in the fourth chapter. All have occasional by-forms of the ordinary conjugation, which are specially numerous in the case of the verbs in -νυμι, though from Homer downwards they do occur in other verbs besides, and justify the conclusion that this class of forma- tions was already becoming obsolete at the earliest period of which we have any record. Many only exist in isolated forms. Perhaps we may find in the fact that a small number of verbs of uncommonly frequent use, like ειμί, είμι 1 τίθημι, ΰ&ωμι, 'ίστημι, ϊημι, φημί, ΰείκννμι, ύμνυμι, re- mained true to the older fashion, the reason why others of less frequent use were able to withstand the main tendency of verbal inflexion and preserve their old forms intact. 2) More extended use in Aeolic. We have the repeated testimony of the old grammarians to the fact that the conjugation in MI was more extensively used in the Aeolic dialect than in the others (Ahrens, Aeol. 134). In single instances we actually find forms with the fuller personal termination peculiar to this conjugation, as κάλη-μι (Sappho, 1, 16), δρη-μι (ib. 2, 11), άσυνετη-μι. (Alcaeus, 18, 1 Be.), άίτη-μι (Pind. fr. 132, 4 Be.), αίνη-μι (Hes. Opp. 683), επαίνη-μι (Simon. Ceus, fr. 5, 19 Be. 3 ). We have too the direct testimony of Herodian (ed. Lentz, ii. 463, 930, 825) to the forms γέλαιμι, πάλαιμι, πλάναιμι, Ιοκίμωμι, and the Boeot. τάρβειμι, ποίειμι, φίλειμι. Most of the examples, however, belong to formations which differ from those of other dialects, not in their terminations, but only in their 1 Brugman's attempt (Stud. ix. 314 ff.) to represent the vowel which stands before the personal terminations in the perfect as a mere production of the nasal, and make the vowel of the 3rd sing. e.g. (fatfana = yiyov* rest on mere formation by analogy, does not convince me in the least. ch. π. FIRST PERSON SINGULAR, ACTIVE. 27 internal vowel, e.g. 3 pi. ίπφρόμβεισι (Sappho 2), ϊεισι (ib. 16), part. ο'ίκεις (Ale. 69), τίθεις (ib. 34), καλήμενος, άπειλητην, σνναντήτην, φομη- μεναι. Here the dialectic difference lies in the different treatment of the vowels which clash in the contracted verbs, and this must be discussed elsewhere. I only mention this fact now because it serves to refute an assumption made by Ludw. Hirzel (' Zur Beurtheilung des aeolischen Dialekts/ p. 56 ff.), and supported by several scholars, that in their ορημι, /σεισ6)α (Hymn, in Cer. 366), for which he holds εχείτθα and σχί/σεσθα to be the true forms. It was certainly bolder still with Bekker to alter the Homeric ΰιΰοϊσθα Τ 270 to ΰίΰωσθα, as this change cannot be supported by the supposition of a mistake in reading an old character, εθέλεισθα or εθέλησθα in the Aeolizin'g poem of Theo- critus (29 v. 4) comes under the same head as the ε χει σθα in Sappho, though it is more likely than this to have been an antiquarian's mistake. The latter supposition must without doubt be made with reference to forms like ήσθας, οίσθαν, which occasionally appear in M.SS., the former e.g. in the Yen. A at Ε 898, but rejected by Aristarchus (Ariston. on A 85, cp. La Roche Homer. Textkritik, 320 ff.). Notwithstanding, Nauck (Eurip. Stud. ii. 71 ff.) has re-adopted these forms and actually tried to 56 introduce them into Euripides. ' Third Person Singular. With regard to Schleicher's view that -ta, the original termination of the 3rd sing., appears sometimes, in the perfect for instance, as -a, some- 4 The assumption of fuller phonetic structures by the side of more slender ones, formed to give greater stress to the word, seems to me specially reasonable in the case of pronouns. Sayce, Principles of. Comparative Philology, London, 1875, p. 2.5, regards ' emphasis, ,' the counterpart of ' laziness,'' as an early factor in the formation of language. The fuller form could then be used subsequently for the termination indiscriminately with the more slender one. 38 * THE PEKSONAL TERMINATIONS. ch. ii. times as -ti, I may refer the reader to what I have said on p. 24 when dealing with the 1st pers. sing. I consider -ti to be the oldest mark we have of the third person, as -mi is of the first, and I see no reason to doubt that it grew out of the pronominal stem ta, which appears in Greek as το. In the language as we know it this stem is not used for. the personal subject, and its place is taken by sa=Gk. ό; but this makes no differ- ence. When we find the stem το in the Homeric and Doric τοί taking the place in the nom. plur. of the later stem ο of common Greek, we may surely venture to assume the like use of the same stem in the singular too to have prevailed in the primitive period in which verbal inflexions had their origin. Analogies to this (Lith. ta-s this, Lat. is-te, Ch.-H. tit) are given by Bopp, Yergl. Gr. ii. 2 132. The termination ti is . treated in three different ways in Greek. Either it remains unchanged, or it becomes σι, or it disappears. 1) tl preserved intact". The only instance common to all Greek dialects is in εσ-τί, where the preceding σ preserved the τ intact, as it did in πίσ-τι-ς, πνσ-τι-ς. The Dorians, however, as elsewhere, retained the r here to a greater extent : see Ahrens,'312. To the forms like ΰίΰω-τι, τίθη-τι, άφίη-τι, "ισα-τι, there adduced from inscriptions, fresh instances have since been added from Delphic inscriptions (Berichte der k. sachs. Ges. d. Wissensch. 1864, p. 223). These forms are of rare occurrence in the poets. ήτί=ήσί, he says, is found in Alkman fr. 139 Bergk 3 , εφίητι in Pindar, but only once 57 Isthm. ii. 9 (Boeckh, Pind. i. 2, 292), St'cWt in Simonides fr. 18 Bergk 3 , τίθητι in Theocritus, iii. 48, νφίητι, iv. 4. Matthia, i. 489, quotes a few instances of άποΰίΰωτι from the Doric prose of Timaeus Locrus. 2) σι. a) There is no need to quote instances of its occurrence in the indie, pres. of the verbs in -μι in forms like ψη-σί, τίθη-σι, δίνω- σι, όνίνη-σι, πίμπλη-σι, άη-σι, εΤ-σ -t, ζενγνν-σι, 6λλν-σι, ρηγνν-σι, which occur from Homer's time on to the latest period of Attic Greek, though even here there are not wanting by-forms which follow the usual mode of inflexion. This formation may therefore be said to be Ionic in the fullest sense of the term, and it even penetrated to Pindar's language, while the Dorians preserved the -τι, and the Lesbian Aeolians knocked the termina- tion off altogether and said τίθη, as we shall see presently. It is only in ή-σί he says (Sappho, 97 Be. 3 ), that we find a trace of this termination even in this stem. b) In conjunctives the Homeric dialect often retains the termination σι. I. Bekker (Horn. Bl. i. 218) gives 75 forms, not reckoning compounds and synonymous pairs of forms. Of presents of both of the main conjuga- tions there are άγτ)σι, aeftyat, έθέ\τ}σι (also in Hesiod), τέρπει, τύμνησι, "ιτ}σι } μεθύσι, ψγσι ; ' of active aorists of every formation, άγάγιρσι, λάρρσι (also in Hesiod), \άθτ)σι, ελτισι, κάμφσι, Ιωττισι ΟΓ Ζφσι, 7]σ<, παύσρσ/, πέμψτισι. We may add the perfect ερρίγιρσι, which Bekker has omitted (Stiei•. Stud. ii. 128, 353). Sometimes there is a clear connexion between these longer 3rd persons and the longer 1st or 2nd persons : CH. II. THIKD PERSON SINGULAR, ACTIVE. ΙθίΚωμι ΙθίΚησθα i6e\rjai ΐϊπωμί (ϊττησθα βάλησθα ςνδησθα (χησθα ϊησθα πάθησθα ς'ίπησι βάλ-ησι €νδτ)σί ΐησι πάθησι άγάγωμι ayayrjai τυχωμι τύχτ]σι 39 The number is not much increased by later poets : ρεζ^σι Hesiod fr. 195 (Rzach Dialekt des Hesiod, p. 438), έρέθρσι Hymn, in Martem (viii.) 14, irjaL Theogn. 94, υποπιμπρήσι. (?) Aristoph. Lysistr. 348, πίπτϊ)<π Plato Com. ap. Eustathium ad Iliad, p. 1161, υποπτεΰτισι Theocr. xxiii., 10, 58 θαλέθρσι Nicand. Ther. 832, κατακτείνγσι Quint. Smyrn. θ 153, ύρσρσι Apollon. Ebod. III. 1039, άμψιίστισί Orpb. Litb. 273, most of which instances I have taken from Lobeck, Elem. ii. 264. Even the forms already used by Homer occur but rarely in later poets. We may add the forms given by Lobeck, Rhemat. 183 from verbs in -«ω: νπηάρσι, σκιάτισι (Arat. 795, 864), and similar rarities of a still later time. Lobeck upholds the view of the ancients that all these forms had arisen by επέκτασις from the ordinary ones, and hence he combats the view ad- vanced as a conjecture by Buttmann (Ausf. Gr. i. 2 352; cf. Kriiger, Dial. § 30, 1 note 4), and more confidently by Thiersch, Gr. p. 352 and Gbtt- Hng on Hesiod Theog. v. 60, that these forms have no right to the ι sub- script. It is true tbat the authority of good manuscripts, and that of Herodian (cp. on Ο 359), is on the side of the retention of the t. But all that this proves is that the theory of even the best grammarians required it to be written. For it is certain that in Herodian's time there was no difference of pronunciation between η and y. The authority of these men then would not prevent us from rejecting the ι here any more than in the case of the 2nd sing, forms in -ησθα if there were really no way of explaining it. In my Tempora und Modi, p. 23, I expressed a de- cided opinion that it ought to be rejected, because I then regarded the epenthesis as a transposition of the ι from the final syllable to the preced- ing one. I have since come to a different, and as I believe, more correct opinion on the subject, an opinion which I have stated in my Principles II. 337. I now regard the epenthesis as an echo of the t in the syllable pre- ceding it, which, as is shown by είνί (from hi), is not necessarily con- nected with the disappearance of this vowel from its original place. * ayprt, άγρσι are therefore forms admitting of quite a simple explanation. They bear to the more primitive * άγητι precisely the relation that the Zend conjunctive form avdi-ti does to the-Skt. άνα -ti (rt. av). The epenthesis is, it is true, not a necessary or irresistible affection, only one which may attack a sound, and one to which long vowels are less liable than short ones. Still Aeolic forms like -γί\αψι for * γέλά-μι and the like, e.g. χρα-ι-σ-μέ-ω from *γρα-ι-σιμο for *χρα-σιμο, prove that the length 59 of the vowel does not exclude the possibility of this affection. Now as we have besides the analogy of the indicative, i.e. *ayijn : *άγρσι *I άγει -ri : άγε-τί, I think we shall do well to follow the tradition and leave the l subscript in undisputed possession. c) An isolated Optative. Such, it seems, is παραψθαίησι Κ 346, which is defended by I. Bekker 40 THE PERSONAL TERMINATIONS. ch. ii. (Horn. Bl. i. 218 note) against alterations (e.g. παραψθήησι Thiersch, La Roche). The form was perhaps invented by a bard who thought -σι an addition that might be used on occasion even in the optative. I could hardly venture to assume in the case of so isolated a form in a lay of the Iliad that is evidently not one of the oldest, that we have in the -σι a real old companion to the -μι of the 1st sing. opt. Syntactically the conjunc- tive is just as admissible here as the optative, and hence it is not impos- sible that, as Joh. Schmidt suggests (Ztschr. xxiii. 299), φθαίησι is a conjunctive of a present * φθαίω that occurs nowhere else. d) Indicatives in -ησι. It is very generally held that there are also indicatives in -ησι from verbs of the thematic conjugation. It is true that Buttmann ( Ausf. Gr. i. 2 498) has pronounced against such forms in the Homeric poems, show- ing that all that were thought to be indicatives might be conjunctives. This applies particularly to r 111 ff. In Hymn. Homer, xxi. 15 the gap that immediately precedes renders a decisive judgment impossible. In Ibycus these indicatives seem to be better established. It is true that Aristarchus by his 'diple' at Ε 6 ναμψαίννσι, η ΰιπλή 'ότι αντί τοί ταμφαίν». πλεονάζει δε "Ιβυκος τω τοιοντω points to nothing but a conjunctive. But the σχήμα Ίβνκεων in the language of the later grammarians and rhe- toricians, especially in Aelius Herodianus (Spengel, Bhetores Graeci, iii. 101), Lesbonax (De Figuris, p. 166 Yalcken.), Heraclides, in Eustath., 1576, 20, was the specific name of a grammatical figure, that is, it was believed that Ibycus used conjunctive forms in the sense of indicatives. Instances of this which we actually find in our fragments are fr. 7 Be. 3 τάμος άνπνος (?) κλιτός όρθρος εγείρησιν άηϊόνας and fr. 9 ψάμις εχησι βροτών and in Bacchyl.fr. 27 γλυκέ! 1 ανάγκα σευομενα κυλίκων θάλπησι 60 θνπόν. We must meanwhile admit with Bergk, with reference to the first passage, that the fragmentary form in which it reaches us precludes the possibility 'of passing a decisive judgment upon it. It is possible after all that the grammarians made a mistake and that they were all the time instances of the conjunctive which was used in the Homeric fashion in a certain kind of relative sentences. But the statement is made so often and in such precise terms that it is hard to believe that they were so mistaken. Buttmann saw no objection to the supposition of such indicatives in -ησι in the ' Dorico-Aeolic dialect,' and appealed to some supposed indica- tives in -ης for -εις and -η for -ει. But the genuineness of such forms is suc- cessfully impugned by Ahrens (Aeol. 91). Corssen's attempt too to find a support for this η in the * vowel-intensification ' of which so much is heard and so little seen, must be held to have failed. 5 It is hard to imagine how a language of such nice distinctions as Greek is, which in all other case's consistently reserves the long vowel for the subjunctive, should in this one instance have been so careless as to lengthen the the- matic vowel in the indicative as well. If then we cannot believe in the actual existence in living speech of indicatives like εχησι we must take one of two courses. Either we must 5 Corssen discusses these forms at great length in his posthumous work, Beitrage zur italischen Sjtrachfamde, p. 479. In his attempt to establish ίχησι and the like as real forms of vulgar Doric, this meritorious scholar (who, however, was too much inclined to make short work of questions of sound-change in Greek) quite forgot that the Dorians never put εασι, and so be a contraction, unless we prefer to class it with the related verbs and to accent it ϊστασι (cp. Skt. tishthanti, Zd. histenti). f ) There remains to be discussed an extraordinary Cretic form. Hesychius has the gloss iyovC ϊχονσι Κρήτες. It would be easy to con- jecture (as Ahrens does, Dor. 293) that he meant εχωτι or εχονσι as a dat. plur. But it is just as possible that this tradition is a sound one, as Boeckh, C. I. ii. 404, Stier, Ztschr.-, vii. 7, suppose, εχονι may have come immediately from *εχοννι (cp. Ιεννος). The *εχονπ thus arrived at may perhaps be taken as derived by assimilation from the ordinary B Cp. G. Stier, Ztschr. vii. 1 if. In this essay, which takes in much matter related to our present subject, the form eSTt = «tVi too is, according to an old precedent, again adduced as Aeolic. This form, though, is only found in Eustath. 1557, 41, quoted from Heraclides, where, however, the whole connexion clearly points to the conjecture made by Ahrens (Aeol. 209), that tlm is a mistake for tttni, a Doric contraction for torn. ch. π. THIED PERSON PLURAL ACTIVE. 49 Doric εχοντι. The Cretans were fond of unusual assimilations. Cp. e.g. their transformation of εκλυσιν to ελλυσιν, Ανκτυς to Ανττος (Hey, De Dial. Cretica, p. 48). We shall meet with this assimilation of ντ to vv directly in the final letters of the secondary forms. 2) The secondary termination -ant, -nt. 73 The Dorians accented every 3rd plur. of the secondary forms as a paroxytone : ελέγον, εφάγον, εφάσαν, έλυσαν. Ahrens (Dor. 28) has col- lected the testimony of the grammarians on this head. No one, I think, will believe nowadays that this accentuation is, as Macrobius (De Differ, p. 310) assumes, ' discretionis gratia,' i.e. meant to distinguish the 3rd pi. from the 1st sing. There was no i discretio ' at all in the case of εφάσαν, έλυσαν. Ahrens saw clearly that the reason lay in the origin of these forms. At the time when nt was still sounded at the end of these syllables the final syllable was long by position, and this affected the accent. The main accent, which had a greater persistency than the final consonant, held fast to the syllable on which it had once established itself. The only question is whether it was the old full nt which was preserved in the Lat. erant (=ήσαν) that produced this effect, or a connecting link between nt and the simple v. The former is the view held by Ahrens, and formerly by myself. Misteli, on the other hand (Ztschr. xvii. 166), has endeavoured to show that the latter is the probable one. He is no doubt light in assuming that between the deducible ήσαντ and the ordinary Greek ήσαν there must have come a form in which ντ had assimilated itself to vv, ή σαν ν that is, and aptly compares with it the Skt. form dsann which still appears before vowels. For the formation of the nom. and ace. neut. of vr-stems in -v also I believe that I have in Stud. ii. 167 been right in conjecturing foims in w, e. g. φέρονν. Since then there can hardly be a doubt that the step immediately preceding ελέγον, εφάσαν was not ελέγοντ, εφάσαντ, but ελέγονν, εφάσανν, it seems advisable to find an explanation for the Doric accentuation in this the nearest etep, especially as the Greek law as to the final syllable was not one of the oldest laws of accentuation and e.g. can certainly not be held to apply to the Graeco-Italic period in which we are forced to admit esant and the like. There is even nothing im- probable in the further assumption that the Doric ν in this place, e.g. in ελέγον — as a kind of fellow to the initial ν arising from ov, and making the ο long by position, in the Homeric άπο νευρήφιν — had even in his- torical times a somewhat different sound from that of the ν in the 1st sing, εφερον, and that this was the reason why the analogy of all 74 verbal accentuation so imperious elsewhere was powerless here. When in the next place the final consonant was completely expelled, Greek reached the same stage as the Sanskrit : ε-φερο-ν=ά-11ιανα-η and Zd. bare-n. A companion to the forms in -ασι after vowels above discussed is the Boeotian άνέθιαν,Ο. I. 1588, i.e. άν-έ-θε-αν (Ahrens, Aeol. 211, Dor. 525). Ahrens is no doubt right in refusing to assume with Boeckh that an σ had been expelled here, for ένίκώσαν (1583) shows that the σ of similar forms remained intact. It is far better to suppose that here, as in the perfect άποΰεΰό-ανθι (1569a, 35), and the Attic ΰώό-ασι, τιθέ-ασι the α is Ε 50 THE PERSONAL TERMINATIONS. ch. ii. an integral part of the personal termination. 1 This agreement between Attic and Boeotian is perhaps not accidental : it may be one of a series of phenomena common to the two neighbouring countries. We have already had occasion to speak of the partial or sporadic part played by composition in this person (p. 48). Few can doubt that ε-φα-σαν, ε-δο-σαν contain the preterite of the rt. as shortened to σαν and destitute of augment (== ε σαν), and the view of the ancients that the shorter formation arose from the longer may be regarded as disposed of. The direct addition of the personal termination to the stem in the case of verbs with no thematic vowel is from the earliest periods of the language onwards rarer than that of the longer -σαν. In Homer we have — earav A 535 by the side of ΐστασαν Σ 346 στάν I 193 „ „ παρίσταναν Η 467 . Ζβαν Α 391 „ „ νπίρβασαν only Μ 469 βάν Δ 209 „ „ ϊβησαν θ 343 ίφαν Γ 161 „ „ Ζφασαν Ο 700 φάν β 337 „ „ φάσαν Β 278 φθάυ only Λ 51 €τλαν only Φ 608 €κταν only Κ 626 πρότιθ€ν α 112 (Aristarchus) by the side of τίθεσαν χ 449 θίσαν A 433 75 - Up Μ 33, μέθΐ€ ν φ 377 by the side of aveaav Φ 537 πρό*σαν δ 681 [hymn, in Oerer. 328, 437 ϊδώον] δόσαν A 162 tbw Λ 263 by the side of ήισαν Κ 197 ϊφυν e 481 „ „ ίσαν A 494 Hesiod uses εΰώον Opp. 139 (M.SS. ΈΜδων, cp. Ezach, Dialekt des Hesiodos, Theogn. p. 439), Uov, Theog. 30, by the side of Uoaav 141, and the remarkable ήν=ήσαν, to be discussed below, p. 99. The short form finds a fair number of representatives in Doric dialects ; we have the Argive ανέθεν (C. I. 29), now found also on the Olympian nike- inscription of the Messenians and Naupactians, απέσταν, Ιίέγνον (Heracl. Tables, Meister Stud. iv. 420), to which may be added 'ilov from the inscription from Tegea, C. I. 30. From hence they made their way into Pindar's poetry : κατεσταν Pyth. i. 35, πίτναν Nem. v. 11, ανέθεν Isthm. viii. 58 (cp. Simonides Ceus, 137), τίθεν Pyth. iii. 65, Uv Isthm. i. 22, εψνν, Pyth. i. 42, by the side of θέσαν Pyth. ii. 39, Uoaav Nem. vi. 10, and indirectly, though very scantily, into that of the dramatists : εβαν Aesch. Pers. 18, άπέΰραν Soph. Aj. 167, κατέβαν Soph. Trach. 504, εσταν Eurip. Phoen. 1246. The corresponding forms of the passive aorists are discussed ii. 323. The termination σαν, which clearly took the place of ν more and more as time went on, has perhaps do counterpart outside Greek except in Old Persian : e.g. ραώ-ι/-άί-8α=προς-ήι-σαν (Spiegel, Altpersische Keilinschriften, p. 168). DUAL. The first person dual, which originally ended in -va-si, as is shown by the Skt. vas, the Zd. va-hi, the Ch.-Sl. v8, the Lith. -va, and which, it is highly probable, differed from the 1st plur. only by the 1 Beermann, de dial. Boeotica (Stud. ix. p. 78), takes another view of this form, which he connects with the Arcadian άιπ/-δ<$α* discussed Verb. ii. 288. ch. π. DUAL, ACTIVE. 51 weakening of the m to ν (Schleich. Comp. 3 653), has disappeared from Greek without leaving a trace. We may perhaps conjecture that the disinclination this language showed to a digamma, especially in the middle of a word, had something to do with this, especially as of the two dialects which did hold to the /, the Aeolic had no dual. There are so many questions that touch both persons of the dual equally, 76 that we shall have to consider the second and third persons together. The termination of the 2nd pers. in the Indo-Germanic period appears to have been -tva-s, the nearest approach to which is the Skt. -tha-s. It would hardly be possible to establish any other analysis of this than that into -tva-si, i.e. ' thou thou/ so that, as already pointed out on p. 44, the second person dual was originally identical with the second person plural. Notwithstanding this it is quite clear that there was a difference between the two numbers in this person before the separation of the Indo- Germanic languages, for the difference is visible not only in Sanskrit and Greek but Lithuanian (2nd du. -ta, 2nd pi. -te) and Gothic (2nd du. -is, 2nd pi. -tit) as well. Now this primary termination has, it is true, dis- appeared from Greek. But there is a corresponding secondary form, the Skt. tam, with which the Gk το ν is identical. How this tam is connected with the assumed primary form *tva-s is not clearly ascertained. If we take the Greek language as our point of view we are inclined to state the relation thus : μεν : μες : : τον : * tva-s. It is true that in the 1st plur. the nasal is exclusively Greek, while in the 2nd and 3rd dual it is Indian as well. In the former case it can be explained by the phonetic ten- dencies of Greek, in the latter it is a mere meaningless phonetic addition that admits of no explanation. We cannot rest content either with Schleicher's conjecture (Comp. 3 660) that we ought perhaps to divide it t-am, and regard the am as an accession ' with no clearly distinguishable function.' I am afraid that we must here and in some other cases leave this m for the present as an χ to be explained in the future. In the 3rd person dual Sanskrit has the primary termination -tas, Zend -to, in which the 6 is the regular and purely phonetic transforma- tion of as. It is clear that -tas .* -ihas '. \ pron. stem ta \ tva (tha), that is, tas originated in ta-si * he he,' as did thas in tva-si. The syllable si then, which is in most cases the mark of the second person, here belongs to the third, being weakened from sa, ' he.' Bopp's conjecture that the s is the remnant of a pronoun sma, which must evidently have arisen from sa + ma, seems to me less likely (Ygl. Gr. ii. 2 280), though on this assump- tion, which would have to be extended to the 2nd dual as well, it might 77 be possible to find an explanation for the enigmatic m of the secondary terminations. It might be that out of this sma the primary *tvas, ta-s kept the s and the secondary tam the m. No specific mark of duality has been retained in either of the two persons. Possibly it was a gradually formed usage which restricted them to this narrower sphere, though at the first they denoted indefinite plurality. The Gk. termination -τον is evidently related to this -tas precisely as the same termination in the 2nd dual is to -thas. By the side of this primary form stands a secondary. Here Indian and Greek show a remarkable accord, the former having -tam, the latter -την. The length of the vowel as compared with the short vowel of the primary form is striking. Misteli in his essay on the terminations of the middle (Ztschr. xv. 329), believes the ground of the lengthening is ■ 2 52 THE PERSONAL TERMINATIONS. ch. ii. to be found in the effort to distinguish it from the 2nd dual. But this effort cannot anyhow have been very strong in Greek, or the primary forms for the two persons would not have remained the same. I am more inclined to think that it was a kind of accident which preserved the long secondary vowel, and that the primary vowel was long too at first, that consequently there stood originally tas tarn side by side, and that this a was of the same nature as that in the -thct of the 2nd sing. We have occasionally to assume terminations with long vowels for impe- rative and middle forms as well. The agreement between Greek and Sanskrit in this point is the more remarkable because Zend here shows the short vowel (Schleicher, Comp. 3 670) : here the termination is -tern, e.g. in (fagae-tem=fiaaKoi-Tnv. The Ch.-Sl. te, both primary and secondary, does not help us much : Gothic and Lithuanian give us no help at all. "While the schema given by Greek grammarians — 2 du. τον τον 3 du. τον την derives powerful support from these considerations, Comparative Grammar has here much to say on the points of controversy which have occupied the students of Greek Grammar in particular. In discussing the strange 78 irregularities which meet us here we may take the middle forms into consideration at the same time, although we shall have to discuss their origin later on. Anyhow the parallel between τον την and σθον σθην is visible at once. The strange thing is that the rules of the schema given us by tradition are so very seldom observed in practice. No doubt this is partly because the dual forms do not occur very often anywhere, and only in Attic with any frequency. It is therefore worth special notice that the regular form in -την for the 3rd dual does sometimes occur in Attic in- scriptions (Corp. Inscr. Att. No. 358 άνεθέτην, 396, 374, and elsewhere έποιησάτην), and that we have in a Boeotian (C. I. No. 1580) and in a Dorian inscription (No. 25) a 3rd dual in -ταν : άνεθέταν, εποησάταν (Ahr. Dor. 298). To the grammatical rule there are exceptions of two kinds, which have been discussed with most minuteness by Aug. Bieber, De Duali Numero (Jena, 1864), p. 20, who follows in the steps of earlier authorities to whom we shall have occasionally to refer. 1) In Homer -τον takes the place of -την as the termination of the 3rd dual in secondary forms. Of this we have three certain instances, recognised as such by the old grammarians : Κ 363. ως τον Τυδύδης ηδ' 6 πτολίπορθος Όδυσσβύί \αου άποτμήξαντε διώκίτον έμμςνές αεί. Aristarchus's attempt to get out of the difficulty here by assuming an enallage temporum has been thoroughly refuted by Friedlander, Philol. vi. 669 ff. No refutation is needed of the view of other grammarians that there is an enallage personarum. Ν 345. τω δ' άμψ\ς φρονεοντε δύω Κρόνου vU κραταιω άνδράσιν ήρώεσσιν ετενχετον αλγεα λνγρά, where there is the entirely unsuitable variant τετεύχατον. 2 582. in a narrative — τω μεν εναρρηζαντε βοος μεγάΚοω βοείην έγκατα και μεΚαν αίμα Χαφυσσετον. ch. π. DUAL, ACTIVE. 53 A passage in Hesiod, Opp. 199, which used to be classed with these, is too uncertain to prove anything, for it is by no means certain that 'ltop, which is itself doubtful, 2 is to be taken as a past tense. I. Bekker, in 79 his review of Wolf's Homer (Horn. Bl. i. 50), wanted to bring the three Homeric forms into accordance with the rule, and proposed to read Ιιωκτην, ετύκτην, λαφύκτην, but since then he has shown a wise caution in not venturing to introduce these creations of his own into the text. In the case of *λαφνκτην particularly this would have been open to objection, as it would have been contrary to all analogy, and has no longer the support even of *συναίκτην. For in Hesiod, Scut. 189, it is no doubt right with the Ε. M. to read συ ναίγΰην, which Kochly and Kinkel have adopted. One or two third persons dual in the middle in -σθον for -σθην are mentioned as variants in our scholia on Ν 613, Π 218, Ψ 506. It is clear from this that the Alexandrians, and especially Aristophanes and Aristarchus, had to fight for their schema, which has since been held the regular one. We need not be surprised at this when we find that there were actually those who held that in Homer the dual might be used for the plural at will, a view which, although not without its supporters even nowadays, may be regarded as disposed of by anyone who will reflect on it (Bieber, De Duali Numero, pp. 39-46). The leaders among the Alexandrians no doubt drew from the superabundance of the forms in -την, -σθην for the 3rd pers. of the secondary forms the correct conclusion that such was the prevailing rule, and Comparative Grammar in this instance endorses their verdict. No one I think will now accept Buttmann's decision that * the distinction between the dual endings ov and ην was not matured till the time of the later poets ' (Ausf. Gr. i. 2 341 note). But the question is how to regard these remarkable exceptions. Thiersch (p. 352 note) thought there might have been a shortening of -ετην to -ετεν for the sake of the metre. In support of this might now be adduced the Zend forms in -tern. Still more artificial and arbitrary is Bollensen's attempt (Ztschr. xiii. 202). We have every reason to be very careful how we alter the text of Homer to suit grammatical rules 80 and theories. When we set against this the fact that the three verses occur in parts of the Iliad which are certainly not among the oldest — the Doloneia, the Shield of Achilles, and in a passage which Bekker finds ground for regarding as an interpolation — we shall be inclined I think to attribute the anomaly to the want of proper linguistic instinct on the part of some late rhapsodist. There are next to no forms in -τον for -την in Attic writers : Plato, Euthyd. 274, εψατον, Thucyd. ii. 86, where Classen, though he follows others in reading the unobjectionable Ιιεχετον, still has a word to say for the ΰιείχετον of the M.SS. In Aristoph. fr. 523 Dind. it is impossible to make the words of the grammarian who cites these verses in the Et. M. a sufficient ground for pronouncing κατανηβολέϊτον and εκμαίνετον unaugmented past tenses. They are clearly historical presents. Such a state of the facts is surely enough to warrant the alteration of Plato's εκατόν into εώάτην, 2 Compare Hesiodea edd. Koechly et Kinkel, who have adopted ϊτην, and Rzach, Der Dialekt des Hesiodos (Jahrbucher f. class. Philologie, Suppl. b. viii. p. 438). 54 THE PERSONAL TERMINATIONS. ch. n. 2) More attention has been paid to the anomaly which is the reverse of this, the substitution of -την (-σθην) for -τον (-σβον) in the second person dual of the secondary forms. Since Elmsley's note on Aristoph. Acharn. 733, and Eurip. Medea, 1041, a small literature has amassed itself on this subject, out of which I will only mention Buttmann, Ausf. Gr. i. 2 341 ; Cobet, Mnemos. viii. 408, Κόντος Λόγιος Έρμης, i. 29 ff., and Fritsche's most sensible discussion on Aristoph. Thesm. v. 1158. Elmsley went so far as to reject altogether the second pers. in -τον in historical tenses even in Homer, and therefore not only preferred at θ 448, Κ 545, Λ 782, Zenodotus's καμέτην, λαβέτην, ήθελέτην to Aristarchus's κάμετον, &c, but at θ 456 altered the metrically estab- lished ΐκεσθον in the most arbitrary fashion, and in direct violation of the syntax, into the conjunctive Ικησθον. This conjecture therefore of Elmsley's has met with no approbation. With respect to the Attic writers the case is different. Here the sagacity of the English critic and his followers has established that in at least six passages in Plato, including one in the Eryxias — Euthyd. p. 273ε ενρέτην, επεΰημησάτην, ib. 294ε ηστην, Legg. VI. 735A εκοινωνησάτην, Symp. 189c ειπέτην, Eryx. p. 199d ε7τεΓελεσά7-»7»', in the scholium on Harmodius and Aristogiton in 81 Athen. xv. 695, Soph. 0. It. 1511 ειχέτην, which is established by the metre, Eurip. Ale. 661 Dind. ήλλαζάτην — in nine passages that is in all, to which may perhaps be added Aristoph. Nubb. 1506, -την for -τον in the second person is fully established. But over against these nine or ten passages stand at least thirteen in which the M.SS. have, -το ν for the second person in secondary forms : Aesch. Ag. 1207 ηλθετον, Soph. O. C. 1379 εαυτόν, 1696 εβητον, 1746 ελάχετον, Eurip. El. 1300 ήρκέσατον, Med. 1073 ευίαιμονοϊτον, Ale. 272 όρωτον, Aristoph. Vesp. 867 ζννέ- βητον, Αν. 112 Thesmoph. 1155 ήλθετον, Plut. 103 εμέλλετον, Plato Euthyd. 273e ff. ε'ιητον, ελέγετον, εψατον (a little above comes ενρέτην), Legg. i. 646b, εψατον. In none of these cases is there the smallest ground for any alteration, and yet Dindorf (for the past tenses at any rate), Nauck, /γε'«7-αι, υΐκέηται, πεποΐ'έαται, εμεμνεατο, κεκλέαται, κεκινέαται (Hippon. 62, 2) and other forms of the kind (Bredow, p. 328) where the corresponding Homeric 99 forms ha-ve -ηαται. 7 And as the New-Ionic dialect has a decided preference for such accumulations of vowels, the -arai, -ατο not only established itself in such forms as merely admitted of it in Homer, e.g. in άπεΰεικνύατο, Ιΰρύαται, and in the above-mentioned κέαται, but even made its way into words in which it followed short hard radical vowels : τιθέ-αται, εκΰιΐό- αται, neither of which are really more remarkable than τιθέ-ασι, ΰιΰό-ασι. a is dissimulated to ε : Bvi'C-arat, έπιστε -arat, πεπτε-αται. But forms like εβονλέ -aTOy κηΰέ-αται for εβονλοντο, κήδονταί must be regarded as apocry- phal, since it is against all analogy that the thematic vowel should be attenuated to ε and followed by -α™*, -oro, instead of -vrai, -ιτο. This is the decision arrived at by Dindorf (praef. p. xxvii.) and Abicht (Philol. xi. 275 8 ), and now adopted by other editors of Herodotus and students of his dialect (Stein, Herodotus, 4th edit., Berl. 1877, p. 57). In Attic Greek -arat, -ατο only survived after consonants, and even then only in the perfect, and served as a distinguishing mark of the older Attic writers, who therein closely follow Herodotus ; so in Thuc. iii. 13 ϊψθαραται, iv. 31 ΰιετετάγατο, v. 6, vi. 4 ετετάγατο, Xenoph. Anab. iv. . 8, 5 αντιτε τάχαται, Plato, Rep. vii. 533 τετράφαται, and it has the testimony of the inscription of Methone which dates from the beginning of the Peloponnesian war (Sauppe, Inscriptiones Macedonicae, iv. p. 7), where we find ετετάχτιτο, γεγράψαται (cp. άναγεγράψαται, C. I. 75, 3). More- over the dramatic poets did not regard the opt. in -taro, the only form in use with Homer and Herodotus (cp. too Simon. Amorg. i. 22, vii. 107, Charon of Lampsacus in Ath. xii. 520 εζεπισταίατο), as too outlandish to be used in their poems in passages where a tinge of antiquity was not out of place, and they found it very useful metrically, especially at the end of the verse. Fischer, ad Wellerum, ii. 418 (cp. Matthiae, § 204, 100 7a), gives the following instances : Aesch. Pers. 451 εκσωζοίατο, 369 ψενξοίαβ' (both in a messenger's speech), Suppl. 754 εχθαφοίατο, Choeph. 484 κτιζοίαθ', Sept. 552 6\olar.o, Soph. Aj. 842 (?) όλοίατο, Ο. R. 1274 όάοίαθ' — γωτοίατο (messenger's speech), Ο. C. ΰεξαίατο 44, πεμψαίαθ' 602, 921 πνθοίατο, 945 fotoiur', El. 211 άπυναίατο (in a choral passage), Eurip. Hel. 159 άντιΰωρησαίατο, Here. fur. 547 ε/νΤίσαίαΓο, Aristoph. Pax. 209 αίσθανοίατο, Αν. 1147, Lys. 42 εργα- σαίατο, Nub. 1199 υφελοίατο. Now that we have examined the extent to which these forms, in 7 The form πειτλήαται, used by Simonides Amorg. 36, stands alone. Cp. Renner, Stvd. i. 2, 24. H Ace. to Abicht (cp. Kiihner, i. 548), though there are any amount of presents in -ovrot, there are only three in -carat, none of which has the authority of the best M.S., the Medicean." Among thousands of past tenses, there are only six forms in -«ατο which are attested by all the M.SS. But as there is not the least probability that a prose writer said lyivovro ten times and changed it the eleventh to iytvitLTo, it cannot be doubted that these forms made their way often into the inferior M.SS., and occasionally even into the better ones, on the false analogy of the pluperfects and preterites, like Ιστ4ατο. ch. ii. DUAL, MIDDLE. 67 -αται, -ατο occur, it remains for us to notice a view of their origin which may appear to some not unwarrantable. In cases where -rat, -ατο come after vowels, and such cases are the majority, it is not unnatural to con- jecture that we have here the result of a composition, and Schleicher (Comp. 3 678) declares this view to be a tenable one. There is nothing surprising in the idea that there should here have been a composition with the 3rd pi. mid. of the rt. as, Gk. ες, which would be εσ-αται, tV uto, since we meet with such compounds in the active, and not only in past tenses like 'i-co -σαν, ί-ψα-σαι , but in ί-σασι=ϊ ιΖ-σαντ l and ε'ίζασι — είκ-σαντι. But for all this it is only the Heraclean γεγραψάται mentioned on p. 64 that could be allowed to be thus compounded". The forms with a 3 like the already mentioned Homeric άκηχέ^-αται,ερηρέ^-αται, ί\ηλά(-ατο, or the Herodotean εσκευάΰαται, κεχωρί^αται, and again forms like the Herodotean άπίκαται, απίκα-ο, exclude all possibility of such an origin. Nor is it any more probable for εστάλ-ατο (Hes. Sc. 288), κφθύρ-αται (Thuc. in. 13). Phonology teaches us that it is only between two vowels that a σ can fall out. Now it is scarcely necessary to point out how unlikely it is that βεβλή -qrai and ειρν-ατο should have arisen in any other way than the consonantal forms. We had to come to a similar decision on p. 48 in the case of the 3rd pi. active. We have just as little ground for conjecturing the loss of a a here as in τώεασι or μεμάασι, γεγάασι. DUAL FORMS IN THE MIDDLE VOICE. The dual of the middle voice has this advantage over that of the active that it has a first person to itself. It is true though that the termi- 101 nation -μέθα ν is by no means connected organically with the corresponding Skt. -vahe. The latter is clearly to be referred, on the analogy of the 1st pi. make, to a preceding va-dha-i, so that here, as in the active, the existence of a weaker form in ν by the side of the stronger in m was• made use of to differentiate the two numbers. In Greek it was other- wise. Between -μεθον and -μεβα there is clearly no greater difference than between the Aeol. -μεθεν and -μεβα. It is true that the existence of the whole form as such has been called in question. Elmsley on Aristoph. Acharn. 741 (733, 698), where he expresses the doubt referred to on p. 54 of the correctness of the usual schema of the active dual, has the merit of having pointed out that the form in -μεθοϊ only occurs three times in good authors, i.e. Ψ 485 hv- ρό vvv rj τρίποδος περϊίωμεθον ήε Χξβητος, Soph. ΕΙ. 950 μόνα ΧεΧείμ- μεθοι; Philoct. 1079 ι ώ μεν ουν όρμώμεθον. In all these three passages it has the testimony of the best M.SS., but is not demanded by the metre, for in Homei• there would be nothing impossible in a hiatus after περιίωμεθα. Moreover Hesychius's gloss περώωμεϋα' σννθωμεθα is evidence that there was such a reading in this passage. To these wo must add the would-be antiquarian Pompeianus in A then. iii. p. 98 — also adduced by Elmsley — who says in the adclress to his slaves πρότερο ν σνντριβησόμεθον, επειθ' όντως άποΧονμεθον. Bieber, de duali numero, p. 18, shows how often in Homer and the dramatic poets the opportunity for using this form was neglected, and that even the 'Ομηρου 'Επιμερισμοί in the Anecdota Oxon. i. 406, actually give τινύμεθα as a dual. For all this Buttmann, A. Gr. i. 2 343, and G: Hermann on Soph. El. 937 (950), and Kuhner, i, 543, are no doubt right in deciding that Elmsley f 2 68 THE PERSONAL TEEMINATIONS. ch. it. goes too far in wanting to reject this form everywhere as an invention of the Grammarians. A rational criticism will always be inclined to see a relic of antiquity in exceptional forms which do not violate analogy. How could such forms have been invented 1 It is easier to imagine that a by-form of the 1st pi. was by local -usage, Or the influence of a gram- matical theory, and not without reference to the -of of other dual forms, 102 transferred from the plural and stamped as a ai\idat r Gortynian inscr. 1. 4, apparently= άιαιλεϊσθαι (Voretzsch ut supra 673). To this we can now add ττοηασ- σαί=ποη'ισαπθαι from the Olympian inscr. of Damokrater (1. 33). These dialectic by-forms entitle us 1 think to assume two things : first, that the θ is not an essential and inherent element in these termina- 104 tions, but grew out of a r ; in the second place r that the σθ belongs to a class of sound-groups which have been considerably changed by dissimila- tion and assimilation. The examination of the 2nd sing, in -σθα (p. 37) brought us to an unaspirated t, and in niauy other instances it looked probable that spirants had had a hand in the formation of the σθ. * The most familiar phonetic laws serve thus to account very simply for the forms of the imperative middle. As Schleicher, says (Comp. 3 676), ' It is possible that these forms sprang from middle terminations in which the initial consonants of the two pronouns of the 3rd or 2nd; person were brought into juxtaposition by the disappearance of the intervening vowels, and that ττ became στ and then σθ.' I think that we can explain the σθ in the 3rd sing, imperat. middle (No. 7), e.g. in <ράσθω, which we shall afterwards refer to an older *φάσθωτ, by supposing it to stand for ττ, that is ψάσβω for ψα-τ-τω-τ,.ίϊΐ which the connexion of the exponent of the 3rd pers., which is doubled for emphasis' sake, with the preceding ~, which is the sign of the same person, gives he-he-himself, and that is the mark of the imperative middle. The same explanation serves for the 3rd plur. The fact that the ν of φάσθων is wanting in the Doric dialect shows it to be no essential part of the termination, and φά-σθω, as will be shown hereafter, stands for *φά-σθω-τ. We will postpone the discussion of the remarkable forms in -οσθω to a later section of this book (ii. 51 ff.). The 3rd pi. of the imperat. middle was evidently not always identical with the 3rd sing., while in the 3rd du. imperat. midd., e.g.. Φάσθων (No. 9) the absence of dialectic by-forms precludes the formation of a definite opinion about the final letter. It is enough for our purpose that we can with great proba- bility set down for the three 3rd persons of the imperative middle the series — ττ στ σθ and explain the σ by dissimilation, and the θ as due to the aspiration so frequent after a σ (Princ. ii. 110). In the Cretan form in 63 we must assume that after dissimilation had done its work a retrogressive 70 THE PERSONAL TERMINATIONS. oh. 11. 105 assimilation took place. This dialect shows other instances of the ten- dency to assimilate a sibilant to a following explosive (Hey de dial. Cret. p. 33). Nearest to these imperative forms come the 3rd duals in -σθον (No. 5) and -σθην (No. 6). We have seen that there was" no original and thoroughgoing distinction between dual and plural forms,. and so we must expect to find the same elements in the σθ of these forms as in that of the imperatives. Here too then the series is ττ στ σθ. Having thus disposed of five of the ten forms, we now turn to the 2nd persons dual and plural in -σθε and -σθον. They differ from the 3rd persons just discussed in much the same way as the 2nd pi. act. in re and the 2nd dual act. in to ι - from the 3rd dual in τον, that is, the first dental sound in these 2nd persons is to be referred to the pronominal stem tva, that of the 3rd to ta. There is therefore nothing to prevent our refer- ring the σθ in φά-σθε, ψά-σθον likewise to a ττ. The three remaining forms are more difficult. The 2nd sing, in -σθα and the 1st plur. in με-σθα, as we saw on p. 37, are again the most nearly connected of the three. The cases hitherto examined will incline us to suppose a similar process of development here. In the first place analogy, on all strict rules of procedure, leads us straight to the supposi- tion that as τσθω and -σθαι arose directly from the authenticated dialectic byrforms -στω and -στα t, so pur -σθα arose from a *-στα. This step brings this termination considerably nearer to the Lat. -stl and the Goth, -st mentioned on p. 36. In the next place we may conjecture that στ arose here, as in the cases already analysed, from ττ. This ττ' again can hardly have . arisen in any other way than by progressive assimilation, and conse- quently from the tv of the pronominal stem tva. It is true that it would be hard to find more than one instance of such assimilation in Greek ; Att. τέτ-αρ-ες, Boeot. πέτταρ-ες (Ahrens, Aeol. 176) from the primary form katvar-as, but this one, which is beyond a doubt, is enough to corro- borate the conjecture that ττ might have come from r/. In this instance, it is true, the ττ remained intact in older Attic and in Boeotian at least, 106 while in the personal terminations it underwent further transforma- tions. But there are other cases in which the tendency to phonetic lightening went further in terminations than in stem-syllables. If we assume then that in the same period of the language which saw forms like that deduced on p. 69 and *<ράττω and *ψάττε, the 2nd pers. sing, was ε-ήί-ηττα, it is not surprising that ε-φησθα should have come from the latter by way of *ε-ψηστα in the same way as ψάσθω arose from *ψάττω and φάσθε, as it appeared, from *φάττε. There remains the infinitive in -σθαι. Here we have in addition to the Locrian forms in -σται above mentioned Hesychius's isolated Ιτθαι ' καθ ίσα ι, for which Mor. Schmidt wants to read καθήσθαι. The r in this form may have come from the radical σ as in εττία=ίστία (Hes.), and this leaves us with -flat for the termination, which bears to the -σθαι of ordinary Greek the same relation that -μέθα does to the Homeric -μεσθα. If, as Ahrens 177 holds, this form were Boeotian, we should expect η instead of at, on the analogy of άπογράφεσθη, δεϊόχθη (Ahrens 187), both in inscriptions. We may follow Bopp (VgLGr. iii. 330) and Schleicher (Comp. 3 446) in making a comparison between the middle infin. termination and the Vedic -tfhjdi, Ζά. -djcii or -dhjai, e.g. in the Skt. ja£a-dhjai=Gk αζε-σθαι, ch. π. 2Θ IN PERSONAL TERMINATIONS. 71 although the oriental termination expresses the action in itself without reference to active or middle meanings. It must be admitted that this difference in meaning does not amount .to so much in the case of the infinitive, which seems from the first to have expressed merely the action as such and not a definite relation to a subject. The inquiry into the origin of the termination we will postpone to a later chapter. Here we have to deal only with the relation of the Gk. -σθαι to this -dhjdi, and I think the simplest explanation of this is the following. From -dhjai, when the soft aspirates generally shifted into hard ones, arose -Ojcu, thence, by progressive assimilation, -θθαι, or as it is probably more correct to write it, -τθαι. It would be not at all surprising if we were to discover on a Boeotian inscription forms like *γράφετθη by the side of the actually existing άττογράφεσβη. For the present we must make what use we can of the above-mentioned Ιτβαι. From -τβαι the next step was 107 to -σθαι by the usual dissimilation. A distant analogy may be noticed in the process which must be assumed to explain the Homeric πέποσθε (ΰ-επ-ονϋ-τε, πεποστε, πέπυσθε). Whether the Elic ποήασσαι came from the ordinary ποιήσασθαι or by assimilation from an earlier form. is a question I cannot decide. Several of the views here presented owe their origin to a paper read many years ago in my ' Grammatische Gesellschaft ' by Dr. Richard Klotz, which he has allowed me to use here. Other related matter received a similar treatment by Allen, Stud. in. 243. I differ from the latter, who moreover deals with a part only of the forms here dis- cussed, mainly in this, that I cannot admit the assumption that θ in these old forms expanded by a purely phonetic process into -0. LIBRARY UNIVERSITY OF iALIFORflA. J 72 THE AUGMENT. CHAPTER III. THE AUGMENT. No special mark is needed to distinguish present time, for the connexion of the stem with the terminations naturally conveys the impression that the subject and predicate are to be thought of as connected at. the time which is present to the speaker. Whether notwithstanding there may have been a period when the same forms served for past time as well, may be left an open question. It is certain that there arose very early a special form to designate a past action, and that this form was characterised by two things, the augment, and the shorter, secondary form of the personal terminations. These secondary terminations have been already discussed in the first chapter, where we conjectured that the weight added to the beginning of the word by the augment furnished the first inducement to the shortening of the terminations. This is the place at which to treat of the augment as the linguistic element which serves in its proper function, and probably served at first solely, to express past time. 108 Our word augment is a translation of the ανξησις of the later gram- marians. Lentz, Herodian I. p. lxxxviii, following Skrzeczka, shows that this term is not to be found in Apollonius Dyscolus or Herodian, who describe the prefixing of the augment much more aptly as a part of the κλίσις or κΊνησιν of the verb, whereas the name, augment implies an unessential and purely external accretion. We shall, however, retain the old term. The augment is only found in Sanskrit, Iranian, and Greek. 1 It consists in the syllable a, Gk. ε — instead of which there are some traces, to be discussed below, of a long vowel — and this syllable is prefixed to the verbal form, and that only in the indicative mood. In the language of the Vedas as in that of the Homeric poems the augment is sometimes left out. To conclude from this that it was not an essential part of these forms seems to me (so too Delbriick, Altind. Verbuin, p. 80) rash, for there are other cases where a language vacillates between a fuller and a shorter form, and where we are right in holding the fuller to be the more perfect of the two. As soon as a language has, with the aid of its meagre store of elements, succeeded in providing a distinct expression for any given notion, one or another of these elements may be discarded again without any loss of distinctness As an instance of this take the way in which secondary personal terminations in many cases take the place of primary ones, or that in which the original 8 of the nom. sing. ' Since Armenian seems proved by Hiibschmann's' investigations to be an independent and individual link in the chain of Indo-Germanic languages, it deserves to be noticed that there are traces in this language too of an augment, and that too in 'the form of e, e.g. e-di~*t-fh\-v, e-tu = *t-Hwv (Hiibschmann, Ztschr. xxiii. p. 34). ch. πι. OEIGIN OF THE AUGMENT. 73 or the s of. the Skt. ace. plur. in ns is lost. It is not to be wondered at that a prefix of such little weight, which moreover, in Greek espe- cially, does not always help the metre in poetry, should early have found its existence precarious, and have been discarded altogether in the majority of the related languages. It is surprising, on the contrary, that this element should have survived as much as it has in languages of three different families. Without the assumption that the augment was at one time present in all forms expressing past time, the organism of 109 tense-forms seems to me inexplicable. If, therefore, differing from Schleicher (Oomp. 3 738), we regard this a as having formed from the first an essential part of the preterite, we have to ask next what view is to be taken of its origin. On this point very different opinions have been expressed, 2 the most important of which we will now proceed to examine. 1) Buttmann in his Ausfuhrl. Gram. I. 2 312 expressed the opinion that the augment was only a curtailed reduplication. Thiersch agrees with him, saying at p. 231 of his ' Griechische Gr. vorziiglich des homer. Dialekts ' that ' the syllabic augment was originally identical with the reduplication;' so to a certain extent does Pott, who (Et. Forsch. II. 1 73) calls the augment ' a variety of the reduplication,' and (Doppelung, p. 226) ' an embryonic reduplication.' There are various points in the Greek use of the augment and the reduplication which at first sight seem to make for the identification of the two. . The reduplication now and then actually assumes the same form as the augment : ΐ-ζητη-κα ί-ζητη-σα, and although again the former seems to belong properly to the perfect, it appears in aorists like λέ-λαθ-α-ν, πέ-πιθ-ο-ν, to be taking the place of the augment. But we encounter difficulties as soon as we look further. In Sanskrit the augment always appears in the form a, while the redu- plication always changes its vowel to suit the stem of the verb : a-tuda-m but tu-toda, d-bheda-m but bi-bheda. This objection, raised also by Bopp (Vergl. Gr. ii. 2 422), might perhaps be met by the not impossible supposition that, as has been sometimes argued, among others, by Nolting, in his essay ' iiber den genetischen Zusammenhang des Aoristus II. mit 110 dem Perfectum II.,' Wismar, 1843, the original vowel of the reduplication was in all cases an a. The Skt. forms ba-bhuv-a (rt. bliu be) and sa-suv-a (rt. su to bring forth), and the Old-Latin perfects pe-posc-i, pe-pug-i (later po-posc-i, pu-puy-i), are in favour of this supposition. A still weaker point in this theory is the conjecture that such a number of initial consonants should have disappeared with no sufficient cause ; and yet this is what we must suppose in order to get from *ba-bheda-t, which we must start from, to d-bhedat, and that too for the early period before the separation of the languages, when the articulation was generally strongly marked, for anyhow there must have been an a then that was completely distinct from .the *ba, *ka, *ta &c. The specifically Greek habit of putting ε instead of σκε, σπε, στε, ζε, where the verb-stem begins with a double consonant, is therefore no adequate 2 Compare, too, the Leipzig doctoral .dissertation of Konrad Koch, Be Augmento apud Homerum omisso, Brunsv. 1868, the introduction to which gives several of the views mentioned below. — To this we may add P. Molhem's careful work, Be migmenti apud Homerum Hei'odotumque iisu, Lundae, 1876, and the accurate examination of Hesiod?c usage in Rzach, Bcr Bialekt des Hesiodos (8th supplem. to the Jdhrbucher fur Class. Philologie), p. 431 f. 74 THE AUGMENT. en. m. analogy. Again, the identification of the reduplication with the aug- ment necessitates in all consistency the identification of all past tenses with the perfect in their terminations as well, and it is clearly no use to attempt that. The most important objection, however, is to be gained from the impress borne by the verbal forms themselves. The augment belongs exclusively to the indicative, the reduplication is excluded from no mood, not even from the participle and infinitive. The augment serves, that is, to mark a pas't tense ; it is the exponent of a grade of time, while the reduplication characterises a tense-stem all through, attaching itself firmly to it, not confining itself exclusively to the perfect stem, but appearing occasionally in the present and aorist as well. From this it is clear that the reduplication was not originally a mark of past time, and that the apparent substitution of the reduplication for the augment in certain aorists is not what it seems, for, as will be shown below, the augment occurs sometimes in these very aorists as a sign of past time prefixed to the reduplication which characterises the tense-stem as a whole : ε-κέ-κλε -ro. Moreover the fact that the pluperfect shows the two united is a clear proof that we have here to deal with two quite distinct linguistic elements. For these reasons we may regard this view as* exploded. It was a natural attempt to explain the more difficult form by means of one which seemed a somewhat- more comprehensible one, but it belongs to a more backward stage of the Science of Language than the present. Ill 2) Hoefer in his ' Beitrage zur Etymologie ' (Berlin, 1839), p. 388, attempts to connect the augment with the Teutonic prefix ga (gi, ge), which seemed in its application to the expression of the perfect to come near to the function of the augment. But the assumption that the initial consonant of this prefix originally varied between a guttural explosive and the dental sibilant, and then disappeared altogether, will scarcely find acceptance with anyone. Besides, this attempt too rests on a confusion of the meanings of the perfect and the preterite. 3) A third explanation is that given by Bopp (vgl. Gr. II. 2 415), who takes it to be the α privative. This is met at the outset by an objection on the ground of the form. The negative prefix is only a- before consonants, but is elsewhere cu-, while there is not a trace to be found of a nasal in the augment. In regard 6o meaning, however, this hypothesis is less • satisfactory still. It is true that past time is not present, but it is highly improbable that language should have marked it as not present. The negative force of the perfect in dixi I have said my say, fuimus Troes and the like, to which Bopp appeals, is by no means enough to prove this. In the first place we have in this usage not a preterite, but that kind of perfect which we may term absolute. The statement of the full com- pletion of an action implies, it is true, that it is no longer continuing, but the preterite, which transports the action to a section of past time chosen at will, does not present a contradictory opposite of the present. So far is it from this that the so-called gnomic aorist actually puts before us something done in the past as a rule that applies to all time : κάτθαν όμως ο τ άεργος ανήρ ο τε πολλά εοργώς. Again, a negation prefixed to a verbal form, as nescio and the like show, negatives the whole assertion, not merely a comparatively accidental qualification like that of time, a qualification moreover which on this showing is itself expressed by no «ι. πι. ORIGIN OF. THE AUGMENT. 75 special external mark. If, then, the a in 4-tudam were negative, it would mean, as opposed to tudd-mi, ί I do not strike,' not ' I struck,' For all these reasons, this explanation, which Bopp himself was so little satisfied with that he proposed another, to be mentioned below, as an 1 1 2 alternative, may be set down as erroneous. 4) Beufey, in Ms Kurze Sanskritgrammatik, p. 85, and the Kieler Monatsschrift, 1854, p. 733, sees in the augment, 'as the original instru- mental case of the pronominal stem (*,' the expression of the relation to another action. He quotes the use of the present in Sanskrit with the particles ]ntrd * before,' and sma, which he supposes to mean ' at the same time with,' and concludes that ' in these cases past time is, properly speaking, only in so far denoted as the action to be thought of as occurring in it is represented as having happened along with, or before ; is represented that is as tempus relativum, which is exactly what is expressed by the old Indo- Germanic imperfect.' Benfey, too, brings the Teutonic ge into the question, and assigns to it a similar function. But there is a fundamental error here. The syllable ge does mean * together,' but by no means the putting one action together with another : it denotes the collection together of all the elements of an action, and resembles the con in conficio. It thus expresses not an external but an internal connexion, and provides the verb with a means of expressing completion, and for this very reason its temporal force is a secondary and not an essential one, and has only gradually become attached to it. The notion of relativity, moreover, would at most only fit in with the mean- ing of the imperfect, but not at all with that of the aorist, and would not be a probable accompaniment even of the imperfect, for in dealing with this tense too we certainly ought to start from its use in simple isolated sentences. The relativity is clearly only a result of the durative force of the imperfect, so aptly represented in the Gk. name παρατακτικός. 5)• There is but little difference between Benfey's view and that of Scherer, and it seems to. me that both are equally unsatisfactory. The latter conjectures" (Zur Geschichte der deutschen Sprache, p. 230) the primary meaning of the augment to have been ' near at hand.' It will not be easy to establish the connexion of this meaning with the pro- nominal stem a. But granting it might mean this, I cannot see how the notion of nearness could be transformed into that of a past time, which is anyhow not next to the speaker. It is true indeed that Scherer believes 113 that this augment was as it were only an accessory indication of what was already implied in the form itself, and ends by translating this a or a by there, and thus arrives at a view which is not far removed from that to which we are coming next. 6) By the side of the explanation given under 3) Bopp mentions another which he thinks a possible one. Though he regards the two explanations as nearly the same, they are really very different. On p. 420 he expresses the opinion that language, in prefixing an a to verbs, may perhaps ' not have been thinking of the negative a, and not have meant to deny the existence of an action in present time, but have used the a as a real pronoun in the sense of that, and so have intended thereby to transport the action " away there " — to the time lying in the distance and behind them.' Schleicher adopts this view, inasmuch as he regards the reference to past time (p. 749) as the function of the a, and so does Richard Garnet in the Proceedings of the Philological Society, Vol. I. 76 THE AUGMENT. ch. hi. (1844), p. 265, where various parallels not all equally apt are ad- duced from other languages, some of which are quite unconnected with the Indo-Germanic stock. It deserves notice anyhow, that languages of a less formed character denote past time, and future too, by particles that point to the distance. If the pronominal stem a meant that one yonder as an actual pronoun, in an uninflected form it must have corresponded to our there, and, temporally, to our thhi. Scherer is unwilling to allow that the stem a points to something at a distance, and mainly for this reason : he recognises this same stem in the a of aham i I.' It may be doubted whether this stem always and exclusively had the force of pointing to a distance. Most of these distinctions between 'this' and ' that,' \ here ' and ' there,' were probably developed antithetically in each separate lan- guage. It cannot be denied, however, that a series of forms undoubtedly belonging to this stem are used to refer to something at a distance. Scherer himself mentions the fact that in Zend athra there is contrasted with ithra here. The Skt. d-tra when used of place means here, but when used of time then, at that time, so too atas thereafter, a-ti out beyond there (=ετι), ά -tha then, therefore, at thereupon, then, while the 114 preposition a, which unquestionably belongs to the same stem, with its main meanings of to, up to,, and as an adverb• hither, further, takes an intermediate position. Perhaps the best representation of the meaning of the particle that is used as the augment is that it is equivalent to the ϊνβα with which the story of the Odyssey begins : cvff άλλοι uei/ πάντες όσοι φνγον αίπνν ολίθρον οίκοι βσαΐΛ A reference is made to some point chosen at discretion ; present time needs no such reference, and it was not till later that the need of denoting future time arose, and so this there became the there of past time. This view is only a shade different from Scherer's final explanation (p. 231). In proceeding to examine the various phenomena connected with the augment, wa will follow the old division, based on the nature of the subject-matter. A) The Syllabic Augment. In the dialect of the Yedas Kuhn (Beitr. iii. 463) points out that a long a sometimes takes the place of the short one. The cases, however, which are referred to by Delbruck (Verb. 79) as well, are few in number, and it may be doubted whether the length is inherent even in them, or whether it is due to a kind of ' position ' effected by the following con- sonant. In Greek there are three verbs which occasionally take η instead of ε for augment : μέλλω? Ζΰναμπι, βονλομαι. The ancients, whose notices of the phenomenon have been best collected by Fischer ad Wel- lerum, ii. 299, call the substitution of η for t in these verbs an Atticism (Moeris s. v. ήμελλυι'). Properly speaking, we can only be sure of the instances from the poets in which the metre testifies to the η, for the M.SS. of the prose writers vacillate much between the two forms. In this way we get the following result : μέλλω has η according to Zeno- dotus — though Aristarchus rejects his reading — at Μ 34, in Hesiod (Theog. 888 αλλ' or*, h) ρ ήμελλε θεάν γλανκώπιν' Άθηνηΐ' τέttσθ^>t)by the side of έμίλλυν at other places (Rzach, p. 430), in Theognis (906 ήμε\λ' εκτελέσας ίΐς Άί?αο πέραν), in Aristophanes (Eccl. 597 τοϋτυ γϊψ ch. in. THE SYLLABIC AUGMENT. 77 ήμελλον έγω λέξειν, Ran. 1039 τον λόφον ήμελλ' επιθήσειν), ΰύιαμαι in Aeschylus (Prom. 206 ουκ ήΰυνήθην), and the comic poet Philippides Mein. iv. p. 472 έπειτα φυσάν δυστυχής chtc ήΐυνω. For ηβοΰλετο only 115 two instances from poetry are adduced, Eurip. Hel. 752 6 θεός ουκ: ηβοΰλετο, and Alexis fr. 256 Mein. ου το -κ pay μ ηβοΰλετο, neither of which prove anything. Since then this phenomenon is unknown in Homer, where ψελλέ -and έβοΰλετο are often established by the metre, and since it is not till the Attic period that its gradual establishment begins, we must be careful how we call it a primitive growth. Anyway the explanation is hard to find. If we take the analogy of έθέλω, θέλω 3 it is only in the. case of μέλλω that we get any help from the etymology of the word, which seems to have lost a σ before the μ (Principles, i. 412), and prothetic vowels are not unusual before double consonants. In the case of βούλομαι, where the β has come from a f t we might appeal to the prothetic ε in έείκοσι for if είκοσι etc., and assume a by-form εβούλομαι, or we might even attribute the η to the lengthening power of the/ itself, which we find at work in ή-είΐη (No. 15 below^ Brugman Stud. iv. 166). The etymology of ΰίναμαι is not clear. Of the a, the usual form of the augment in Sanskrit and Old-Persian, a few traces have been preserved in Greek. Whether such a trace is to be found on an inscription is not .certain, for the right- to-left superscrip- tion on a helmet found at Olympia (C. I. G. no. 31) 90ΙΟΣΜΑΠΟΕΣΕΦΥ . . . may either be read μ' άπόησε with Ahrens (Aeol. 229), or μα πόησε with Boeckh. The latter reading with the .augment missing assumes that the inscription is in poetry. But it is quite possible if we suppose the rhythm to be dactylic, especially if with Boeckh we take the first word to be Κωος and the two last letters to be the beginning of a proper name. It is true that μά for με is unparalleled, but πάρ for περί is also known to us only through the one old Rhetra of the Eleans C. I. no. 11, and πατάράίον πατέρα only through the lately discovered Locrian inscription, and other peculiarities of the same isolated nature are being discovered every year. Hence this can only be said to be a possible and not a certain instance of a as an augment, and we have to go for further cases to the glosses of Hesychius άΰηρεί'Ι εΰειρεν, άβραχεν' ήχησε); άσβεσθε' διέφθειρε Κρήτες. The doubts expressed about these and the alternatives proposed \\Q for them seem to me of little weight, but it must be admitted that this lexicon has no claim to infallibility, άσβεσθε I take to be a preterite like εσχεθοι; from the rt. σβες (σβέννυμι, άσβεστος), meaning exstinxit. Forms with the syllabic augment are witnesses in many ways to the older initial of the verbal stems concerned. Cases of this kir.d fall into two main classes. 1) Double consonants following the augment. It is well known that p is almost invariably doubled after the aug- ment. This fact has long since been compared with the same pheno- menon in compound nominal forms, and it is impossible not to see the parallel between ΐ-ρρεε and περί-ρρυτος, ερρηζα and άρρηκτος. Buttmann 3 Cp. Buttmann, Ausf. Gr. i. 317. 78 THE AUGMENT. CH. III. (i. 84) thought the reason lay in the fact that an initial p was pronounced like a double consonant, and appealed in proof of this to the rough breathing written over it. Since, however, the same doubling which is the rule with p occurs sporadically in Homer in the case of other initial consonants as well, we are compelled here, as is now pretty generally admitted, to 'regard the doubling of consonants in the great majority of cases as an assimilation (Ahrens, Formenl. § 85 j Hoffmann, Quaest. horn. i. 135). The verbs beginning with ρ are exhaustively discussed by Leo Meyer (Ztschr. xv, 1 ff.), where, however, he is supporting the, as I think,• erroneous theory that in the case of verbs which can be shown to have onge had / before p, not only has Homer's language traces of this Bound to show, but the sound itself. The assertion made by the same scholar (p. 3) that ' it is extremely improbable that the Homeric Jr was ever assimilated to a following p,' appears to me altogether un- founded. In post-Homeric Greek the doubling of the μ is undoubtedly to be explained in the way suggested above. It should be noticed that of the verbs which begin with ρ many can be shown to have lost a consonant, 117 i-e. either a f or a n. A /is established by clear traces in ρέζω (rt. hpy Principles, i. 221), ρένω (ib. i. 437), rt. ρε by the side of ερ, hp (ib. i. 428), ρη-γννμι (ib. ii.. 159), 'ριγέω (ib. i. 438), ριζόω (ib. i. 438), ρίπτω (i. 437); a σ in ρέω (i. 439), ροψε-ω (i. 368),ρώομαι, ρωνννμι (i. 440). On the other hand, there is hardly one Greek root beginning with μ whose representatives in the other languages begin with a simple r too. Roots which do begin with r in these languages generally correspond to Greek words in which a vowel is prefixed to the ρ : ερενγω (cp. rue-tare Princ. i. 222), έρημος (Skt. ram i. 404), ερεύθω (i. 312), όμέγω (i. 226), ώρύω (rumor i. 444). Thus we see that the• doubling of the p is of great antiquity. Compared with it the appearance of λλ, ι>ρ, μ μ after the aug- ment is an isolated phenomenon, which is to be explained partly in the same way, as due to tbe fact that the root once began with two con- • sonants, e.g. in εννεαν Φ 11 (Princ. i. 396), partly as due to a mistaken imitation of such forms made in the conventional spirit of Epic language. On the precedent of ελλίσσετο, ελλισάμην, έλλιτάιευε, which are perhaps correctly formed (Hoffm. Qu. horn. i. 145), writers ventured upon ελληβε (Princ. ii. 145), εμμαθε (ib.* i. 387), while Apollonius Ehodius ii. 1032 was the first to allow himself the use of ελλιπε. Much that relates to this subject has been treated by me at greater length in Stud. iv. 479 ff. Lastly, the double consonant is to be explained in a few instances to be• due to an assimilation which has affected the consonant succeeding the initial, as in σαω (Princ. i. 465) and εΰΰεισεν (Stud. viii. 465, Princ. ii. 308 note). 2) Syllabic augment before a vowel. When we find a syllabic augment before a vowel, apparently, that is, taking the place of the temporal augment, we may* conclude that a con- sonant has fallen away, and that the consonant is one of those three spirants which Greek phonetic laws always condemned between vowels. Buttmann (i. 324) recognised this fact as far as the digamina was concerned, only he was obliged to leave a few cases doubtful, which we are now able to understand more clearly. The forms in question here can again be divided into two classes, according as the syllabic augment ch. in. THE SYLLABIC AUGMENT. Α Λ / γ 79- 4 is in each case present in its integrity or only felt in its results. To the forms with the syllabic augment intact we have a remarkable analogy in 118 Old-Persian, i.e. α-ϊ -sta-td (Spiegel, Altpers. 165), for * a-hi-sta-td and still older * a-si-sta-td, which would correspond to a Gk. * έ-ι-στα-το for the regular Ί-στα-το. We shall see below that where the initial was / < originally a vowel Old-Persian formed the augment in another way. ' ν a) Syllabic augment intact before a vowel. The following forms admit of a very simple explanation : — 1) k-ky -ην Ν 162 εν καυλω εάγη οολιχον }όρυ (Ρ 607), Aristoph. Yesp. 1428 και πως κατεαγη της κεφαλής μέγα σφόδρα', εαζε Η 270 ε'ίσω ff άσπίό' εαξε βαλών, Λ 175 της Γ εξ αυχεν εαξε λαβών etc. by the side of ήζε r 539 πάσι κατ αυχένας ηξε καϊ εκτανεν (Ψ 392). The /, 'which is con- firmed by the perf. taya (Aeol. ΐέαγε), is clearly established, especially by the Hesiodic καυάζαις (Opp. 666, 693). Princ. ii. 158, 188. 2) έ-ΰαΰ-0-ν, where the υ arose from f or a still earlier σ}, IS 340 επεί vv rot εύαΰεν ευνη, Ρ 647 επεί νΰ τοι εΰα^εν όντως, by the side of the perf. εάΰα (εάΰότα I 173). In the imperfect the form we should expect, ίάνίανε, occurs in Herod, ix. 5, 19, though we find ήνίανε A 24, 378, Σ 510 etc. (cp. επιηνΰανε'ν 16 and elsewhere in the Odyssey), and also εηνίανε Ω 25, γ 143 to be discussed below). From the same verbal stem comes εΦ-έ-ασ-θεν ' εγέλασαν, ΰιεχύθηόαν Hesych. i.e. εόήσθηααν. The uncompounded εασθεν is conjectured by Mor. Schmidt with Pearson in the gloss έαθεν εχώρησαν, which both scholars, with the alphabetical arrangement on their side, write εασθεν εχάρησαν. For other traces of the } see Princ. i. 282. 3) ε-άλ-η-ν Ν 408 τη υ~6 πας εάλη, Υ 16"8 $ονρ\ βάλη, εάλη ΰε γανών, 278 Αινε/ας <Γ εάλη, corresponding to which we get Σ 447 Τρώες επί πρΰμνησιν έείλεον and perfect forms like έελμένοι Μ 38. Clear proof of the / is given by the Doric form εγ-βηληθίωνπ.=εζειληβώσι on the Heraclean tables (Meister, Stud. iv. 404). 4) ε-άλ-ω»' first found in Attic writers, e.g. Aristoph. Yesp. 355 οτε Νάξος εαλω, while at χ 230 we have σ /j δ* ?/λω βουλή "Πριάμου πόλις, Herod, i. 78. The Lesbian ευάλωκα given in the Anecd. Oxon. iii. 237 points to a/ (Ahrens, Aeol. 37), and on this is founded the etymological combination given in Princ. ii. 170, which connects άλίσκομαι with the above-mentioned rt. ϊαλ, /ελ, which shows most clearly in άλυσις, chain, 119 the notion of shutting up or fettering. 5) έτάνασσε Alcaeus fr. 64 Be. 3 και πλείστοις εάνασσε λαοΓις, \vhere others less correctly write ίΐάνασσε, for a / in the middle of the word must have become υ in Lesbian. On the $ of the stem cf. Princ. ii. 182, Angermann, Stud. iii. 117. 4 6) έ'-ειξε Alcman fr. 31 Be. 3 τω $ε γυνα ταμία σφεάς εείζε χώρας. The / of είκω is established (Princ. i. 166). 7) ε-ειπον, εειπες, εειπε (απίειπΐ, με-εηπε, προσέειπε), very frequent in Homer, e.g. Ε 683, I 173. The f is clearly proved both by Aeolic forms and by the comparison of the related languages. Princ. ii. 57. 5 4 Έοφδάλη • e7rA77(n'a(rej/ Hesych. cp. βαρΰην ' rb βιάζ^σθαι yvvouxas Άμπραιαώται (cp. Mor. Schmidt s. v.) is too uncertain to be brought into the list. 5 "Etpyov and the cognate forms I omit, because the present iipyw (jcpyrf/Kcroi Ν 525) shows the initial e to be prothetic. 80 THE AUGMENT. ch. hi. 8) ε-έσσατο or ε-έσσατο only | 295 ες Αιβνην μ έπ\ νηος εεσσατο ποντο- πόροιο, where the scholia give έφέσσατο as the reading of Rhianus and εφείσατο as that of Zenodotus. The M.SS. according to La Roche seem all to point to the simple verb. As καθεϊσε often occurs in Homer, e.g. 22 204, εεσσατο need not surprise us, related to.it as εαζε to ήξε. The σ of the rt. εΰ is beyond a doubt Princ. i. 297. Op. Mayhoff de Rhiani Studiis Homericis (Dresden, 1870), p. 36 9) ε-έσσατο from the rt. Ϋες (εννυμι) Κ 23 αμψί 2* έπειτα Ιαφοινον εεσσατο Γέρμα λέοντος (=Κ 177), £ 529, άμφι ΰέ γλαΊναν εέσσατ άλε£ά- νεμον, by the side of άμφιέσασα, έπιέσασθαι, the latter found even in Xenophon, έπιειμένος, plupf. εεσΓο Μ 464. The ,f of the root is as certain as any. Princ. i. 470. 10) ε-είσαο from the rt. A3, where the } (Princ. i. 299) needs no confirmation : I 645 πάν,τα ri μοι κατά θυμόν εε'ισαο μυθησασθαι, έείσατο ε 398 ώς Ό?υσε 1 άσπαστον εείσατο γαία και ΰλη, while in tlcor the aug- ment is obliterated by contraction with the stem. 11) ε-ηκε from the rt. 1 originally it seems je,ja (Princ. i. 500), A 8 Τί'ς τ άρ σψωε θεών εριΰι ζυνέηκε μάχεσθαι, Α 48 μετά <Γ ιόν εηκε 1 Μ 221 άψάρ δ* άφέηκε. 120 12) ε-ονρησε. ένεούρησε is pronounced jbo be Attic in Cramer's Anecdota Oxon. i. 446, 17, and supported by a quotation from Eupolis Ajitolycus (Meineke, Comici, ii. p. 444) : apa σφόΖρ ένεούρησεν εξώλης γέρων ; Hence in Demosthenes 54 too, four good manuscripts have προσ- εοΰρουν. A corresponding formation is the perfect-form έι-ε-ουρηκότας Aristoph. Lys. 402. The etymological connexion with Skt. vdri water and the Lat. ur-tna is discussed Princ. i. 436 ; it is clear therefore that this verb too had once a / after the augment. Only we ought probably to follow Ebel here (Ztschr. iv. 166) in starting from ε-βορ-ησε, in which the fo as in ουρ-ανό-ς for the original var-ana-s (Princ. ii. 209), was transformed to ου. The ε held its place even after this transformation, in the same way as did the a in άϋτμην for cJW-p/r. 13) ε-ώθουν hymn, in Mercur. 305 χερσιν έώβει, ε-ωσα Π 410 καΰ ο' άρ επί στόμ εωσε, ι 181 καΐ Βυρέης άπέωσε (but άπώσε Ρ 649 and else- where), frequent in Attic : Aristoph. Pax 637 τηνΐε μεν ΰικροϊς έώθουν την βεον κεκράγμασιν, εοσα, εωσάμην, έώσβην with the corresponding perfect εωσμαι, later Ιωκα as well. The comparison of the Skt. vadh ferire (Princ. i. 323) establishes the digamma. Ebel sees here too only the effect of this spirant— vadh becoming ώθ. And we actually find in εννοσίγαιος and εινοσίφυλλος traces of an assimilated /, so that we are entitled to give ίοθ as the root, while we may at any rate attribute the lengthening of ο to ω to the operation of the gradually retiring spirant (Brugman, Stud. iv. 174). 14) έ-ωνούμην first to be found in Attic : Eupolis Maricas fr. 15 (Meineke, ii. p. 505) Κ(>ούων γε μην αυτας εωιουμην εγω. The perfect εωιημαι belongs to the same period. In the stem ύνο the length of the ω seems to have nothing to do with the disappearance of the digamma, for corresponding to it we find the Skt. vasnd and the Lit. veno also with a long vowel. We must look for the source of this lengthening to the lost 8, and so regard Ρωνο as the stem (cp. ωμο-ς — όμσο-ς), Princ. i. 400. 15) In ιμίίης X 280 εκ Δίος ηείίης τον εμον μόρον, ι 206 ούΰέ τις αυτόν | ήείΐη Ιμώων we have a second instance of what is apparently η for ch. πι. THE SYLLABIC AUGMENT. 81 ε before a vowel. There can be no doubt here as to the rt. hh We should therefore expect Ηείΰης, έείοης. As in τοκήος the disappearing .F 121 has produced a prolongation of the preceding vowel. The similar ήϊκται ήϊκτο will be dealt with when we come to the perfect. The following forms stand by themselves : — 16) ij-ttpe belonging to f φω Κ 499 συν V ηειρεν ίμάσι καϊ έζι)λαυιεν ομίλου. The perfect εερμένος σ 296 and εερτο' εκυημνα (Hesych.) prove that a consonant has been lost. But the etymology of the word tells us of the loss of two consonants (Princ. i. 441). We may take the root to be σβφ, which in Latin became ser (sero), and has preserved its sibilant in σειρά, while after a vowel most probably σ first assimilated itself to the /?, and then the fuller sound of the f effected the lengthening of ε to ?/. 17) εώρων, not found till the Attic period (Herod, ώρων), by the side of εώρακα (εόρακα), εώραμαι. 18) εωΐ'οχόει Δ 3 νέκταρ εωιοχόει, ν 255 καΧόΐς εν καιέοισιν έωνο-χόει Ιέ Μελανθείις, but the form without an ε occurs also, and it was written by Aristarchus with ο (oh οχόει A 598), by others ωνοχόει (La Roche, Homer. Textkritik, p. 324). 19) άνέωγον Π 221 \η\οϋ ο' άπο πωμ άνέωγεν, Ω 228 ή και φωριαμών επιθήματα /τάλ' ανέωγεν, άιέωζε κ 389 θύρας & άνέωζε συψειοϋ (Herod. ΰνοίζί) to be compared with Ω 455 τρεΊς δ' ά> αοίγεσκον μεγάλην κληΐΰιι θυράων, with the Attic peif. άνε'ωγε, άνέωκται, ανεωχθην, but the word is a trisyllable at IS 168 n)v $ ov θεός άλλος ίνώγεν (like ωιοχο'ει), more correctly άνοιγε v. The peculiarity of the last three verbs is that after the syllabic augment there seems to be the temporal as well. The digamma is again clearly the cause of the lengthening. When it fell out the following vowel was prolonged, as in βασιλέως for βασιλεΡ-ος. The nature of these vowels has been discussed by Ebel in the essay often referred to above. Whether the spirant had from the first the power of lengthening the preceding or the following vowel indifferently is a difficult question, to which Ebel's essay does not seem to me to have given a satisfactory answer. I am not sure that, as τοκηος, πόληος, νηός, ηος are of earlier occurrence than τοκίως, πόλεως, νεώς, εως, it was not the general rule that the backward influence of the spirant was the first step, while a somewhat later period reversed the relation of the long and short vowel. The Attic εώρταζον can hardly be explained in any other way than by sup- 122 posing it to come from ήόρταζον (cp. Buttmann, i. 326) ; έώλπειν {φ 96) and εωκειν (3 174) owe their ω, as is shown by εοικα, to the augment, that is, ηο became εω, while the reverse of this process in the Homeric εήνίανε which the metre proves to have four syllybles at Ώ 25 and γ 143, and which we must assume to have come by metathesis of quantity from ήάνΖανε (cp. ήεφε), is not so easy to understand. In any case the form is peculiar, for the analogy of τοκεα would lead us to expect ά and not η in the second syllable. May not the η owe its existence to a mistaken doctrine of πλίονασμός 1 The form εώργει, t, 289 τρωκτηι ος h) πολλά κάκ ανθρώπους εώργει, is very singular indeed. I. Bekker here reads εΐώργει (cp. ειώθει), while La Boche in the face of the M.SS., reads άνθρώποισιν εώργει. The length of the first syllable might be explained to be the result of *ε-Ρί-Ηργει *έε-όργει, but there would then be no reason to be seen why the ο which is short in εοργα should be long here, and we should have to write εΐόργει. Brugman G 82 THE AUGMENT. CH. III. (Stud. iv. 167) defends the reading εωργει by supposing that from the primary *ε-ΐε-) : όργει there came first, by the influence of the disappearing ϊ (cp. ϊ}-εΊύΎ\), *ή-ε-Ρόργει, then *ή-όργει. The latter form he conjectures to be the true reading at ξ 289, ή-όργει having become ε-ωργει by metathesis of quantity. In the case of forms found in Homer it is well to remember that in the old writing there was no difference between ο and ω, οι and ω, and that in consequence the authority for the latter is always extremely small. In any case it is worth notice that the New-Ionic dialect has a decided dislike to these forms with an ω. Whatever may be the case as to the successive stages of these • pparently anomalous phenomena, of this we may be sure, that the augment points in all instances to the loss of a spirant. 20) Another form of a peculiar nature is the Homeric εάφθη, the origin of which is a much debated question. 6 It occurs but twice : Ν 543 123 and S 419 €κΚίνθη δ' ir4pa>o~e <άρη, eVi δ' άσπ\ς έάφ&η καϊ κόρυς ' άμφ\ be οι θάνατος χύτο θνμοραϊστης, χ(ΐρ6ς δ' €κβα\βν εγχος, eV αντω δ' άο~π\ς ίάφθη κα\ κόρνς • άμφϊ δε οι βράχς τςνχεα ποικίλα χάλκω. Aristarchus wrote εάφθη with the spiritus asper, and interpreted it, as Herodian tells us in a note on the former passage, by επηκολοΰθησεν, and consequently derived the form from επεσθαι. Herodian (ed. Lentz, ii. 309) followed Tyrannio in regarding απτεσθαι as the etymon, and inter- preting it οιονεί σννηφθη. Both views have found supporters among modern scholars. Buttmann, in the Lexilogus, ii. 138, took the side of Aristarchus ; Spitzner, in the 24th excursus on the Iliad (vol. i. sect. iv. p. xvii), followed the other two grammarians j and since Buttmann in his Ausf. Gram. ii. 117 forsook the side he first embraced, the latter view has become the prevailing one. Buttmann was quite alive, however, to the difficulties in the way of the derivation from αντω. First and foremost among these he placed the irregularity of the augment as compared with ήπτετο, ίφήπται. The rt. of άπτω, άφ, is probably to be connected with the Latin ap-iscor, ap-tu-s, ap-ex. If we except ααπτος, to which we shall return, it shows no trace of a con- sonantal initial. Spitzner, it is true, is able to set this objection aside at p. xxiv in a characteristically superficial way ; simply saying that the littera incertissima, as he calls the /, never means much any way. As to the sense of the passage, Buttmann was undoubtedly right in the Lexilogus in translating it * and the shield fell on to him,' which exactly suits the context. Heyne's interpretation, adopted by Spitzner, 'and the shield stuck to him,' is quite wrong. In the first place, the aorist επ'-Ιάφθη can never mean the same thing as the pluperfect εφηπτο. Since it denotes the commencement of an action, Μ — εάφθη, if it did belong to έφάπτεσθαι, could only mean ' fastened itself.' This is how Tyrannio and Herodian take the word : οιονεί σννηφβη αντω. They appear to have taken this σννηφΒη in the sense * bound itself, united itself to,' not a very apt meaning, it seems to me. What connexion was established • I have treated this form at greater length in the Commcntalio ae Forma Homerica ίάφθη that is prefaced to the list of the doctors created in the philoso- phical faculty of Leipzig in 1869-70. ch. in. THE SYLLABIC AUGMENT. 83 between the head and the shield that fell on it? Then the whole idea, especially in the first passage, ' he bent his head, to which the shield and helmet stuck fast/ does not seem natural, and this is why Buttmann, on returning in his Ansf. Gram, to άπτομαι, preferred to see 124 in this verb the meaning ' vnfiicta est,' * struck him/ a meaning, however, which cannot be extracted from it. The discussion then stands, I take it, as follows : in favour of Aristarchus's derivation from έπομαι there is first the context, secondly the augment, which can be explained, though not by f, still, as Buttmann saw, by the σ which is to be seen in ε-σπε-το, σπέσβαι, as in the Lat. sequi, while all that makes for the derivation from άπτομαι is perhaps the α and the common aorist form ήφβη. Both points certainly give rise to some difficulty. Still for the α in the face of an ε everywhere else we can adduce the Homeric έτάρφβην (I. Bekker on ε 74) by the side of τέρπομαι, and τραφθηναι (ο 80) by the side of τρέπω, while Homer has στρεψβέιτε (Ε 575) where Herodotus has an α (κατεστράφθησαν I 130). And from επω Herodotus has, though in a passive sense, περιέψθην v. 81. And might not άαπτος itself, for which some wonderful derivations have been devised, possibly mean, not ' not to be touched/ but ' unsociable, un- approachable, intractabilis ' 1 For έπειν means tractare, to deal with a thing, μεβέπειν (ο 175) ' versari, be present' (Princ. ii. 58). We should then have a parallel for the a in εόφθη, and a reason, for the absence of the ν in άαπτος. If our conclusions are correct, the apparent irregularity of the augment too in the case of εάφβη is satisfactorily explained. 21) εείσατο belonging to ειμί, Ο 415 άντ Αΐαντος εείσατο, 544 τω ρέν έεισάσθην cp. χ 89, we must, it is true, admit to be an anomaly, as there is no hint of a consonantal initial here. But then it stands alone as such. At Princ. ii. 207 this form is discussed, and attention is called to similar mistaken formations in Homer. b) Syllabic augment discernible in a contraction. The old grammarians seem to have regarded the ει, which a number of verbs show in the stem-syllable instead of the η which was lo be expected, as hardly an anomaly, but only as a not very unnatural variation. In the scholion above referred to on Ν 543 Tyrannio says αδιάφορων τά από του i αρχόμενα ρήματα ε'ίωθε κΧίνεσθαι και δια της ει διφθόγγου κατ 125 αρχήν καϊ δια τοΰ ή. Buttmann (ii. 323), who derives the temporal augment from the contraction of the ε with the initial vowel of the stem, was consistent in regarding ει as the earliest form of the augment, but he does not tell us why in other cases η took its place. No explana- tion whatever was given of the fact that the Greeks said έιχΊ>ν. but ϊ'ϊβελον. It was from Comparative Grammar that the first help came here. Pott. Et. F. ii. 1 71 gave a few suggestions. The first, as far as I know, to give clear expression to the correct principle, was Savelsberg in his doctoral dissertation ' Quaestiones lexicales de radicibus Graecis ' Berol. 1841, where, at p. 7, after an examination of the several forms, he puts it as follows : ' ε augmentum in istis exemplis omnibus, ubi cum ε prima radicis vocali in ει contrahitur, vere est syllabicum, cum ejus- modi contractioni eae tantum radices sint obnoxiae, quae aut σ literam aut digamma in initio amiserunt.' Savelsberg only omitted the third spirant j. This was the very view which I afterwards established in my Tempora g 2 84 THE AUGMENT. ch. hi. unci Modi, p. 136 ff., as did Ahrens Formenlehre, § 83, Ebel Ztschr. iv. 167 ff. Now, I suppose, no one doubts ii;. Kiihner at p. 498 of the new edition of his Ausf. Gr. mentions it as self-evident. It can, in fact, be hardly accidental, that of the fifteen verbs with εί in the augment- syllable twelve show unmistakable traces of having originally had a consonant at the beginning of their stem. In strict Doiic the difference between the augment in ft and that in η was unknown, because here εε regularly contracted to »?, and they said ήχον, ήλκον as well as ήσθων, i'ldeXov (Ahrens, Dor. 202). .The several forms are asjfollow : 1) είυσα, είων. Both forms are Homeric (Ω 684, Σ 448), and are joined later by ειάβην, the perfect-forms ε'ίακα, είαμαι also show the diphthong in the reduplication-syllable. The Homeric present-forms ε?ώ Δ 55 by the side οϊΐώ.θ 42$, είωσ' Β 132 by the side of εώσι δ 805 (ονΰε εώσι), 7 the conj. είωμεν ψ 260, the form εβασον (=εασοΐ') said to be Syracusan an# Laconic, for which Gregorius Corinth. 354 also 126 writes εΰιισον, as also ευα = έ'α (Ahrens, Dor. 49), point to the loss of one or more, spirants after the ε, so that the diphthong would seem to be the .result of a compensatory lengthening. This consideration has, it is true, not led as yet to a certain etymology. Ebel's .-(Ztschr. iv. 169) derivation of εάω from ε'ϋς• seems to me improbable as far as meaning goes, Kraushaar's attempt (Studien, ii. 430 ff.) to connect it with thert. as throw, from which come s-ino and — as is pointed out by Bugge, Fleckeisen's Jahrb. 1872, p. 95— the Old-Latin de-sivare (desinere Paul. Ep. 72), must be admitted to be acutely reasoned out, but his proof assumes too many unauthenticated steps to be convincing. If, as we must suppose, the fi is the result of a compensatory lengthening, it is accidental that Attic Greek kept the diphthong only in the augmented forms, and there was formed, at a time when, as in HcJmer, ίίάω and εάω existed side by side, the somewhat arbitrary rule of saying είων but εάω, which appeared to bear to each other the same relation as Λχον and έ'χω. Strictly speaking, therefore, We have here to deal with no augment- syllable at all (cp. Kiihner, Ausf. Gr. i. 2 499). 2) εϊΐον, ordinary Greek along with 'ίΰω, 'ίΰοιμι, iZetv »etc, is one of the clearest cases, for no one will doubt in the face; of the proofs of a f in this root that it stands for ε : Ρώο-ν (Princ. i. 299). The form with a vocalised f (cp. εΰαδοι ) ει /idop occurs in the poem of Balbilla, C. I. Gr. 4725, 1. 10 (Ahrens, Dor. 578)— Βαλ/3/λλα δη κάμςν οισι πόνυις γρό'ππατα .σαμαίνοντά τ οσ €Vibe κωσσ (ςάκσυσ*. These verses were written a.d. T50. But Bergk (Lyr* p. 879) is in all probability right in reading also in fr. 2, 7 of Sappho — a>s yap (ΰώον βροχίως σ€, and Nauck is perhaps right in conjecturing (Melanges Groco-Romains, Bulletin de l'ac. de St. Petersb. 1863, p. 409) that in several instances where we now read ύςιΐι in Homer the original reading of the text was ευίί (e.g. S 13, Σ 235). The expulsion of the/ and the contraction of fieov in the Homeric poems is, however, in some places established by the metre : λ 162 ovh' είδες ενι μεγάρυισι -γυναίκα, Λ 112 (Jfo, ΰτ εΙ"\ΐης, while ticot> or \' hlov is often possible. Bekker's Ρείΐον is, as Nauck saw, indefensible. 7 The awkward hiatus is here and elsewhere got rid of if we read βΐάκπ and the like. ch. in. THE SYLLABIC AUGMENT. 85 3) ε'ίθιζον, ε'ίθίσα, ειθίσθην first found in the Attic period — though 127 there is a certain variation between ει and η in the instances preserved — to which may be added ε'ίθικα and ε'ίθισμαι, once had a J? in its stem, as is shown most conclusively by the Aeolic perfect ευεθωκα • έίωθα (Hesych.). This form points to ΐεθόω, a by-form of £ εθίζω, which without its f was known also to the Dorians (εβωκα, ήβωκα, Ahr. 340): In very early times there was a σ before the f, so that we get a root σ£εβ whose initial double consonant moreover gives the best explanation of είωβα (Princ. i. 311). 4) εΐλισσον. The only testimony of Homer to this is doubtful, as the reading at Μ 49 varied even in antiquity (cp. Schol. A.) between — ως"Εκτωρ αν* ομιΧον Ιων «•ίλισσ6#' εταίρου* τάψρον Ιτζοτρννων hiafiaivkpev and ελλι'σσεθ', which is now the general reading. In the Attic period are found ε'ίλιξα, ειλίχθην, and είΚιγ μένος is found as early as Hesiod, Th. 791. The diphthong occurs also, it is true, in the unaugmented forms in Herodotus (ii. 38), in the tragedians, in Plato and elsewhere, and also in the undoubtedly related εϊλλω or ε'ίλλω. We must here probably regard a prothetic ε as the source of the diphthong. No one can doubt the connexion of these forms with the Lat. volv-o, by which the digamma is established (Princ. i. 447). 5) εϊλκον, unknown in the Iliad and Odyssey, where the only form is ελκον, first occurs in the hymn, in Cerer. 308, then in Herod, (i. 31 and elsewhere), and is common in Attic writers, from Sophocles- (O. C. 927) onward, as too the aorists εϊλκυσα, εϊλκύσθην, ειλκνσάμην, to which are to be added the perfects ε'ίλκυκα, ε'ίλκνσμαι. With these forms ηλκησε, which is given by good M.SS. at λ 580, is in strange contradiction. But La Poche is no doubt right in reading ελκησε, which he conjectures to have been the reading of Aristarchus. Not much weight, ft is true, attaches to the lengthening of the previous yap : Λητώ γαρ έλκη σε, as a trace of the P, as eXxetv shows no similar traces anywhere else. But the witness of the related languages to this initial — in Lithuanian (yelku I draw) and Slavonic — is distinct, and αυλαζ=ά-Ρλαζ confirms it (Princ. i. 167). 6) είλον, εΐλόμην, common in Homer, e.g. Γ 35 ωχρός τέ μιν είλε 128 παρειάς, Δ 406, ημείς και θηβης εοος είλομερ επταπνλοιο, Λ 697 ε'ίλετο, and in ordinary Greek from that time onwards. The traces of a $ in this stem are not very numerous. See Princ. ii. 180. 7) ει-μεν, ει-τε, ει-σαν, εϊ-μην, εϊ-θη-ν from the rt. ε (Ί'ημι). In Homer these forms have no augment — Ι μεν, εσαν &c. — except at Ω 720 τταρα F εισαν άοώούς, and at Ψ 868 παρείβη. From Herodotus onward (vii. 122 6 στρατός άπείθη υπό Βερζεω) the augmented forms are in common use : at -ε'ι-μεν Aristoph. Yesp. 574, άψεΊσαν Thuc. vii. 53. εφ-ε~ι-το Soph. Phil. 619. We have already seen a trace of an initial consonant here in the form ε-η-κε discussed on p. 80. We have the same thing in the perf. εΐ-να, εΐ-μαι. The difference between ηκα and ε\-μεν is easily ex- plained by supposing that the former was contracted from «fica as ήΧων from εά\ων, ?]ζε from έ'αξε. The η is not due to the augment, but, like that in ε-θη-κα, is part of the formation of this anomalous aorist. There is nothing, therefore, to contradict the derivation of the verb given above from the rt. ja. 8) ειποι-, ειπόμη)• Λ 706 ημεΊς μεν τα έκαστα ειείπομεν, Γ 447 αμα δ' 86 THE AU.GMENT. ch. hi. εΐπετ ακοιτις, Ε 591 Τρωών εϊποντο φάλαγγες. The middle occurs in ordinary Greek ; and the origin of the ει is made as clear as can be by ε-στό-μην, i.e. σε-σεπ-ό-μην, which gives us a rt. σί7τ=1^ΐ. sequ in sequ-or. It is hardly necessary to refer to Princ. ii. 57. 9) εφγαζόμην, ειργασάμην, ειργάσβην. The Homeric poems contain no certain instance of the £/, for though at γ 435 the M.SS. have οίσίν re χρνσον είργάζιτο, the extraordinary lengthening of the -ov points, not to Bekker's ϊίψγάζετο, which is impossible, but to ϊεργάζετο, and this is borne out by the reading adopted at ω 210 : τοί οι φίλα εργάζοντο. But Hesiod Ορρ. 151 has είργάζοντυ. In Herodotus's dialect είργάζετο and the like (Bredow, 301) are unknown, but the unaugmented form is extremely frequent. Among the Attic writers again the ει is very common, though in later times η sometimes takes its place. (Hager de G-raecitate Hyperidea, Stud. iii. 105, Wecklein Curae epigraphicae, 36.) The ει is clearly due to the I of the rt. }εργ (Princ. i. 221). 10) εϊρττον μ 395 as a 3rd pi., with this exception not earlier than 129 the tragedians; the comic poets also use the aor. είρπυσα. The ει is due to the σ, with which the word originally began, and which the Lat. serpo shows intact (Princ. i. 329). 11) εϊρνσα. As we shall see later, the stems ^ερν(ί) draw, and ίερυ guard, are to be carefully distinguished from each other. To Ρερν{ς) draw belong είρϋσαν θ 226, είρϋσε β 389, είρυσάμην κ 165. The traces of a / are pointed out by Hoffmann, Quaest. homer, ii 49 ; probably the Lat. verro (for vers-o) is related. [But cp. Corssen Beitr. p. 403.] The appearance of an ει in unaugmented forms, e.g. in εφΰμεναι Hes. Opp. 818, είρυηον Soph. Trach. 1034, is to be explained in precisely the same way as in the case of ελίσσω (No. 4). In Attic prose ερυω is unknown. 12) Λσα, Δ 392 πνκινον λόχον flaav άγοντες, β 472 ε'ισε <)' ίφ' nvroy μέσσω ΰαιτνμόιων, cp. Hesiod. Theog. 174, and then in Herodotus and the tragedians, who also recognise the middle ε'ίσατο (εγκαβείσατο, Eurip. Hippol. 31). The diphthong is to be explained as due to the original σ of the root σε£, Μ, from which too came the form είσσατο, discussed on p. 80. The strange thing is that it appears in forms that have not the force of a past tense, in Homer only at η 163 είσον ίιναστησας (by the side of ε σα c, εσσαι), then in Herodotus (iii. 126 νπείσας, i. 66 ε'ισάμειοί). At Thuc. iii. 58 Bekker and Classen follow good M.SS. in reading εσσαμί- νων. The ει must have been due here to a confusion, aided apparently by the related ϊζω, Ισα, κύβισα, καθισάμενος (Cobet, Variae lect. p. 88). 13) Λ είστήκειν, the Attic form of the pluperfect as contrasted with εστηκειν, which is the only form in Homer and Herodotus, is found first at Hes. Scut. 269 ειστηκει, at Eurip. Here. fur. 925, and constantly in prose. Its origin from ε-σε-στηκ-εί-ν explains the diphthong (Wecklein, Curae epigraphicae, 36). 14) ειστ'ίων, ειστίασα, ειστιάβην by the side of the perf. είστίαμαι not found before the time of Attic prose, but occurring there constantly (Lys. 19, 27, Xen. Cyr. i. 3, 10), is explained by the fact that εστία belongs to the same root as the Lat. Vesta (Princ. i. 496). Traces of the ϊ are to be seen in Doric, but not in Homer {Ahr. 55). 15) «Ιχο»', ιϊχόμην need no reference to special passages, as they are universal from Homer (Γ 123 etc.) onwards. The forms ε-σχο-ν, σ\η~σω, ϊΐω etc. show clearly that the root is σεχ, and consequently that tl\oy stands for ϊ-σεχ-ο-ν (Princ. i. 237). ch. πι. THE TEMPOEAL AUGMENT. 87 A review of all these forms shows us this result. In seven verbs the 130 . loss of a } is demonstrable, in five that of σ, in one {άμεν etc.) the evident loss of a spirant, probably of a j. In είλον, as we saw, the loss of a spirant is not clearly established, in ε ίων the ει is not really due to the augment, and in no single case can it be established that the root originally began with a vowel. If we reckon these fourteen cases (in- cluding (!\<>y, along with the twenty-one cases treated under a), in which the syllabic augment remains intact, we obtain a total of thirty -five cases, in which the augment has something to tell us about an original con- sonantal initial, and this .is a circumstance not to be overlooked in con- sidering, as we shall have presently to do, the question of the persistency of this element of inflexion. B) The Temporal Augment. Buttmann was able to see that the temporal augment was originally identical with the syllabic, but he did not state the fact correctly when he said, at p. 323 of vol. i. of his Ausf. Gr. : ' From all that has gone before it is clear that the Augmentum temporale is nothing else than the Augmentum syllabicum έ that has been contracted with the initial vowel of the verb, e.g. άγω ί-αγυν ήγον ', and it thus appears that the augment of the verbs in Text 3 is the original form : εχω, ε-εχον, ειχον, while in the rest of the cases of amplification contained in this paragraph this original contraction has been replaced by a simple lengthening of the main vowel.' This change of procedure would be inconceivable in the case of the large number of verbs beginning with α in Doric, and of those beginning with ε and ο in Ionic. For ε-αγον, for instance, the only possible contraction in Doric would be ήγον, like κρης for κρέας, while what we actually find is άγον, άρχόμαν, ανάγγειλαν, άϊ,ίονν (cf. the Lesbian σννάγαγε, the Arcad. νπάρχε, the Cypr. αιωγοι•, Ahrens, 129) ; and in Ionic ε-εσαν could only produce εισαν, ε-ορ-το ονρτο. The temporal augment therefore points undeniably by its form to an older linguistic period in which the augment had not yet turned to e but was still a. As to the period at which the rules which hold for Greek were settled there are two possibilities; either this happened on 131 Greek soil at a time when the augment was still a, while the stem- syllable had already got the vowel which was the prevailing one at a later time. On this hypothesis the augment in the case of a verb be- ginning with α would be explained by the following steps : α-αγον άγον Ion. ηγον but not in that of verbs in ε and ο j for though in Ionic ά-ορ-το might give ώρτο, in strict Doric it would give άρτο, as βοάοντι gives the Dor. βυάίτι (Ahrens, 197), while *ά-εσαν would giye*apro. The η shown by Ionic in the place of α is evidently of late origin, and reminds us of the way in which in the nouns the uniformity of the a- declension is marred by the way in which the Ionians sometimes put an η into the place of an α : δίκη ΰίκης /iu or ή ω ν as 1st sing, as the exact coun- terpart to the Skt. ajam, the imperf. of the rt. i, and ήιον as 3rd pi. as that of the Skt. djan (for *ajant). To ήσαν there is moreover an exact 133 parallel in the Old-Persian atiy-disa they overstepped, patiy-aisa=iroTi- ήα αν (Spiegel, D. Altpers. Keilschriften, p. 188, cp. 168). But how is the long vowel to be accounted for in these forms ? It has been thought that it might be taken as a proof that the augment originally consisted in a long a. It would be strange, if this were so, that this should be almost the only instance of a. Schleicher (Comp. 3 738) is of opinion that there was in the Indian forms no contraction of a + i, a-\-u, which must have -given e, δ, but only an approximation of the two letters, the result of which would have been cd and du. This explanation would not suit the Greek forms anyhow, for in Greek the approximation of t + ι very often leads to ει, as it did in the above- mentioned eldov. I should be more willing to believe that the vowels i and u produced a spirant before them, which made itself felt later in the length of the α. ήια would thus stand on the same footing as χπυσίμος, άιβρωπηιος, and other forms of the kind which I have discussed at Stud. ii. 187. However this may be, the agreement between ήια and the Sanskrit forms in the matter of augmentation may be set down as established. In all other cases the rule is that the initial vowel is simply lengthened. In inquiring into the origin of this apparently remarkable rule we must notice, to begin with, what it is easy to overlook, that the whole amount of cases affected by the rule is by no means a large one. In Homer there are only four or five instances of an ι made long by the augment : %ιίνετο ΰ' νΰωρ κ 359 (by the side of 'ϊαίνων and the like), Ηινχον (μέγα 'ίαχοτ, επίαχε Σ 29), έπίηλεν χ 49 (by the side of έπϊάλλων), 'ικανέ (ες Χρύσην "ικανέ Α 431, καρτταλίμωι, c)' 'ίκανε βοάς cwt νήας 'Αχαιών Β 17, by the side of 'ϊκάί'ω, 'ϊκάνει etc.), 'ίκετο (Ίκετο πένθος Α 362, ίκετο Ιώμα Φ 44 by the side of 'ϊκέσθαι etc.). The stems Ιάχ and Ίκ once had a F, so that the augmented forms cannot be of a very ancient date. • The later periods will not add very much to this list, as the number of verbs be- ginning with ι is small. We get e.g. Ίκέτευσα, Eurip. Med. 338. Several of this small class of verbs, e.g. the derivatives of ϊΰιος Ihodv, ΙΙιάζειν 134 etc. hardly occur in poetry. Some derivatives of ιερός, like ιερεΰω and again Ιθαίνω, ιμάσσω (in Homer' there is only Ίμασεν), might possibly furnish instances, but I have not been able to find any. Of verbs in ν there is not a single instance of an augmented form in Homer, and even in later Greek they need a great deal of looking for, as the number of such verbs is small. The following are certain : Aesch. Prom. 558 και λέχος συν 'νμεναίονν, Anthol. vi. 265 Νοσσίϊος 'νψανεν θενφιλ\ς ά Κλεο'χος, to which Nauck (Melanges Greco-Bomains, iv. 5) adds three more from υφαίνω, one from υγιαίνω (ουχ υγίαινε Com. anon., Meineke, iv. 182), and 'ϋλάκτει κΰων (Aristoph. Yesp. 1402). 'ύτφ•ζες, Eur. Tro. 1020 and the like do not count, as here the υ may be long in the tragedians in unaugmented forms, so that 'ύβριζον might stand 90 THE AUGMENT. ch. hi. on the same footing as ιισσωμην. On the whole I do not think it is going too far to say that the rule which all grammars give is established by barely a dozen verbs altogether. This fact puts the difference between Greek and Sanskrit in a completely different light. The Greek usage is evidently the result of a comparatively late development, due to the analogy of the verbs "beginning with a hard vowel. Owing to the lack of primitive stems beginning with t and ν the old tradition was apparently quite interrupted, and i\ia, which took an anomalous position instead of serving as a pattern for the rest, stands alone as witness to the old rule. Another anomaly of the augment which has not had much notice bestowed on it is the change of the position of the aspirate. Inasmuch as the temporal augment originated in the syllabic, in the contraction, that is, of the a with the initial vowel, we should expect to find this form of the augment always with the spiritus lenis. The asper shows that the linguistic sense had no very lively recollection of this contrac- tion. Hence even Homer has ήρει (Ρ 463), ήπτετο (Υ 468), ήρμοσεν (Ρ 210), ώρμαινε, ώρμησε, though in verbs in which the spiritus asper had arisen from σ, F , or j, the contraction could not have been of very long 1 35 standing. The sense of the connexion between the preterite and the other verbal forms was probably too strong to allow of such a difference as we can imagine might have existed between άπτεται and *ήπτετο. Still more surprising than the aspirated temporal augment is the aspirated syllabic in forms like εάλωτ, εέσσατο, εώρων (cf. above pp. 79 and 81). The grammarians defend this strange usage by the peculiar supposition that the ε is not εκ κλίσεως, not inflexional, that is, or, in other words, not an augment, but εκ πλεονασμοί) (Herodian, i. p. 542). They had cases like εεϊνα in their minds. We may learn two things from these forms : first, the fact that the spiritus asper was of a movable nature, and had no prominent position as a letter even in early times ; and secondly, the power of analogy which was here the means of ob- scuring so ordinary a linguistic instrument as the augment. There is moreover a noteworthy exception to this surprising rule in the case of a temporal augment, i.e. the Homeric άλσο, αλτο, with its spiritus lenis. The forms are attested beyond any doubt at Π 754, A 532, Γ 29, Δ 419 etc. Herodian expressly prescribes the lenis at A 532, and gives some marvellous explanations of it. Other witnesses to the fact are collected by La Roche, Homer. Textkritik, 185. Since, as Buttmann saw (ii. 109), the circumflex points to a contraction, it is best to set down the lenis as a relic of the old pronunciation, and take αλτο to have come from ε-άλ -ro, or, more properly speaking, *ά-άλ- to. If this is the right conclusion we have here a completely isolated instance, which can only have arisen at a time when the α still remained intact, though the original σ of the rt. άλ (Princ. ii. 167) had already passed into the spiritus asper, the order of the changes being different to that in the cases of the above-mentioned εέσσατο, εάφθη, ε\ρπον, whose predecessors, *ε-σέσσατο, *ε-σάφβη, *'έ-σερπον, seem to have known a time when the a of the augment had been weakened to f, but at which the σ, which was afterwards volatilised, still remained intact. Some doubts might certainly arise about αλτο from the fact that &\μ ενός often appears with the lenis, which does not admit of the same explanation. But might this not have arisen from a mistaken imitation of αλτο ? Even μετάλ- ch. πι. ABSENCE OF. THE AUGMENT. 91 μένος Ε 336, εττάλμειος Η 260, and elsewhere, might be accounted for by the instances which, though .not plentiful, do occur, of an Ionic preference 1 36 for tenuis instead of asper, such as ενίστων ζ 265, αντόΐων θ 449. It is conceivable therefore that there once was a άλμενος corresponding to άλτο. The attempt to explain the lenis in ήμβυοτον in the same way would find an obstacle in ημάρτανον Ω 68. The etymology, and the re- lated άβοοτάζειν, rather point here to the lenis as the original initial (Princ. ii. 350). C) Absence of the Augment. At the very beginning of our investigation of the augment we en- countered the question whether and how far it is an integral and original part of the preterite. 9 The fact that the augment is very often left out in the Vedas (Benfey, Vollst. Gr. p. 362 j Kurzc Gr. p. 85), that its omission is not unknown even in epic Sanskrit, and is the rule for all periods of the language after the particles ma (μη) and sma, taken in conjunction with the other fact that the augment is unstable in the Homeric poems, has led to the precipitate conclusion that the old original Indo-Germanic language stood on the same footing in this respect as the two oldest texts which have come down to us from India and Greece. The course of the development of our science teaches us caution, I think, most emphatically. How many centuries do we suppose passed from the time when the Indo-Germans, as we imagine, lived as one nation in the table-lands of Asia, to that when the Indians composed the oldest of their hymns which we possess, or to that still later period of the Homeric' poems 1 We do not rush in other instances to the conclusion that because two languages agree in the absence of some element, it must therefore have been absent in the primitive language. The Indians, even the oldest of them, said s-mas for as-mas, and the Romans said su-mus ; but a glance at ίσ-μεν is enough to show us that the loss befel these two languages independently, as on the other hand erant teaches us that the loss of the t in the Skt. dsan and the Gk. ήσαν is of no very ancient 137 date. Who could deny that the language of the Yedas has itself been subject to most material alterations'? Aphaeresis is of pretty frequent occurrence in prepositions in Sanskrit — api, for example — and in aphae- resis we may find a satisfactory explanation Of the loss of the augment. And the Homeric language too has its own special weaknesses which are sometimes corrected by the language of a later time, and the Dorians and Aeolians especially preserve many older forms than Homer. We may perhaps even venture to maintain that it is the peculiarity of old periods of language that in spite of all the treasures they preserve for us from a preceding period, they always show certain signs of degeneration which disappear again as the consciousness of the rule grows more defined. For instance, it is only in Homer that we find pa and ap by the side of άρα, and in Homer, though, in this case, not in Homer only, we find νέρθεν for ενερθεν, though it is unmistakable that it came from εν. Again, without accepting the superficial doctrine of 9 I am pleased to find that Delbriick, Altind. Yerhum, p. 80, agrees un- reservedly with my view, that the augment was from the first a necessary part of the preterite. 92 THE AUGMENT. CH. III. earlier times, which made the metre responsible off-hand for all possible kinds of license, we must admit that where pairs of forms existed in the spoken language, the poets eagerly availed themselves of the fact. Every additional mode of expression gives additional facility in the fabrication of the verse. If then, as we assume, at the time of the formation of Homeric language, or perhaps of its predecessor, the lan- guage of those stiffer Epic songs which must have preceded Homer, βή was said now and then as well as εβη, βάλλε, as well as εβαλλε, how convenient this must have been found by the not over deft versifier of those early times ! And how could even the more elaborate Epic of a later time afford to abandon so productive a source of useful alternatives in the arrangement of the words ! εβαλλε could not begin a line, how con- venient to have /3όλλε at command as well, and the same in other like cases ! Against the assumption that the augment existed from primi- tive times till a little before the period of the Homeric poems, as a kind of movable prefixed particle, sometimes present and sometimes absent, decided objections may be found in the history above given of the tem- 1 38 poral augment. We saw the temporal augment to be a syllabic aug- ment which in an earlier period, before Greek was a separate language, had lost its original shape. But if it lost its original shape it cannot have had a separate existence of its own : the two things are incon- sistent. Moreover the preservation, in spite of phonetic difficulties, of the ε after the disappearance of initial spirants, whether it remained unchanged or was contracted, goes to show that the spoken language was by no means in the habit of dismissing the augment off-hand. It seems to me best on all grounds to suppose that shortly before the rise of the Greek Epic the augment became occasionally exposed to the same tendency towards wearing away ( Ver witter ung), which the a of άρα. and the ε of ενερθε could not always withstand j that there were, in short, pairs of forms then in use, one with the augment and one without. This assumption too will be found to suit the special condi- tions under which the augment fell away, in reference to which the following facts are to be noticed : 1) The syllabic augment is never wanting anywhere but in poetry, with three exceptions. These are χρήν, which from Herodotus onward is more used than εχρή*', iteratives, with regard to which we have only Herodotus to deal with, and pluperfects. In the case of the last- men- tioned the loss is quite explicable, and was due no doubt to the difficulty experienced in the attempt to retain the augment always when coming before a reduplication, a difficulty which made itself felt in the same way in the case of the reduplicated aorists, which, however, all but ήγαγον and fhroy, were confined to poetry. The iteratives would anyhow have no great need of an augment, as they are preterites which have no cor- responding presents or modal forms. We may notice specially however the well-attested ί<τ*ί=ε<τ*ί in Alkman fr. 72 B 3 . For χρήν, which is post-Homeric, Ahrens conjectures an origin from χηη 7ιν=-χρεω Jr. (On the verbs in -μι cf. Nauck, Bulletin de l'Academie de Petersb. p. 28 ; Kiihner, Ausf. Gr. I 2 667.) However this may be, these exceptions are, when compared with the thousands of forms that have an augment, so insignificant that they help rather to establish the general rule than to confute it. The fact that the living spoken language, as far as we can see, as good as never neglected the augment in its completest form, is ch. πι. ABSENCE OF THE AUGMENT. 93 a strong confutation of the view which represents the augment as an unessential element in the word. 2) The omission of the syllabic augment in Homer was purely a matter of choice. After all the laborious investigations of Grashof (Programm of Diisseldorf, 1852), M. Schmidt (Philol. ix.), La Roche (Homerische 139 Text-Kritik i. Alterth. p. 423 if.), and others, very little else can be said than was said by Merkel, Praefatio ad Apollon. Rhod. p. 107 : ' de augmento verborum molestissima est ac fortassis inextricabilis quaestio.' Herodian has told US (on θ 161) that 'ίση η -χρήσις παρά τω ποιητή (similarly on Β 808). Beyond this we shall hardly advance. No doubt conventional considerations of the structure of the verse and of euphony were in many instances used as a guide, but it is scarcely possible to re- duce these to the shape of definite rules, and the ingenuity of scholars who tried to unearth them would be better employed elsewhere. 3) Post-Homeric poetry adopts the power of dispensing with the syl- labic augment as an inheritance from its predecessor^ and makes the greater use of it in proportion as it is removed from the language of ordinary life. Hence it is that, as is shown by the careful investiga- tions made by Renner (Stud. i. 2, 18 ff.) the omission of the syllabic augment is extremely rare in iambic, and far more common in elegiac and lyric verse. Hence, as is shown (Stud. i. 2., 259) by Gerth, in the dialogue of tragedy the range of this license is very limited indeed, while the majority of instances of it occur in the slightly Epic style of the messengers' speeches, or still more commonly in lyric passages. 4) The case of the temporal augment is altogether different. Owing, no doubt, to the phonetic difficulty with which its pronunciation was accompanied, it was at no time preserved with strict consistency. In Herodotus, as is shown by the careful investigations of Lhardy (Berol. 1844) and Bredow, the temporal augment is very frequently absent, especially before double consonants : αρρωίεον, ερΖον, ερζαι; έψησε, άπαλ- λάσσετυ, just as in Homer it disappears particularly often in similar cir- cumstances. So too before diphthongs : είκαζε, εΰζατο, ενϊον, αίρεε, ανζετο, υ'ίκτεφε. It is evident that the same reasons are at work here as made the Ionians say εσσων for ησσων, κρεσσων, μεζων for κρείσσων, μείζων, απόίεζις for άπόΰειζις. Here too we have not as yet discovered clearly denned rules, especially as the M.SS. fail us sometimes, as was to be expected. The disinclination to heavy diphthongs occasioned even in Attic Greek forms like αναινον, ε'ίκαζυν (by the side of γκαζόν), εύροι•, 1 and the disinclination to long vowels before double consonants produced 140 ε ζ ετο and ελληνίσβησιιν. In all these cases it was not felt to be in any way a Homeric or poetic usage to leave out the augment. The best expression therefore of the important difference between the omission of the syllabic augment and that of the temporal is this : the former is a poetical and archaic license, the latter is a sacrifice to con- venience of articulation, and was more or less common to all periods. Both omissions fall under the head of weakening, and at no time ^Jid the 1 Cp. Lobeck. ad Phryn. p. 140, and bis note, directed against Elmsley's crav- ing for uniformity (ad Med. 190), on Ajax, v. 120. In a note on this verse Din- dorf remarks that the augment in evpov is unknown to the La. M.S. of Sophocles. Wecklein (Curae Epigrajyhicas, 33) finds inscriptional evidence to τηύρέθν from the year 01. 95, 3, but no similar evidence to forms without augment till a later date. Still, the number of cases is on the whole very small. 94 THE AUGMENT. ch. iti. Greeks lose the sense that the augmented form was the complete and the correct one. D) The Position of the Augment. We shall not enter here into the individual peculiarities in respect to the position of the augment in compound verbs. The statistical state- ment or even the general review of these does not fall within the province of the genetic consideration of the Greek Yerb. We may however call attention to the fact that nice distinctions were observed, and definite rules arose for its position. The guiding principle was clearly this, that in the preterite . the augment was bound immediately to precede the real stem of the verb. Prepositions are transitory elements, defining the direction of the verbal notion either in its original or in its metaphorical application, and were not regarded as belonging to the proper substance of the verb. For this• reason they stand before the 141 augment, and outside the frame of the verbal form, 2 and in like manner complete freedom of position is allowed them in other instances in the older language. Forms then like προςέειπε, περιέβη, and the corresponding Sanskrit forms like pratj-a-volcat, parj-a-gdt, prove incon- testably that the verbal form had become far more closely united to the augment than to the preposition. There are exceptions and irregularities in both languages. But the mere fact that such definite laws arose, shows how far both languages were from regarding the augment with indifference. If the augment really had for centuries, and up to Homer's time, been felt to be an entirely unessential element in the verb, it would be inconceivable that such laws should have arisen, and that men should not have taken the short and simple course of leaving the augment out altogether. Our Sanskrit grammars give us very sparing information as to the practice of that language (Benfey, Yollst. Gr. p. 361). We can see this much however, that it was a refinement peculiar to Greek to distinguish prepositions in this respect from other prefixes, such e.g. as εν and ΰύς. . The structure of the language shows in this something of a glimmering consciousness of grammatical categories which was not developed till much later into a clear recognition of their nature. The irregularities in Greek are specially instructive in two ways. By far the greater part consist in this, that even such verbs as have been derived from nouns already compounded with prepositions which have become an integral part of the word, allow the augment, contrary to the fundamental principle to follow the preposition. The rule is satisfied in ήναντιαΰμψ• (Thuc. iv. 89), but not in άπ-ε-ΰήμησε, νπ-ωπτενον, προ-ε- θυμε~ιτο, and many other like cases, on which the reader may be referred specially to Kiihner, Ausf. Gr. i. 2 516 ff. These exceptions show that the linguistic sense of the Greeks came in the course of time to follow external instead of internal analogies, and took refuge in the simplest statement of the rule, i.e. to put the augment always after the preposi- 2 It might even be said that the preposition is not compounded with the verb-stem, but with the particular verbal form. This view resembles that expressed by Apollonius Dyskolos (Herodian, Ed. Lentz, ii. 790) : Μ των άπΛ ■προθίσεω* αρχομένων γίνεται σύνθεσις κλίσεως, i.e. composition with the inflected, in this case, the augmented form, τουτίστι κατά χοόνον γίνεται η ούνθεσις, καϊ ws &ν tis eiirot, πρώτον κΚίνεται καλ οΰτω συντίθεται, οίον από του γράφω γίνεται κατα- γράφω, κα\ anb του ίγραφον κατεγραφον. ch. ιπ. POSITION OF THE AUGMENT. 95 tion. They carried this so far as to push the augment into the middle of the word when the word only apparently began with a preposition, saying ΰίηκόνονν for the older ιΖιακόνονν^ Ζητωμην, and the like. The second main exception is to some extent an outcome of the principle itself. In cases where the preposition had so far lost its force as such that it ceased to be recognised as a preposition, where, that is, 142 the compound verb formed a practically indivisible whole, the augment took its usual position : yjoiew, Ικάθένδον,ήμφΙεσα. In the case of such subtle differences it is only. natural that there should have been devia- tions from the rule, of which again there is one kind which most deserves our attention. A considerable number of verbs have a twofold augment, one at the beginning and one in" the middle: ήιτεβΰλησε (Arist. fr. 101 Dind., while we read αντεβΰλησε at Λ 809), κατ-ε-ύιή-α, ηνωχΚονν. As a rule the second of these augments is the older one, and the first was added when the preposition had almost ceased to be felt as such. Here again the almost anxious solicitude not to overlook the augment, even in forms of this kind, shows how little it was felt to be dispensable as a mark of past time. 96 PRESENT STEMS WITH NO THEMATIC VOWEL. CHAPTER IV. PRESENT STEMS WHICH HAVE NO THEMATIC VOWEL. Now that we have considered the two expedients most widely used in the structure of the verb, the personal terminations which appear at the end of all verbal forms, and the augment which is attached to the beginning of a portion of them, and in so doing have become acquainted with the frame which is common to forms of the most various kinds, we next turn to the manifold systems which group themselves round the various tense-stems, beginning of course with the simplest. Now the simplest verbal forms are those of the so-called conjugation in -μι. Forms like ψα-μέν, ε-θε-μεν cannot be said to have anything beyond the barest necessities in the way of formal elements. In this respect they stand apart both from such forms as ΰρά-ο-μεν, ε-λίπ-ο-μεν, and from such too as πίμπ\α-μεν and Ιείκ-νυ-μεν. For the purposes of a general review of the Greek verbal structure, however, it is better to keep the old twofold division which treats as a single class all verbs which know nothing 143 of the vowel — whether an o-sound or an e-sound — which we call thematic. The present-stems which have no thematic vowel fall into two main divisions. I. Monosyllabic. II. Dissyllabic. We shall begin with the first division, as it is the simplest. I. MONOSYLLABIC PRESENT-STEMS. A) YOWEL-STEMS. In two of the verb-stems of this class, φη and i, we notice an alterna- tion between forms with a short stem-vowel, like φα-μέν, ϊ-τε, <ρά-*ίι, ϊ-τω, φά-μειο-ζ) Ί-μεναι, ε-<ρά-μην, ε-φα-το, and others with a long one, like the Dor. φά-μί (Aristoph. Ach. 736)= Att. φη-μί, φά-τί (ib. 771)=Att. φη-σί, ει-μι, Dor. ε Ύ ι-τι (Hesych. εζειτι • εζελενσίται)=Αϋ. ε\-σι. Of the related languages Sanskrit shows a most decided agreement in this, but only in roots in i and u. e^mi (from ai-mi) i-mds eshi i-thas i-tha e-ti ir-tas j-anti. Roots in a, on the other hand, whose vowel counts as radically long in Sanskrit, keep this long vowel throughout, so that though bha-mi and φά-μί coincide bha-mds and φά-μές do not. Some traces of the same phenomenon appear in a like place in Zend : aei-ti=Oor eT-r«, 2nd sing, imperat. i-di='i-di (Skt i-hi) ; while in Lithuanian the diphthong goes ch. iv. MONOSYLLABIC STEMS. 97 through all the persons : ei-ml, ei-si, ei-ti, ei-me (Schleicher, Conip. 3 783). The whole process gains special importance from the fact that it is repeated in the dissyllabic stems among the verbs in -μι, and that to a far greater extent and with far greater regularity, and that some traces of it are to be seen in the perfect-stem as well. Various kinds of explanation have at times been attempted of this. To an earlier period in which it had become habitual to regard vowel -intensification as a dynamic process, arising from the effort to attach greater emphasis to 144 certain peculiarly significant syllables, succeeded a complete revolution of ideas on this head, represented most markedly by Grein (Ablaut, Reduplication und secnndare Wurzeln, Cassel, 1862), and his views are more or less shared by Benfey and Holtzmann, and also by Kuhn (Ztschr. xii. 143). The teaching of this school is that all intensification of sound is of purely external, and hence mechanical, origin, and is effected solely by the accent. This view relies for support on what is certainly a pretty formidable series of facts drawn from the verbal in- flexion in Sanskrit, which prove that there existed in this language a far-reaching though certainly not an absolutely unconditional connexion between intensification and accentuation. To establish this theory it is necessary to suppose that Sanskrit has in all those cases preserved nothing but the original accent, that, on the other hand, all the other languages, and in particular Greek, which elsewhere shows much agreement with Sanskrit in matters of accent, shifted the accentuation completely, and not only so, but that this shifting had no power to effect any change in the length of the different syllables, which we must assume to have been by that time definitely settled. Besides, if we are to be convinced by this doctrine, the process of noun and case formation must be thoroughly examined with reference to this point, and a like agreement must be shown to exist throughout these regions, or in a great portion of them, between the forms assumed by vowels and the position of the accent. Even granting though that this could be done, 1 this would-be explana- tion would not be a real explanation after all. For no answer would be found to the question why the accent shifted in such a capricious way. Why not hold the opposite to have been the case, and find in the inten- sification itself the reason why the accent clings to the stem-syllable, and suppose that where strength was not thus added to it the final syllable 145 had power to attract the accent to itself, somewhat as λιπών as compared with λείπων may have owed its accent on the last to this circumstance 1 ? In this way one assertion would be met by another. Bopp, again, did not altogether accept the theory of the mechanical production of the changes here under consideration. He attributed the difference between the form with the stronger and that with the weaker stem to the ' weight of the terminations/ The tendency to strengthen the stem only makes itself felt before the lighter terminations of the singular, while it is counteracted by the heavier terminations of the 1 Since I wrote this, the inquiry into the effect of the Indo-Germanic accent on the form of the words of the several languages has been conducted with decided result and from quite new points of view by Karl Verner {Ztstchr. xxiii. 97 ff.), and succeeded by still further investigations by Osthoff (Beitrage ztir Gesch. der deutschen Sprache, iii.) and Brugman (Studien, ix.)• These labours, however, which are still fast proceeding, deal more directly with other provinces than with that which occupies us at present. Η 98 PKESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. dual and plural, which, as we have seen above, had never less than two syllables, and also by those of the middle ; or, to use Bopp's own words, it is 'revoked before the heavy terminations.' Whether we are to suppose that the fuller form of the stem did actually exist at one time in the dual and plural as well, and that it was afterwards exchanged for the lighter, we are not told. Such a revocation as this would in fact be extremely rare, and could hardly be supported by analogies. Holtz- mann's accentuation hypothesis, which is closely bound up with the still bolder theory that the whole phenomenon of the Guna owes its existence to the influence of an a either present or supposed to have once been present, is controverted by Bopp, Yergl. Gr. ii. 2 480, at which place he expresses his belief that in the weight of certain terminations is to be found at once the cause of the shifting of the accent and that of the 'revocation' of the intensification. This view has at least this much in its favour, that it contains a uniform principle. But even so we are left with a mass of difficulties, e.g. in the imperative, where we should have to suppose that the Skt. dhi had the force of a heavy ter- mination. For this reason Schleicher has been very guarded in his expressions on this point. Without going into the reasons of the pheno- menon he merely notices (Comp. 3 750) the fact that ' the intensification only holds before the personal terminations of the indicative sing, act.' I would only suggest that instead of 'holds' — by which word it is implied that in other forms the intensification has lost its hold — we should say ' takes place.' For no one will be able to show that in the place of i-mas men ever said ai-mas, or in that of 'ί-μες εϊ-μες. For the 146 same reason we shall have no right to place this sporadic intensification of the root- vowel without qualification on the same footing with that permanent 2 intensification which provides the distinguishing mark of the second class of thematic verbs (λείπω, φεύγω). We should perhaps do better to seek for a parallel in the occurrence of fuller stems by the side of shorter ones, which is to be noticed here and there in the declension of nouns, where we find e.g. stems like πολι and πολει, άστυ and άστεν changing places with each other just in the same way, though no definite reason has as yet been found why there should be two forms. It is undeniable that the linguistic instinct took pleasure, at some time early in its history, in adding greater versatility to inflexion by supplementing the effect of the various terminations by giving the stem itself the power of appearing in slightly different shapes. We may be sure though that the weight of the terminations had something to do with this change of quantity. The effort to dwell on the stem in pronunciation was checked by the greater weight and number of syllables in the terminations, and so far, it seems to me, Bopp's principle is correct, though it cannot be said to explain everything. In any case I shall not be wrong, I believe, in saying that this sporadic intensification is not an expedient employed specially to form 2 Fr. Miiller, Dis VocaUteigemng der indogermanischen Sprachen, Vienna, 1871 {Sitznngsber. d. k. Altad. phil. Mstor. CI. vol. lxvi.), has within the shortest possible compass enunciated propositions of a very general nature about these phenomena. With one of these, i.e. that the Indo-Germanic period knew but one stage of intensification, I entirely agree. But I fail to see sufficient grounds for the assumption that in the case both of present-stems in vu and of noun-stems in i and w, we ought to start from the fuller forms na-va, aja, and ava. ch. iv. MONOSYLLABIC STEMS. 99 the present-stem. That it is not so is clearly evident from the fact that ε-ώη-ν, ε-φη-ς, ε-φη bear exactly the same relation to the rt. φα that ε-βη-ν, ε-βη-ς, ε-βη do to the rt. βα, and that as the long vowel does not hold its place in the imperfect so firmly as in the aorist, it has evidently nothing to do with the expression of a continuous action. Strictly, then, ε-φη- ν is no imperfect, but a preterite, and lacks the marks which distin- guish the imperfect from the indie, aorist. H. L. Ahrens (Formenl. 2nd edit. p. 92) says, ' In truth this supposed present ought more 147 jDroperly to be called a 2nd aorist with a primarium, and this is made clear by the purely aoristic force of the preterite.' ' Primarium' is the term used by Ahrens for the indicatives of the present, perfect, and future ; as a rule aorists have none, but here is a case in which such a tense occurs. It is just the same, however, with ειμί, whose participle ιών actually follows the analogy of the aorist in its accent. Here in the indicative the momentary action in present time shifts its ground to the future, a process to which we shall find analogies below in the thematic conjugation. The forms from consonantal stems, moreover, which we shall take next, are formations of a similar kind. The presents belonging to this class are as follows : 1) ψη-μί, conjugated through present and preterite from Homer onwards. For ε-φα-ν (Ν 89), φά-ν (Ζ 108) Homer has also the longer form ε-φα-σαν (Ο 700), φά-σαν (Β 278), formed, as we saw on pp. 12 and 48, by composition with -σαν=εσαν, ήσαν. Besides these, Homer has the middle forms φά-σθε (ζ 200), φά-μενο-ς (Ε 290), φάσθαι (Ι 100), εφάμην (Μ 165), and often ε-φα-το, φύ-το, φά-ντο. — The 2nd sing, φης is peculiar : the t subscript has the authority of all the old grammarians with the exception of Apollonius Dyscolus, who preferred to omit it. The authorities may be found collected in La Roche's Horn. Text- kritik, 374. ψψς is most likely of a like formation with the Ionic 2nd sing, εις, that is, the t of the original termination -σι, makes itself heard in the stem-syllable preceding, exactly as in λνεις for *Χνε-σι or in the Aeol. γέλαις. — If, as Apollonius assumed, the 3rd sing. pres. in an obscure fragment of Anacreon (Be. 3 41) really was φή, the termination -σι must have fallen away after the Aeolic fashion (Ahrens, Aeol. 138). 2) εΐ-μι. The only forms which have no expansion of stem beyond that already discussed, which is confined to the singular, areε7-/it, 2nd sing, εΐ (Soph. Trach. 83) or είσθα (Κ 450), εΐ-σί, ϊ-μεν,. \-τε, 'ί-ασι, 3 and in the preterite 'ί-την (A 347), y -μεν, y -τε (both Attic), ?/-ισαν (Κ 197), εττ- 148 ί/σαν (τ 445), "t -σαν (Γ 8). The expansion by means of added vowels will be discussed below. 3) Of precisely similar formation again is a third present which only occurs in the middle, the real nature of which has not as yet been generally recognised : επί-στα-μαι. The old grammarians were greatly perplexed by this word. In Choeroboscus's discussions, given by Lentz in his Herodian (ii.. 839), we find three different explanations of it. Apollonius identified it with εφιστά μαι, assuming an Ionic psilosis. Such an explanation as this could not find favour with anyone except 8 A remarkable Latin form of apparently like stamp, i.e. ' int iropeuo/rat,' has been brought to light by Gust. Loewe in his Prodromus corporis glossariorum Zatinorum, -p. 421, from the ' glossae PMloxeni ' ; Is, lmus y Itis reveal themselves by their long vowel as contracted from *e-i-s etc. and are therefore thematic. η 2 100 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. the old grammarians, who saw no difference between the psilosis in άπ-ηλιώτης (Princ. i. 497), which is accounted for by its etymology, and that at the beginning οίΐ-στα-μαι (for σι-στα-μαι). A second derivation from *επ-ισα-μαι the middle of 'ίση-μι I know, * πλεονασμοί του r,' may perhaps find acceptance again with the modern advocates of pleonasm, but not with us. The view of Philoxenus, who starts from the verbal adjective Ιστός (rt. hi), in order to get first to *ιστάω and thus to *'ίστημι *'ίσταμαι, cannot be reconciled with the shortness of the o, which points decidedly to a primitive formation. Buttmann (Lexil. i. 278 note) thought the case hopeless, and pronounced ίπίσταμαι to be an original word. The correct solution is given by Pott (Wurzelwbrterb. i. 341) : ' With respect to the form, he (Buttmann) has failed to notice that, though ίπίσταμαι cannot be identified with the reduplicated t- ί&τα-μαι, it may very well be related to it as containing an un- reduplicated form of the same root, like the Lat. 8to, which would be *σταμαι.' The Lat. 8to is not completely analogous, for it has become thematic, and would be more exactly represented by a Gk. *στάω mid. *στάομαι. In the participial στά-μνο-ς tub (Germ. Stander) however we have in Greek itself an analogy for such a formation, as again in the O.H.G. stdn and the Ch.-Sl. sta-ti. As to the meaning acquired, we are justly reminded of the German ver-stehn (O.H.G. firstdn, M.H.G. ver-stdn) and the English understand. The three verbs seem to have had in common the meaning ' step up to, apply oneself to something,' the English word, like the German 1 sick unterstehen,' in the sense of venture on, being suggestive of 149 υποστήναι or subire, while έπίστασθαι is not to be separated from επιστάτης. It thus contains the notion of superiority. There is no very close connexion between the German vorstellen (represent), sick vorstellen (conceive), or the active εφιστάναι τον νουν and έπίστασθαι. From Homer onwards (επίσταμαι ν 207, επίσταμεθα Ν 223, επιστάτη Ε 60) it is always used in the metaphorical sense, and hence the consciousness of its connexion with the rt. στα was so completely obscured that it altogether ceased to be regarded as a compound, and had its augment from Herodotus's time onward at the beginning : ηπίστατο etc. 4) While the three verbs already mentioned leave the stem-vowel in many instances short, the stem κει never occurs in any but the intensified form, κεϊ-ται corresponds exactly to the synonymous Skt. ge-te and Zd. $ae-U (BoppVgl. G. ii. 2 339; Pott Wurzelwtb. i. 543, Princ. i. 178). It is doubtful whether the rt. hi ever appears without intensification. Two forms which apparently belong to it, the Cret. κίαται (Bergmann, Inscr. Cret. p. 11, line 22) and Hesychius's κίασθαι must, I think, on closer inspection be taken otherwise. In various Doric dialects ι replaces an ε of the other dialects before vowels (θιός, άμίων) (Hey de dial. Cret. p. 12 ff.). Consequently κιάται corresponds to the Homeric κίαται, to which belongs the pluperf. κέατο. The t is here, like the e, a weakening under- gone by the diphthong ει before vowels (cp. κείαται, κείατο), like that in βοΐον= βόειο ν (Ahr. Dor. 121). — The New-Ionic κεεσθαι (κέεται, εκεετο) stands for κείεσθαι, and has thus become thematic, as is the case also with Attic conjunctives like κέηται, 5) χρή, used from Homer onwards (A 216) as a 3rd sing, with no personal termination, finds its only analogy in the above-mentioned φή, ch. ιν. MONOSYLLABIC STEMS. 101 but this analogy does not extend to the accent. In Attic writers there are also the conj. xprj, Opt. χρείη, Inf. χρή-ρια (οιίχρήν cp. Nauck, Eurip. Stud. i. 7), preterite εχρην with irregular accentuation and a ν ephelk., or χρήν without augment, whi]e the part, χρεών, also with an anomalous accent and indeclinable, shows a thematic vowel. These numerous anomalies induced Ahrens in his essay on the Conj. in μι, p. 28 (cp. his Formenl. § 195) to pronounce χρή to be a substantive of the same mean- ing as the Homeric χρειώ, χρεώ, and χρτ) to be for χρή 5, χρείη for χρή είη etc. This, however, by no means obviates all difficulties. It is a 150 difficult question — cp. too W. Dindorf in Steph. Thes. viii. 1645 — and is still in want of a thorough examination. B) Stems ending in a Consonant. 6) Et. ες. We have here to deal of course only with those forms which contain absolutely nothing but the root with the terminations, and possibly the augment. The forms of the rt. ες have been discussed at length by Leo Meyer (Ztschr. ix. 373 ff., 423 ff.), and the 3rd pi. pres. ind. in particular by G. Stier (Ztschr. vii. 3). Among the forms of the pres. ind. we may here notice especially the following, as giving rise^fco controversies in one way or another. 1st Sing. It was so easy, even without the help of the Skt. asmi and the Lith. esmi, to arrive at the primary Greek form *εσ-μί from a consideration of the Greek word alone, that even the old grammarians got as far as this. Choeroboscus says (Lentz, Herod, ii. p. 833, 8) Sin αται δε το εσμεν άπο του εσμι εϊιαι. From this primary form we arrive at the Lesb. ψ μι (Sappho* 2, 15). What the strict Doric form was we do not know. It may be doubted whether the Ionic and moderate- Doric ειμί (Ahr. 318) came straight from the primary or from the Aeolic form; most likely from the latter. In that case ειμί bears the same relation to εμμι that ενειμα does to the Aeol. ενεμμα, from which it must have sprung, because in all cases of compensatory lengthening there must have been a consonant lost immediately after the lengthened vowel. εσ-σί, εΤ, ε'ίς have been discussed on p. 33. 1st plur. Here the Attic έσ-μέν surpasses all other forms in antiquity, owing its preservation clearly to the special and often shown liking of the Attics for σμ in the middle of a word. What surprises us is that the 1st sing, did not keep the σμ too. Unfortunately, we do not know what the Aeolic form was, and there are some doubts even about the authenticity of the Doric έσμές (Ahr. 320). The Dor. εψές (Pind. εΐμέν) is of frequent occurrence, and perhaps in stricter Doric it was ημες. Homer knows nothing but εΐμέν, and it is the same with Herodotus. Herodian (περί μον. λεξ. ii. 930 Lentz) has a form εμέν shortened from εσμεν, or directly perhaps from an imaginable Aeolising *ίμμέν. The passage by which he supports it is elsewhere assigned to 151 Callimachus. For all this, good manuscript authority and an entirely satisfactory sense are not able to secure acceptance for ίμίν at Soph. El. 21. Cp. Stud. viii. 322.- Phonetically the loss of the σ is no more inexplicable than in the Homeric πύματο-ς for πύσ-ματο-ς (Princ. ii. 385), in ημαι by the side of ησται, and ή μεν by the side of ήστε. 3rd plur. Here, as has been already pointed out on p. 48, there are two primary forms which share the various dialects between them : 102 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. *εσ-αντι, whence comes only the Homeric earn, and *εσ-ντι, the source of the Doric (Ahr. 321), but also Boeotian (Ahr. Aeol. 211) Αντί, from which again arose the New-Ionic and Attic είσί. A Gk. form cor- responding exactly to the Skt. s-anti, Lat. s-unt, would give *άντι, *άσι, or *άσι. Among the forms of the imperative we may notice εσθι, attested by Herodian ii. 355 from Hecataeus. This is the primary form of the usual ϊσθι. We find εσσο the middle to this in Sappho, i. 28. In Homer εσσ occurs but twice, and in the same form of words (a 302, γ 200). Ahrens (Formenl. 2 101) without any reason sets down the form as an imperat. fut. It is very strange that this middle form should be so isolated when έστω, εστον etc. are so common. The participial -stem ε-ντ offers a strict parallel to the 3rd pi. ε-ντι. Both forms are Doric : Alcman fr. 64 Be. 3 παρεντων, tab. Heracl. i. 104 ειτ-ασσι, 117, 178 εντες. Of the preterite forms the following belong here : 1st sing, ή*-ν, where the σ was expelled by the same necessity as in the forms just discussed •; 2nd sing, ήσθα, where the σ of the root was forced to identify itself with that of the termination ; 3rd sing. r)c=the Yedic as, long since vouched for as Doric, and especially Sicilian (Ahr. 326), and now shown by the inscription of Tegea to be Arcadian as well (Gelbke, Stud. ii. 40). ή ς is 1 of course for ήσ-τ. as the entirely identical Yedic as is for ds-t. On the other hand, ή ν as 3rd sing, cannot be placed in this list because of its v. — In the dual and plural there is a distinction between the forms which preserve and those which have suppressed the σ : ήσ-τον ήσ-την ή*σ-τε ήσ-αν and ή-τον ή-την ή-μεν ή-τε ή-ν. The dual forms with σ are accounted good Attic; ή στη ν is Homeric (Ε 10), and adopted from one M.S. by Memeke fit Theocr. 8, 3. In the 3rd pi. the form with the σ — in which the α is of course to be regarded as the same α as that in ε-ασι — is the only one in common use along with the unaugmented εσαν. 152 The fact that the σ was no more expelled here than in the sigmatic aorist is no doubt to be explained by supposing that σαν had established itself within a wide area as the termination of the 3rd plur. There is but small authority for the form εϋσαν, which is found in the M.SS. H. and J. at ζ 224 instead of the ήσαν of the remaining M.SS., and has been received into the text in two passages of Pindar (Nem. 9, 17 ; 01. 9, 53), where the M.SS. read εσαν or ήσαν. These points and the frag- ment assigned by Bergk to Alcaeus (fr. 91 Be. 3 ), ' Άρκάδες εσσαν βαλανηφάγοι,' are discussed by Nauck, Melanges, iv. 81 (cp. Stud/viiL 326). Since we agree with Bopp in regarding this σαν as a shortened εσαν, and thus regard it as an auxiliary verb added to the verb-stem,. έσ-σαν would on this showing be a compound made by the rt. ες with> itself. Such processes are not impossible, and cannot be denied to have taken place in the fut. εσ-σομαι and the Lat. es-sem, but here the fact is not an established one. Not a trace is to be found of *7ισ-μεν, which would be the form- analogous to εσ-μεν, and the Dor. 7\-μες and the usual Jj -μεν are all for, which evidence exists. For the 2nd plur. however ή*σ-τε, which is set down as good Attic in the Lexicon Vindobonense (ed. Nauck, p. 98), and quoted from Plato (Conviv. p. 176), has survived in two passages in Aristophanes (Pax, 821, Eccles. 1086), and Ahrens accepts it in Theocr. i. 60, on the testimony of some M.SS. The form in general use from ch. it. MONOSYLLABIC STEMS. 103 Homer onwards (Π 557) is ήτε. Leo Meyer (Ztschr. iv. 425) maintains that the forms where there is no σ before the r are contracted for ήε-τον ήε-τε. It seems more natural to refer the loss of the σ to the analogy of the forms in which the σ had been driven out by phonetic influences, like ή-ν ή-μεν and the middle form ήμην, which occurred once or twice in the Attic period, and became later very frequent. The 3rd plur. ήν, vouched for by Hesiod Theog. 321, 825, Aristoph. Lysistr. 1260 ήν γαρ τώιΰρες ουκ ελάσσως τάς Ψάμμας^ and several passages of Epicharmus (Ahrens Dor. 326) bears to ήσ-αν exactly the same relation as ε(σ)-ντι does to *έσ-αντι. 7) Rt. r/c=Skt. as (Princ. i. 472). It seems incomprehensible that Kiihner (Ausf. Gr. i. 671) should hold by the erroneous opinion that η μα ι has anything to do with the rt. έ^, while he at the same time compares the Skt. as which is a long way from sad the Skt. equivalent of the rt. εί). It would be quite impossible to explain the 3rd plur. εαται, ear ο from the rt. εΰ, as c does not fall out between vowels. The final ς of the rt. undergoes exactly the same treatment as in the dual and plural of 153 ήν. The 1st sing, ησ-μαι is only attested as a Dorism by Anecd. Paris, iv. 22, 8 (Ahrens 574), elsewhere it is ημαι, ή μη ν like ή μεν. For the 1st plur. and the participle there are no forms with σ, but only ψμεθα, ij -μενος; ησ-ται, ήσ-το are the usual forms in all dialects, also καθήστο (New-Ionic κατήστο), while in the present the compound never has the σ : κάθηται. There is also a preterite καθήτο. These forms throw light again on those of ήν. The 3rd plur. has only once in Homer the ter- mination beginning with ν : ψ>τ (Γ 153), elsewhere it has always arai. aro. The formation corresponding to ε-ασι would be *ήσ-αται, *ήσ-ατο, and with the regular loss of the σ η-αται, ή-ατο. In the place of these our texts of Homer have ε'ίαται, εϊατο. These forms, favoured by I. Bekker (Homer. Bl. i. 64) were regarded by the ancients as lengthened from the likewise Homeric ε-αται, ε-ατο. Thus Herodian ii. 268, 497, ' πλεονασμώ τον ι.' We shall hardly be satisfied with this, and shall be more inclined to hold that the old HE AT A Ι, Η Ε ΑΤΟ ought, where the metre requires a long first syllable, to be written ήαται, rjaro ; where a short one, εαται, εατο. εαται is related to ηαται as νέα is to νήα, βασιλέα to βασιλήα. Hence, like these noun-forms, it is New-Ionic (κατέαται, κατέατο). 8) Rt. ή say (Princ. i. 496). The forms preserved are, 1st sing. ή-μί (Aristoph. Ran. 37, Nub. 1145), 3rd sing. Dor. ή-τί Alcman fr. 139 B. 3 , Aeol. Ion. ή-σί (Sappho fr. 97, Hermippus Com. ii. 382, 6 Meineke), 1st sing. Pret. %-v in the Attic ήν δ' εγώ, 3rd sing, ή in Homer and in the Attic phrase if δ' or, ή δ' ή. The stem ή stands quite alone in having lost a guttural after its vowel. 4 The Skt. ah-a I spoke points to an original gh, which has been regularly shifted to g in the Lat. ad-ag-iu-m (Princ. i. 497). The final consonant χ which we could before only infer to have existed, I have shown (Stud. iv. 208) to have been actually pre- served in ήχ-ανε-ν είπε ν (Hesych.). Since the Dorians and Aeolians had ή and not ά in this stem, the Greek root must be given as εχ. Such an expulsion of an explosive is of rare occurrence in Greek. We may however see an analogy in the loss of the same consonant in the form σπλήν 154 4 I see no foundation for Westphal's conjecture (Formenl. ii. 112) that pos- sibly ή arose from fa. 104 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. which we cannot but connect with σπΧά-γχνο-ν. The comparison of the Skt. plihan with σπΧήν gives a primary form σπλαχαν, and with loss of the second α σπλαχν (Ascoli, Ztschr. xvii. 269). C) Isolated and Doubtful Forms. Three vowel present-stems present the appearance of a non-thematic formation, but it is extremely probable that it is appearance only. That οιμαι, ωμην, and εΚονμεν, Χονται, Χοννται, Χονμενος once had the thematic vowel is universally acknowledged, only Buttmann (Ausf. Gr. ii. 236) prefers to refer the contracted forms of λούω to λόω, though the con- traction from Χοΰομεν, Χονεται etc. is phonetically legitimate. Good reasons are advanced by Leskien Stud. ii. 104 ff. in support of Butt- mann's view. Another form of the same kind is σεν-ται, which only occurs at Soph. Trach. 645. Nauck follows Elmsley in writing σουται (cp. σούσθω Soph. Aj. 1414). The stem συν corresponds to Xov, and so the Imp. m. 2nd sing, σον make haste (Ar. Yesp. 209) with Hesychius's Χοϋ ■ Χοΰσαι, σοννται Aesch. Pers. 25 with Χοϋνται, and the uncontracted Doric σωοντο • ώρμώντο (Hesych.) with the uncontracted Κωοντο (Callim.). A present σώω would stand in exactly the same relation to the rt. συ, which appears in σν-το, εσσντο, as 7τλώω does to the rt. πλν. We have no hesitation however in classing στενται (Γ 83 etc., Aesch. Pers. 49 instead of the 3rd pi.) and its imperfect στεϋτο (Σ 191) along with σενται. I believe that I was right in referring (Princ. i. 267) these forms to a stem stav expanded from sta, and preserved also in σταν-ρό-ς, and στο-ά (for στ^-ιά), for which there are analogies in the related languages. To the same stem has rightly been referred στύ-εσθαι stand stiff (Fick, Indogerm. Wtb. i. 3 246). στενται then probably stands for στενεται. The Doric μώται (ζητεϊ) along with μώνται, μώμεθα, μώμενος (Soph. Ο. C. 836) is quite analogous to σώμαι. μω can only be a by-form of μα (μέμαα, μαιμάω). The isolated active μώ as 1st sing. pres. is sufficient evidence of thematic formation. If the verb really followed an older 155 fashion we should have to expect *μώμι here. The separate forms are discussed by Ahrens, Dor. 349 f. There are on the other hand three isolated forms from consonantal roots, in the case of which I see no ground for the assumption that the lack of the thematic vowel is of later origin than the forms themselves. First εγμεν ' εχειν Hesych., completely isolated, inasmuch as there is no other similar inflexion of the rt. σε χ as distinct from the transposed σχε. Next εέμεναι, of frequent occurrence in Homer, and that with a de- cidedly present force (cp. Δ 345, Ε 203). This mode of inflexion is here proved to be primitive by the Skt. ad-mi, 3rd sing, at-ti, Lat. es-t by the side of ed-it, Ch.-Sl. ja-mi or έ -ml, Lith. id-mi (Princ. i. 296). We cannot be so sure of this in the case of the 2nd plur. Imperat. φερ-τε (only I 171). For here, though the Greek inflexion is supported by the Lat. fer-te (fer-8, fer-t etc.) and the Skt. has bhar-ti=fert (Princ. i. 373), the prevailing mode of inflexion is decidedly the thematic. We are not obliged, however, to regard φέρτε as syncopated from φέρ-ε-τε. — Ιεχαται (Μ 147), on the other hand, along with Ιέ-γμενος (I 191, more frequently Tortciy μένος), cp. προτίΰεγμαι' προσίεχομαι. in Hesych., is, to use Lobeck's expression (on Buttmann ii. 21), an ' unarticulated ' pre- ch. iv. REDUPLICATED STEMS. 105 sent-form. Lobeck points out at the same place that the late poets made other forms like it, e.g. άμειπτο=άμείβετο (Npnnus). II. STEMS OF TWO OR THREE SYLLABLES. We start here with the present-stems whose formation is the most transparent, those which have two syllables in the present only. In these cases the force of the expanding syllable is clear. This syllable is what we may call a present-expansion, and it serves to characterise the continuous action by a greater amplitude in the form. A) Stems which are made dissyllabic or trisyllabic by means 156 OF THIS PRESENT EXPANSION. a) Eeduplicated present stems. We have already seen on p. 8 that reduplication is one of the oldest methods employed to expand the verbal stem, and thus give it addi- tional emphasis. The same method was used to strengthen the stems of other tenses as well, especially those of the perfect and aorist, so that we are not able to find in reduplication an original and specially distinguish- ing mark of the present- stem. It may have depended somewhat on the individual meaning of the separate verb, whether it should use this method in one way or the other, or possibly in several. Nice distinc- tions were however set up in Greek between . the reduplication for the present tense and that for the perfect or aorist. The present-reduplica- . tion is always characterised by the t-sound, and this is observable, though only sporadically, in Sanskrit as well (Delbriick, Altind. Yerb. 104 f.). 1) The rt. βα forms its present-stem in this way in the participle βιβάς, which has only survived in certain phrases, such as μακρά βιβάς (Ο 307, Η 213), ί/ψί βιβάΐ'τα (Ν 371), κραιττνά ποσι προβιβάς (Ν 18), νπασπίΰια προβιβάντος (Π 609), in all of which an intensive force is discernible. The same may be said of the Laconic βίβαη, which denotes a kind of dance (βιβασις) in a somewhat obscure passage in Pollux iv. 102. What Ahrens (Dor. 312) says about the t-sound is not satisfactory. It seems however to establish the Doric pres. βίβαμι. A complete parallel is furnished by the Skt. gi-gd-ti from the rt. gd go, by the side of which there is $a-gan-ti from the synonymous root gam. By a change to the thematic conjugation there arose the form βιβώ (part, βιβών, an old variant for βιβάς in Homer). Cp. La Roche, Horn. Textkritik, 216. 2) The rt. £ε bind is found only in two reduplicated forms belonging to the present-stem ; 3rd plur. Imperat. εν ΙεσμοΊσι ΰώέντωι•, Aris- tarchus's reading μ 54, instead of Ιεάντων, and the 3rd sing, imperf. &£$, Λ 105. 3) Common to all Greek is Ιίΰωμι from the rt. £o, corresponding to the Skt daddmi, Zd. dadhami. The long vowel in the imperat. ΙίΙω -di (γ 380) is remarkable, as is the diphthong in IlIovvcll (il 425). The 157 variations in the length of the stem- vowel are often irregular in the Epic dialect (cp. τιθημενος, τώημεναι). 4) Ιί-ζη-μαι is the Epic for the later ζητεω, which only occurs once in Homer (Ε 258). It is natural to conjecture that h -ζη bears to ζη -re 106 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. the same relation that βι-βα does to (αμφις)βη-τε (Princ. ii. 262). That <5t is a reduplication is made more clear by the Aeol. ΰί-σΰη-μαι (ζητώ Hesych.). The latter form bears exactly the same relation to the stem σΰη that the Skt. ti-shthd-mi does to the rt. sthd, and is similar to that of κε-σκι-ο-ν sieve to the rt. σα. There is no more analogy for δ as a representative of ζ in a reduplication than there is for κ for £ or π for ψ (cp. φε-φαλο-ς by the side of ψόλο-ς). But we cannot doubt this explana- tion of Ιίζημαι. For the length of the stem-vowel we may compare κείμαι. If it were not for the ι in the reduplication which clearly marks the present, it might be thought to be a perfect. Ιίζηαι (λ 100), Ιιζημενος (Ε 168) are found in Homer, and other cognate forms in Herodotus, Theognis, Aeschylus (Ιίζηνται Suppl. 821). The future Ιιζησόμεθ' 1 (π 239) does not upset our theory, as is shown by the isolated Homeric ΙιΙώσω. By the adoption by the stem-vowel of the analogy of the the- matic formation there arises Ιίζω (even in Homer Π 713). 5) Ί-η-μι. I have discussed the origin of the verb at Princ. i. 500, where I adopted the derivation, first given by Bopp, from ji-jd-mi. It is evident that 'ί-ε-σαν is related in the same way to ε-σαν as ε-τί-θε-σαν is to ε-θε-σαν, and Ίέμενος to εμενος as τιθέμενος λ& to θέμενος. Butt- mann (i. 521) recognised this fact. The reduplication syllable is both long and short, and I have discussed this in connexion with the partition of the different meanings between active and middle, and the traces of an initial consonant in Philol. iii. 5 ff. It explains everything if we start from a rt.ja, Gk. fe, which when reduplicated gives jji-fe. The hiatus before ιέμενος (οίκάδε ίεμένων Β 154) is accounted for by the surviving effect of the initial j, and the length of the ι by that of the second j, and the same explanation, as we saw above, p. 79, accounts for the syllabic augment in the aor. εηκε. Out of the meaning ' go,' which belongs to 158 the rt.ja in Sanskrit, was developed on one side the causative * make to go, send,' on the other in the middle the meaning ' strive, wish.' None other of the derivations attempted for ΐημι is so in harmony with both its form and its meaning as this. Pott's energetic attack upon it (Wurzelwtb. ii. 2, 288) is mainly based on the difference of meaning seen in the active. Though he cannot deny the analogy of 7- στ η- μι and si-sto to 'ί-η-μι and all the reduplicated forms, he finds a stumbling-block in the fact that the non-reduplicated forms have a causative meaning too, and certainly άφ-ε-σαν * they let go or sent ' does not bear com- parison in this point with άπ-ε-στη-σαν they went off, nor άφε~ικα with άφέστηκα. But seeing the causative meaning extends in the case of the rt. στα from the present-stem to the future and the sigmatic aorist, it is not so very much to suppose that in that of the rt. ε it extended a little further still. The German verbs for * send,' schicken and eenden, are causatives too, the former from the O. H. G. scehan (Germ, ge-schehen happen), the latter (in Gothic sand-jan) from *smd-an (A.S. sinnan) go ; and πέμπειν, whose etymology is obscure, shows in πομπή procession, and especially in the Homeric ΰυς-πέμφ-αλο-ς, an epithet of the sea, ' ill to traverse,' a meaning which points to an originally intransitive force. In the inflexion of ΐημι it is the imperfect which presents most diffi- culty. The 2nd and 3rd sing, ΐ-ει-ς ί-ει are easily explained as having followed the analogy of the contracted verbs, so that they are contracted from '/'t-t-c, t-e-f, but this explanation accounts only for one form of the ch. iv. REDUPLICATED STEMS. 107 1st sing. i.e. 'ίουν, but not for tup which appears occasionally (Herodian, ii. 835), and is countenanced by έτίθειν on an inscription (C. I. 3605, 33). Boeckh puts the inscription as late as the first or second century before Christ. I. Bekker may therefore perhaps be right (Hom. Bl. i. 61) in rejecting πρυίειν (ι 88, κ 100) in favour of προίην, and in declining, in Plato, Euthyd. 293 a, to follow the one M.S. which has ήφίειν. (Cp. also Αόγιος Έρμης i. p. 355 f.) Possibly it is only that the 2nd and 3rd persons have dragged the 1st after them. The analogy of γειν and the pluperfects may also have had something to do with it. 6) ϊ-στη-μι, like ΰίοωμι and -ίθημι, common to all Greek. Its spe- cially causative meaning has been noticed already. We may perhaps 159 conclude from si-sto, which is identical with ϊστημι, that this meaning dates from a time when there was a s yet no difference in the vowel be- tween si-sti-mus si-sti-tis on one side and Ί-στα-μες Ί-στα-τε on the other, but when *si-sta-mas *si-sta-tes were the forms in use. The peculiarly Greek transformation of the s to the spiritus asper needs no further dis- cussion. The Skt. ti-shtha-ti and the Zd. hi-gta-iti differ from "ι-στη-σι in that they are intransitive and leave the vowel short after the fashion of the thematic verbs. 7) Kiy -κμα-μι Dor. Ahrens 346,, Hesych. κι-γκρψ' κψνψ, Sophron. fr. 2 εγκίκρα. 8) κί-χρη-μι, of rather late occurrence, i.e. not earlier than the pseudo- Demosthenic speech προς Νικόστρατον § 12 {των κτημάτων σοι των εμων κίχρημι 6 τι βοΰλει). The rt. χρα shows a similar meaning in χρέος, and the Lat. usura illustrates the relation between χρήσθαι and these words. From Hesychius's gloss κιχρήσει• δανείσει we can see that, as in the cases given under No. 4 (ΰίζημαι), the reduplication extended beyond the present- stem. 9) πίμ-πλη-μι, from Homer onwards (Φ 23 πιμ-π\ασι), in use along with π\ήθω as a present form of the rt. πλα (Princ. i. 344). The Skt. pi-par-mi, I fill, agrees exactly with the Greek, except that in the latter a nasal is put in to strengthen the reduplication syllable as in κά•γ-κα-νο-ν, Ιεν-Ιρε{Ρ)ο-ν, just as it is in a number of Sanskrit intensives. Rich. Fiitzsche (Stud. iv. 310 f.), who is supported by Joh. Schmidt (Vocal, ii. 228), conjectures that this nasal is the remains of a .stronger consonant. If the μ of a preposition precedes the word there is only a simple ι : εμ- πί-πλη-μι. The long root-vowel in εμπίπληϋι Φ 311 is just like that in ΰίΰωθι. 10) πίμ-πρη-μι, a present-formation unknown to Homer (who has ενέπρηθον, i. 589), but in all other respects a parallel to πίμπΧημι, even in respect of the movable nature of the nasal : ίμπιπράσι (Thuc. 3, 74), συμπιπράναι. 11) πί-ψρη-μι, only found in Aristotle Hist. Anim. v. p. 541 b, 11 : η (read ήι) εςπιψράναι εις τον μνκτηρα της θηλείας. In common use in Attic are εκ-φρε-ς, επ-εις-φρώ, ΰια-ψρη-σονσι etc., on which I cannot accept Nauck's view (Bulletin de Γ Ac. de St.-Petersb. T. vi. p. 424 ff.), for the reasons given by me at Stud. viii. 327 if. Herodian (i. 463, 160 1. 14) says φρές is the same as φέρε. πί-φρη-μι may be compared with the Skt. bi-bhar-mi I carry, bring, in everything but the metathesis. 12) r/-077-/jt=Skt. dd-dhd-mi (Zd. 3rd sing, da-dhdi-ti), treated in every respect, even in that of the ει which takes the place of the η, like 'ίημι. τι-θη-μεπα (ψ 83) and τι-βή-μενος (Κ 34) have been noticed under no. 3. 103 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. 13) τί-τρη-μι I bore, not found before Galen and Appian (ϊιατίτρησι, ^lartrpcu αι) for τιτράω or τετραίνω. Cp. Lobeck on Buttmann, ii. 304. We must add here two verbs which are singular in every way, i.e. 1 4) t-7rra -μα t, a late by -form of πέτομα t, first occurring in the spurious part of Euripides Iph. Aul. (160S), and then in Babrius, Plutarch, Moschus etc. [Aristotle Hist. An. v. 9], but apparently an old form, as there is no clear model from which it could have been taken. I clearly stands for τι here, as ε in ε-ψω for πε (Prine. ii. 375). 15) όν-ίνη-μι which, though it is trisyllabic, and consequently must have been reduplicated from a stem which had already two syllables (ώα), may for brevity's sake be discussed here. Here, as in όπ-ίπ-άς and όπ-ϊπ-ενω (όπιπτεύω) the second vowel has become ι. Buttmann, Ausf. Gr. ii. 74, compares άτιτάλλω as well, only here the origin of the word is obscure. Similar phenomena are displayed by the reduplicated aorist in Sanskrit, e.g. dp-ip-a-m from the causative dpajdmi (dp ob- tain), όίήησι in Homer only at Ω 45, dm άσο Plato Phil. 58. For its origin see Princ. ii. 397, Gustav. Meyer, Nasale Prasensstamme, p. 42. The form hdpavai on the other hand is spurious, though Westphal has brought it up again, notwithstanding that at Thuc. iv. 46 the read- ing now universally adopted on good authority is άποΐραναι, and there is no other credible testimony to the word. b) Present-stems formed by the addition of the syllable w. b The syllable vv, which in a considerable number of verb; distin- 161 guishes the present-stem from the verb stem, corresponds to the nu of Sanskrit, where the verbs belonging here form the fifth class. In the Iranian languages too there are individual instances of this kind of present-formation (Schleicher Comp. § 293 iv. a). These expanded pre- sent-stems are to be regarded, as we saw above (p. 10, cp. my Zur Chronpl. 2 p.-4f> f., Benfey,. Ztschr. viii. 94), as noun-themes of a kind which has been preserved in Sanskrit outside the verb as well. We saw above that in all probability such a theme in nu had, when these forms took a definite shape, the meaning of a nomen agentis, such as has sur- vived, e.g. in the above-mentioned Skt. dhrsh-nu-s bold (which would correspond to a Gk. * βαρσ-νν) and in grdh-nu-s eager, from the rt. gardh. In Greek there are no nouns of this meaning of such a formation. In form however θρή-ιν-ς is similar, from the rt. dhra (Skt. dhar hold, support), only here the meaning i holding, supporting/ has shifted to that of ' holder, supporter,' especially of the feet, a foot-stool. It is clear that the suffix nu is very closely related to the suffix na. Present-stems in nu and in na (9th class) are very often developed side by side from the same verb- stems. The syllable na appears in Sanskrit in one special class of verbs, the 7th, in the middle of the root : ju-na-(f-mi from the rt. jutf. Not unfrequently too there appears, especially in the Sanskrit and Latin, a simple n : ju-n-(j-mds 1st plur. to the sing, ju-na-tf-mi, Lt&t.juny-i-mus, • Rich materials for the discussion of these', as for that of all present-stems formed with nasal syllables, are contained in Gustav Meyer's work, Die mit Nasalen gebildeten Prasensstamme des GriechiscJtcn, Jena, 1873, which for brevity's sake I shall refer to as « G. Meyer n. Pr.' ch. iv. STEMS WITH A NASAL AFFIX. 109 in which case forms with and forms without a thematic vowel are often interchanged. There are several points in these processes which have not yet been clearly made out, and such of them as do not occur at all in Greek, as is the case with the peculiarity of the Skt. 7th class, we may here pass by. But thus much is unmistakable, that the linguistic instinct mixed these nasal expansions of the present-stem very much with each other. It is therefore important when dealing with the fol- lowing Greek formations to notice anything at all analogous either in Greek itself or in the related languages, and specially in Sanskrit and 162 Latin. The formation of the expanded stem in vv is often accompanied by an intensification of the root- vowel : ζνγ ζενγνν, ΰικ ΰεικνν, παγ πηγνν, pay βηγνι, kl κϊνν, τι τινν, but it cannot be said to be the fixed rule that such is the case. In Sanskrit (cp. Delbruck, Verb. 154) there appears a faint trace of such a procedure in the Yedic ddg-no-ti from the rt. dag (cp. Ιάκνω). Other peculiarities in the formation of these verbs are noticed by Lobeck on Buttm. ii. 68. While in Sanskrit not only roots ending in a vowel but also those in all kinds of consonants (cak, dp, tarp, dharsh, ag) belong to this class, in Greek it is only roots ending in gutturals, nasals, λ ρ σ, and vowels, that develope present-stems of this kind, and the λ in ολλυμι assimilates to itself the following r, and conversely the ν of the expanding syllable assimilates a preceding σ (εν-ιν-μι). This limita- tion of an expedient which the language might have employed generally if it had wished, is a feature peculiar to Greek, and one which we shall often encounter. Elements used to expand stems, which in their origin had no closer connexion with any particular final letter of a stem than that of derivative suffixes, are found afterwards under the rule of purely phonetic analogies. By far the larger number of the verbs in -νν-μι have numerous and sometimes much more generally current by-forms in ννω. 6 Even this phenomenon, which does not look at first sight like a very old one, has analogies in the East. Bopp, Ygl. Gr. § 519, and Skt. Gr. § 343, notices corresponding changes of form in Sanskrit and Zend : Skt. d-du- nv-a-sva (du torment) for d-du-nu-shva, as it might be *Ιιικ-νν-ε-σο ΰακννον for £fo:-j ν-σο, Ζά. kere -ηυΰ (primary form har-nv-a-s) for kere- naus thou didst make. Leo Meyer, Goth. Sprache, p. 203, gives from theBigveda 6, 2, 6 rnvati (=*όρννει) instead of the usual rnoti (=op- νυσι). Many more cases of the kind may be found collected by G. Meyer n. Pr. 39 and Delbruck, Verb. 158. We may safely follow Leo Meyer and Adalbert Kuhn, the discoverer of this connexion (Ztschr. ii. 460), 163 at least in some instances, when they refer Teutonic strong verbs in nn to a similar source, so that Goth. Ηηηαη=*όρννειν. The individual stems may be best arranged in the following three main groups : I. vv affixed without material change in monosyllabic stems, and that a) to consonantal stems ; β) to vowel stems. . II. vv with a ν before it. 6 Lists to show the comparative frequency of forms in -ννω and -νυμι are given by La Roche, Ztschr. f. Men: Gymn. 1876, p. 584 ff. 110 PRESENT STEMS WITH NO THEMATIC VOWEL. I. First Group. 1) i'ty-w -μι (break) rt. /αγ. The present-stem is pretty common in poetry from Homer onwards (M 148 αγη/τ-οι•, Π 769 άγνυμενάων), rare in prose. If we were right (Princ. ii. 158) in comparing the Skt. bhartij with the rt. At γ, both verbs agree in taking the nasal. 2) αρ-νν-μαι (get) rt. ap, the same which is the basis of the noun- forms άρ-ος gain, μίσθ -ap-to-c, ερ-ιθο-ς (Princ. i. 425). A 159 τιμήν αρνΰμενοι (cp. α 5), άρννσθην 3rd du. Χ 160, τοιάΰ* άρννμαι Soph. Ant. 903, in prose only used in the phrase μισθον άμννσθαι (Plato Prot. 349 a). It has no connexion whatever with αίρω, άείρω. 3) αχ-νν-μαι (grieve), one of the many present forms of the rt. άχ, which makes also αχ-ο-μαι (σ 256, τ 129), άχ-θ-ο-μαι, άχ-εν-ω, and in a causative sense άκ-αχ-ίζ-ω (cp. ηκαχον). αχνντο hi σφιν θυμός & 38, commonest of all άχννμενος, which shows a relic of itself in άχννμενως in Soph. Ant. 627. Late poets formed from άχνυ a passive aorist άχνΰνθην Anth. vi. 343. It is a purely poetical verb. Numerous instances of the insertion of a nasal in the middle of this root (Skt. ah-u-s, Germ. eng narrow, Lat. ango) are given at Princ. i. 234. 4) ΰείκ-νυ-μι, one of the few verbs of this formation which is common to all Greek. The pure rt. Blk occurs in Greek only in Ζίκ-η, but is just as evident in the Skt. dig and the Lat. die (causi-dic-u-s), while the 164 verbal forms are made throughout from the intensified ΰεικ (Princ. i. 165). The latter is in the New-Ionic dialect partially attenuated again to ΰεκ (δε ζω, εΰε£α), but preserved in the present-stem : ϊεικνύς Herod, ii. 78, so that here there is a fresh pair of short and long roots, $εκ and δεικ•. There is not a trace of nasal expansion anywhere but in Greek. In Greek itself we may compare feiK-ar -ά-ο-μαι. 5) Ιεχ-νν-μι, quite a late by-form of Ιεχ-ο-μαι. Parthenius Bekk. Anecd. 1385 b, Orphica Lith. 692, mostly found in the imperative ΰέχνυσο. 6) εφγ-νν-μι by the side of ε'ι'ργω. In Homer we find the imperf. iipyvv κ 238 κατά σνψεοίσιν εέργιν, where it is impossible to say whether the first ε is an augment or a prothetic vowel before a lost digamma. In any case εϊργννμι, which occurs in Herodotus and in Attic prose (Plato, Tim. 45), presupposes k-hpy-w -μι. The corresponding Sanskrit root var$ likewise makes its present by a nasal expansion : vr-nd-g-mi (Princ. i. 222). 7) ζενγ-νν-μι, like Ιιίκννμι in frequent use with Greeks of all times and all tribes, is the primitive Indo-Germanic verb for the yoking and harnessing of horses (Princ. i. 223). A present formed by nasal expan- sion occurs in Skt. (ju-na-$-mi) and Lat. (ju-n-g-o) as well. 8) θόρ-νν-μαι (copulate) first occurs in Nicander Ther. 130, as a by- form of θρώσκω, the conj. θορρνωνται, which might belong either to θόρννμηι or θαρννομαι, in Herod, iii. 109. 9) καβίγννσθαι. So Lobeck (on Buttmann, ii. 68) writes the word which occurs in several passages of Hippocrates (Mul.), and always with the var. lect. καθίνινσθαι, and others follow him in so doing. No one can doubt that the word with the meaning κατακλίνειν, καθίζιιν, is con- nected with the rt. εδ. In spite of this Lobeck thinks the connexion ch. iv. STEMS WITH A NASAL AFFIX. ] 1 1 with ιγννς, ιγνύα, poples, probable, in the sense of ingeniculare. The two views are incompatible. For the latter words clearly belong to yorv, and no evidence can be shown for the change of the undoubtedly primitive δ in the rt. eh to γ; and besides the breathing of l -γνΰς does not suit that of the verb in question. Everything points therefore to the reading καθίνννσϋαι with ι for e as in ΙΙ-ρΰ-ω and Ιζω (cp. Westphal. 165 Method. Gr. ii. 163). To this we may add Hesychius's glosses Iwviv (M.S. ιννοιει)' εκαθέζετο, and Ιννεσθια ' κοσμέϊν, ίΰμΰεσθαι. There is no analogy to this formation in the related languages, and moreover this would be the solitary instance of the addition of the syllable vv to a dental explosive. May it not even be possible that καβίν-νν-μι arose out of καβιζ-νν-μι ? A similar addition of a nasal expansion to the already ex- panded present form took place in καθιζάνω. In that case the verb would be analogous to the comparatively late forms in vv to be discussed below. I cannot see the force of Joh. Schmidt's new explanation of the form (Ztschr. xxiii. 298). 10) μίγ-νν-μί, the later present-form of the rt. μιγ, first found in Pindar, then in Attic writers, for which Homer has μίσγω. The ι, as Lobeck shows (Paralipp. 414), was long, and who can say that Μειζ,ίας, Μειζιάΰης, for which there is the testimony of inscriptions, were not the older forms of the names 1 μείγ-νν-μι : μιγ : : ΰείκ-νν-μι : ΰικ (Princ. i. 417). 11) ο'ίγ-νν-μι, Β 809 πάσαι δ' ωίγ-νν-ρτο πνλαι, then Attic along with ο'ίγω in compounds. The origin has not been clearly ascertained ; all that we can be sure of is that / was the original initial. Cp. above, p. 81. 12) ομ-νυ-μι, in general use from Homer onwards, though there are frequent by-forms in νυ-ω. The origin of the word is obscure. 13) ό-μόργ-ι>ν-μι. Ε 416 απ ιχώ χειρός όμόργνν, λ 527 δάκρυα £' ώμόργί'νντο, of isolated occurrence in Attic. Compare the Ved. 3rd plur. mid. mr-n-tf-a-ta, Gust. Meyer n. Pr. 19. The rt. marg has by means of a difference in the vowel and in the formation of the present, split up into two stems of different meaning, Princ. i. 226. 14) ό-ρέγ-νν-μι, only A 351, X 37 χείρας όρεγννς and in later poets (Moschus, Anthol.) όρεγννμενος, while όρέγω (along with όρεχθεω) re- mains the usual form of the present. Another kind of nasal expansion of the rt. is to be seen in opiy -νά-ο-μαι (Hes. Scut. 190, Eurip. Theocr.), by the side of which there is a doubtful trace (Pollux v. 165) of an όρίγ-να-μαι or όρίγνομαι. opiyi άομαι bears to όρέγννμι a relation similar to that of ΰεικανάυμαι to ΰείκννμι. The corresponding Skt. rt. ar$ forms the 1st sing. mid. r-n-g-e in an analogous way (Princ. i. 226). 15) υρ-νν-μι from Homer onwards (ορνυΟι, όρννμεναι, ώρνντο) in 166 poetry, entirely corresponding to the Skt. r-no-mi (Princ. i. 432). In όρίνω there is another kind of nasal expansion, which however we ought probably to refer to όρ-ι-ρν-ω. Other present-formations are όρ-έ-ον-το (cp. oriri), όρονω, όρο-θ-ύνω. 16) πήγ-νυ-μι. Homer, who constantly has έπηζε, επάγη, πέπηγε, has no forms of the present-stem, πήγ-νν-σι Aesch. Pers. 496 and after him repeatedly. The by-form πηττω is not found till after Alexander. The Lat. pa-n-g-o shows an analogous expansion of the present. 17) π\έγ-νν-μι. The only form found is πλεγ-νν-μενας in Oppian Cyneg. iii. 213, Halieut. i. 311. Elsewhere the pres. is always πλέκω. In spite of this the Skt. pr-na-K-mi (Princ. i. 203) suggests the conjecture 112 PKESENT STEMS ΛΥΙΤΗ NO THEMATIC VOWEL. ch. iv. that the late form rests on an old tradition, for πλεγνυμι : 'prndU-mi \\ ζεΰγννμι '. junatfmi, 18) πλήγ-νν-μι, only preserved in (κπλήγννσθαι Thuc. iv. 125, and yet it must be of early origin, since a nasal appears also in the related πλάγχθη (Princ. i. 345) and in the Lat. plango. 19) πτάρ-νν-μαι, Attic; Homer has only επταρον; a later present is πταίρω. The Lat. ster-nu-e-re, frequentative sternutare, proves that this expansion of the present is of old standing. This is the only instance in which Latin has preserved the syllable nu (Princ. ii. 372). Cp. Bugge Ztschr. xx. 37. 20) ρήγ-νν-μι, occurring from Homer onwards (3rd plur. ρηγινσι Ρ 751, ρηγννμενος, ρήγνντο, ρήγνυντο). ρήσσω, used in Homer (Σ 571) and elsewhere in the sense of pulsare, has certainly nothing to do with it (Lobeck on Buttmann ii. 287), but in Hippocrates and Diodorus ρήσσω occurs with the meaning of ρήγννμι (cp. πήττω, πλήσσω). If it is right to connect ρήγννμι with frango (Princ. ii. 159, otherwise Fick Wtb. i. 3 772), we have in the latter an analogous expression. 21) στόρ-νν-μι is related to the Skt. str no-mi (by-form str-na-mi) exactly as ύρ-νυ-μι is to r-no-mi. Add to these the Lat. ster-no (Princ. i. 265). In Homer καστορννσα p 32. The verb is found besides in the tragic and comic dramatists, in Herodotus and in Xenophon, with the Attic by-form στρώνννμι with metathesis like that in the Lat. strd-tu-s, 167 while στορ-ε-ννυ-μι was developed at quite a late period (only found in the Schol. on Theocr. 7, 59) from forms like εστόρεσα. 22) φράγ-νν-μι, an isolated by-form of the usual φράσσω, Thuc. vii. 74 άπεφράγννσαν, Soph. Antig. 241 κάποφράγννσαι (Dind. άποφάργιυσαι, cp. Siegismund, Stud. v. 159). In Homer there is no present-stem to φράζε, ψράζας, εφράχθην, cp. Princ. i. 376. 23) φώγ-νν-μι, only found in Dioscovides (ψώγννται), elsewhere φωγω. β- 24) αί-νν-μαι only in poetry, from Homer onwards : άπο-αίννται Ν 262, ρ 322, αινύμενος ι 429, χ 500, άποαίνντο μ 419, άτταίνντο ϋ 595. Beyond the present-stem no forms are in• use. The very general notion that this verb belongs to αιρέω is hardly tenable, for the root of αίρέω is fap (by-form ΐελ), and it is hardly possible to get from *ΐαρ-νυ-μαι to αϊνυμαι. The hiatus in άποαίννμαι does certainly entitle us to give the word an initial /, but εζαιτος, αΐτέω (a frequentative to αΐννμαι) also be- long, so that we cannot say more than that the stem was either ni or fai. 25) γά-νν-μαι, in poetry, from Homer onwards, γάννται Ν 493, γάνννται μ 43. The present-stem is also used to form the future : γαννσ- σεται Ε 504. There is no doubt that the root is the same as that of *γη-θ-έ-ω, γα-ί-ω (cp. Princ. i. 211), perhaps also as that of γά-ι-ος and γα-νάω shine. 26) γί-νν-μαι on an inscription from Aegosthena, edited by Boeckh, Monatsber. der Berl. Akad. Nov. 1857 (Princ. i. 215). Cp. below κτίνννμι. 27) lai-w -μι, Homeric (faci -νιτα I 3, laivv 2nd imperat. and 3rd imperf.) in the middle (δαύ νται Ο 99) it occurs also in other poets and in Herodotus, 1 aor. ίαησα, ί^αισάμην, related to ία/ω, ΰαίομϋΐ, δα/ς, dot- ch. iv. NASALISED STEMS. 113 τυμων (Princ. i. 285). The rt. is da, and the ί is itself an expansion of the present-stem to which the w is subsequently added as a second. 28) καί-νυ-μαι only Epic (γ 282 ος εκαίνυτο φνλ' ανθρώπων νήα κυβέρ- νησαν, άπεκαίνντο θ 127). The more generally diffused forms κέκασμαι, εκέκαστο, κόσμος, lead us to the conclusion that the root had a consonant at the end. Still it is by no means clear what relation the diphthong bore to the a. 29) κί-νν-μαι. Horn, κϊνύμενος Κ 280, κ'ινυντυ Δ 281, and then not 168 till late poets. The form in general use is κινέω. The Skt. ci-nu-te is identical with κί-νυ-ται except that the root vowel in the Skt. word is short. The present κί-ω (by-form κί-α-θ-ο-ν) is of a shorter formation. 30) τά-νν-ται, just like ya-vv -ται, only at Ρ 393, but τα-νν-ω is much commoner. The syllable vv makes its way into the structure of other tenses as well : τάνυσπα (τάννσα), τετάνυσται, τάννσθεν. τά-νυ-ται corre- sponds exactly to the Skt. ta-nu-te, though as a rule this is divided thus, tan-u-te (8th class). But, as Bopp saw (Skt. Gr. § 343), the present- expansion does not consist in u but in nu. Whether, as he supposes, the root had a final n as well, so that ta-nu-te is for tan-nu-te, or whether the root of this word is the shorter ta, may be left an open question (Princ. i. 267). In Zend too there is a 1st sing. pres. tanva or tanava (Justi Handb. 131), exactly identical with τανΰω. The more general forms of the present are τείνω and τιταίνω, each with a slightly distinct meaning. 31) τί-νυ-μαι. τίνυνται Τ 260, Γ 279, τϊ-νύμεναι Eurip. Or. 323, τίννσθαι Hdt. v. 77. The length of the ι is of just the same kind as in κίΐ'νμαι (29). Perhaps this word had originally the diphthong, which is actually found in τειμη, τείσω, on old inscriptions which show no itacism of any kind, and in many other words as well (Ίποτεισεϊ tabb. Heracl. i. 109, Meister Stud. iv. 387). Buttmann's view that we ought to write τίνννμαι (Ausf. Gr. ii. 69) rests on mistaken premises. We may com- pare the related Skt. Jci-no-mi, and ci the corresponding present in Zend (cp. Princ. ii. 93). τίνω, with the simpler τίω (Arcad. άπυτειέτω, cp. Joh. Schmidt Vocal, i. 142), is closely connected with τί-νυ-μαι. 32) φαί-νν-μι, only known to us from Hesych. : φαι-νυ-ντες. φωμί- ζοντες, cp. φα 1-μα ' ολίγον, φαί-νν-σμα' ολίγον, φι-νΰ-θ-ιο-ν' φλανρον, φεΐ-σαι » φωμίσαι. The root is the same as that of the verbs φάω, φαίω, and the adj. φΐ-λό-ς. Cp. Ιαίνυμι no. 27. II. W WITH AN ν BEFORE IT. a) Where the first ν is evidently assimilated from σ. 33) βΰέν-νν-σθαι, found in Suidas with the meaning βΰεΐν. The rt. is βΰες, cp. Princ. i. 284. 34) εν-νν-μι. In Homer this present-form (κ 543, ζ' 28, ξ 522) is sometimes replaced by ε'ί-νν-μι (καταείννσαν Ψ 135). The latter, which is also New-Ionic (Hdt. επείννσΰαι), came by compensatory lengthening from εννυμι as είμα from the Aeol. Ημμα. There is the same incon- sistency between εννοσίγαιος and είνοσίφυλλος. With Attic writers the older form held its ground in άμφιένννμι. The rt. is Ρες (Lat. ves-ti-s). Cp. Princ. i. 470, and Leskien, Stud. ii. 85. The corresponding rt. in Skt. vas forms its present-stem in a different way. Whether in the ι 169 114 PEESENT STEMS WITH NO THEMATIC VOWEL. ch. ι v. Zend ioanh the nasal, which remains throughout, is to be compared with the syllable w I leave an open question. 35) ζέν-νυ-μι as a by-form of ζέω is only found in Alexander of Aphrodisias and Dioscorides. The rt. i&jas (Princ. i. 471), and there is no analogy for this kind of expansion of the present-stem. 36) σβίν-νν-μι. Homer's forms are all from other stems than that of the present (σβέσεν, σβέσσαι, εσβη, άσβεστος), but from Hesiod onwards σβένννμι (σβεννύω) is in general use. The ει in the gloss ζεί^υμεν σβέννυμενΉ,β^γοϊι. is of the same nature as that in ε'ίνυμι, while αποζίννυται * αποσβένονται (cp. εζίνει' άπεσβέννυεν) reminds us οϊ'ίννυμι (no. 9), and κτίννυμι (no. 38). The root is obscure (Joh. Schmidt, Ztschr. xxiii. 300, Fick, i. 3 603). β) The remaining dissyllabic present-stems in ν v. 37) ζώνννμι, from Homer onwards (ζώννυνται ω 89, ζώννυτο Κ 78, ζωννΰσκετο Ε 857). A shorter form of the present appears in ζούσθω * ζωννύσθω (Hesych.) i.e. ζο-έ-σθω. The root is apparently ζως (cp. ζωσ- τηρ), and this to be compared with jas, which is deducible from the Zend and Slavonic words given at Princ. ii. 263. The intensification of the root- vowel is like that in πηγνυμι, ϊείκνυμι. 38) κτίν-νυ-μι, post-Homeric, often with the variant κτείννυμι (Kiihner, p. 854) in good M.SS. It is clear that the more usual κτείνω 170 has helped to make confusion here. Buttmann (Ausf. Gr. ii. 227 note) is wrong in pronouncing for κτείννυμι and comparing the ει to that of ΰείκννμι, for the latter has "arisen by intensification from t, while κτίννυμι has for its root κτα {απ έκτα, κτάμεναι), κταν, κτεν (Αβοΐ. κτέννω). κτεί-νν- μι could only be a formation made from κτείνω after this verb had come into use (cp. ιζάνω). That κτίννυμι, however, is a very early form is proved by the synonymous Skt. kshanomi (Princ. i. 192) ; it must stand then for *κτέν-νυ-μι, with weakening of ε to ι as in πίτ-νη-μι and similar formations. Cp. Gust. Meyer, n. Pr. 33. 39) ρώννυμι. The forms of the present-stem are not found before Hippocrates, the Locrian Timaeus and writers of the Roman period. As present to the universally common ερρωσα, ερρωμαι,ρόψη Homer has only ρώεσθαι wave, strive, for the derivation of which from the rt. ρυ I have given reasons at length at Princ. i. 440. At the same place attention is called to the difliculties which are in the way of Kuhn's comparison of ρώννυμι with the Skt. rddh-no-mi (thrive, succeed) in respect of the meanings of the two words. We can hardly venture, therefore, to discern in this present-form an analogy to that of the Greek. Is it possible that ρω came from ρως, and is thus an expanded form of the intensified rt. ρυ ? Cp. Princ. i. 83. We have not yet arrived at a certain explanation of the vv. 40) στρωννυμι cp. στόρννμι, no. 21. We are still further here from finding a reason for the first v. Possibly the double ν owes its existence here and in ρώννυμι and in the next verb only to the analogy of ζώννυμι. 41) σώννυμι is only inferred from σωννύω with the meaning of σώζω, which is given in Bekker's Anecd. p. 114 from the Sicilian comic poet Deinolochus. The form is remarkable inasmuch as σωννύω as well as σώζω (for σω-ίζω } and hence it should be written σψζω) is a denominative" from σώος. ch. iv. NASALISED STEMS. • 115 42) τρωνννμι for the usual τιτρώσκω, only inferred from τρωνινω given by late grammarians (Lobeck, Rhem. p. 208). 43) χρωννυμι occurs first in Lucian and Themistius as a by-form of the usual χρω ζω. 44) χώίτνμι for the older χόω (Hdt., Thuc.) from the rt. χν (χεω cp. χους) seems to be not older than Hadrian's time (Arrian, Dio Cassius), 171 while χωνννω is found as early as in Polybius. It is thus in just the same case as χρωννυμι. As both verbs have a σ in the passive aorist (εχρωσθην, εχώσθην), it is possible that the first ν is a transformed sibilant. A quite isolated position is held by — 45) ολλνμι. It comes, as everybody admits, from όλ-νν-μι, and thus has experienced a forward assimilation, the reverse of the backward assimilation noticed in εν-νυ-μι. The present-stem όλλν is in general use from Homer onwards (άλλύιτά, όλλύντας, όλλυμειονς). By-form - όλεκω. The Lat. ab-ole-re is related (Fick, Ztschr. xxi. 3, Gust. Meyer n. Pr. 34). y) vw added to stems already dissyllabic. In answer to the conjectures of Kuhn (Ztschr. ii. 469) and Benfey (viii. 93), who connect the following class of verbs with Sanskrit adjec- tival stems in snu or shnu, Leskien (Stud. ii. 110) has rightly pointed out that all these forms are post-Homeric, and in some cases — like several of those treated already — of very recent date, 7 and for that reason not likely to reproduce types of primitive antiquity. It is clear, however, that the first ν in these stems stands in some relation to the σ which appears in the structure of the other tenses of all verbs of this class, κορεν-νυ-μι is related to εκορέσθην as σβένννμι is to εσβέσθην. In the case of χρωι- ννμι, and χωννυμι too, we felt compelled to see a connexion between the first ν and σ. We shall have to deal in Chap. XXI. with the growth of a sigma at the end of stems. This class is composed of the following six verbs : 46) κεράνννμι from Plato onwards. Homer's presents to aorist- forms like κέρασσε (ε 93), κεράσασα (κ 362), by the side of επι-κρη-σαι (η 164), are κερα-ί-ω (κέραιε I 203), κεράω, κέραμαι, κίρνημι (κίρνη ζ 78), κιρνάω (εκίρνα η 182). On its derivation and its connexion with κέρα- μο-ς cp. Princ. i. 181, Fick, Wtb. 204 (Skt. sam-kir-d-ti he mixes). .47) κορέινυμι. Yeitch (p. 333) cites Themistius as the oldest witness 172 to this form, while the fut. κορέω, the aorists κορέσαι (κορεσσαι), κορεσθήναι and other forms are in use from Homer onwards. The present has a by-form κορεσκω (Nicander, Hippocr.). 48) κρεμάννυμι, since Plato, while κρέμασε, κρεμάσας are as early as Homer; the middle κρεμα-μαι occurs from Homer onwards (εκρέμω Ο 21), and Pindar, Eurip. and Aristoph. have κρήμ-νη-μι. There is nothing in the origin of this verb to explain the formation (Princ. i. 190). 49) πετάνννμι since Aristophanes (Lys. 733) as present to the already long-established πέτασα, πετάσσας, πετάσθην, but generally it prefers to follow the analogy of the thematic conjugation (πεταννΰω). The by- 7 Gust. Meyer n. Pr. 37 calls attention to κατ-*-κςντάνν\ηο ( = καπκζντέίτό) and ϊττι-ιτωμάννυμι ( = 4πιιτωμάζω), remarkable coinages of a late age found by Lobeck (Rhem. 208) in late prose writers. ι 2 116 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv form πίτιημι is as early as Homer, the middle πέταμαι is first found in Pindar. The noun πέτασμα is note-worthy on account of its σ. 50) σκε 2άν> νμι : the first sure instance is in Xenophon, while the Homeric present to the fut. σκεΰώ, and εκέΰασσε (Ρ 285), έκέΰασθεν is σκίΐ-νη-μι (ΰιασκώνάΰι, κ&ναται). Hesych. has the gloss κεΰαται ' σκεΰάν- ννται. κεΰάομαι may have been formed in the same way as κεράομαι (κεράασθε γ 332). The rt. σκεΰ comes in its vowel nearest to the Skt. shhad and Wiad, but Mid is undoubtedly nearly related (Princ. i. 305), which has nasal expansion in Skt. (Jchi-nd-d-mi) and Latin (sci-n-d-o). 51) στορίνννμι has been already mentioned at nos. 21 and 40. 'χρήνννμαι, which is cited by Lobeck (Rhem. 208), has no authority whatever. It rests only on a conjecture in Theophrastus, Char, v, for χρ?) νυν αεί. c) Present-stems made by the addition of the syllable να. The 9th class of verbs in Sanskrit, which, according to Delbriick (Altind. Verbum, p. 151) includes 44 regularly inflected roots, affixes the syllable na to the root as its characteristic distinction. It is only in the 3rd plur., as a rule, that it appears unaltered, e.g. str-nd-nti (cp. Dor. πίτ-να-ντι), in the singular the na is lengthened to na, e.g. str-nd- mi (cp. Dor. πίτ-νά-μι), and in the remaining persons there appears the 173 specially Indian form n%, which is regarded as a weakening of na, e.g. 1st plur. str-ni-mds. The considerable number of such present-stems in Sanskrit is confronted by a very small number of presents in Greek — only seven — in which the syllable va really seems to make the present- stem, and to these may be added two verbs in which this syllable, like the w in several cases cited above under nos. 25 and 30, and like the isolated instance of the reduplication in the Homeric ΙιΙωσω, has become fast attached to the verb-stem for all tenses. Only two verbs (4 and 9) correspond to an Indian verb with a similarly formed present. Still there are several cases of coincidence with Sanskrit verbs which follow the 7th or the 5th class. If Bopp is right (Vergl. Gr. § 497, Schleicher Compend. 3 752) in conjecturing that the 7th class in Skt. owes the na which is inserted in the middle of its root to the transposition of this syllable from the end of the root which was its original place, that con- sequently e.g. Jchi-nd-d-mi arose out of *khid-nd-mi, the Greeks would then have preserved in σκίΐ-νη-μι the older form, and this would make the coincidence more complete still. The agreement between the lan- guages is increased moreover by the verbs with a thematic vowel like κάμνω, τέμνω, which, as will appear, arose out of verbs of the formation now under consideration. Finally the thematic verbs in ανω also dis- tinctly reveal their relationship to those of this class, inasmuch as a number of verbs of the 9th Skt. class make by-forms in -ana or -ana instead of -na, mostly imperatives (2nd sing.), e.g. ag-dna eat thou, but here and there indicatives as well, e.g. from the rt. ish fling, brandish, 2nd sing. pres. ish-nd-si, while the 2nd sing. impf. is ishana-s, 3rd sing. ishana-t, 3rd plur. mid. ishand-nta (Delbriick, p. 154). Both na and ana are common suffixes used apparently here, as was the case with the present-stems in nu, to denote the person acting. On this supposition we have in dam-anas the counterpart of the Lat. dom-inus. If we suppose a by-form damna (cp. τοϊ,ό-ΰαμνο-ς Eurip.). ?άμ-νη-μι will then ch. iv. NASALISED STEMS. 117 bear a similar relation to the forms made directly from the rt. ΰαμ (e.g. ε-ΰάμ-η-}) as that of dominari to domare. Benfey is mistaken, I think, in his attempt to make out that the forms in -nd-mi are real. denomina- tives, that is, are derived by the help of the syllable ja from nominal stems and are consequently curtailed from -na-jd-mi (Orient u. Occident, i. 425). Greek distinguishes forms like Ιάμνάται, κιρνάμεν, πίτναντο in 174 the clearest possible way from contracted forms which correspond to those of the Skt. verbs in ajdmi. I see no necessity whatever to regard forms of so antiquated a stamp as already violently reduced in length, and forms like Ζαμνάω, πιτνάω etc. on the other hand, whose general acceptance we can see to have been of gradual growth, as the older of the two kinds. The analogy of Ιιίκνυμι and ΰεικνίω also makes against Benfey's view. Schleieher Comp. 3 pp. 779, 784, 788 gives the Slavonic and Teutonic analogies to the present stems in na, in which the vowel, as in κάμνω etc., has become entirely thematic (Ch.-Sl. sta-na f Lith. ei-nti, Goth, fraih-na). t a) ν a in the present-stem only. 1) Ζάμ-νη-μι. Common from Homer onwards (ΰάμνημι Ε 893, ϋάμνησι a 100, ΰάμναται £ 488, ΰάμνατο A 309) in poetry with by-forms in -ΐ'άω (ΰαμνρ λ 221, εΐάμνα Ε 391). There is also a present δαμάζω, in Skt. ddm-jd-mi, dama-jd-mi and in the Bigveda an isolated daman- jd-mi, the latter form (cp. άλιταίνω) with a similar nasal expansion and the syllable ja as well (Princ. i. 287). 2) κίρ-νη-μι, a present to κεράσαι etc. of an older formation than κεράνιυμι under which (no. 46 in the preceding class) it has been noticed already. Attic poets used these forms, e.g. εκίρνατο, even after those in ννμι had come into general use. 3) κρημ-νη-μι, κρημ-νά-ντ-ων Pind. Pyth. 4, 25, κρημναμενάν νεφελαν Aesch. Sept. 229. Cp. no. 48 of the last class. The η of the root- syllable suggests κρημ -vc-c, between which then and κρήμνημι there is a relation like that which we supposed between *dam(a)na-s and ΰάμνψμι. 4) περ-νη-μι, poetical from Homer on wards (περ νά ς Χ 45, πέρνασκε Ω 752, περνάμενυς Σ 292). The remaining tenses have to choose between the stems πέρα (επέρασσε, περάσας) and πρα (επράβην). Another present is πιπράσκω. πρ-ία-σβαι too is clearly related. An analogous formation is the Skt. pana-te for par-na-fe, he buys, wagers=7T6p»'arat (Princ. i. 339). Hesych. has the gloss, πορ-νάμεν πωλεί»', apparently from an Aeolian dialect. 5) πίλ-να-μαι, only Epic (πίλναται Τ 93, πίλνατο Ψ 368). By-forms 175 πελάω, πελάθω. "Weakening of ε to ι as in κίρνημι and the two following verbs. 6) πίτ-νη-μι, poetical from Homer onwards, πιπάς X 392, πίτναντο Χ 402, πίτναν=.επίτνασαν Pind. Nem. 5, 11. The impf. πίτνα Φ 7 belongs to πιτνάω, cp. κίρνα, εΐάμνα. 7) σκίΰ-νη-μι has been already noticed under σκεΰάνννμι (no. 50 of the last class) as an older present. Of the Attic writers Thucydides alone appears to have admitted it : άποσκίΐνασθαι vi. 98. β) να which has firmly established itself in the verb. 8) lu-va -μαι, eommoil to all Greek (3rd plur. impf. Hdt. εΐυνέατυ) 118 PRESENT STEMS WITH NO THEMATIC VOWEL. ch. iv. and from Homer onwards with firmly established να (έΰυνήσατο, δυνατός, Ζυνασις, ΰύναμις, έΐυνήθην and έΰυνάσθην, the latter at Ψ 465, μέγα Ιϋναμίνοω only α 276, λ 414). On the analogy of all the words of this class the root syllable must be ΰυ. Is it not possible that it may be identical with that of δύω, ΰΰνω, ΰυιέω (Hdt.), and that δύνομαι is really equivalent in its original meaning to νποοϋναι, υποΰνεσθαι in the sense of subire, put oneself under an obligation, undertake Ί When Xenophon says (Oec. XIV. 3) ή και ταντην την ΰικαιοσυνην σν νποΰΰει διδάσκειν ; νποϊύει is not so very far from Ιΰνασαι. A similar suggestion was made by Damm in his Lexicon Homericum s. v. ΰΰναμαι. Delbriick, p. 79, compares the Skt. ju draw to oneself, get into one's power. 9) μάρ-να-μαι, poetical from Homer's time (μάρναο Ο 475, μαρνά- μενος, μάρνατο, μάρναντο), only in the present-stem. Hesychius has the presumably Aeolic by-form μόρ-να-μαι (cp. πορ-νά-μεν). A precisely similar formation is the Skt. mr-na-mi maim, strike dead (Princ. i. 406). Finally we ought perhaps to add βερνώμεθα' κληρωσώμεθα Αάκωνες which looks just like a conjunctive to the indie. *βερνάμεθα, and at the same place we find the enigmatic infinitive βέρρεαι' κληρώσαι, possibly a mistake for fiippai i.e. hp-ffai. The words strongly suggest the Skt. var (vr-no-mi) choose. 176 B) Stems which are always Dissyllabic. What distinguishes this division from the preceding is the fact that in the case of the stems now to be discussed the special present-formation has nothing whatever to do with making the stems dissyllabic. It is true that we. can here often arrive from the. dissyllabic stems at a monosyllabic root, but we must seek some other origin for the second syllable. What this origin is it is often hard and sometimes impossible to say. We will first enumerate the stems of this kind, arranging them according to their final letter, and try at the end if not thoroughly to explain how they come to be dissyllabic, at least to classify them accord- ing to the main features of the process by which they became so. a) Stems in a. 1) ay a, one of the commonest stems of this kind from Homer on to the time of the Attic prose- writers (άγαμαι Ψ 175, ήγάμην Plato Rep. 367 e). The present has the by-forms άγαίομαι (αγαιομένου κακά έργα ν 16), άγάζω (Aesch. Suppl. 1062), άγάομαι (οϊ τε θεαΊς άγάασθε παρ' ανίράσιν ενναζεσθαι ε 119). Buttmann Lexil. i. 236 has given a good account of the history of this stem. The fundamental notion, as in μεγαίρω (from μέγας) is rate high, esteem, hence the further meaning of marvel, wonder at, but also of envy, grudge. Throughout the remain- ing formations the second a remains short except in άγητός. This fact and the double σ in άγάσσατο Ρ 71, leadLeskien, Stud. ii. 11 3, to assume a stem άγας. Anyhow αγαμαι must be related to άγαν and άγανός, άγαυ-ρό-ς. Cp. Princ. i. 211. 2) δεα with the by-form loa, both, as I have shown at Princ. ii. 195 f., to be referred to the primary form SiFa, which in turn is to be re- garded as an expansion of the rt. div appear, shine. Ιεάμην εΐοκίμαζον, ίΐόίαζον (Hesych.), Ua-ro ζ 242, Uaroi conj. pres. on a Tegeatic inscrip- tion (above p. 61). Aor. ϊυάσσατο Ν 458, Ε 23, e 474 etc. ch. iv. DISSYLLABIC STEMS. 119 3) εα for εσα is the basis of some of the forms of the preterite of the rt. ες, for instance of the 1st sing. ?)a or έ'α (=Skt. dsa-m), ήα Ε 808, εα Δ 321, Hdt. ii. 19, whence the contracted old- Attic 1}. For the 2nd sing. Hdt. has έας (i. 187). In this way alone is it possible to account 177 for the ν of the 3r and dha-t from the rt. dhd. Moreover forms of the 3rd pi. like d-du-s must have gone through a stage in which they had a short a. Cp. Delbriick, Verb. p. 87 ff. ch. v. MONOSYLLABIC STEMS. 127 compounds, Lacon. κάβασι, i.e. κατ-βάθι), βήμεναι, βήναι, βάς. The ν'τ preserved the original quality of the vowel, even in Ionic, both in the participial stem βαντ and in the 3rd pi. *ε-βα-ντ, though it also made the vowel short : βάντ-ος ε-βάν. The same effect was produced by the ι in the optative : βα-ιή-ν. It is to be noticed that Homer varies the •quantity : εβητην Ζ 40, βητην Μ 330, Ξ 281, 285— βάτην Α 327, Ε 778, Ι 182, 192, Κ 469, εβησαν θ 343, Λ 460, Μ 16, but also υπερβασαν Μ 469. 2) ε-Ιρά-ν (Hdt. εΖρην), Homer has only the participle άπυΰράς π 65, ρ 516. The 3rd pi. άπέΰραν occurs Soph. Aj. 167, and from this time onward it was in general use. The vowel in άπείραμεν (Aristoph. fr. ii. 1155 Mein. Com.), άποΐράναι etc. is always long. 3) άπο-κλά-ς, only in Anacreon fr. 17 Β : ήρίστησα μεν Ιτρίου λεπτού 188 μικρόν άποκλάς. κλά-ς bears to the pres. κλάω just the same relation as that of ΰύς to ?ύω. 4) πλή-το, (filled oneself) in Homer and Aristophanes : Φ 16 πλητο ρόος κελάΰων έπιμίζ 'ίππων τε και ανΰρών, Φ 607 εμπληντο, Aristoph. Yesp. 1304 ένέπλητο πολλών κάγαθών, opt. εμπλΐ]το, imperat. εμπλησο, part. εμπλημενος also in Aristophanes. 2 5) κατα-πτη-την, only in Homer: θ 136 τώ& 'ίππω ΰε'ισαντε καταπτη- την υπ' οχεσφιν. The stem πτη occurs elsewhere only in the perf. part. πεπτηώτεο, the pres. πτησσω is formed from a stem πτακ. — άποπτάμενος (p. 120). 6) ε-στη-ν, like ε-βη-ν in universal use, and carried through all moods with the infin. and part., Dor. and Aeol. imperat. στάθι, Dor. inf. προστάμεν. The vowel is only short where it is succeeded by ντ, in the 3rd pi. εσταν, στάν which are Homeric (also έστησαν Ν 488) and Doric (tabb. Heracl.), and in the part, στάντ-ος, and before the ι of the opt. σταό/ΐ'. 3 7) ε-φθη-ν. λ 58 εψβης πεζός εών η εγώ συν νηί μελαίνη, Π 314 εφθη, Λ 451 ψθή, 3rd ρΐ. ψθάν Λ 51, part, υποφθάς. The long vowel is as persistent as it is in nos. 1, 2, and 6, and so we have the Attic εψθημεν, εψθησαν, the Homeric conj. φθήη, the inf. ψθήναι Hdt. and Attic. In the middle, however, we have φθάμενος (Ν 387), υποψθαμένη ο 171 in contrast to πλήμενος. b) Boots ending in an e. 8) 1st pi. εΊ-μεν etc. On the augment see above p. 85. "Without the augment: κάθ-ε-μεν ι 72, άν-ε-σαν πρό-ε-σαν 3 681, άψ-έ-την Λ 642, opt. άψ-ε-ίη, imperat. πρό-ες, πρυ-έ-τω, Έυν-ες, part, υφ-έ-ντ-ες, inf. μεθ-έ- μεν } mid. ε- ντο, ΐ,υν-ε-το,. all Homeric, imperat. εί-ε-ο Hdt. There are many forms of this kind in compounds in Attic. Except the epic conj. rjrj, άνήτι along with εφείω etc., of which we shall have to speak later, and the inf. είναι,. where also the diphthong needs a closer examination, the vowel is short throughout. 9) 1st pi. ε-θε-μεν etc., common to all Greek, 3rd pi. άν-έ-θιν C. I. 189 29 (Argolic), mid. ί-θέ-μην. The same may be said of the quantity of the vowel as was said of that of no. 8. 2 Ι-πτα-το, απο-πτά-μ*νος etc. have been given already — under no. 10 on p. Π9. I cannot accept Joh. Schmidt's attempted identification (Ztschr. xxiii. 300) of the rts. πτα crouch and πτα fly. 3 We might add here the reduplicated aorist *4-φη- ίφάνη Hesych. Cp. the Skt. bha-ti he appears. 128 AORIST-STEMS WHICH HAVE NO THEMATIC VOWEL. ch. v. 10) ε-σβη-ν. ε-σβη I 471, inf. κατα-σβηναι and the like (Hippocr. άποσβείς) from Herodotus onwards. The vowel is long wherever it can be so (cp. perf. ε-σβη-κα). If, as is made probable by σβέν-νυ-μι, the Homeric σβέσ-σαι (ε-σβε-(Τα), σβ(<τ-σω (σβέ-σω), and the Homeric α-ΰβεσ- τος, the root originally ended in a sibilant (cp. Princ. ii. 197), we should have to suppose compensatory lengthening here. Doric has the α :* άπέσβα Theocr. 4, 39 (cp. Joh. Schmidt, Ztschr. xxiii. 300). 11) 2nd sing, imperat. ^οε'-ς-, Com. anon. Meineke, iv. 651 ένθα σταθείς φρές μ' ώς το μειρακνλλιον, φοές το φέρε Herodian, i. 463, 14 Lentz, part, έπειςφρείς Eurip. (Phaeth.) fr. 781, 46 Dind. μή riv' "ΙΙφαιστος χόλον ΰόμοις επειςφρείς μέλαθρα συμφΧέζτ) πυρί, inf. είςφρήναι ' εΐξάϊ,αι, ενεγκεΊν Hesych. Nauck, ' das Verbum φρέω ' Bulletin de l'Acad. Imp. de St. Potersb. Tome vi. p. 424 ff., tries to prove that the stem φρε is nothing else than a compound of προ and ε, and that con- sequently φρές is the same as προές etc. But ingenious as are his argu- ments it is none the less impossible to find a precedent for the expulsion of the ο before ε. Moreover we have already on p. 107 seen that the present πι-φρά-ναι is a well-attested form. I follow Herodian in regard- ing the stem φρε as identical with φερ, and I believe that φρές bears to φέρε precisely the same relation as that of σγέ ς 4 to εχε, σπές to (ϊνν)επε. The verbs έκψέρειν, εϊςφέρειν are by no means far removed in their usage from the forms under consideration. The compounds of the Skt. bhar anu-bhar, ava-bhar mean bring in, introduce (e.g. into the body), just like εςπιφράναι and είςφρεϊν. The same root which in Skt. forms at one time bhar a-ti— φέρει ) at another bhar-ti, as it were *φερ-τι, has survived in both forms in Greek : φερ with a thematic vowel and φρε without one, and the latter became an aorist like σχε. From φρε again was 190 formed φρέω. As the connexion of φρές with φέρω was lost sight of there arose a fresh aorist of a different kind. Cp. Stud. viii. 327 ff. c) Eoots ending in o. 12) 'i-yvoj-v common to all Greek from Homer onwards (ε-γνω-σαν κ 397, γνωτηΐ' φ 36, γιώμεναι, γνώνάι). The vowel is always long when it is possible, ο occurs only in γνο-ίη-ν, γνονς, γνύντος and the 3rd pi. εγνον, which is demanded by the metre at Pind. Pyth. 4, 120, and has been adopted elsewhere in Pindar by Ahrens and others as against the variant lyviuv (Isthm. 2, 23, Pyth. 9, 79), while the latter form, which anyhow stands isolated among these primitive aorists, holds a place to which it probably has no right, in hymn, in Cerer. Ill ovV έγνων χαλεποί 2e θεοί θνητοΊσιν οράαθαι (cp. Nauck Molanges, iv. 25 f.) — There is, accord- ing to Veitch, only one instance of a corresponding middle form συγγ νοϊτο Aesch. Suppl. 216. 13) ε-Ιο-μεν etc., in general use from Homer onwards. 3rd pi. l-lo-v Hes. Theog. 30, also Laconic C. I. no. 1511, middle ε-ΰό-μην etc. The vowel is short throughout. The Aeolic πώ-θι will be noticed as a by-form under πΊθι. — The word εΐέτρω (έζεβλάβη, έζεκόπη ?/ κΰησις) only found in the Ε. M. 347, 48, and 4 Joh. Schmidt's assertion {Ztschr. xxiii, 301), that the s of ο•χ«, in a middle sense and therefore coming ch. υ. MONOSYLLABIC STEMS. 131 nearest to ήΖομαι. The σ is remarkable, as the analogy of ϊΰμεν, όΰμή would make us expect the radical δ in Homer. In κεκασ μένος too, how- ever, the sibilant appears in contrast to the Pindaric κεκαομένος. 30) ykv -το he took hold of, only in Homer: γέντο δε Ζονρε Ν 241, ■γέντο V ιμάσθλην θ 43 (Ν 25), yivro Ζε χειρι ραιστήρα κρατερήν, ετέρηφι ie γέντο πυράγρην Σ 476, from a rt. γεμ, which appears most clearly in Hesychius's glosses άπό-γεμ-ε' άφελκε and σύγγεμος' συλλα/3»/. — Cp. yivro he became under no. 25. 31) ΰέκτο he received Β 420 αλλ' ο γε ΰέκτο μεν Ιρά, Ο 88 ΰέκτο ΰέπας, Ιπέΰεκτο he undertook I 480, ϊέζο receive Τ 10, δέχθαι Α 23, Pind. 01. 2, 49 γέρας εΰεκτο, while εΐέγμην ι 513, μ 230 and the part, λεγμένος belong to the meaning await which will have to be discussed in connexion with ΰέΰεγμαι, ΐέχαται when we come to the perfect. 32) έλέλικτο belonging to έλελίζειν whirl round (cp. έλελιξάμενος) A 39, Ν 558 cp. Buttm. Lexil. i. 138, Fick, Ztschr. xix. 252. 33) ευκτο only in the fragment of the Cyclic Thebais quoted by the scholiast on Soph. Oed. Col. 1375 : ενκτο Δα (3ασιληι και άλλοις άθανά- τοισιν. 34) ίκτο only Hes. Theog. 481 ένθα μεν Ικτο (Koechly Ίκτο), but the part, 'ίκμενος ονρος Α 479 and elsewhere, literally ' that has come ' (cf. * welcome'), an apt epithet for a favourable wind that has at last come in answer to the prayers of the sailors. Ahrens's connexion of ίκμενος with εοικα seems less natural. 35) κεντο only Alcman fr. 141 with Doric ν for λ, belonging to 194 κέλομαι. . 36) λεκτο he counted δ 451 λεκτο h' αριθμόν, a middle to the impf. λέγε in line 452, and there is also πέμπτος έλέγμην ι 335 with a passive meaning. 37) λεκτο he laid himself (rt. λεχ, λέχος) δ 453 έπειτα ΰέ λέκτο και αυτός, κάτέλεκτο Ι 662, ν 75, παρκατέλεκτο Ι 565, 664, προςέλεκτο μ 34, with the meaning of an imperfect in Hesiod Scut. 46 πανννγιος V ap ελεκτο. Also the imperat. λέξο Ω 650 (but at I 617, κ 320 λέζεο from the sigmatic aorist), καταλέχθαι ο 394, καταλέγ μένος λ 62, χ 196. 38) μ-ικτο Λ 354, Π 813, εμικτο α 433. ' 39) op -σο Δ 204, Ε 109 (but ορσεο Γ 250 etc.), ώρτο Ε 590, Hesiod Theog. 990 etc., also επώρτο, ένώρτο, ορθαι θ 474, ορμενος Λ 572, πάλιν όρμένω Λ 326, and certain of these forms occur also in Pindar and the tragedians. 40) πάλ-το Ο 645, εκτταλΓ' Υ 483, ανέπαλτο θ 85, εκκαΓέτταλτο Τ 351. — There is what seems to be an active to this form metathesised in Hesychius's gloss εΐέπλη' έΐέπεσεν, formed like ζνμβλήτην. Or can έττλ'τ; be identical with επλε (Μ 11) which counts for an imperfect, and so belong to the rt. ττελ I 41) πέρθαι for περθ-σθαι Π 708. 42) πηκτό, κατέπηκτο Λ 378. Forms which only occur in late poets, such as ελειπτο Apoll. Rhod. I 45 ου Ιέ μεν "Ιφικλής Φυλάκη ενι ΰηρον ελειπτο, and άμειπτο Νοηη. Dion. xliv. 241, and which are condemned at once by their diphthong as wrong forms, or rather as mistaken imitations of Homeric forms which were not understood, have not been admitted into this list. Buttmann i. 318 regards them as pluperfects without reduplication. κ 2 132 AORIST-STEMS WHICH HAVE NO THEMATIC VOWEL, ch. v. B) Metathesised Aorists. 43) Ιυμ-βλη-την φ 15 τω l• 1 εν Μεσσήντ) ζνμβλήτην άλλήλοιιν, ε-βλη-το by the side of βάλλω, εβαλον Λ 410, βλητο Δ 518, ξύμβλητο 3 39, with the conj. βλήεται, opt. 2nd sing, βλείο (Bekk. βλί/ο), inf. βλήσθαι, part. βλήμενος, ζνμβλήμενος. 44) πλή-το & 438 αντις δ' εϊ,οπίσω πλήτο χθονί, επληντο Δ 449, θ 16, πλήντο & 468 by the side of πελάζω, έπέλασεν, πελάω. πλήνται with present meaning, Parmen. v. 8 is doubtful (Stein Symbola Philol. Bonn, p. 803). So too πλη-σίον by the side of πέλας. Cp. Princ. i. 345. 195 45) άπο-σκλή-ναι, Aristoph. Vesp. 160, άποσκλαίη' άποξημαίνοιτο, άποθάνοι (Hesych.), by the side of σκέλλω. Other similar forms, also a κατασκλήναι,Ήΐ late prose (Nauck Melanges iv. 26). 46) ενί-σπε-ς Λ 186, Ξ 470 by the side of ένίσπε, from the rt. σεπ, which occurs without syncope in εν-νεπ-ε for εν-σεπ-ε. Buttmann (Ausf. Gr. ii. 168), who finds a supporter in La Boche (Horn. Textkritik 256), will not allow the form, because ενί-σπε-ς can only be explained as a compound, while he wants to establish that έννέπω with all belonging to it is nothing but a simple verb, an idea which, in the face of the Latin in-sec-e and the fut. ενι-σπη-σω (ε 98), I find it impossible to accept. It is true that ενί-σπε-ς only occurs at the end of a line, and that the only other place, except in Apollon. Bhod., in which it occurs, is a strophe of Sappho which from its subject is rightly held to be spurious (Bergk 3 26), but in its formation, which the best of the old grammarians rightly compared with έπί-σχε-ς there is nothing remarkable whatever (Herodian ed. Lentz i. 467, 25, ii. 127, 29, 137, 10). The indicative ένί-σπε, which Herodian is careful to distinguish from the imperative, bearstoij'iW7r£-c exactly the same relation that ε-σχε does to σχέ-ς. The by-form for the imperative ενί-σπε is naturally to be explained in the same way, as also are παρά-σχε and the like which often occur in M.SS. but are generally rejected by editors (Veitch, p. 257). That is the ε which from the beginning belonged to the root has followed the analogy of the thematic vowel. That the old grammarians should have ' invented' a form like ένί-σπε-ς seems to me altogether unlikely. 47) σχέ-ς is related to the rt. σεχ in exactly the same way as σπέ-ς to the rt. σεπ and is as isolated a form as σπέ-ς, for all the other aorist- forms like ε-σχον, σχοίη-ν, σχεΊν (σχέ-μεν θ 254) etc. follow the analog} 7 of the thematic conjugation. There is all the less reason then to chal- lenge the isolated παρά-σχε κατά-σχε, noticed under no. 46, out of sheer purism, when there is good evidence for them. In the accentuation of this form I follow Gottling Allg. Lehre vom gr. Accent, p. 45. 48) ε-τλη-ν (Dor. έ-Γλά -y) common in poetry from Homer onwards, 1st pi. τλή-μεν, 3rd pi. ε-τλά-ν, opt. τλα-ίη-ν, imp. τλή-τω, τλή-τε all Homeric, τλή-θι, τλή-ναι, part, τλά-ς in the tragedians. Homer however 196 has πολν-τλα-ς which, like τάλά-ς, is rather regarded as an adjective. There is no present in use as a verbal form, but 7-άλά-ς like τάλ-αντ-ο-ν, τάλ-αρο-ς, ταλα-ό-ς (Princ. i. 272), points to ταλά-ω from which it was originally a regular participle of the Aeolic type. The epic εΥάλασσα (cp. επέλασσα) likewise points to ταλ, while τέ-τλά-μεν, τέ-τλη-κα come from the metathesised τλα. 49) ε-βλω - εψάνη, ψχετο, 'έστη (Hesych.) along with the participle άγχιβλώς' άρτι παρών has already been placed by Lobeck on Buttm. ii. ch. v. DISSYLLABIC STEMS. 133 12 in the list of aorists of the old type and connected with ε-μολ-ο-ν, μέ-μβλω-κα. 50) ε-βρω' εφαγεν, εΰιικε, Ζιέσπασεν Hesych., and therefore belonging to the rt. βορ, βιβρώσκω. Hymn, in Apoll. 127 κατέβρως αμβροτον εϊΰαρ, Callim. hymn, in Jovem 49 επι δε γλυκύ κηρίον εβρως. ο) The following form stands quite alone : 51) ε-πλω-ν επέπλων Hes. "Ejoy. 650, επ-έ-πλως y 15, άπ-έ-πλω £ 339, παρ-έ-πλω μ 69, part, έπιπλώς Ζ 291. As πλώω has come from *7τλέ/ ; ω, and the ω can hardly be otherwise explained than from of, it may be said that a defaced consonant stem forms the basis of ε-πλω-ν. Whether the case is the same with ε-τρω mentioned on p. 128 after no. 13 (cp. τραύ-ω) I cannot decide. Π. DISSYLLABIC AOMST-STEMS. Besides these 51 aorists, which must be mostly of a primitive formation, there are a number of dissyllabic forms, akin to them in inflexion and use, which clearly only owe their position as aorists in the verbal system to the fact that the corresponding present stems have base-forms differing from theirs. Properly speaking these dissyllabic aorists belong as little to the above-mentioned monosyllabic ones as do the contracted verbs of the Aeolic inflexion to the verbs in μι. Since however the said forms are of various kinds, are occasionally of obscure formation, and only resemble each other in this, that they are to be referred to dissyllabic stems of aoristic force, and that they have no 197 thematic vowel, it seemed best to give them their place here. They are as follow : 52) Ι-αλω-ν, in common use from Homer onward (ind. ήλων cp. p. 79) with the conj. άλώω contr. ά\ώ, opt. άλωην later άλοίην, inf. άλώναι, part, άλονς, almost exactly the same as ε-γνω -v- with its moods. The inchoative άλ-ίσκο-μαι does duty as present. Cp. Princ. ii. 170. ε-άλω-ν is apparently a contracted preterite from the stem άλω, inflected like the Aeol. ε-ΰοκίμω-ν from the stem Ιοκιμω. Accordingly the vowel is long wherever it is possible. 53) εζ-ήμβλω a late form, not occurring before Themistius, instead of the older ημβλωσε. It is quite enough to show us that the power of making forms of an old-fashioned stamp on the analogy of old forms lasted a long time, εξαμβλοϋμεν in Eurip. Androm. 356 is a present. 54) άρπά-μενος not earlier than the poets of the Anthology : υψαρπα- μένη (active) Agathias ix. 619, άρπαμένης (passive) Macedonius xi. 59, evidently therefore like Ιστά-μενος from ιστα, one of the many evidences to the relationship between the verbs in αω and those in αζω. 55) άπ-ούμα-ς. ελών yap έχει γέρας αυτός άπούρας Α 356 and often elsewhere in Homer. Pindar too has άπούραις (Pyth. 4, 149). Ahrens's view of this form now finds considerable and deserved acceptance. It is that its stem is really dissyllabic only in appearance, its ultimate form being the monosyllabic fpa, the relationship of which to the rt. hp has been discussed at Princ. i. 431 (cp. Sonne Ztschr. xiii. 434). The only representative of the middle is άπουράμενος Hes. Scut. 173 : άπουράμενοι ■φυχάς. In all the forms then, as in εναδε, καυάζαις and other forms of 1 34 AORIST-STEMS WHICH HAVE NO THEMATIC VOWEL, ch. v. the kind, the f has turned to a vowel. We may add to them the fut* άπουρήσω, which is the reading of some good M.SS. at X 489, though the most and the best M.SS. have άπονρίσσουσιν from άπ-ουρίζω=άφυρίζω~ άπονρήσονσι suits the sense far better, and is adopted by Buttmann and I. Bekker. There is no hint of an *άπαυράω. Following Sonne, Brug- man, at Stud. iv. 166, conjectures that the η has been lengthened from ε by the influence of the /, and that άπηνρα came from *άπεϊρα, which would bear the same relation to άπονρας as does άπέΰρα to άποΰράς. It is hardly probable however that the / should have had the power of 198 lengthening the vowel after it had become v. As regards the accent of the participle Westphal, Formenl. i. 2, 285, is perhaps right in sup- posing that it ought by rights to be on the final syllable. 56) ε-βίω-ν. Homer has βιώτω θ 429, βιώναι Κ 174. The remain- ing forms are in common use in Attic : conj. βιίΰ,. opt. βιωην, part. βιονς. Apparently ίβίων, like εάλων, is the preterite of a stem which has nothing aoristic about it, so that there is no sort of inherent and original difference of stem between these aorist-forms and the present- forms βιοΊ, βίων, βιοϋν. The present is moreover generally supplied by ζάω. 57) ε-γήρά-ν, from Homer's time (εγήρα Η 148, γηράς Ρ 197) in very general μββ : inf. γηράναι, hardly γηράναι, which is sometimes written, and is maintained by Cobet, Mnemos. xi. 124. Dat. part, γηράντεσσι Hes. Opp. 188. In its formation this stem is just as much a contracted one and just as little of an aorist as άλω, βιω. The present to it is γηράσκω. 58) 3rd sing, ουτα Ε 376, Ζ 64, inf. οντάμεναι Φ 68 οντάμεναι μεμαώς, and also ουτάμεν, mid. part, οντάμενος, κατ ονταμένην ώτειλήν. — νε-οντα- τος (Σ 536) by the side of άοντος. There is a present ουτάω (ονταε χαλκά χ 356) with οντησε, οντήθη, and an οντάζω Υ 459 (cp. Hesych. βωτάζειν βάΧΚειν) with ουτασε Ο 528, ουτασται Λ 661, ουτασμένος λ 536. Forms of the latter kind are not unknown to the tragedians. The stem is a very peculiar one. οντά suggests έκτα, and this would point to a final ν, ουτά- μενος might be compared with κτ ά μένος, or on the other hand with the late αρπάμενος. The formation of the presents evidently points to an *ovraju). The only related Greek words «are ώτειλή (Hesych. γατειλαί* ονλαί). Conjectures have been made as to related words -in other lan- guages by Fick Wtb. i. 3 769. 59) ε-πριά-μην has been already discussed on p. 120 no. 11. 60) ώνη-μην, represented in Homer by the forms άπόνητο Ρ 25 and elsewhere, άπόναιο Ω 556, imperat. ονησο τ 68, όνημενος β 33. — In later poets (Theogn. Eurip.) ώνημην (by the side of ωνασθε Eurip. Here. f. 1368) ώνήμεθα, in both poetry and prose όναίμην, ovaadai. The quantity of the vowel is thus a varying one. ova bears a regular relation to 6vtva κτιζο κταν (26) >> κτεινο άλ (27) J> άλλο eXeXiy (32) »> ζ\ς\ίζο παλ (40) J> παΚΚο βλα (43) >» βαλλο σκλη (45) »> σκ(ΧΚο 8 in all. Moreover the rt. yav, yev vacillates between the first and third classes. Perhaps too ΰυ (17), λυ (20), κλν (19), ψν (23) belong to this 3rd class by reason of their vacillating quantity — a question we shall have to consider later. But a small number of the stems form their present thematically with an addition to the vowel-sound — only the following 4 : πνν πνεο (21), συ σευο (22), χν χεο (24), άΰ ηΰο (29), for the last of which there is άνΰανο as well. There are 7 presents which are characterised by the thematic vowel alone, and consequently belong to the 1st class of thematic verbs : $εκ ΰεκο or ΰεχο (31), ενχ ενχο (33), λεγ Χεγο (36), περθ περθο (41), σεπ σεπο (46), κελ κελο (35), σεχ σεχο (47). For the last there is l -σχο as well. πλη (44) and πελαζο, τλα (43) and ταλαο stand by themselves, and 203 γεν (30), λεχ (37) had no present-stem at all. The conclusion then to which this investigation brings us is, that beyond a certain preference discernible for the reduplicating method, there is no fixed principle which can be said to have ruled the formation of the presents. We shall often make this same discovery later on, and we may formulate it thus : the tense-stems of the Greek verb are by no means mutually boimd together by fixed and pervading analogies, but they combine in the freest manner to form a whole, to form, that is, a single system of verbal forms. 138 UNEXPANDED THEMATIC PRESENTS. CHAPTER VI. THEMATIC PRESENTS FORMED WITHOUT ANY FURTHER STRENGTHENING OF THE βΤΕΜ. In the introduction to this book we found that the vowel which distin- guishes τί-ο-μεν, τί-ε-τε from 'ί-μεν, ί-τε is an element belonging to the present-stem of a great number of verbs. At p. 9 we attempted to discover its significance as an element in verbal structure, and the name we give it as the result of this investigation is thematic. Contrasted with other vowels which in dealing with other verbal forms we have seen occasionally arise and make monosyllabic roots into dissyllabic stems, the vowel which we name thematic κατ εξοχήν is an α -sound which from the first was prevailingly short, but was occasionally lengthened, and was subjected in every Indo-Germanic language to the most definite laws both as to its quantity and its quality. In Sanskrit the only change it undergoes is one of quantity. The a is short; as a rule, and only long before m and ν : Sing, bhar-a-mi bhar-a-si. bhar-a-ti. PI. bhar-a-mas bhar-a-tha bhar-a-nti. Dual bhar-d-va bhar-a-thas bhar-a-tas. 204 This change of quantity can, as Bopp saw (Vgl. Gr. ii. 2 290), hardly have been primitive. In the 1st pi. and 1st du. the long vowel appears only in Sanskrit and the nearly related Zend. All the other languages leave the vowel short in the plural ; ψερ-ο-μες, fer-i-mus, Goth, bair-a-m, Ch.-Sl. ber-e-mU, so do the Lithuanians and Slavonians in the dual ; Lith. vez-a-va, Ch.-Sl. vez-e-ve. In the 1st sing., as we showed on p.29f., the length of the vowel in φέρω and the Lat. fero has to do with the loss of the termination. It would be a remarkable thing that in this one form the vowel should have been lengthened for no discernible reason before the personal termination fell away. For this reason we adopted Ascoli's thoroughly established assumption that we must start from a primary *bhar-a-mi which in Greek was once *φέρ-ο-μι. In respect of the quality of the thematic vowel, we have to notice a regular interchange of e and o. In this respect Greek and Latin almost entirely coincide in the ind. pres. act. We may assume the primary forms to have been — *Xeyo-/xi lego-m(i). λέγο-μ^ς lego-mas (?). *\iyt-vi lege-s(i). \cy€-re(s) lege-tes. *\iye-ri lege-t(i). Xeyo-irt lego-nt(i) (cp. p. 46 f.). The e-sound then shows itself before 8 and t y and in this Gothic (vigis, vigith, vigith) and Church-Slavonic (veze-$i> veze-ti, veze-te) coincide as well, the o-sonnd before nasals, while in all cases the vowel remains an a ch. τι. UNEXPANDED THEMATIC PRESENTS. 139 with the Indians and Iranians. The same is the case with the two dual- forms λέγετον, έλεγέτην, with the forms of the preterite έλεγες, ελεγε(Γ) and ελεγον as 1st sing, and 3rd pi. The only, doubt that could arise is with reference to the 1st plur. in Latin, since the prevailing vowel here is i. It might be thought, in the face of the Skt. vdhdmas, Goth vigam, Lith. vezame, Ch.-Sl. vezemii, that the Lat. vehi-mus came from vaha-mas by way of the intermediate stage *vehe-mus, and that it never went through the stage *veho-mus suggested by the analogy of λέγομες at all. But it can hardly be accidental that volu-mus, su-mus, quaesu-mus have a u before the termination. It is probable rather, considering the tendency shown in historical times to attenuate a u before m in the middle of a word to i, that we ought to assume a *vehu-mus, and a still older *veho- 205 mus, or perhaps even *veho-mas, for it is hard to say what the vowel of the final syllable was at that time. The relation of vehimus to εχομες would then be the same as that of Septimus to εβίομος. We get the same simple rule for the Greek middle voice from a comparison of — φίρο-μαι with *φίρε-σαι. εφέρο-ντο with φερε-σθε. ε-φερό-μην" „ *ε-φερε-σο. φερό-μενος „ φερε-σθον. φερό-μεθα „ φερε-ται. ε-φερε-σθην φε ρο-νται ,, ε-φερε-το. and φερε-σθαι. Here again the participle φερόμενοι is confronted by the Lat. ferimihi, for which, as for vehimus, we are inclined with equal probability to assume an older form with a heavier vowel, on the ground of forms like alu-mnu-s vertu-mnu-s. Attention should be paid to the deviation from this rule that ο comes before a nasal shown by the Homeric infini- tive active έλθεμεναι, είπέμεν, by which means the fine sense of the Greek tongue was able to distinguish in the easiest way between active infini- tives and middle participles. It is true that the forms we have here assumed as primary are not always evident at first. In the 2nd and 3rd sing, an ι has become attached to the thematic ε. What is the relation of — λέγεις to the Lat. legis and Skt. vdha-si and of λέγει to the Lat. legit and Skt. vaha-ti ? This is a question to which we must now give the answer we deferred to give when dealing with the personal terminations. As the forms of the various Greek dialects have an importance for the proper settlement of the question, we will first give them as far as we can. For the 2nd sing, there is good testimony for only the Doric by-form in ες, i.e. in συρίσίες Theocr. 1, 3, άμέλγες 4, 3, and that of Apollonius περί άντωνυμ. 119 to ποιες. In the last word the widespread tendency of the Dorians to shorten final syllables in spite of contraction that has taken place and consonants that have been lost (ίαρές —Att. Ιερείς, ΐεσπότάς from ϊεσπότάνς) has actually left only ες out of the original εεσι. Other branches of Doric, if we may judge from Aristoph. Lysistr. 206 180, 1013, had the Attic form. — It is established by Anecd. Oxon. i. 71, 22 that the Boeotians pronounced it λεγις (Ahrens, Aeol. 189). On the other hand we cannot be sure about the Lesbian ης which the M.SS. sometimes give, in particular at Alcaeus fr. 52 Βε. 3 : πωνης, which Ahrens (Aeol. 91) is most likely right in rejecting, though Bergk (Lyr. 3 p. 931) again defends the ης on the ground of an obscure and ve'ry 140 UNEXPANDED THEMATIC PRESENTS. ch. vi. corrupt passage in Apollonius Dysc. Bergk regards both η and ει as lengthenings compensatory for the loss of the final ι, in which case the series would be ισι ες ης, εις. How untenable this assumption is we shall try to show later on. For the 3rd sing, ει was Doric as well, and is shown by ν ει (Ale. 34, 1) to have been Lesbio-Aeolic, so that άΰικήη=άΰιι;ήει, an isolated form occurring at Sappho I, 20, cannot be trusted, and has been rightly exchanged for άΰικήει. Lastly the Boeotians pronounced it λέγι. These dialectic forms prove conclusively that the primary Greek forms were — *λέγεσι *\eyeri. λέγεσι lost its ι and became λεγες as *τίθη-σι became τίθης. The Doric λέγες of which the final syllable has lost the mark of a primary tense, exactly corresponds to the Lat. legis and the Goth, vigis. The Dorians however preserve a trace of the lost ι in the accent, provided the tradi- tion is to be trusted which gives us άμέλγες, συρίσοες. λέγεις on the other hand, though it likewise has lost its final ι, did nq,t lose it, as Bopp saw, till epenthesis had taken place. Midway then between *λέγεσι and λέγεις came *λέγεισι. In the epenthesis or anticipatory sound (Vork- lanf) we have the same process by which ενί became είνί and ειν (Princ. ii. 334 ff.). This explanation has it is true been attacked by Bergk, who maintains (Philol. xxix. p. 319) that in the old alphabet the Greeks wrote ΔΟΚΕΣ, and since then Bergk has actually called attention (in Fleckeisen's Jahrb. 1878, p. 190) to a form ΚΑΛΕΔΟΚΕΣ in an inscrip- tion on a vase found at Locri, in Southern Italy, and already edited at Corp. Inscr. Graec. iii. no. 5770. But so isolated a spelling, occurring in a private inscription, is no foundation for the explanation of a form, especially as during the period immediately preceding the introduction of the new alphabet into Attica — and this is in Bergk's view the date of the vase — the letters Ε and EI, representing, as they did, sounds which had then at all events become very much like each other, were often interchanged (Cauer, Stud. viii. 230). Moreover, the explanation that 207 the ει owes its existence merely to the ' endeavour to secure compensa- tion' — 'ut damnum resarcirent' Bergk says in the note on the Lyric poets quoted above — is one which utterly fails to content us, inasmuch as we now regard compensatory lengthening not so much as of an external mechanical nature, and can assign more of a definite limit to its action by laying it down that a vowel is never made long except by the influence of consonants directly following it which have been gradually fading away (Brugman de productione suppletoria Stud. iv.). An utterly different theory as to the relationship of these forms has been propounded by Corssen after Alb. Dietrich (Ztschr. f. Alterthw. 1847, p. 710 ff.) in his work on the Pronunciation of Latin, i. 2 p. 600 ff. He believes that in the Gk. forms in εις, ει we have parallels to the forms ecribls, aglt which, though very rare and therefore impugned by some, and otherwise explained by others, do occur in Latin, and since it is impossible to explain these Latin forms by epenthesis — for Latin knows nothing of epenthesis — he regards the long syllable in both languages as the result of ' intensification.' There are however good grounds to be given for not adopting this view. In the first place, there is no reason why the thematic vowel should be intensified at all. It is certain that ch. vi. TJNEXPANDED THEMATIC PKESENTS. 141 of all the formal elements in the verb it is that which has least signifi- cance, and for this reason we have regarded its lengthening, when, as in d-mi d-mas, Gk. and Lat. δ, it has actually taken place, as by no means an intensification due to the effort made by language to emphasise par- ticular syllables, but as a mechanical lengthening, one, that is, which has been brought about by the influence of the succeeding consonants. Conscious that ει merely viewed phonetically would even in Attic be a remarkable intensification of ε, Corssen has recourse to Doric and Aeolic. But there is nothing to be gained by the production of what he calls the Doric and Aeolic η, for, as we have seen, the η in these persons has but a slender authority. In the 3rd sing, we actually find a Doric ει abundantly established by inscriptions, and this strict-Doric ει of the Heracleic tables (e.g. έψορεύει tab. Heracl. i. 122, αποτεισεϊ ib. 109, φντενσεϊ 114, έμβαλεΊ 115, εζεϊ 130), as well as forms like βρέμει, σίζει, φέρει in Epicharmus would anyway be incomprehensible as an intensifica- 208 tion of an ε. 1 Then the assertion (p. 602) that ' there are other instances in the Attic dialect of the appearance of an ει where the Doric and Aeolic dialects have η, as the regular intensification or lengthening of ε ' is decidedly incorrect. Intensification of an ε produces η in all Greek dialects except Boeotian, which shifts every η to ει — e.g. rt. μελ perf. μέ-μηλε, rt. έδ έ£-ήΰ-ο-κα. The Homeric forms θεης, θείγ for θέτ)ς, θέη, which Corssen brings forward, prove nothing, for we shall see below what slender support these forms have. In θείς, τιθείς to which he also appeals, ει has arisen by compensatory lengthening, and hence Argolic has -εν ς, Heracleic -ης. And granting that there were cases in Homeric Ionic in which ει had taken the position of an older η which arose by intensification, this ει would be absolutely inconceivable in strict Doric, for it is one of the many characteristics of this dialect to hold fast by an η even when it has arisen from compensatory lengthening or contraction, and all the more when it has arisen by intensification. We dealt on p. 40 f. with the 3rd persons sing, in -ησι which are properly speaking not ' Doric ' but are exclusively peculiar to Ibycus, and which are at •variance with all the testimony of inscriptions as to this personal ter- mination. The Boeotian λέγϊς λέγΐ do not suit Corssen's hypothesis either. Where could an Z, which the Boeotians substitute only for a truly diphthongal ει, be an intensification of an ε 1 It is ει that is the Boeotian representative of a primitive Greek η. There still remains a conclusive argument against Corssen's view. If we are not to suppose that the ει of λέγεις, λέγει arose from an echo of the ι which once was in the final syllable, what explanation is to be given of the η in the con- junctives λέγης λέγη 1 I do not think anyone will say that η has been 209 'intensified' to rj. The ι here is so firmly established, e.g. in the Heracleic dialect, that besides νέμη, φέρη, λάβη there are also instances 1 The warm defence of his theory which Corssen has put into his Beitrage zur italisclien SprachJtunde (p. 484 fE.) contains no fresh grounds for it. The sug- gestion that the et which appears at this place among all Dorians with the single exception of the above-mentioned forms in -ησι used by Ibycus — which moreover are suspicious by reason of their σ — is due to < Attic influence,' of which there is (e.g.) in Epicharmus not the faintest trace besides, will convince no one. The statement that tyopeva on the Heracl. tables is ' completely isolated' has no foundation. It is quite a mistake to call ϊθεικς an • Aeolic,' i.e. a Lesbio- Aeolic form. It is Boeotian. The 3rd pi. ϊεισι in Sappho he has confounded with the 3rd sing. etc. 142 UNEXPANDED THEMATIC PKESENTS. ch. vt. of conjunctives with what is apparently y shortened to εί, like γράφει, αποθάνει, which serve as confirmation for the indicatives. 2 There is no other possible explanation of this ι except that it is an anticipatory sound, an echo of the following t, and when we consider that λίγης corresponds to λέγεις, λέγη to λέγει precisely as Χέγητε does to λέγετε, Χέγησθε to λέγεσθε, it cannot be doubted that the ει of λέγεις, λέγει arose in the same way as the rj in λέγης, λέγη. This brings us to an explanation of the 3rd sing, which differs from that which we have adopted. Schleicher, Comp. § 275, assumes, in his account of the change from *φέρε-τι, φέρη-τι to φέρει, φέρη an intermediate *φέρε-σι φέρη-σι. This hypothesis, again, can be easily refuted by a refer- ence to the dialects. It is only the Ionic dialect that shows the weakening of r before ι to σ. Here then a *φερε-σι is conceivable, and φέρη-σι and the like do occur over and over, again in Homeric Greek. But we must take quite another way to account for the Doric forms. How are we to iihink that the same stems which produced φα-τί, ϊσα-τι and the like were faithless to their r here 1 Consequently Schleicher's intermediate *φέρεσι *φέρησι cannot be allowed for the Doric φέρει, <ρέρη. φέρησι is an old form which is specifically Homeric and came directly from φέρητι. In other cases the final ι, after it had exercised its influence upon the preceding syllable, disappeared, so that the series was as follows : — *φέρε-σι *φέρη-σι *φέρς-τι *φέρη-τι J I J J *φέρ€ΐ-σι *φέρτ]-σι *φερ€ΐ-τι *φ4ρτ}-τι φέρ€ΐ-$ Φ*ΡΏ~* *φ4ρ€ΐ-τ *φέρτ]-τ φέρει Φ € 'ρν• '10 This explanation accounts completely for all the sounds in the various forms. The assumed *φέρειτι corresponds exactly to the baraiti of Zend, and the conj. *φέρψι to an avaiti (he may go) in the same language. A Greek parallel to the phonetic changes we have assumed is to be found in the -form ποΐ (= Zd. pai-ti) as is pointed out by Allen Stud. iii. 271. In the Et. M. 678, 44 we read : ποΐ πάρα Άργείοις αντί τον ποτί, αφαιρέσει του τ, είτα σννόΰω, and there is something very like it in the Delphic Ποιτρόπιος the name of a month meaning Προςτρόπιος (suppli- catorius). Three instances of the form ποΐ are given in the collection of inscriptions edited by Foucart and Le Bas (vol. iii. no. 157). We thus get the following proportion: ποΊ : ποτί :: φέρει'. *φέρετι. — Finally we must, on account of the similarity of their formation, call attention to the Aeolic forms of the 3rd sing, of the verbs in μι, discussed by Ahrens Aeol. 137. τίθη, ΐστη, δίδω, which are given by grammarians, have, like ψέρει, λέγει, lost the consonants of the personal termination. Since a final ς is not in the habit of disappearing, we must undoubtedly refer these 2 The inconsistencies in the use of the t adscriptum; discussed by Ahrens Dor. 294, prove nothing more than that this sound began to vacillate early. There is a genuine old witness to the t in απόθανα in the Laconian inscription of Xuthias in the old alphabet Άρχαιαλογική Έφημ^ρϊε Β, ή. Who will believe that the * has made its way in here on the analogy of some indicative or other? If the 6i of the indicative really arose as Corssen thought, it would of necessity have been written Ε in the old alphabet, and in the conjunctive an EI would be more inexplicable still. ch. vi. UNEXPANDED THEMATIC PEESENTS. 143 to *τίθητ, *Ίστητ, *2/cW, forms which would stand on the same footing as *<ρέρειτ. The Lesbio-Aeolic γελαι, if, as is probable, it owes its ι to epenthesis, would exactly con*espond to φέρει, λέγει. The special character of this kind of present-formation appears most clearly in the case of roots which end in a consonant. In present- forms like άγω, πλεκό-μεν, πέτο-μαι, βλίπε-τε, μένει we are sure that nothing else has been added to the root to form the present-stem but the thematic vowel. The number of such roots is large. According to the computation given below it is 109, and these may be arranged in groups according to their final consonant. Present-stems with a long vowel have been nearly all excluded from the list, because it is possible that the long vowel might be due to intensification. Eoots in κ. Ιέρκομαι *εκω, in the part, εκών, έλκω κρέκω ολέκω πλέκω and (only in Hesych.) τνκω (ετοιμάζω). — Total 7. Eoots in y. 211 άγω άμέλγω άμέργω έ'ιργω λέγω (άλε'γω) ννγει (τω κέντρω πλήττει Hesych.) όρέγω στέγω στέργω τέγγω φέγγω (Aristoph.) φθεγγομαι φλέγω ψεγω.— Total 14. Eoots in χ. άγχω άρίχεται άρόχεται explained by γλίχεται by Hesych., άρχω βρέχω γλίχομαι Ιέχομαι (Ion. Ζέκομαΐ) ελέγχω εύχομαι εχω *Κεχω or λέχομαι, deduced from Hesychius's καλέχες' κατέκεισο Πάφιοι, for which perhaps Meineke is right in reading κα-λέχεο i.e. καταλέχεο, μάχομαι όρύχω (ορύχοιεν Arat. 1086) ρέγχω σπέρχω στίχω (Lobeck Khemat. 67) τρέχω. I have omitted Ιάχω, because it might be thought a reduplicated form, also 'έρχομαι and others because their χ has apparently arisen from the σκ of the inchoative class. — Total 16. Eoots in r. άντομαι λίτομαι (hymn. Horn. 16, 5, 19, 48, Aristoph. Thes. 313) πέτομα. — Total 3. Eoots in δ. α'ίΰομαι άρΰω εΰω ελΰομαι μέΰω πέρΰομαι σπένΰω τένΰω (Hes. "Ejoy. 524) νΰω (λέγω Hesych.) and the doubtful ψε?ω (έντρέπω, φροντίζω Hesych.). — Total 10. Eoots in θ. αχθομαι εθω εχθομαι οθομαι πέρθω, while those whose θ is or may be 144 UNEXPANDED THEMATIC PKESENTS. ch. yi. the characteristic of the present — e.g. έρέχθω by the side of έρείκω, εσθω by the side of εθω have been omitted. — Total 5. Boots in π. βλέπω ΰρέπω εΧπομαι (εν)έπω επω έρπω θάΧπω Χάμπω Χέπω μέΧπω πέμπω πρέπει ρέπω σκέπω (late by-form of σκεπάζω) τέρπω τρέπω. — Total 16. Boot in β. βΧάβεται only Τ 82, 166, ν 34 and Anacreontica 31, 26 Be. Boots in φ. γΧάφω γΧνφω γράφω γρίφω (Hippocr.) ΰέφω (άπο)ΰρνφω (?) έρέφω μέμφομαι στέφω στρέφω τρέφω ψέφει (δε'δοικτε, λν7τε7 Hesych.), to which we may add νήφω, which a closer examination (Stud. II. 440) shows to be not open to the suspicion of being intensified. — Total 12. Boots in v. o-jo μένω π&ομαι σθένω στένω, while άνω (cp. άννω) is omitted because the ν is possibly a characteristic of the present, φθάνω because its ν is certainly such. — Total 4. Boots in μ. βρέμω γέμω and another γέμω in Hesychius (άπόγεμε ' άφεΧκε), treated of at Princ. ii. 246, ΰέμω θέρμω, which occupies a peculiar position as a denominative (cp. θέρω and θερμός), νέμω τέμω (Ν 707) τρέμω. — Total 8. Boots in ρ. ΰέρω θέρω, more often θέρομαι ορομαι (ορονται, οροντο £ 104, γ 471) πτάρω (Aristot.) for which some editors want to read πταίρω, φέρω στέρομαι. — Total 6. Boots in λ. βόΧομαι (Horn.) θέλω κέΧομαι μέΧω πέΧω. — Total 5. Boots in σ. ερσω (Nicander) τέρσομαι. — Total 2. Boots in ξ. άέϊ,ω άΧέϊ,ω οδάξω. — Total 3. Boots in ψ. ίε'ψω εψω. — Total 2. ch. vi. THE PEEV AILING VOWEL. 145 I have omitted ερρω because the origin of the pp is ambiguous, as also σφίγγω in the γ roots because the nasal looks as if' it were a present strengthening, though this is by no means so clear in forms like τέγγω, φέγγω, φθέγγομαι, ελέγχω, πέμπω, μέμφομαι. Our list might possibly however suffer a loss from this reason. It is very remarkable how the vowels are divided among these stems. Lobeck noticed this. Though he does not confine himself to the class of verbs now occupying us, he says, while further developing the remarks of some old grammarians (Rhemat. 50) : ■ Nulla sunt verba quae alpha breve in penultima pura habent, antecedente et succedente consona sim- plici, perpauca crassioris structurae βλάβω, γράφω, φΒάνω, nam plerum- que assumitur consona auxiliaris Ζάκνω, άπτω. — Creber vero secundae vocalis [ε] usus λέγω, στέγω.' As we should put it the statement would run thus : ' Verbal-stems with an α hardly ever make a present-stem with no further addition than that of the thematic vowel, unless the stem contains heavy groups of consonants, and hence ε is the more frequent.' As a fact 84 of the stems just given have ε, while a occurs 213 only 13 times and the remaining vowels almost exclusively in out-of-the- way and isolated specimens. The linguistic sense of the Aftics seems to have discerned this affinity between ε and the present stem, and hence ετραπον became the aorist and ετρεπον the imperfect. The Dorians pre- ferred an α next to a ρ : τράπω, τράφω, στράφω, τράγω, Locr. φάρω (Ahrens Dor. 117, Allen Stud. iii. 219). Greek shows herein a note- worthy agreement with Latin. In Latin too presents like emo, veho, tero, are extremely common, while presents like alo, molo, coquo are rare. We can enumerate 16 presents in which the two South-European lan- guages show the same vowel; 13 with e: βρέμω=/τβηιο, c:i^=depso 9 if this is not a borrowed word, ε$ω=βάο 9 εννεπε^ίη -sece, έπομαι =sequo7', Χέγω —lego, ορέγω^ -rego, πέρΰομαι=ρβάο, πέταμαι=ρβίο, ρέπω cp. repens, στέγω=ίβρο, τρέμω=ίτβ7)ΐο, φέρω=/βνο ; two with a in both languages: αγω —ago, άγχω=αηρο ; one with o: Homeric βόλομαι=νοΙο. Seeing that the North-European languages, which we cannot examine here, show a widespread tendency 3 to change a primitive ft in a similar position to e (Goth, ai) or even to i (Goth. oatra=Ch.-Sl. berq, Goth. ita, Lat. eclo), we may venture to conjecture that even in the period before the separation of the European languages there was in these cases no pure a but either an e or an a that tended to turn into an e. Cp. my essay 'Ueber die Spaltung des A-Lautes.' Ber. d. k. sachs. Ges. d. Wissensch. 1864. It is somewhat surprising, at first sight, to find that in present stems, which in general are prone to fuller forms, it is the weakest of the hard vowels that prevails, and that this prevalence is specially prominent in Graeco-Italic. The influence of the accent, which has only in Sanskrit a direct effect upon the formation of the pre- sent, will be found quite inadequate to explain this phenomenon, at all 214 events in Greek and Latin. It would be absurd to pre-suppose a *φερώ =*fero or a *φερώμι=*/6τό'ηι\ in order to get from the Sanskrit bhdrami 3 I am indebted to the kindness of A. Leskien for the fact that in Church- Slavonic out of 73 verbs with an unintensified «--sound 57 have e, 6 a, 4 o, and 6 u as the vowel of the present-stem. — Armenian, in which Hiibsch- mann's investigations (Ztschr. xxiii. 33) have shown the vowel system to agree widely in other points with the Indo- Germanic, has but rarely— e.g. level bear — an e at this place. L 146 UNEXPANDED THEMATIC STEMS. ch. vi. to φέρω and/ero. I should be inclined rather to venture on the assump- tion that the thematic vowel, which began early in most though not in all forms to weaken itself to e, exercised an assimilating influence on the vowel of the root. Then the weakening would have gradually extended itself from forms like *φέρεσι *φέρεπ. φέρετε — as is also conjectured by Broal, M^moires de la Socio te de Linguistique ii. 169 — to such forms as φέρω for *φάρω,'φέροντι for *ψάρονη. If such was the genesis of the ε in the root it would be easy to see why strong groups of consonants were able to preserve the old vowel e.g. in αγχω, άρχω, άντομαι, λάμπω. Even so, it is true, we do not get an adequate explanation for all in- stances and this is a thing we can hardly hope to do. But it is all the clearer as a fact in the history of language that at this place e was from very early times the favourite vowel. Besides the consonantal there are apparently a very considerable number of vocalic roots which follow this rule in their present-formation. Omitting entirely the denominative verbs with the wide-spread derivative terminations αω εω οω ενω, which the comparison of the related languages shows undoubtedly to have lost a j, we have so-called pure verbs like ΰράω ζέω κίω λύω, which the old grammatical theory reckoned as pecu- liarly primitive verbs, so that e.g. Lobeck puts them at the head of his Rhematicon, under the belief that the fuller forms grew gradually from simple formations of this kind by the accretion of sounds and syllables. The wider views opened out to us by the comparative study of languages oblige us on the contrary to hold it far from probable that two vowels, the radical and the thematic, especially if they were both a at first, stood next to each other from the beginning. It would be hard to find forms in Sanskrit and Zend which could be compared with δράω as a primitive formation. Such a juxtaposition of vowels is probably always due to the loss of spirants. The dropping of spirants between vowels is one of the most extensive and fundamental characteristics of the Greek lan- 215 guage. And for a number of such verbs we can clearly establish such a loss, though this is not the only process that has been at work. ■ It is often the case in the life of language that what looks primitive turns out on closer inspection to have been already defaced, and so it is here. A. portion of these presents can be shown to have lost a present- strength- ening which they had at an earlier time. Lastly we have to consider the cases in which the transition has taken place from the conjugation in -μι to the thematic conjugation. We will discuss the forms in ques- tion from these three points of view. We can be sure of the loss of a σ in the following present-forms : 1) βίεω compared with the Lat. vis-io, Lith. bez-du (Princ. i. 284) and the substantive βδέσ-μα. 2) ζέω compared with the Bkt.jas (Princ. i. 471), O.H.G.jesan, and ζέσ-σεν, ζέσ-μα, ζεσ-τύς. 3) νέομαι compared with the Skt. nas (Princ. i. 391) and νόσ-το-ς, νίσ-σο-μαι (for νεσ-^ο-μαι). 4) τρέω compared with the Skt. trdsd-mi (Princ. i. 277) and τρέσσε. 5) λάω I wish, compared with the Skt. lasha -Ίηί, Lat. las-civu-s, Goth. Iu8-tus (Princ. i. 450). 6) The forms of the rt. ές, which, like ίων, conj. εω, opt. εοι have assumed the thematic inflexion. ch. vi. TKACES OF LOST CONSONANTS. 147 In other cases the same loss is at least very probable, e.g. in χρίω, which shows a σ in χρίσμα, χριστός and agrees completely with the Skt. gharsh (Princ. i. 251), in ερνω draw with ερνσσα, for which, as I think I have shown at Stud. vi. 265 ff., we get a stem ΐερνς, which only differs from the Lat. verro vers in having developed a ν between the p and the f . Leskien (Stud. ii. 85 f.) conjectures the same final ς for other verb-stems besides, on the ground of the sigma which appears in aorists and nominal forms, as for θλάω, κλάω, μνω (Princ. i. 419), ζέω, Ι,νω, πτύω, έράω, σπάω. Even though this classification, the etymological investigation of which we cannot proceed with here, should be doubtful in many cases, of this much we may be sure, that a not inconsiderable portion of the pure verbs owe the juxtaposition of the vowels to the loss of a σ, which naturally took place at an early period when such a juxtaposition was as yet not found disagreeable. The loss of a β is not so often demonstrable. It would be the case 216 with αω satiate if it really belongs to the Skt. rt. av (Princ. i. 483, cp. Fick Wtb. 3 i. 24). The Homeric Xabw (Princ. i. 452) must have come from *Χαΐ-ων, and φά-ε from *φαΐ-ε (Princ. i. 369). In some other cases we can conjecture the loss of a j, but the j is of quite a different character. Under this head importance attaches to the statement of the Et. M. 254, 14 το φνω. ΑιοΧικώς φνίω φασϊ και το άΧνω άΧνίω, relying on which Ahrens (Aeol. 98) has received ψυίει into the text in Alcaeus fr. 68 (Bergk 3 97). Support is given to this formation, as Schleicher (Beitr. iii. 248) was the first to recognise, by the Umbr. fuia ( = Lat. yWii) a,nafuiest with the meaning erit, and perhaps in the phenomena from Keltic and Teutonic languages which Schleicher there discusses. This as good as proves that φνω arose in the same way from φνίω as τιμάω from τιμα^ω, φιλέω from φιΧε^ω. In my Studien iii. 398 I have called attention to the fact that the fluctuations in vowel-quantity shown by the verbs in question must be due to the after-effect of a lost spirant. This very fluctuation is observable in φνω. Homer knows only the short ν in the present stem, but at Aristoph. A v. 106 we read — Trrepoppvet, κατ ανθις έτερα φνομεν, and there are more instances in other Attic poets. It is only apparently therefore that φνω belongs to our present class ; really it belongs to the t-class, inasmuch as it once had the syllable ja added to its root as a present-expansion. The fact thus established will serve as an analogy for similar conjectures about other verbs in -νω, of which we now deal only with such as are root-verbs. θνω has in Homer a long υ as a rule : ο id ματ ι θνων, ΰάπεΰον δ' άπαν αίματι θνεν, θνε δ' Άθήνη, but at ο 260 it is short : επεί σε θνοντα κιχάνω. On the quantity in Attic poets cp. Ellendt's Lex. Soph. (2nd edit.). Here too the fluctuation of quantity is explained if we start from θνίω, and this is not only a presumable but an actually occurring form. Hesychius has the gloss εθνιεν εμαίνετο, ετρεχεν, and in virtue of this Ruhnken, at hymn, in Merc. 560 reads θνίωσιν instead of the manu- script βνίσωσιν, and in this later editors have followed him. It is from this stem too that the θνιάΐες get their name. In accordance with all this we shall not scruple, even where the 217 form with t is not to be found, to refer irregularities of quantity to the L 2 148 UNEXPANDED THEMATIC STEMS. ch. yi. same source, and especially in the case of λύω and £ύω. That the root- vowel in λύω was short is manifest from λε'λΐ/κα, λέλνμαι, λντο, έλνθην. In the present, by the side of Όδνσεύς λύε μώννχας ίππους Κ 498, Homer has ανΖράσι νείκεα λύει η 74. With the Attics the long vowel prevails in the present. (Cp. Ellendt, Lex. Soph. 2nd edit. s. v.) The case clearly stands thus : the present was originally *λυί'ω after the manner of the i-class, and the long vowel due to the diphthong was extended to the future and the aorist which are always prone to follow the present in the matter of quantity. In the case of δύω the long vowel is some- what more persistent. There remains finally a special class of forms of the kind ; those which have in the course of time abandoned the primitive method of the so-called conjugation in -μι, for thematic method of formation which was increasingly becoming the rule. Forms like ϊω, 'ίοιμι, Ιών can hardly have existed from the first by the side of such as είμι, "ιμεν, "ιτε. The α of ηια, the ε of Ιέναι, Ιείην, with which we had to deal at p. 121, was perhaps not distinct from this ο originally, that is, perhaps the latter arose from the former. Above all however we may conjecture a propor- tionally late origin for the thematic vowel where, as e.g. in έράω by εραμαι, ερύω by ερυσθαι (cp. above, p. 122) and most of all in ΰεικνύω by ΰείκνυμι, όμννουσι by όμννασι, Kipvq. by κίρνημι, the said vowel added itself to an already dissyllabic stem, thus making it trisyllabic. This is a plain case of the gradual spread of an analogy. We may say that the thematic vowel has in such cases quite another formative or etymological value from that which it has in φέρω, έχω and other presents of the kind. Present-forms such as used to be given as themata from Philoxenus's time, were not quite so rare in the usage of the various dialects as might be supposed. Hesychius has φαν λέγειν which Nauck (Melanges, iv. 29) regards as an infinitive like the rare forms ΰϋν, lovv discussed by me in Chap. XV. Of the same kind is προστάν (=προστήναι) in an inscription 218 from Erythrae. Lobeck (Rhem. 5) discusses the traces of a βάω which appear most clearly in the Heracleic έπιβϊ} tab. Heracl. i. 68. We may be sure of the forms προβώντες Cratin. Com. ii. 88, and έκβώντας in a Doric contract in Thuc. v. 77. θέω=τίθημι occurs, notwithstanding all that has been said about it, in the much disputed προθέουσιν A 291, and perhaps in άνέθει C. I. no. 1195. ε'λάω is abundantly attested as Doric (Ahr. 341). Whether or not there are many more pure verbs that do not fall into any of these four categories I cannot say. It might be hard any way to prove that presents like ί,λύω, νει, ΰίομαι are not just as genuine and unmutilated formations as λε'γω, άγει, only even here we occasionally find by-forms which at least make it credible that a consonant should have been lost. To κλύω correspond the Lat. duo and clueo, to ΰίομαι the Skt. dljdmi. It is possible of course that in both cases a j has been lost which we should have to regard as a formative element. Still less credible must it appear that such a harsh hiatus as that in Wj to πτ, and the impossibility of getting from fij to πτ. Where there is a in the stem we should certainly, on the analogy of χθες mentioned above, expect φθ instead of πτ. It is true Ebel at p. 43 says it is possible that the sounds

ος and the like are older. The undeniable connexion with κόλαφος quoted from Epicharmus (κολαφίζειν NT.) entitles us to set down φ as the final consonant of the root. 9) κρΰπτω common from Homer onwards (η 205). The φ of the stem is shown in the post- Homeric κρνφτ}, κρνφα, κρνφαίος, κρί/φος, κρνφιος and in the form κρυφής which is established by La. at Soph. Aj. 1145. Later Greek shows numerous by-forms with a β : κρΰβω, κρυβώι , ίκρΰβην, κρνβήσομαι. Cp. Lobeck on Aj. 1145, Princ. ii. 141. The case of καλύπτω is similar. Probably 7r was the primitive final letter of the root. 10) κύπτω, from Aristophanes onwards. The φ has the support of the Homeric κνφός (β 16), κνφος, and the Lat. cumbere, cubare. Still here too perhaps (Princ. ii. 142) the root originally ended in a p. 11) λίπτω long for, first occurring in Alexandrine poets as a present to the isolated λελιμμένος of Aesch. Sept. 380. The only traces of a ψ are to be seen in the remarkable λιφερνοϋντες discussed at Stud. iii. 199 and in the analogous Skt. rt. lubh, Lat. lubet etc. Princ. i. 459. 12) ράπτω in all kinds of Greek from Homer onwards (π 422), ίρράψην from Euripides s time. Cp. ραφή, ραφεΰς, ραφίς. Bugges con- jecture, mentioned hove on p. 161, that ράπτω exactly corresponds to ch. νιπ. LABIAL STEMS. 167 the Lat. sarc-io, and thus stands for *σραπ-]ω, with π = c, lacks corro- boration. C) Stems in /3. 243 1) βλάπτω (cp. /3Xa/3<-ra t p. 144) 294, α 195, ν 22, I 507, Τ 94,- O 724, later in general use. The β appears in εβλάβην (as early as Ψ 461), βλαβήσομαι (Attic), in the noun-forms βλάβη, βλαβερός, βλάβυς and in the derivative βλαβύσσειν βλάπτεσθαι (Hesych.). Still the β is apparently not primitive. Cp. Princ. ii. 153 and Sophus Bugge Stud. iv. 326. 2) καλύπτω in use from Homer onwards (P 243, 3 114). The β perhaps only in καλνβη (Hdt., Thuc.) and its later derivatives καλνβιον καλυβίτης. It is probable that the final letter of the root was originally p. Cp. Princ. ii. 154. A few forms show a <\> as well : περί κάλυψη (Plato), άκάλυψος (Dk)g. Laert.). Cp. κρύπτω. 3) νίπτω, a by-form of νίζω. It is doubtful if it is Homeric, as in the only place where it stands in our texts, σ 179, άπονίψασθαι is proved by Apollonius Lexicon 23. 15 to be the old reading, and not άπονίπτεσθαι which most of the M.SS. have (Nauck Melanges iv. 31). νίπτομαι Hippocr. Strabo. χερ> ίπτου Aristoph. Pax 961 (Homeric χερνίψαντο). That the stem had a β is shown by χίρνιφ (ace. χέρνιβα), and that the β came from a g by the Skt. nig (Princ. i. 395). In the Septuagint we find νιώήσομαι. D) Stems with an Undeterminable Labial. 1) γνάμπτω poetical from Homer onwards. Cp. κάμπτω. 2) ΰάπτω the same, probably an expansion of the rt. £α (ΰαίω) by means of π. Cp. δαπάνη. 3) δαρΰάπτω the same, from *ΰαρ-(!άρπ-τω (cp. δρέπω) with broken reduplication, as has been shown by Eritzsche Stud. vi. 297. 4) ΰυπτω, a late by-form of δΰω, probably expanded by π, the oldest authority for which seems to be Antimachus (Schol. Apollon. Khod. i. 1008). 5) ερέπτομαι, esse, Β 776, Ε 196, ι 97 only ερεπτόμενοι, ερεπτομενους, cp. Aristoph. Equ. 1295. The active νπέρεπτε occurs at Φ 271, κονίην h' υπίρεπτε ποΰοαν. Perhaps έρέπτω is related to rap-io, άρπαζ, αρπάζω. 6) Ιάπτω. From the present stem Homer only has κατά χρόα καλόν Ιάπτγς δ 749 (cp. β 376). Aesch. Sept. 544 ιάπτεσβαι, Soph. Aj. 244 501 λόγοις Ιάπτων, and in late poets. Ια-π is apparently expanded from la = Skt. jd go, in a causative sense by the determinative ρ just as the Lat.Ja-c is by k. 7) κορΰπτω in late poets. Theocritus, besides the aor. (μι) τυ κορΰφη 3, 5), has the noun κορυπτίλος, and Hesych. has κορνπτόλης ■ κερατιστηϊ, and ενοιάΰες' (ΐ) αίγες, αϊ μη κορΰπτουσιν. Connexion with κορυφή is probable, especially as κυρνπτιάν means to carry the head high. 8) λάπτω. The present-stem is not found before Aristotle. The perf. λέλαφα in Aristophanes. λαφύσσω is evidently related, though possibly the root is lap, which has softened its ρ in the Latin lamho. Princ. i. 453. 9) σκηρίπτομαι. ει ποθί τοι ρόπαλον τετμημένον εστί σκηρίπτεσθ' ρ 196, σκηριπτόμινος λ 595. The active in Apoll. Khod., origin obscure. 168 THE T-CLASS. CH. VITT. 10) χρεμπτομαι, clear the throat, from Eurip. onwards. Clearly an expansion of χρεμ (χρεμίζω, χρεμετίζω Princ. i. 250), so that the root probably had a π. 11) χρίμπτω. εγχρίμπτοντο Ρ 413. Then in the tragedians and late poets, επιχρίμπτει Bacchyl. 36 Be. 3 The rt. χριμ-π '. χραν (χραίνω) 1 1 χρεμ-π Ι χρεμ. Π. GUTTURAL STEMS, 1) τίκτω, a rare present-form of thert. πεκ, has already been discussed on p. 162. From the same root are formed πέκω (only in the gram- marians, e.g. Herodian i. 435, 21), πείκω (σ 316, Hesiod Opp. 775), and πεκτέω (Aristoph.). Cp. πόκο-ς, ποκάς, ποκίζω. 2) τίκτω (ύ 86, τ 113), common to all Greek. Rt. τεκ, whence ετεκον, τεκονμαι, τετοκα, τέκος, τέκνον, τοκεΰς, τόκος. The explanation given of τίκτω by Ahrens and others (Kiihner Ausf. Gr. i. 629), i.e. that it is an originally reduplicated form and stands for an earlier *τι τ(ε)κ>,> has no analogy to support it. Moreover the number of thematic present- stems that are reduplicated is too small to make this conjecture probable. On the weakening of ε to ι see Princ. ii. 378 ff. 3) φάρκτεσθαι explained by Photius by' φμάττεσθαι, also φάρκτον φνλακήν σκεύαζε Hesych , so that it must be a by-form of φράσσω, ψάργινμι, and correspond to the Latin farcio. 245 . III. VOWEL STEMS. 1) ανντω an Attic present-form, first in Aesch. Ag. 1123 Ζννανύτει, for the Homeric and Herodotean άνΰω. Cp. ήνυτο p. 122, where the form with the rough breathing was also noticed. 2) άρΰτω, an Attic though not frequently attested by-form of upvu>, Plat. Phaedr. 253 αρντωσιν, Pherecr. fr. 124 άρντεσθαι. αμΰσσονται Hdt. vi. 119 as a present stands alone, and L. Dindorf conjectures άψΰσσονται instead. It origin is obscure. It would be possible to start from the rt. ar obtain, win, which appears in ap-yv -ται. If in conclusion we review the whole extent of this present-formation, we shall find that the class contains 48 labial, only 3 guttural, and 2 vowel- stems, in all, that is, 53. In very many cases, especially in that of ΐρεπτω, ενίπτω (no. 4), πέπτω, ερεπτω, βλάπτω, ) ίπτω, πεκτω, άνύτω, άρύτω •this present-formation confronts an older one of another kind. In not a few cases the forms of this class could only be found in late Greek. The number of verbs of this class therefore, which were actually in general use, is not exceedingly large. THE NASAL CLASS. 169 CHAPTER IX. THE NASAL CLASS. Next to the present-stems which, have been strengthened by a r it will be best to place the widely ramifying formations in which the element of expansion is a nasal, and for this reason, that the syllables which contain the nasal are apparently of precisely similar origin with the syllable in the present-stem containing a r. Here too we have to deal with a stem- formation not unlike the formation of nominal stems, or more correctly ^*" speaking, of exactly the same character, and in the first instance pro- ceeding likewise by means of a simple suffix. We have already met with expansion by means of a nasal syllable at p. 109 if. in dealing with the present-stems of the non-thematic verbs, and it is clear that there is the closest relationship between the formation to be discussed here and those discussed above. To this point I have called attention on various occasions, and especially on p. 116. All nasalised present-stems which have a thematic vowel can be distinctly proved to have arisen from corresponding present-stems which had none, by the action of the tendency to uniformity spoken of on p. 148, which permeated the European languages in particular, and which led on all sides to a gradual retrenchment of the older, non- thematic conjugation. This gradually effected change proceeded in one of two ways. Either the thematic vowel took the place of the final vowel of the stem in να, νυ,οτ it was affixed to this stem as it was. Eor instance in the 1st plur. — -νά-μεν might become either -νο-μςν or -να-ο-μεν, -νΰ-μεν might become either -νο-μεν or -νν-ο-μ*ν. The former of these two changes is somewhat surprising in the case of present-stems in w, because the phonetic change of ν to ο and still more to ε as would have to be the case e.g. in the 2nd plur. -re-re by the side of -w-re, is not natural to Greek. We shall have to enquire therefore whether the facts of the case may not admit of a different explanation. In the case of the stems in α however both explanations are equally feasible. The oriental branches of the Indo-Germanic stock as a rule know nothing of formations of this class. There are however single instances of a tendency in this direction to be seen in connexion with the forms given above on p. 116. For instance the impf. a-grh-na-m which accord- ing to Delbriick first occurs in Epic Sanskrit, bears exactly the same relation to the old prevailing formation shown by a-grh-na-m, that the Gk. ΐ-καμ-νο-ν does to such a form as ί-σκίΐ-νη-ν. Spiegel in his grammar 170 THE NASAL CLASS. ch. tx. 247 of the Old Bactrian language p. 243 notices similar processes in this quarter. The Zd. imperative perena (fill) e.g. bears to the Skt. pr-nl-dhi a relation quite similar to that of τέμ-νε to πίλ-να-θι. In Sanskrit itself this verb has developed forms like prna-ti, prna-te, and nirna-ti he maims, for which in that case a root mrn is given, as compared with mr-na-ti is quite analogous. The 3rd pi. prna-nti, which might belong to either of the two formations, corresponds to the Old-Latin (ex)-plenu-nt =explent, and is only distinguished from the Doric πιμ-πλάνο-ντι, which the Homeric πιμπλάνεται warrants us in supposing, by the absence of the reduplication. Another link between the classes of verbs in use in Sanskrit and those in question here is the connexion that exists between the 5th or 9th olass and the nasalising verbs of the 6th, as seen e.g. in tr-mp-a-mi (Kv.) by the side of trp-no-mi (τέρπω), ma-n-th-d-mi (shake) by the ^ide of math-na-mi, cr-n-th-a-ti by the side of grath-na-mi (plait). The last-named forms come nearest to Latin forms like ru-m-p-o, fi-n-d-o, pi-n-g-o. — Much light is thrown on the subject of the mutual interchange of the various nasal present-formations by the Old-Persian a-di-na-m I took away (Spiegel, Altpers. 203), by the side of the Zd. zi-nd-t (conj.) and the Skt. hi-no-mi I throw, with which the Iranian words have no doubt rightly been connected. We thus see that there is by no means an absence of precedents for the present-formation which is now under discussion, and which extends more or less to all the European languages. We will in the first instance adduce only such Greek presents of the kind as have connexions of a similar stem-formation in another language : ΰάκ-νω Skt. dqqa-mi (no authority given, but there are other forms with a nasal in the Pet. Diet.). τί-νω „ Ki-nd-mi (Princ. ii. 93). φθί-νω „ kshi-no-mi (Princ. ii. 370). κλί-νω „ ςή-ηά -mi, Zd. ni-qiri-nao-mi (Joh. Schmidt Voc. ii. 251). κρί-νω cp. Lat. cer-no. άλίνω (αλβ/φω Hesych.) „ li-no (Lob. Rhemat. 123, 238). ιστά-νω (Polyb.) Ch.-Sl. sta-na. (στα-ννω C. I. 2556). χαίνω (rt. χα, χαν) Oh. -SI. zi-nq (cp. Ο. H. G. gi-n-e-m). 248 A remarkable agreement between Italian and Teutonic is to be seen in the Umbrian imperat. pers-ni-mu pray (rt. pers for prek, cp. Lat. prec-dri), as compared with the Gothic fraith-na, 0. 'N.freg-na, A. S. ' frigne and fringe (Schade Wtb. 143), in which the metathesis strongly suggests scindo as compared with the Gk. σκίΰιη-μι. The Gothic presents in -na are distinguished by a special intransitive and passive meaning, in the former of which they agree with the similar presents in Church- Slavonic and Lithuanian, though in Lithuanian (Schleicher, Comp. 3 784) the nasal holds a different position. In Greek and Latin no such limita- tion of the meaning is apparent. Now that we have taken a summary view of nasalised thematic presents in general, it becomes our duty to classify the Greek verbs belonging to this class, and to enquire into the origin of each kind. We must divide them into five main classes, each with various sub-clas«es. We shall here discuss each class as a whole, preparatory to the subsequent ch. ix. CLASSIFICATION OF NASAL STEMS. 171 arrangement and verification of the separate verbs of which each class is composed. I. PRESENTS IN -νω, -νο-μαι. This class comprises only verbs in which the root has received no further addition than this, e.g. πί-νω, $άκ-νω. They are to be compared to Latin verbs like li-no, si-no^ 0. Lat. da-nunt, and imp. da-ne (?) (Neue, Formenl. ii. 2 412 f.), -i-nunt in ob-i-nunt, red-inu-nt, which exactly agrees with the Lith. einu I go, and, as Gust. Meyer (Stud. v. 337) conjectures, with the Skt. inva-mi, I press in, make myself master of. These forms fall into two divisions, according to their origin. Some come, we may conjecture, from an older -νά-μι -ta -μαι. This origin is clearest in the case of ΰάμ-νει' δαμάζει, εΰαμον' εΐάμαζον (Hesych.), and in that of ετητνον Hes. Scut. 291 = *επιτναν or επίτνασαν. If the 1st pi. *Ιάμ-νο-μεν had survived it would have borne exactly the same rela- tion to ΰάμ-να-μεν as that of the Lat. ster-ni-mus to the pre supposable *star-na-mas (Skt. str-nl-mas). Fick 3 i. 57 connects κάμνω with the Skt. άπαξ λεγόμενον ςαηι-ηά -mi, and Delbruck (Yerb. 216) justifies the comparison. — A second kind are clearly related to verbs in -νν-μι, which, as we have seen on p. 109, have numerous by-forms in -νυ-ω. In this way we have side by side τί-νν-ται (p. 113) and tip». φθι-νύ-θω (cp. Skt. kshi-no-mi) „ φθίνω. πτάρ-νν-μαι (p. 112) „ πτάρνοισθε (Aristot. Probl. X. 18). 249 e-(T&€vw-ev „ ς'ζινεν (Hesych. cp. άποζίνννται ' άποσβίνννται) . Skt. dhu-no-mi „ θννω (Princ. i. 321). The clearest trace of the old w is that contained by έλαύνω, i.e. ελα-νν-ω (cp. the Sanskrit forms in -nv-a for nu given on p. 109), where it is clear that the υ of the ο ν was anticipatory of the ν in the next syllable, from which it afterwards quite disappeared (Princ. ii. 338). Our attention must further be directed, as it has often been already (cp. especially p. 121), to άννω (αΐ'ύω) on account of the forms 'άνεται and 'a t olto, which are as early as Homer. In this case we have the series, which we only suppose in the case of other stems, complete : ανν-μαι άνν-ω ''άνω 'ανω. As to the etymology of the verb, we may leave it an open question whether or not the ν was part of the root, in which case we should have to deal not with a w used as a present-expansion, but with ν used to expand the stem, a possibility which gained a certain amount of proba- bility from the comparison of the Skt. rt. san (san-o-mi) of the 5th class. — As an instance of the phonetic process assumed by us it holds good either way, and entitles us, even when the series is not as complete as that given for άνυμαι, to fill up the gaps in it on the analogy of that series, and in particular (cp. p. 113) in these two cases — Ti-vv -νται τϊ-νν-μ^ναι (Eurip.) ^τι-νύ-ω 1 *τι-νω τΧνω *φθί-νν-μι *φθϊ-νν-ω φθι-νω φθΧ-νω. 1 I write these forms with the asterisk, although I am quite aware that in Plutarch and Diodorus forms like rivvovres or vivvvoptss are of isolated occur- rence. But from what L. D. in Steph. Thes. s. v. τινννω sa\s, I doubt if the 172 THE NASAL CLASS. ch. ix. Importance here attaches to this fact in the history of the forms : that the t before the -w was sometimes short, while before -ιω Homer has it always long, and it does not begin to be short till Pindar's time. We may conjecture that from the time at which the digamma was in use in all Greek dialects there arose beside *πνΰω a form *ταΡω, from which by assimilation there next came *n ν iu>, and then, by compensatory lengthening, τίνω, until at last all trace of the spirant was swept away 250 with the shortening of the vowel. Precisely the same phonetic series may be actually seen in the Dor. ζέΐ'Ρος (Corcyr. inscr.), Aeol. £twoc, Ion. Οίνος, Att. ζέιος. The assumed stem *ψϋινν gets special support from the Homeric φβϊνυ-θι.>, φβί-νν-θο-ν. No Aeolic present-forms for the stems φθι and π have been preserved, but opirvv and σίννονται are well attested (Ahrens Aeol. 53), and these have clearly come from *όριιΐω, *ση\Ρυιται. This assumption is confirmed by the relation of όυίνω to opi'v/ii. It will be enough to point out how natural such a series as the following would be : — op-w -μι *ορ-ννω *6ρ-ι-ννω *opivfa> ορ'ιννω όρινω. 2 The last form but one would stand on the same level phonetically with the Gothic rinna, though the meaning of the latter is intransitive. On the other hand, we must abide by the old view of κρίνω, κλίνω (Aeol. κρίννω, κλίννω), i.e. that they have arisen from κρι^ω, κλινβω, because it is only in this way that we can explain the futures κρϊνέω, κλϊιέω, which point as clearly to a verbal stem κρΐν, κλϊν (beside <ρι κλι) as φανώ does to 0a v. If, then, we attribute the gradually disappearing length of the vowel before ν in τίνω, φθίιω, to the after-effects of the old syllable i'v 3 , it seems a most natural thing to conjecture that φβά-νω, whose quantity underwent a similar change, arose in the same way. Finally, it is possible that we have a still less obvious trace of the present-forma- tion here in question preserved in two presents which are just like each other, βηνλομαι and the Homeric ονλύ μένος. For βούλομαι we should be entitled by the Aeol. βολλα=βουλή (Ahrens, 59), the Cret. βώλομαι (Hey de dial. Cret. p. 25), and the Homeric ' ε βόλο ντο α, 234, βόλεται Λ 251 319, to assume an Aeolic βάλλομαι. This form however is actually to be found in Theocr. 28, 15 (ίβολλόμαν). βάλλομαι, it is extremely probable, arose by progressive assimilation from *βολ-νο-μαι, as did ολλν-μι from *όλ-νυ-μι. Now the jo of *βολ-νο-μαι would bear to the no of the Skt. vr-no-mi for *var-nd-mi (I choose) exactly the relation of that of τί-νο-μαι to the no of the Skt. Ki-no-mi (cp. Brugman, Stud. iv. reading in these cases is correct. It is possible too anyway that it was coined afresh in late times, and that it was only a chance that it resembled the old form. 2 Of the anaptyxis of an ι after a liquid we should have an excellent example in iK -ϊ-νύ-ω, if we could be more clear about the etymon of the verb, whose mean- ing • rest, loiter,' does not come so very near to that of the root f «λ turn, revolve. It is possible on the other hand that the ι is of the same kind as the i in the Lat. orior (Gk. bpiorro), i.e. a present- expansion. * Gust. Meyer n. Pr. 45 regards the long vowel as an intensification of the same kind as that seen in τβί» τ/ω (cp. above, p. 113). But the• contrast between φθΧνύθαν and ψθίι /tiv in Homer and the existence of "άνω by the side of ivvm seem to speak for my view. The same scholar does not like the derivation of -νω from -»ί>«, and will only recognise -nd-mi as a forerunner of -vw. But there is often not a trace to be seen of such a formation. en. ix. CLASSIFICATION OF NASAL STEMS. 173 121). With βονλσμαι, however, I have at Stud. v. 218 compared the Homeric participle ονλόμενος, which as clearly "belongs to the rt. o\ (ολλυμι) as it is distinguished by its active meaning from middle aorist- forms like ώλετο and όλέσθαι. I conjecture, then, that by the side of *ολ-νυ-μι there existed a middle *όλ-ννο-μαι, which early became *6\- νο-μαι, *ολ-λο-μο», and was finally made into *ουλομαι by compensatory lengthening, and, no longer thought of in connexion with ολλνμι, only survived in the restricted ονλόμηος. In an aorist participle the ου would be inexplicable, while, on the other hand, the shortened όλόμενος (Eurip.) is completely explained by the analogy of the Homeric βόλοντο. To these may be added είλλω, which we shalj have to discuss on p. 176. For the remaining present-stems of this division there are no criteria to show whether the syllable vo'(vt) came from na or by way of an inter- mediate wo (ννε) from w, or, in other words, whether they stand in direct relation to the 9th or to the 5th class of Sanskrit verbs. II. PKESENTS IN -α-νω, α-νο-μαι. The traces in Sanskrit of a similar formation to that of the very numerous Greek verbs in -ανω have been already noticed on p. 116. The Yedic ish-ana-t there mentioned is in formation not at all distin- guished from Greek forms such as e.g. the Dor. ηνί-ανε, nor the middle impf. ish-ana-nta from such forms as ήχθ-άνο-ντο, while the more frequent forms of the 2nd sing, imperat. act. in -ana, e.g. grh-ana catch hold of, ac-dna eat, are to be compared with Greek forms like ίκ-άιε. Of a precisely similar character are the Armenian presents in -anem given by Htibschmann Ztschr. xxiii. 406, e g. Ikh-anem = Χιμπ-άνω. — Latin presents a number of parallels which are adduced by Neue Formenlehre ii. 2 412 f. and Corssen i. 2 420 : sol-ino consulo (Fest. p. 351), while sol- inunt, according to Fest. 162, occurred for solent, and at Fest. p. 352 in- ser-ln-untur is given from Livius," Odyssia Latina. It is at first sight 252 somewhat surprising that the i in the last form should be long (Kitschl . Monum. epigr. tria, p. 18). But it is not impossible that the long vowel ought to be explained as originating in *ser-io a possible by-form of ser-o formed like/er-io (ferinunt Fest. p. 162). Such a present would exactly correspond in formation to the Gk. έ'φω for *σερ-βω. Thereupon the vowel of the i-class would have united with the initial letter of the nasal suffix -mo (for -dno) to form the long i, somewhat as in ίζάνω, φνζάνω which are for *kh-j-arw * = (Jingo), diyy -άνω, I cannot help thinking too elaborate. 176 THE NASAL CLASS. ch, ix. related languages. The conjunction of the nasal class with the incho- ative is to be seen, only in the reverse order to that of the Greek, in the» Lat. fru-n-isco-r (Lucilius ed. Luc. Muller xviii. 2) and in nanciscor which apparently arose by metathesis from *nac-ni-sco-r (Joh. Schmidt Ztschr. xxiii. 270). Greek formations like ?αρ-θ-άνω exactly correspond to the Lithuanian in -d-inu, of which Schleicher gives a list in his Lit. 'Gr. p. 165, e.g. ly-din-ti to make it rain, p\-din-ti to get plaited. The causative meaning which attaches itself to these forms as well as to those in -inti is no hindrance to our comparison. The nature of these verbs renders it probable that the second subdivision of verbs in -ανω is of comparatively later origin than the first, and this conclu- sion will be confirmed by the more minute investigation of the several forms given below. 256 ΙΠ. PRESENTS IN -ναω, -ναομα,ί, AND -αναω, -αναομαι. The verbs in -νη-μι treated of on p. 116 ff. have in many cases by- forms in -ΐ'α-ω : e.g. Ιαμ-νά-ω, κιρ-νά-ω, πιτ-νά-ω. Apparently the rela- tion of -να-ω to -νη-μι is the same as that of -νυ-ω to -w -μι. Schleicher Comp. 3 p. 765 assumes that a j has fallen out between the α and the thematic vowel, just as in the ordinary verbs in -αω e.g. ΰαμά-ω. In sup- port of this we can certainly mention ΰαχνάζομαι (Aesch. Pers. 571), which is unmistakably a derivative from Ιάκνω made by means of -αζω= ajdmi. And it looks as if the Sanskrit forms• in -nd-jd-mi discussed by Benfey Or. u. Occ. i. 427, iii. 217, like pa?idjdmi=nrepvaw were also in its favour. But since we not only have no ground for assuming in the verbs in -ννω just mentioned the loss of a /, or in other words for assum- ing that the mark of the i-class was added to that of the nasal-class, since moreover we have in other instances repeatedly seen the simple thematic vowel added to the stem of verbs in -μι, e.g. in forms like Ί-η-ίμι, ίων, ιστφ (Hdt.), πψπλέω (Hes v Theog. 880), I do not know whether we ought not to prefer the simple to the more elaborate explanations. The case, however, is somewhat otherwise with the verbs in -αναω, -αναομαι. These verbs have no such prototype as is provided for verbs in -ναώ by those in -νημι. What are we to say then to the Homeric Ζιικα νόωντο, ισχανόωντο (cp. ισχανέτην)1 TOr ϊεικανάω we have in 3ft- κννμι at least something like a related form, and can conjecture that there is between the former and the latter a relation similar to that between όριγνάομαι (p. Ill) and όρέγνυμι. A transition to the analogy of the a- conjugation is here unmistakable. It might perhaps be conjectured that these verbs are properly denominatives, and for κνρκανάω a noun κνρκάνη is actually given. But the linguistic sense certainly regarded them as little different from verbs in -ανω, and besides they only occur in the present-stem. In Zend (Schleicher Comp. 3 761) the verbs in -nao-mi (=Skt. ndmi) have by-forms in -navd-mi, e.g. 2nd sing, imperat. kere- nava (make). Might we venture to compare the enigmatical Greek 257 formations with these 1 όριγνάομαι would then stand for *όριγναί<>-μαι with the old intensifying α retained. The sounds in the two cases agree, but the comparison is doubtful all the same, especially as α is extremely rare in such a position. It is with greater confidence that I compare the Latin verbs in ch. ix. CLASSIFICATION OF NASAL STEMS. 177 -inare : car-ina-re (Enn.), which is explained by jurgare, obtrectare and referred to a root which is perhaps connected with κείρω, coquinare (Plaut.), a by-form of coquere. de-sti-na-re occupies a position by itself, inasmuch as the i has evidently been weakened from a radical a. It bears, then, a similar relation to coqu-ina-re that παμ-φα-νάω, assuming that this form has come from the rt. Φα not 0a- v, does to fcttc -ανάω, and has its counterpart in the -Cretan στα-νν-ω as also in the form ί-στάνω first found in Polybius. IV. PKESENTS IN -νεω, -νεομαι. These not very numerous formations are evidently closely connected with those just discussed. As to the origin of the vowel ε we shall hardly avoid the uncertainty which arose in the last division about the a. The syllable νε along with the thematic vowel often attaches itself to the same roots which also take other nasal strengthenings, so ικ-νίο-μαι by the side of Ικάνω bv -νίω (Hdt. ?) „ „ δύνω φ βυ-νίω (Hes. Scut.) „ „ θννω (ντή-ίσχ-νεο-μαι ,, „ Ισχάνω, Ισχανάω. In κινέω the syllable νε has passed into the whole verb-stem, but the re- lation to κίνυ-μαι is unmistakable. The verb άγϊιέω, which occurs in Homer by the side of άγινέμεναι, αγίνεσκυν, stands by itself. The t is evidently the same as that in όρίνω, and here too we get a form without this vowel. Similar to the relation of ορ-νυ-μι to όρίνω is that of the Cretan άγνέω (Hesych.) to άγϊιέω. V. PRESENTS ΓΝ -αινώ, -αινομ,αι. These presents are due to a union of the marks of the nasal class and the j-class. We can hardly be surprised at such an accumulation of 258 expansives after having already encountered several instances of the union of the marks of the nasal and inchoative classes. The verbs in αινώ, if we omit those which, like άγριαίνω, αναίνω, κερΰαίνω, υγιαίνω, come from Greek noun-stems in common use, fall into two divisions ac- cording as -ν-ιω or -αν-ιω is joined to the root. This variation is evi- dently due to the same reason as that between -νω and -ανω. We may conjecture, therefore, that the first division is closely connected with the verbs in -νω, the second with those in -ανω. βαίνω is the only instance of the first division, while the second is represented by υφ-αίνω, άλιτ-αίνω. In καγκαίνει' θάλπε», Ιηραίνει (Hesych.) i.e. καγ-κα(Ρ)-νι-ει, we find the same inserted nasal as in the likewise reduplicated πιμπλά-νω. A shorter by-form occurs in καγκομένης (ξηράς τω φόβω Hesych.). τε- τρεμ-αίνω is likewise reduplicated, and it is noticeable about the second part of the word that αν-βω has been added. This formation too has its Indian prototypes. In the Yedas there occurs as a causative present of the rt. dam among others the dam-an-jd-ti already mentioned on p. 117, a form which would exactly correspond to a quite possible Greek *%αμαινω, and which Delbriick, who gives a collection of such presents from the Yedas at p. 207, regards as a denominative formation from da- mana-m taming. (Cp. G. Meyer n. Pr. 99.) Most of these presents 178 THE NASAL CLASS. ch. ix. actually have nominal stems of that kind by their side. Still there are some that have no such stems, e.g. rishan-jd-ti he is in fault, which stands by the side of the synonymous resha-ti just as, say, the Greek κροαίνω by the side of κρούω. The verbs in -αΐ'ω and -αινώ have this peculiarity in common, that in many instances they have, besides the shortest and the longest stem, a third, which is used for the formation of the -compound tenses and even of the perfect : μαθ μανθανο μάθε (μαθησομαι) αχσθ αίσθανο αίσθς (αίσθήσομαι) άλιτ άλιταινο άλιτε (άλιτημενος) 259 and by-forms of the last kind are not unknown even in the formation of the present : e.g. the Homeric νφόωσι by the side of νφαίνειν. The verbs which belong to this widely ramifying class are the follow- ing. They are arranged in the subdivisions given above. I. PRESENTS IN -νω, -νο-μαι. 1) *άλ/ νω only given by grammarians and explained by αλείφω. We follow Lobeck Bhem. 123 in connecting it with the Lat. li-no (le-vi, U-tu-s). Connected are Hesychius's glosses άΧεΊναι ' το έπαλεϋφαι τοίχω, άΧϊναι' επάλειψα*, άΧίνειν (cod. άΧινεΊν)' άΧείφειν. άΧίνονσιν given by Bekk. Anecd. 383 as Sophoclean hardly belongs here, since it is ex- plained by λεπτνίΌυσίΐ', but rather to άλέω grind, and is probably to be regarded as a formation similar to όρίνω. It must be admitted therefore that this άΧίνω=Ιίηο is not beyond suspicion as a present-form. The aoiist άλϊναι shows that the ν extended here beyond the present-stem as in κοϊναι, κΧΊναι etc. 2) άνω has been discussed on pp. 121 f. and 171, where the Homeric forms will be found, ανοις Aesch. fr. 156 Dind. with ά. ανειν Plato Crat. 415 a. 3) βερνωμεθα' κληρωσώμεθα Hesych. was given on p. 118. It is possible that it belongs here as well. 4) βύνω a rare by-form of the usual βυνέω in Hdt. Ιιαβΰνεται ii. 96. 5) ΰάκνω, the forms of the present-stem, which are afterwards in general use, are wanting in Homer. Cp. lay κάνω. In Skt. there exist side by side the rts. da$ and dq^. Cp. p. 170. 6) ΰάμνω (?) only in Hesych. : Ιάμνει (cod. ϊαμνεϊ) δαμάζει, cp. p. 171. 7) Ηνω, Aeol. Ζίννω, elsewhere Ζινεω, Hes. Opp. 598 Δημητέρος ιερόν άκτην Ζινέμεν, άποΐίνωντι tab. Heracl. i. 102, Meister Stud. iv. 433. 8) Ιύνω from Homer onwards (Ιυνοντες λ 579, ΰΰνε Ε 845, άπέδυνε χ 364, Ζϋνον Λ 268), but seldom in Attic prose. 260 9) ελαΰνω ordinary Greek from Homer onwards (M 62), a shorter present-form ελών (?) Ω 696, ειςελάω κ 83 and elsewh3re in poets, rare in prose, tab. Heracl. i. 127 επελάσθω. Cp. p. 148. 10) ε-ζινε-ν εσβέννυεν Hesych. Cp. ζείνυμεν (cod. ζείνομεν) ' σβεν- ννμεν. 11) θύνω. θϋνε (impf.), Ε 87, by the side of oXoiyai φρεσι θί/ει 342 ; also in Pindar and later poets, θννέω Hes. Scut. 210. ch. ix. PRESENTS IN -νω. 179 12) ίστάνω, a by-form of ϊστημι, known to occur from Polybius onwards. 13) κάμνω, from Homer onwards (μ 280, Τ 170, κάμνε Ε 797) in general use. Cp. p. 171. 14) *κέρχνω is only quoted from Hippocr., κέρχνει (cod. κερχνεϊ)' τραχύ νει Hesych. By-forms κέρχω, κερχνόω. 15) κιχά-νω. In Homer with snort ι and long α : κιχάνει Τ 165, κιχάνετε Ψ 407, later the quantities are reversed; κιγ-χά-νει (above p. 175). Cp. κίχημι(ρ. 121). 16) πφ-πλά-νε -rm only I 679 ; discussed on p. 170. 17) πίνω, Aeol. πώνω, ordinary Greek, by the side of the fut. πίομαι, aor. επιο-ν, πΊθι. The related languages all know the rts. πο, -κι, origi- nally pa, but know of no nasal present-formation from this rt. 18) *ττΓάρνομαι. The Aristotelian πτάρνοισθε (Probl. x. 18) was mentioned above on p. 171, πτάρνυμαι on p. 112. 19) *στάνει' στείνεται, σνμβίβνσται, Hesych. anyhow related to στενός, though the meaning makes it doubtful whether it is also related to the Cretan στανύω (statuo). 20) τέμνω (D.or. Ion. τάμνω), ordinary Greek, τάμντ) Γ 105, εταμνον Δ 155, τάμνοντο Σ 528, έκτάμνειν Λ 515, only at γ 175 τέμνειν, Hdt. π. 65 τάμνονσα, Heracl. tables Ιιατάμνειν (i. 12, ii. 65); Homeric by-form τέμει, Ν 707. 21) τ'ινω. With ϊ in Homer : ζωάγρια τίνειν Σ 407, τίνειν ουκ εθέλωσι Γ 289, τίνων Β 193. Later the short vowel gradually gains ground : Solon fr. 13, 31, — — έργα τίνονσιν, Pind. Pyth. 2, 24 τίνεσθαι, Theogn. 204 τίνονται, but at 740 άντιτίνειν, Soph. O. C. 635 ου σμικρόν τίνει etc. On the relation of the word to τίννται see p. 171 f. If at Princ. ii. 93 the Skt. Mind-mi is rightly compared, we have therein not only the same root, but also a similar present-formation. By-form τίω. 22) φθάνω, φθάνει U τε πάσαν επ αίαν Ι 506, cp. Φ 262. But οι/ 261 φθάνοι θνησκων τις αν Eurip. Or. 941 Dind., ουκ αν φθάνοις Aristoph. Eccl. 118. Common to all Greek by the side of εφθην, 'έφθασα, φθάσω. Bugge Ztschr. xx. 39 compares the Zd. Jsdnajaiti he sets going, and infers the existence of an Indo-Germ. 8ρά-ηα^ά-τηί=*φθα-νάω. 23) φθίνω, φθίνυυσιν νύκτες τε και ηματα λ 183. φθϊνέτω ε 161, φθί- νοντος τ 307. — But Pind. Pyth. 1, 94 ου φθίνει, Isthm. 7, 46 κατέφθινε, Soph. Aj. 1005, at the end of a trimeter, φθίνεις. Cp. p. 171 f. 24) φυνω only in the list of verba barytona in -νω. (Herodian ed. Lentz i. 450.) Two of these forms, 12 and 16, have come from reduplicated present- stems. Besides these forms, which have all retained the characteristic nasal, there are three which apparently had it originally, but have lost it by a process of progressive assimilation, i.e. 25) είλω, press hard (ειλομέιων θ 215), Aeol. έ'λλω (άπέλλειν άπείρ- -γειν Hesych.), Dor. ΐηλω (γηλεσθαι' κατέχεσθαι Hesych.). Brugman Stud. iv. 122 conclusively infers from these forms a primary /ελλω, and shows by a reference to the by -form εΐλλω (for foXkjv) that it is probable that /ε'λλω came from ΐε\νω. On 26) βουλομαι and 27) ουλόμενος cp. p. 172 f. n2 180 THE NASAL CLASS. ch. ix. II. PRESENTS IN -ανω, -ανο-μαι. A) Those with Nasalised Root-syllables. 1) άνΐάνω, poetical from Homer onwards (ανΰάνει β 114, fyvlav* Ο 674). Rt. σϊαΐ, by-form τβομαι. 2) γρνμπάνω. A rare word, translated by έπικάμπτειν, γρυπονσθαι (Hesych., Bekk. Anecd. p. 228). By-forms γμυπαίνω, γρνπανίζω, γρύπτω, clearly related to γρυπός, curved, bent. 262 3) δαγκάνω, a by-form of ΰάκνω, only given by grammarians, who mention a form ΰήκω as well (cp. p. 156). 4) έρνγγάνω, used by Attic poets instead of ερεΰγαμαι, of which there is older evidence, Eur. Cycl. 523, also in Hippocr. and later writers. 5) θιγγάνω, in Attic poets (Aesch. Sept. 44, Soph. 0. C. 328), and here and there in later prose (Aristot.). A similar present-formation is to be seen in the related Lat. fingo (Princ. i. 223). 5b) κυνθάνει' κρύπτει, Hesych. (G. Meyer 92). 6) κλαγγάνω, Soph. fr. 782 Ώ. οπού τις Όρνις ουχί κλαγγάνει (cp. Aesch. Eumen. 131), elsewhere only twice, compounded with έπ -avu. A by-form κλαγγέω in Theocritus, and, in a somewhat different meaning, κλάζω. Cp. κέκλαγγα (by the side of κέκληγα), κλαγγη and the Lat. clangor. 7) λαγχάνω. In extensive use from Homer onwards (t 160 ες Ιέ εκάστη ν εννέα λάγχανον αίγες). The nasal passes also into the perfect λέλογχα. If Fick 3 i. 748 is right in comparing the Ch.-Sl. po-lac-a (λαγ- χάνω) and the Lith. per-lenk-i-s, a man's due, we have in these words additional testimony to the early presence of this nasal. 8) λαμβάνω. Present-forms occur from Pindar onwards (01. 1, 83). Importance attaches to the Herodotean forms Χάμφομαι, καταλαμπτέος, έλάμψθην, but not to the late καταλημφομαι, άνελήμψθη, and the like. Cp. λάζνμαι, λάζομαι. Job. Schmidt, Vocal, i. 118, gives a most minute discussion of the traces of similar nasal formations in Sanskrit. Cp. above p. 174. 9) λανθάνω. Common to all Greek from Homer onwards (ελάνθανον Ν 721, λανθανόμην μ 227), by the side of ληθω (Dor. λαθαΛ, ληθάνω. 10) λιμπάνω. Once in Thuc. (viii. 17, καταλιμπάνονσι), and then in later writers, λείπω is immeasurably more frequent. Cp. the Lat. linquo, Skt. ri-nd-U-mi, and the Goth, af-lifna-n. Princ. ii. 60. 11) λνγγάνομαι, sob (Hesych.). Cp. λνγζ. 12) μανθάνω. Common to all Greek from Sophocles onwards. Cp. μενθ-ηρη' ψροντίς Hesych. Princ. i. 387. 13) πανθάνω, late and rare: Schol. Eurip. Hec. 1130, άντ\ τίνος ταύτα πανθάνεις ) Apollonius and Herodian gave the form (ed. Lentz 263 ii. 545). πανθάνω : πένθος \\ μανθάνω : μενθηρη. Reasons in favour of the connexion with πένομαι, πόνος may be found at Princ. ii. 365. 14) πνιθάνομαι. Homeric, along with πεΰθομαι (πυνθάνομαι β 315, πυνθανόμην ν 256), new-Ionic and Attic. The nasal is probably only to be found in the related Lith. bundu, I am awake, and bMinu, I wake (trans.). Princ. i. 325. 15) τυγχάνω. Common to all Greek from Homer onwards (παρετύγ- χανε Λ 74, ξ 231 τύγχανε) with τεϋχω in a different meaning. Cp. the Lith. tenku, I fall to the share of, Ch.-Sl. Hkna-ti, figere. ch. ix. PRESENTS IN -ανω. 181 16) φνγγάνω. From Aeschylus onwards of pretty frequent occurrence as a by-form of φεύγω, especially in compounds. 17) ψΧνί'ΰάΐ'ω. εκ<ρΧυνΙάνείν=εκ<ρΧΰζειν, έκψΧύειν, break out, of sores (Hippocr.). 18) χανίάνω. From Homer onwards (Ψ 742 yavluvt). The nasal recurs in κέχανία and the Lat. pre-hendo (Princ. i. 242), and perhaps in the Ch.-Sl. fydati, to want (Joh. Schmidt Vocal, i. 73). B) Presents in which the Root Syllable has not been Nasalised. With the single exception of Ικ-άνω all the root- vowels are long either by nature or by position. For the rest the stems are of the most various character, and this variety will necessitate a further subdivision of these presents. a) ανω added to stems which show no present expansion before it. 5 1) άλφάνω. In Attic poets, Homer knows only ήΧψον } αΧψοι, αΧψοιν. 2) ανζάνω. In use with Herodotus, Attic writers and others by the side of ανζω (Homeric άέζω), which holds its own all through (cp. Veitch, p. 101). Another but a late and rare by-form is αυΐέω (cp. ανζήσομαι etc.). The latter reminds us by its formation of the Lat. aug-eo just as αν£άΐ'ω does of the Lith. augin-ti and the Goth, bi-auk-nan increase. ανξω itself has a stem already expanded by σ. 6 3) ερνκάνω. Imp. ερΰκανε κ 429, 2nd imperat, κατερνκανε Ω 218. 264 ερύκω has been developed from ερυ (ερνμαι) by κ. 4) ευδά νει a doubtful reading in Lycophron 1354. 5) Ίκάνω, poetical from Homer onwards (ίκάνεις ημέτεροι• ΰώ Σ 385, χρειώ γαρ ικάνεται Κ 118). Cp. 'ίκω, Ίκνέομαι. 6) κΰΐάνω. Only in Homer : κυΖάνει & 73, κΰΐανον Υ 42 by the side of κυμαίνω, κνΰιόων. 7) οιΰάνω. οιΰάνει Ι 554, οΐΰάνεται Ι 646, οΙΙάιοντ Aristoph. Pax. 1166, elsewhere οιΐέω (even as early as ε 445 : ωΐεον cp. οΙΒήσω, ωδησα), later οιΰάω, οΐΰαίιω. 8) όφλάϊω. όφΧάνειν όψλισκάνειν, όφείλειν Hesych. Cp. Phot. Lex. By-forms οψΧω (late), οφείλω, όφΧίσκω (Suid.), όφΧισκάνω. b) ανω added to forms of the lengthening class. 9) ήχάνω, to be inferred from Hesychius's ήχανεν εϊπεν from the rt. ux=Skt. ah Lat. ag (djo). Elsewhere there is only ήμί, pret. ή-ν (p. 103, cp. Stud. ix. 463 ff.). 10) θηγάνω. θηγάνει• όζΰνει Hesych., and accepted on this authority by Herm. and Dind. at Aesch. Ag. 1535. Elsewhere θηγω. 11) κενθάνω. εκεϋβανον Γ 453, elsewhere κενθω {κυιβάνω, ρ. 180). 5 On the Homeric frdave see Chap. XIII. § 3. 6 βρατάνει • βάιζει airb rrjs νόσου Hesych. is connected by G. Meyer with the root vart (vertere), so that it would mean ' he takes a turn,' i.e. for the better. 182 THE NASAL CLASS. ch. ix. 12) Χηθάνω. Χηθάνει η 221, causatively, ' he makes to forget,' cp. Χήθω, Χανϋάνω, both in Homer. c) ανω united with reduplication, cp. ίστάνα, πιμπΚάν^τα^ p. 179. 13) ισχάνω. By-form 'ίσχω for σι-σχω. Homeric: Ισχάνει & 387, ίπχανέτην Ρ 747. Cp. Ισχαιόωντο. d) ανω added to stems expanded by r. 14) αμαρτάνω (cp. p. 163). Common to all Greek from Homer onwards (ημάρτανε Κ 372, λ 511). 15) βλαστάνω. From Aeschylus onwards (Sept. 594). 16) βΧυστάνω a by -form of βΧύζω used by ecclesiastical writers. Cp. αναβΧυστάνω (Hesych.), with the apparently apocryphal by-form άνα- βΧνπθάνω. 16 b) όπτάνομαι in late prose in the sense of ορώμαι. 265 e) ανω attached to forms of the t-class. 17) άζάνω, only αζάνεται hymn, in Yen. 270, elsewhere, including Homer, αζω, άζαίνω. 18) ιζάνω pretty frequent from Homer onwards (K 92 Ίζάνει, καθίζα rov ε 3, trans, ίζανεν ενρύν αγώνα Ψ 258). Also 'ίζω, εζετο. 18 b) ψυζάνω (?) Hesych. ψυζάναι' φυγείν, ΰει,Χιάσαι. f ) ανω added to the inchoative σκ. 19) άΧυσκάνω, only άΧυσκανε χ 330, cp. άΧύσκω, άΧνσκάζω. • 20) άμβλισκάνω (Pollux iii. 49) a by-form of άμβΧίσκω. 21) όφλισκάνω, Attic by the side of οφείλω. Cp. 6<\>\άνω no. 8. 22) ενριγισκάνω. ενριγισκάνειν' ένριγοϋν Hesych. g) ανω after an expanding θ (cp. p. 175). 23) αισθάνομαι, in general use from Sophocles onwards, αίσθομαι is a questionable by-form. Clearly the word is connected with άιω, which in επαίω shows similar meanings. 24) Ιαρθάνω. Compounded with prepositions in Attic prose : καταίαρ- θάνειν, επικαταΰαρθάνειν (Plato), ΰαρ-θ is certainly to be referred to a shorter Sap, Ipa (Princ. i. 288). 25) απεχθάνομαι, an Attic word, may find a place here although the origin of the word and all its kin (εχβος, εχθρός) is obscure, άπεχθάνεαι 0202. 25 b) όλισθάνω a thoroughly Attic present, later όΧισθαίνω. In Homer there is only ολισθε. Contrary to all analogy are the apparently denominative forms : 26) μελάνει, Η 64 μεΧάνει Η τε πόντος ί/7τ' αυτής, κυΰάνω, which Lobeck (Rhem. 235) compares with it, differs from it in not having a distinct noun-form with the same stem. Is it possible that μελάνω is a ch. ix. PRESENTS IN -νάω AND -αναω. 183 verb formed straight from the root in the sense of ' to grow turbid ' 1 μόλος, μολύνω are certainly related (Princ. i. 461). Cp. G. Meyer 86. 27) φασγάνεται' Ιίφει αναιρείται. This strange word Mor. Crain (Philol. x. 582) conjectures to have stood for *σφαγ-σκ-άν ε -rat, and the subst. φάσγανο-ν for *σφαγ-σκ-ανο-ν. If he is right — and a defence might easily be found for the aphaeresis of the initial σ — we should have 266 a formation like όφλ-ι-σκ-άνω. But the conjecture is a doubtful one. There are two isolated verbs in -ινω which may find a place in an appendix to the verbs in -ανω. (Cp. pp. 172 and 177.) 1) άγίνω preserved in άγινέμεναι υ 213, άγίνεσκον ρ 294, elsewhere αγω, άγινέω. (Princ. i. 208.) 2) όρίνω, poetical from Homer onwards, όρίνει Λ 298, όρίνονται A 525. Lesb. όρίνιω. III. PRESENTS ΓΝ -νάω AND -ανάω. A) IN -νάω, AND CONSEQUENTLY BY-FORMS TO THOSE IN -νη-μΐ. 1) ΰαμνάω cp. p. 116 f. 2) κιρνάω εκίρνα it 356, Kipvf Hdt. iv. 66. 3) κρημνάω. κατεκρημνώντο hymn. Homer. 7, 39, other forms not till late prose. 4) πιλιάω. 3rd sing. act. πΐλνς. Hes. Opp. 510, 2 sing. mid. πιλν$ hymn, in Cerer. 115. 5) πιτνάω cp. p. 117. B) In -ανάω. 1) βρανκανάσθαι' επι των κλαιόι>των παιείων λέγεται, ώς μίμημα φωνής Hesych. cp. βρνκανησομαι' βοήσομαι, Nicand. Alexipharm. 221 with the scholia. 2) Βεικανάομαι. ΰεικανόωντυ Ζεπασσιν Ο 86, ΰεικανόωντο επεσσιν ω 410 (cp. σ 111) in the sense of greet, while on the other hand the act. Ιεικ- αιάασκεν Theocr. 24, 56 ' showed/ for which reason Buttmann, probably rightly, connects the forms with ΰείκνυμι (cp. I 196 ΰεικνύμενος προςέψη) and έείΰεκτο. Hesych. has ΰεκαναται' ασπάζεται, perhaps formed from the rt. foic (cp. ΰέχνυμαι p. 110), just in the same way as the other from ΰεικ. 3) ερνκανάω, only α 199 υϊ που κεϊνον ερ^Αταΐ'όωσ' άίκοντα and in Q. Smyrn. Cp. έρυκάνω. 4) ισχανάω' ισχα\ άασκον Ο 723, ισχανόωντο Μ 38. Ορ. Ισχάνω. 5) κραυγανάομαι. Only Hdt. i. Ill παιΤίον κρανγαι ωμενον. 6) κϋκανάω Aristoph. Thesm. 852, ib. v. 429 κνρκαναν) both related 267 in some hitherto unexplained way to κυκάω, κνκεών, 7) όριγνάομαι cp. p. 111. 8) παμφανάω, epic from Homer onwards, an intensive from φαίνω like the related παμφαάω. Only in the participle αίγλη παμφανόωσα Β 458, τενχεα τταμφανόωντα Σ 144. 9) νφανάω. Only in Maneth. 6, 433 φάρεά θ' υφανόωντας. A by- form therefore of υφαίνω. 184 THE NASAL CLASS. IV. PEESENTS IX -νέω. 1) άγιέω. άγνεϊν αγειν Κρήνες Hesych. The form is a connecting link between άγω and άγινέω. 2) βυνέω Ιιαβυνέονται Hdt. iv. 71, εβύνουν Ar. Pax 645, cp. above p. 178 Ιιαβυνεται. Its origin is altogether obscure, βύσω, ϊβνσα show that νε is only a present tense expansion. In Aristotle there is the shorter present-form βύω, and still later βύζω. 3) Ζυνέω. A rare by-form of ϊννω (cp. p. 178), ενίννίονσι Hdt. in. 98. 4) θννέω. εθννεορ only Hes. Scut. Here. 210, 286 cp. θύνω. 5) ικιέομαι. Only twice in Homer ι 128 Ικνενμεναι, ω 339 ικνενμεθα. In common use with Herodotus and Attic poets. 6) -ισχ-νέο-μαι. άμπ ισχνού νται is an ill-attested reading in Aristoph. Av. 1090, which has been altered to άμπισχούνται or άμπίσχονται (Dind.). Perhaps though the form is a true one after all, as it stands in complete analogy to υπ-ισχ-νεο-μαι, which is in use from Herodotus onwards. The syllable has here been added to the reduplicated stem Ισχ for σι-σεχ. Cp. παμψανάω πιμπΧάνω. We have met with the same reduplicated stem in Ισχάνω and ισχανάω* 7) *κϊνεω only partly belongs here, inasmuch as the present-expansion has taken a permanent hold in the verb (κινήσω etc.). But κινίω is unmistakably founded on κίννμαι. 7b) *κοννέω* Aesch. Suppl. 9 κοννεϊς, 164 κοννώ for κοΖ-νέω. A present to the aor, ε-κάμεθα- ήσθόμεβα, rt. κοί (Princ. i. 186). 8) κύΐ'έω, Homer {κΰνεον φ 223), Attic poets and late writers. Homer 268 us es κνσσα, εκυσα, but it is usually the ease in the simple verb and always in the compound προι.κυνείν that the νε pervades all tense- stems. 9) οιχνέω•. Poetical by-form of οϊχομαι : y 322 οιχνεϋσι, ι 120 ε<ς- οιχνεϋσι, also οίχνεσκον, once in Pindar (Pyth. 5, 86 οιχνέοντες) and occasionally in the tragedians. 10) πιτνεω. This form which since Elmsley ad Eurip. Heracl. 77, Med. 53 has been expelled from the texts of the tragedians, is well attested by Herodian ad II. Π 827, where the connexion absolutely requires the reading πιτν&ν, which is defended by Lobeck, Lehrs, and Lentz, all the more that we know from other sources that Herodian approved of this form (Herodian, ed. Lentz ii. 1, 290). Since however there appear to be in one or two passages forms with short vowels in a present sense, especially Soph. O. C. 1754 προςπιτνόμεν (the M.SS. have the unmetrical προςπίπτομει), Buttmann is certainly right in his assump- tion that both forms, πίτνω and νιτνέω (cp. above p. 177), existed side by side, and this does not exclude the possibility of an aorist επιτνων which is demanded at Soph. O. G. 1733, with a part, πιτνών. Besides the tragedians Pindar appears to be the only writer who knows of these forms. Following the verbs in -νέω comes a single verb in -ινέω, just in the same way as a few verbs in -iva came after those in -ιω and -arm. 11) άγ'ινέω (cp. p. 183 and άγνέω p. 184) pretty often in Homer (άγινεΊ £ 105, ayirtov Ω 784) and Herodotus, elsewhere only here and ch. ix. NASAL PRESENTS WITH TRACES OF A j. 185 there. In the Homeric hymns there is also the fut. άγινήσω in which the present-expansion has gone further than the present. V, PRESENTS IN -νω WHICH POINT TO AN OLDER -νιω. Here again we must recognise two subdivisions : such verbs as add -ν-ιω, i.e. a conjunction of the nasal present-expansion of the first main class, and the mark of the i-class, to a stem ending in a vowel, and such as take an α as well before the -ν-ιω. άλθαίνω, i.e. άλ-θ-αν-ιω bears to βα -i'-tw exactly the same relation as that of άλφ-άν-ω to φθά-νω. A) -r -ίω ADDED TO YoWEL ROOTS. 269 The only verb we can put here with certainty is — βαίνω, in common use from Homer onwards (βαίνει Δ 443, εβαινον ο 145). The ν of the primary βα-ιιω appeal's in Greek only as a portion of the present-expansion, while the n of the corresponding ven-io (for gven -ίο) goes all through the verb. Cp..Princ. ii. 73. The other verbs of kindred origin (cp. below p. 215 f.) retain their ν in some at least of their tenses : κρίνω from κρι -v-ja) fut. κρϊνώ, aor. kk-plva, though the perf. is κ'-ψϊ -ra, κέ-κρϊ-μαι, so that strictly speaking the r is not in them to be regarded as an element of present-expansion, but rather as a root-determinative, the j alone, in conjunction with the thematic vowel, being the present-expansion. For this reason we place these verbs in the J-class. From the point of view of the historical development we must not be blind to the fact — seen in the relation of βαίνω to venio — that the processes of root-determination and present- expansion are often almost indistinguishable, as we have already had occasion to remark with reference to the ^T-class. Β) -αι-ιω ADDED TO CONSONANTAL ROOTS. 1) *άλαίνω, wander, Eurip. and later authors. It may be doubted whether the verb, like άλάομαι, is a denominative from άλη, or has come straight from the root, especially as there are no tenses but those of the present-stem. 2) άλθαίνω, Hippocr. and late poets, while in Homer άλθομαι (fut. ίπαλθησομαι) is in use. By-form άλθησκω with the variant άλθίσκω in Hippocr. The corresponding Skt. rt. ardh also forms the present occa- sionally by nasal additions : rdh-no-mi, rnadhmi. 3) άλιταίνω. Only άΚιταίνεται Hes. Opp. 330. Also ήλιτον, άλι- τέσθαι, άλιτημενος, all Homeric. 4) εριίαίνω Β 342 αντως γαρ επεεσσ εριίαίνομεν by the Side of Ψ 792 ποσσ\ν ερίίησασθαι (old variant εριζήσασθαι, as I. Bekker writes it). Owing to this fluctuation between the stem with ν and the stem without, this verb, although a denominative, is given in this list, while e.g. in the 270 case of ερώμαίνω (irritate, cp. ερεθίζω) ονερνθαίνω (ήρύθηνα Apoll. Bhod.) there was no reason for so doing. 5) καγκαίνω. Only καγκαίνει' θάλπει, ξηραίνει Hesych., cp. καγκο- μένης ' ξηράς τω ψόβω. Brugman Stud. vii. 205. 6) κηοαίνει' μερίμνη. Hesych. Μ. Schmidt is perhaps right in holding to be a mistake for κηραίνει which is rendered by the same verb. 183 THE NASAL CLASS. cir. ix. 7) κροαίνω, only κμοαίνων Ζ 507, Ο 264 and again in late• writers, clearly connected with κρούω. 8) κναίνω only in Hesych., who renders κυαίνων by εγκνος ώρ. 9) XtratVw, only Eurip. El. 1215, cp. λιτανενω^ λιτανεία. 10) νηψαίιω a by-form of νήψω in Eustathius. 11) όσφραίνομαι, from Euripides and Aristophanes onwards, fut. όσψρήσομαι, aor. ώσψροντο, Aristoph. Ach. 179 and elsewhere, clearly a compound, whose shorter forms are evidently of later origin than itself. One of the most remarkable of these is ώσψραντο (Hdt. i. 80). 12) τετρεμαίνω, only in Attic comedy, Aristoph. Nub. 294 τετρεμαίνω κα\ πεφόβημαι. In the case of several of these verbs it is impossible to determine whether the nasal syllable is to be regarded as an expansion of the present tense or of the verb-stem. The sum total of the verbs of the nasal-class is thus seen to be a very considerable one. The first division (νω) contains 27, the second (ανω, ινω) 50, the third (νσω, αναω) 14, the fourth (νε<>>) 12, the fifth (νβω) 12, in all 115. Of these 7 verbs it is true occur in 2 of these divisions and 1 in 3. Anyhow there are more than 100 verbs, and if we add to these the 50 verbs in νν-μι and the 8 in -νημι, some of which however occur over again in the thematic nasal class, we find that about 150 verbs make use of nasal elements in the formation of the present. Corresponding formations are to be found in great numbers in Sanskrit, and more particularly in the Vedas. In modern Greek presents in -νω have become far more numerous still (G. Meyer 48). THE INCHOATIVE CLASS. 187 CHAPTER X. THE INCHOATIVE CLASS. While the verbs we have to deal with in this chapter are by no means 271 without parallels in the wider domain of the related languages, still this kind of present-formation is more than others confined to the Graeco- Italic circle. It is only here that they have developed to any great extent and with any regularity, so that in the common possession of an in- choative class is found to be one of the strongest arguments for. the specially close connexion of Greek with the Italian languages. From another point of view the present-expansion under discussion is distin- guished from all others by the fact that there is attached to it, though here again almost' exclusively within its own narrower circle, and even there not universally, a clear and definite meaning. In one set of in- stances this meaning is as unmistakable in Greek as it is in Latin, and this fact is one which may throw light on the process of present-expansion generally. For if the present-expansion has in this particular instance to express a special meaning, it is natural to suppose that the variety shown by other forms of such expansion was no mere idle and fortuitous complexity, but was bound up with the effort to distinguish the con- tinuous action not only quantitatively, so to speak, but also qualitatively from the momentary. In the verbs of the class now before us it is the action that is gradually coming into being which is distinguished from that which has attained its full realisation. We therefore call this class, notwithstanding the fact that the meaning as above described does not in all cases stand out with equal distinctness, the inchoative class. The mark of this class from the beginning was the syllable -ska. Traces of this syllable in the same or a like application are to be found in most of the related languages. No one it is true will now venture to adopt Bopp's assumption (Vergl. Gr. iii. 2 104) that the numerous 272 Sanskrit desideratives are to be identified with the present-formation now under consideration. Desideratives have two things in common with the inchoatives, the sibilant and the tendency to reduplication. It is conceivable therefore that some one should hit upon the idea of connect- ing the Skt ;). 2) άέ-σκο-ντο (also αίέσκοντο)' άνεπαυοντο, έκοιμώντο Hesych. (αέσκω is cited by Herodian i. 436), a present to the Homeric aor. άεσα. 282 3) αλθη-σκω intrans. heal, only in Hippocrates, with the variant άλθίσκω (like Class IV.). Other present-forms are άλθο-μαι and άλθαίνω (trans, cp. p. 185). αλθεξις Hippocr. 4) άλν-σκω. άλνσκων χ 363, 382, elsewhere only in Apoll. Rhod. The forms άλύζω, ήλνζα in Homer and the tragedians suggest that αΧΰσκω has come from *άλυκ~σκω. But κακόν μόρον έζαλύοντες hymn, jn Bacch. v. 51, and the common forms άλεύω, αλέομαι point to a vocalic stem άλν, which must have been developed from άλ (cp. άλη, άλάομαι) in a similar way to that in which fcpv watch, discussed on p. 122, was developed from Ρερ, the Skt. var, and l\v from Ρελ. The guttural stem therefore must have been made either independently of the inchoative present-form or else out of it, by the repression of the sigma. 5) βιώ-σκο-μαι, only Aristot. Meteorol. i. 14 έτεροι τόποι βιωσκονται, άναβιώσκεται Plato .Symp. 203 e, besides which there is in late prose the active άναβιώσκω. The corresponding aorist-forms are of more frequent occurrence. 6) γανύ-σκο-μαι first cited from Themistius. Cp. γάΐ'ν-μαι above • p. 112. 7) γενειά'-σκω Plato Symp. 181 d, Xenoph. to get a'beard. By-form γενειάζω, γενειάω. • 8) γηρά-σκω common to all Greek from Homer onwards (γήρασκε Ρ 325, γηράσκει η 120), by the side of γηράν in the same sense (Xen. Cyr. iv. 1, 75). Cp. έγήράν p. 134 and Lobeck on Buttm. Ausf. Gr. ii. 393. 9) ηβά-σκω from Euripides onwards, in pretty much the same sense as ηβάω, although Moeris p. 198 Be. says : ηβάσκειν επι των παίδων των αρχομένων ηβαν ως επι το πλείστον Αττικοί. Eurip. Ale. 1085 ιΰν y εθ % ίιβάσκει κακόν, Xenoph. Anab. iv. 6, 1 πλί)ν του νιου του Άρτι ηβάσκοντος. ch. x. PRESENTS IN -ι-σκω AND -ε-σκω. 195 10) ηλάσκω. Β 470 at τε κατά σταθμον ποιμνηιον ήλάσκουσιν, cp. Ν 104, by-form ήλασκάζω also epic. 11) ίλά-σκο-μαι I propitiate Ζ 380, Α 472 and later, always tran- sitive. In the same sense Ί'λαμαι hymn. Horn. 21, 5, Homeric Ιλάομαι (cp. p. 119), Aesch. Suppl. 117,127 ιλέομαι. ιληκτ)σι on the other hand stands at φ 365 in an intransitive sense. The κ in this form is of the same nature as the guttural which appears in άλύζω. 12) μεθύ-σκω, intoxicate, more commonly μεθύ-σκο-μαι in Herodotus 283 and Attic prose. . 13) πωτάσκεται (?) only Orac. Chald. xxviii. p. 23 according to Lobeck Rhem. 249, though according to Steph. Thes. s.v.the M.SS. have πωτάσκετο, which may be an iterative. 14) σελά-σκων λάμπω ν Theognosti Canones, Cramer Anecd. Oxon. iv. pp. 11, 19. 15) τερύσκεται' νοσεί, φθίνει Hesych. to be compared with τερύσκετο' έτε'ιρετο and τέρν (better τερυ)• ασθενές, λεπτόν. 16) τρωπασκέσθω' μεταβαλλέσθω, έπιστρεφέσθω. IV. -σκω, -σκο-μαι AFFIXED AFTER THE ADDITION OF A SHORT VOWEL. 1) ά\θ-ί-σκω a variant to the άλθή-σκω in Hippocrates, mentioned at III 3. 2) άλ-ί-σκο-μαι from Pindar onwards in poetry and prose. Homer has only ήλων, άλώμεναι etc. The root (Princ. ii. 169) must be fa\, βελ (ά\ν-σι-ς, ε'ίΧλω). 3) άμβλακ-ί-σκω, a Doric present to the aor. ήμβλακο-ν (Archil, fr. 73 Be. 3 ), which in Pindar and the tragedians is ήμπλακον. άμβλακίσκω is only attested by two passages of the Pythagorean Theages in -Stobaeus Floril. i. 67, 68 and one of the Pythagorean authoress Phintys Stob. Floril. lxxiv. 61, and άμπΧακίσκω not at all. 4) αμβΧ-ί-σκω from Plato onwards, especially in compounds. The forms from the st. άμβλω are more frequent. Eurip. (Androm. 356) has also εΐαμβλοϋμεν in a causative sense. Hesychius gives the by-form άμβλν-σκει' έζαμβλοΊ. κυρίως δε έπι αμπέλου, καΐ εκητρώσκει. Σοφοκλής ΆνΰρομεΊτι. άμβλωσκω, attested by Suidas, is cited in Steph. Thes. from Galen and other late prose writers. Perhaps άμβλΰ-ς or a by-form *άμβλο (Princ. i. 406, ii. 396) is the word which furnished the stem for the verb. 5) άναλ-ί-σκω from Pindar onwards in poetry and prose by the side of άναλόω (e.g. άναλοϋν C. I. A. I. 55, 3). 6) *άπαψ-ί-σκω only λ 217, έξαπαφίσκων Hes. Th. 537 *(v. 1. εζαπα- . -ίσκων). The aorist-forms occur pretty often in poets. 7) *άραρ-ί-σκω only ζ 23 αυτός δ' άμφ\ πόΰεσσι εοΊς άράρισκε πέΰιλα, and in imitation of this passage Theocr. 25, 103. 8) άρ-ε-σκω, from Herodotus onwards in poetry and prose ; Homer 284 has only the aorist-forms in a somewhat different meaning. 9) (επ)-αυρ-ί-σκω. του ΰέ τε πολλοί επαυρίσκοντ άνθρωποι Ν 733. The active occurs in Theogn. Ill in a causative sense ol δ' αγαθοί το μέγιστο ν επαυρίσκουσι παθόντες, and thus Bergk's scruples about this passage fall to the ground. By-form έπαυρέω Hes. Opp. 419. o2 196 THE INCHOATIVE CLASS. ch. x. 10) (έκ)-γαμ-ί-σκω Ν. Τ., a by-form of έκγαμίζω. 11) γε γεγων-ί-σκω in the tragedians and Thucydides, so that it is a .posthumous present to the Homeric perf. γέγωνε, plupf. έγεγώνει. By- form γεγωνέω.^ 12) ευρ-ί-σκω. Of the present-forms, which occur everywhere from Pindar onwards, ευρίσκω only occurs in Homer, and that but once (r 158), while εύρε is of frequent occurrence. 13) κορ-έ-σκω, a late present to εκόρεσα, εκορέσσατο etc., attested by passages from Nicander e.g. Alexiph. 415. Cp. κορένννμι. κορεσκων seems also to have been used in the sense of εξυβρίζω ν (Hesych. s. v. κορέων). In Hippocrates περί αδένων p. 271, 31 of the Geneva edition there is κορίσκονται πολλής υγρασίης in the sense of abundant. 14) κυ-ί-σκω, a by-form of κυω, κυέω attested from Hippocrates. The middle occurs in Herodotus (ii. 93) and Plato. 4 15) όφλ-ί-σκω, only mentioned by Suidas as an alternative for the Attic όφλισκάνω. 16) ρυ-ί-σκο-μαι, only found in Heliodorus and Eustathius in the sense of ρέω. έρυίσκετο ' ερρεεν, έχεϊτο Hesych. 17) στερ-ί-σκω, Soph. Ο. C. 376 άποστερίσκει, Thuc. ii. 43, the middle rather more common with the older Attic writers, by the side of στέρο- μαι. 18) τελέσκω complete, only Nicander fr. 74, 10 τελέσκων, cp. O. Schneider Nicandrea p. 96, while others write τελίσκων. The form with 285 an ι is given by Hesych. τελισκόμενος' πληρούμενος, τελεωύμενος. It does not occur before the Christian period, though Herodian i. 436 gives the active. 18 b) τιεσκομένβι C. I. 3538, in a metrical oracle 12 (cp. Nauck Melanges iv. 36). 19) χλοιΰ-έ-σκω, only in Hesych. in the gloss χλοιΰέσκουσαι ' γαστρί- ζουσαι i.e. filling the belly, fattening, so that it is a by-form to χλοώάν ΰιέλκεσθαι και τρνφάν, the perfect to which is κέχλοιΰε' ΰιέλκετο. 20) χρη-ί-σκο-μαι, only Hdt. iii. 17: χρη-ί-σκονται (M.SS. χρήισκοντο) τψ νοατι, by-form to χράομαι in an iterative sense. V. -σκω, -σκο-μαι -ADDED IMMEDIATELY TO CONSONANTAL ROOTS. 1) *οι-Ιά-σκω, which only belongs here on the assumption that the rt. £αχ (piObably from Ιακ, and so;=Zd. dakh-sh teach, Lat. die in disco, di-dic-i, doc in doc-eo) is at the bottom of the whole verb, and conse- quently that the present-form which was common to all Greek from Homer onwards came from *οΊ-ΰαχ-σκω or *ΰι.-ΰακ-σκω and not directly from the rt. δα (δέ-ΰα-εν, ε-Ιάη-ν). Cp. Princ. i. 284, Fick i. 3 611. We met with a corresponding fluctuation between a vocalic root and one which had been expanded by a guttural in the case of άλΰσκω (iii. 4) and Ίλάσκομαι (iii. 11). The fact that forms like εδίδαξε, δεδίδαγ/ζαι are as old 8 Nauck discusses this verb in detail, Melanges iv. 41 ff. He denies the ex- istence of a perfect yeywva, and prefers to call yiyatvs an aorist. But the ω does not suit this view (cp. Ch. XIII. ii.). * μάχ^σκω is only mentioned by Herodian i. 436. The alphabetical arrange- ment at the place is faulty, and so it is quite possible that this form has usurped the place of another. ' ch. χ. PEESENTS IN WHICH THE -σκ IS TEANSEOEMED. 197 as Homer goes to prove that ΰώαχ was in quite early times recognised as the stem. 2) *ΰει-()ί-σκο-μαι or hrdi -σκο-μαι γ 41 ΰειΰισκόμενος (ο 150 ΰεΰισκό- μενος) Ιε προςηνΰα, σ 121 Ιέπα'ί χρνσέω ΰειΒίσκετο (cp. ν 197). The pre- cisely identical use of ΰεικανόωντο and οεικνύμενος (cp. above, p. 183) undoubtedly entitles us to refer the word to the rt. ΰικ (οείκννμι). 3) *έ-'ί-(Τκω, 'ί-σκω (ρίσκων, "ισκουσα) a causative present to εοικα, 'ίκελος, almost confined to Homer (impf. ήισκον, ΐίσκον), so that the rt. is clearly Ik. — On the impf. ίσκε he said (r 203, χ 31) from the rt. σεκ (σεπ) the reader may be referred to Princ. ii. 68. 4) Η-ά-σκ-ειν' άγειν Hesych. which Lobeck Rhem. 249 and M. Schmidt found obscure, but which may probably be very simply explained as an inchoative present from the rt. άγ. It stands then for l -αγ-σκω. 286 "We met with I as a reduplication of a in l -αΰ-ω. . Cp. Princ. i. 484. 5) κλώ-σκω. Only κλώσκων έπικλώθων Hesych., so that it is for κλωθ-σκω. On its origin cp. p. 157. 6) πινυ-σκω, a denominative, apparently, from πινντό-ς, since it means just ' to make wise, prudent.' So Aesch. Pers. 830 προς ταϋτ εκείνον ενφρονεΊν κεχρημένον πινύσκετ ενΧόγοισι. νονθετημασιν, and the aor. occurs at }Ξ} 249 ηίη yap με και αΧλο τεη επίνυσσεν εφετμή. Cp. Simon. C. fr. .12 Be. 3 7) τι-τν-σκο-μαι. Γ 80 Ιοϊσίν τε τιτνσκόμενοι Χάεσΰί τ εβαλλον, τιτΰ- σκετο θ 41, Ν 23, also used in the sense of τεΰχειν. The active is used by late poets in the latter sense. Other inchoative forms from an equivalent of this stem are άποθύσκειν (M.S. άποθύκειν)' αποτυγχάνει ν, ενθνσκει ' έντνγχάνει, with a remarkable shifting of the aspiration, which was lost before the σκ, to the initial, τετύσκων εμφανίζων, τετνσκετο' κατεσκευάζετο, all in Hesych. 8) γά-σκω found from Solon onwards (fr. 13, 36 Be. 3 ), apparently for χαν-σκω, cp. χαίνω, aor. ε-χαν-ο-ν, pf. κέχηνα, while the Lat. hi-sco shows no nasal. άμφίσκοντες' ενΰυόμενοι .(cp. επαμφίσκω), which might at first sight be taken for an inchoative of the rt. /ες, so strongly resembles the form . άμπίσχω (more commonly άμπισχνεομαι) that we may certainly follow Steph. Thes. in regarding it as merely a by-form of the latter, αμψίσκω : *άμφίσχω '. \ σώθητι '. *σώθηθι. We have thus in this form a second instance of the progressive dissimilation in the case of neighbouring aspirated syllables. VI. A TEAKSFORMED -σκω. 1) ΰύ-σγω' άποΰύω Hesych. The γ, as in μίσγω, has arisen from κ. Cp. ΰιΰΰσκω above, p. 193, and μίσγω. 2) ερ-χο-μαι common to all Greek from Homer onwards (N 256). Cp. Princ. ii. 366. 3) μί-σγω used extensively from Homer onwards both in the active and the middle. The Lat. mis-c-eo shows most clearly that the γ has been weakened from κ. This is confirmed by the Skt. mik-sh, mi-mik-sh (Princ. i. 417), from which it might almost be inferred that the course of the expansion was as follows : mik (Skt. mic r rd-8 mixed), rnik-s (Skt. 287 mik-sh), mik-sk (Lat. misc-eo). A glance at Ιΰσγω is enough to teach us that we ought not to explain the y of μίσγω in the way taken by Ahrens, 198 THE INCHOATIVE CLASS. ch. x. who at Formenl. p. 123 says that ' the remarkable γ has been brought about by the transformation of the κ.* It would hardly be possible to find an analogy for such a softening influence. It may even be asked whether the original ksk had not already been softened to sk before the softening of the κ between vowels informs like έμίγην, μιγάς etc., so that from the primary *mik-skd-mi would have come even in Greek a *μισκω. It would not in itself be improbable that the softening of the simple k to γ in forms like εμίγην should by the force of analogy have had something to do with the softening of the sk. Joh. Schmidt however (Vocal, i. 123) makes the very plausible conjecture that the well-attested natural length of the vowel in μίσγω, μϊξαι, μϊκτο is due to the after-effects of a nasal, so that we should have to assume a primary form μιγγ-σκω, in which the first γ would have arisen from the syllable w in μίγννμι. For the softened δύσγω too there occurs a nasal formation in Ιύνω (cp. above, p. 178). So it may be that its primary form was *όνν-σκω, and that the nasal was in both cases the real source of the softening. 4) πά-σχω common to all Greek (αλγεα πάσχει Υ 297). Cp. Princ. ii. 365 f., where objection is taken to the wide-spread assumption that the loss of a θ is the cause of the aspiration. The comparison• of the Lat. pa-ti-o-r and of πέν-ο-μαι, πόνο-ς points conclusively to the assumption that the root-syllable was πα (apparently for σπα, cp. σπά-νι-ς and Princ. ii. 356). ε-πα-θυ-ν, and πέ-πον-θα are expanded by a Θ. The Sicilian perfect π έπασχα (Ahrens, Dor. 351) has been formed in striking analogy to the present. The different view of this verb taken by Joh. Schmidt (Voc. i. 93) fails to convince me. Synonymic differences such as have arisen not only between πένεσθαι, πόνος on one side and πάσχειν, παθέιν on the other, but also between πάθος and the undoubtedly related πένθος (mourning) ought not to induce us to separate the stems πεν and παθ. . In πεν-ι-χ-ρό-ς (y 348), πέν-η-ς, πεν-ίη (ξ 157) we have a modifica- tion of meaning in the case of πεν precisely similar to that which has prevailed in παθεΊν and πάσχειν. Still less am I inclined to separate 288 the Lat. pa-ti-o -τ, which in the wide ramifications of its meaning is completely equivalent to παθεϊν, and πή-μα (πήματα πάσχειν) from the form of the stem which ends in Θ. pa-ti-o -τ comes from the rt. pa, as po-ti-or from po (cp. fa-te-o-r). Besides these I have at Princ. ii. 365 f. tried to make good the assertion that the following 8 verbs, whose stem ends in χ even outside the present tense, owe this consonant to a softening from σκ. γλίχομαι in Herodotus, Aristophanes, and Demosthenes. Forms belonging to other stems than the present are of quite isolated occur- rence, e.g. εγλιζάμην Plato Com. ii. 695 Mein. γλίσχρο-ς (Princ. i. 458) perhaps contains the sibilant which we assume to have existed before χ, whiJe γλί -u, γλοι-ό-ς seem to give the root. εύχομαι, common to all Greek from Homer onwards, accompanied by a plentiful noun-formation : ευχή, ευχωλή, ευχετάασθαι and forms like ενζομαι (Soph.), ευκτο (above, p. 131), η'ύζατο (Pind. Aesch.). The Skt. vdnJcha-ti he wishes, desires (for van-ska-ti) thoroughly corresponds to the meaning wish, as does vdnttlia wish to ευχή and the O.H.G. lounsc. Both words are derived in the Pet. Diet, from van wish for. 5 ευ=να 6 Roth (ZtscJir. xix. 220) however connects the stem «δχ with the Skt. vfyh-at, the offering, presenting one, making vagh the rt. So too Fick» i. 7G5. ch. x. PKESENTS IN WHICH THE -σ /c IS TRANSFORMED. 199 may be plainly seen in ευρύ-ς from *varu-s (Skt. uru-s Princ. i. 431), and in t&vtrc bereft=Skt. vanja, una (Bugge Stud. iv. 328). j'j /χω Homeric (νηχέμεναι ε 375), also νήξομαι ε 364. σμηχω ζ 226 εσμηχον by the side of σμάω in Hdt. and Aristoph., νεό-σμηκ-το-ς Ν 342, ΰιασμηχθείς Aristoph. στεν-ά-χω Π 391 and elsewhere by the side of στενέ Κ 16. Here there are no forms with a ξ. The iterative στενάχεσκε Τ 132 is no objection to the view that χ stands for an earlier σκ, as is shown by βοσκέσκοντο. τρύχω. τρνχόμενος a 288, cp. τερύ-σκω above, p. 195. Still as early as ρ 387 there is τρΰζω. φήχω by the side of ψάω, post-Homeric, ψρ Soph. Tr. 678 by the side of έψηκται ib. 698. ψύχω άνεψνχον Ν 84, ψνξασα Υ 440; ψ"χ>?, ψυχρός, ψϋχος also show the χ, and it is only φϋ-σάω and the other forms put with it at Princ. ii. 117 which are to be referred to a vocalic stem. If this conjecture is correct there is no other explanation left for the 289 2; except that either, as we assumed in the case of άλΰζω, Ιλήκησι, £ί£άξω, they are to be referred to stems which have been expanded by κ, or, and this seems to me the simpler view, that the ξ made its way into the future and aorist on the analogy of ελέζατο by the side of λεχος, ελέγξω by the side of ελέγχω, εΰέζατο by the side of Ιέχαμαι etc. The sum total then of the inchoative verbs in Greek is made up as follows : in the first division there are 13, in the second 6, in the third 16, in the fourth 21, in the fifth 8, in the sixth 12, that would be in all 76. Since however τερΰσκω in Hi. was originally identical with τρύχω in vi., and άλθήσκω (iii. 3) with άλθίσκω (iv. 1) we must subtract two from this number, and this gives us a total of 74, not quite half the number, that is, of the verbs of the nasal class. It must be remarked moreover that very many of these presents .do not occur till late, and that not a few, though given in our grammars as the regular forms, are of quite isolated occurrence. This is especially the case with ΰεοίσκομαι (i. 4), ΰιΰνσκω (i. 6), θράσκω (i. 7), κατ-ε-κίκλασκε (i. 7b), πιπίσκω (i. 9), πιπράσκω (i. 10), βλώσκω (ii. 1), βιβρώσκω (ii. 2), ρήσκομαι (ii. 8), άάσκω (iii. 1), άέσκοντο (iii. 2), άλθήσκω (iii. 3), άλύσκω (iii. 4), γαννσκομαι {iii. 6), ι/λάσκω (iii. 10), πωτάσκεται (iii. 13), σελάσκω (iii. 14), τερνσκω (iii. 15), τρωπάσκω (iii. 16), τιεσκόμενοι (iii. 18b), άμβλακίσκω (iv. 3), άπαφίσκω (iv. 6), άραρίσκω (iv. 7), εκγαμίσκω (iv. 10), κορέσκω (iv. 13), ρυ'ίσκομαι (iv. 16), τελέσκω (iv. 18), χλοιΰέσκω (iv. 19), χρηίσκομαι (iv. 20), ιάσκω (v. 4), κλώσκω (v. 5), δΰσγω (vi. 1). After the subtraction of these 32 rare forms there remain about 40 verbs in which this present- formation was actually in constant use. Finally, as regards the meaning of this present-expansion, if it had not been for the abundantly attested inchoative meaning in Latin verbs like adolescere, reviviscere, pubescere, senescere, clarescere etc., it would perhaps hardly have occurred to anyone to ascribe even partially the ex- pression of the same notion to the Greek verbs of like formation. As a fact out of the 74 present-forms of this class only 5, i.e. the Hero- dotean ΰιαψώσκείΐ' {illucescere), άναβιώσκομαί (revivisco), γενειάσκω, γηράσκω (senesco), ήβάσκω (pubesco) have an unmistakably inchoative 290 meaning. When once awake to this fact we shall perhaps go on to admit that the action also in γιγνωσκω (gnosco), μφνησκω (reminiscor), 200 THE INCHOATIVE CLASS. ch. x. ϊιΰάσκω, τιτνσκομαι, and possibly in βάσκω, οιΛω and κικλησκω is repre- sented as gradually arriving at completion. A number of these verbs have in the present stem, and that partly in contrast to the rest of the verb, a decidedly causative meaning. This is specially the case with επιβάσκω, δεΰίσκομαι make frightened, πιπίσκω, μεΟύσκω, εκγαμίσκω, έίσκω, πιννσκω. άναβιώσκομαι is used sometimes in a simply inchoative, some- times in a causative sense. That the inchoative meaning veered straight round to the causative we are hardly entitled to assume. It was rather that the operation expressed by some of. these verbs was from the first a gradual operation. Gradual upspringing and gradual operation met in the same form, and this is by no means the only case where one and the same form is made the vehicle for an intransitive and a causative meaning. Later on usage, as it often did also in the case of intran- sitive inchoative forms, allowed the notion of gradualness to fall away, and so nothing but. the causative meaning was left. For βάσκειν e.g. we suppose the primary meaning to have been 'to get gradually into motion,' and to this was added the causative ' to set gradually in motion/ and hence for the compound with επί ' to bring gradually nearer to something.' The distinction between that which comes about and that which is brought about formed, to begin with, no more of a special expression here than, say, in στή-σω, ε-στη-σα, εβη-σα, as con- trasted with ε-στη-ν, ε-βη-ν, or in verbs like ελαννην, αγε«ν. After this contrast between επιβάσκω and επιβαίνω, επιβηναι etc. had been once developed, the special expression of gradualness which had really been the primary meaning of the form, fell quite into abeyance, and επιβάσκω in this way came to be a purely causative verb. Thus viewed these very causatives are also witnesses to a period in which the σκ was a present-expansion with a definite meaning. The vulgar dialect of Rome, as Lowe (Prodromus corp. glossar. 362) points out, shows the same change of meaning in e.g. 'ferascit ferum facit, pravescere depravare. In the case of the great majority of the verbs of this class it must be admitted, it is true, that all recollection of this early state of things had as entirely disappeared as in the case of the Latin verbs nancisci, pacisci, 291 ulcisci, proficisci, pascere. Among the Sanskrit forms which we re r cognised above as belonging here in form, at least two have an unmistakably inchoative meaning, i.e. uMhd-ti illucescit and murlcha-ti it curdles, grows firm, stiff. We may see in this a remarkable trace of an inchoative meaning in the syllable ska (Jcha), a meaning originally existing, we may assume, in the Indian languages as well. ch. χι. THE I-CLASS. 201 CHAPTER XL THE I-CLASS. There is probably no discovery made by Comparative Philology which has contributed so much towards a clear understanding of the structure of the Greek verb as the disco\ T ery of the i-class. Buttmann, who so often showed a deeper insight than his contem- poraries, got no further than the perception, expressed under the head of ' double themes ' (Ausf. Gr. i. 2 367) with reference to presents like φαίνω, βάλλω, τάσσω, φράζω, that ' in a large number of verbs the stem of the word ' appears * in the present in a longer, fuller form, produced sometimes by a long vowel or diphthong, sometimes by the addition or the variation of consonants.' Least of all was this a satisfactory account of the presents in -σσω and -ζω, as in fact the ' variation of the con- sonants ' was left quite incomprehensible. It was not for a moment suspected that it might be possible to explain the four verbs selected above as examples, and those like them, on a single principle, notwith- standing that it would have been possible to arrive at the truth merely from a close examination of Latin verbs in -io in connexion with the alterations manifest in the comparatives in -ιων, without any aid from Sanskrit. Bopp Vgl. Gr. i. 2 211 acknowledges that it was the analysis of the Greek comparatives which first led him to discern the connexion between Greek verbs in -σσω and -λλ<»> and the Sanskrit verbs of the 292 fourth class (1st sing, -jo-mi), and this is why, in my ' Tempora und Modi,' I devoted such a considerable space, — and the condition of the science at the time made this quite necessary — to the parallelism between the formation of the comparative and that of the present. Since that time the analogies from Greek have by Bopp himself, by Schleicher and others, been placed in so clear a light that no doubt on the main points is any longer possible. Controversies exist only on a few side questions and single points, and on the origin of the whole class. In respect to these questions I will deal only with such ground as has not been already covered by me in my ' Principles of Greek Etymology.' Our main task here is to demonstrate the original unity of the whole mass of the present-formations, apparently so diverse, which belong to this class. Such a result can be welcomed even by one who still feels some doubts as to the origin of the whole phenomenon. It is a settled fact that the primitive Indo-Germanic language distin- guished a large number of present-stems from the verb-stem by affixing the syllable jo. As j and i are constantly interchanged before vowels, we may expect at starting to find io as well as jo in the various indi- vidual languages, and to find both forms of this one element represented by such substitutes as the phonetic laws of the single languages would lead us to expect, jo can be clearly seen in 4 families, in Sanskrit, where 202 THE I-CLASS. the class of verbs characterised by ja is given as the 4th, in Zend, in Slavonic, and in Gothic : Skt. hip pres. kup-ja-mi ' I become agitated. Zd. veres „ verez-ya-mi I do. Ch.-Sl. zna „ zna-jq, I know. Goth, haf „ haf-ja I heave. ia appears in the Latin verbs of the so-called third conjugation in -io : fug fug-io. In Lithuanian we have the same interchange between ja and ia which we shall presently see to have taken place in Greek. The ia occurs in 293 rt. ar pres. ar-iu I plough, the ja both in derivative verbs e.g. Uiido-ju I bury, and in primary verbs with the phonetic change of j to z, which is pronounced like the French j : rt. sed pres, sM-Zw I sit. We are accordingly entitled to expect to find the forms in -ja-mi represented in Greek sometimes by a vocalic -ιω, or perhaps (but of that later) -εω, sometimes by the old -ju and all the transformations to which such a syllable would by Greek phonetic laws have been liable. We derive the most material assistance here from the analogy of the com- parative, the suffix of which is to be referred to the primary form -jans. Compare : st. ήδυ com „ πλβ , par. ήδ-ίων , πλε-ίων a ad rt. (σΡ)ιδ , ,, δα pres >> ίδ-ιω. δα-ίω. „ μαλ , , μάλλον ρ „ βαλ »> |3άλλω. „ άμεν , , άμΐίνων >» „ τεν j» τείνω. ν Χ*Ρ „ η* „ ελαχν , , χειρών , ησσων , ίλάσσων „ rep ,, λβνκ „ ταραχ τε'ιρω. λενσσω. ταράσσω. „ κρατυ , „ βαθύ , ,, ολίγο , , κρείσσων , βάσσων , (v7r) ολιζων 1 > > „ λιτ ,, κορνθ ν Τ Ρ ι Ύ λίσσομαι. κορνσσω. τρίζω. It is only for the change from ξ; to ζ which we have to assume for ε£&μαί, Ίζω that we have no analogy among the comparatives, though this lack is fully compensated for by parallels in other directions e.g. ά(ογυ/Ε>ύ-7Γ£^α, compared with the feminine of the Lat. acu-pediu-s (Princ. i. 161), Lesb. £a=ordinary Greek 2ιά. The Sanskrit 4th class of verbs is one of great extent. According to Bopp it contains 130 verbs, to which have to be added a few roots in a, which are classed by the Indian grammarians as roots in e and o. Boehtlingk in a note to his Sanskritchrestomathie p. 279 was the first to bring tins last fact to light. Consequently e.g. the rt. dhd suckle, pres. dha-ja-mi, rt. ςά sharpen, pres. ζ -jd-mi belong to this class. This makes the total a still larger one. The Skt. med-ja-mi, from the rt. mid get fat, is the one solitary instance in that language of intensification of the 1 On the newly found Attic form ολείζων (C. I. A. 1 B, 33 etc.) cf. Cauer Stud, viii. 254. CH. XI. EXTENT AND ANTIQUITY OF THE CLASS. 203 root- vowel combined with the addition of the syllable ja, as in the Latin 294 mejo for meig-io from the rt. mig,. and in the Gk. πλήσσω rt. πλαγ. In Zend there are, according to Justi, not so very many of these verbs to be found. From Old- Persian Spiegel (AJtpers. Keilinschr. p. 166) knows of only a single instance. In Latin there are the following 15 verbs which belong directly here : cap-io, cup-io, fac-io, fod-io, fug-io, grad-io-r, jac-io, lac-io, mor-io-r, quat-io, par -ίο, pat-io-r, rap -ίο, sap-io, spec -ίο. But there are some more to be added ; for, as Struve (iib. d. lat. Declination und Conjugation p. 199) has well shown, the boundary line between these verbs of the so-called 3rd conjugation and those of the 4th which, \\kefarc-io,fulc-io, or-io-r, sal-io, are saddled with an i only in the present-stem, is not very clearly drawn. In the very earliest Latin there appear forms like parlre=parere, 7iiorlri-=mori, cuplre, desipire and the like. It. was only in the course of time that the fashion became established of regularly expelling the i of the stem in certain verbs before a short er, and keeping it everywhere in the form of a con- traction in others. The difference between the two sets of verbs is not enough to constitute a. difference of conjugation ; we ought rather to place all verbs whose i is movable (as contrasted with that of audio, audivi etc.) in this class. And even in cases where the i goes right through all forms, e.g. in mug -ίο, we are no more excluded from the sup- position that it may in the beginning have been a present-expansion, than we are in the case of the nasal of jungo and other formations of that class. The class-characteristic is not so evident at first sight in ajo, jiiejo, which no one who looks at nid-jor=mag-ior can doubt to have originated in ag4o, meig-io. Gothic has only 8 verbs in which the syllable ja characterises the present-stem as such : bid-jan heg,frath-jan understand, haf-jan heave, Jdah-jan laugh, rath-jan count, skath-jan injure, skap-jan shape, make, and vahs-jan wax, grow (Leo Meyer Goth. Sprache p. 350)., Here as in so many other cases Greek surpasses most of the other languages in the abundance of the forms preserved, though all kinds of transformations have so modified the original formation that it is almost undistinguish- able. The cases in which the formative syllable ja can be shown to have been affixed to the same stem in more than one Indo-Germanic family of languages are the following 20 : sal-io. . 295 spir-iu (inf. sjnr-ti Princ. i. 358). d-jd-mi (cut). dii'-iu (flay). sid-zu. sub -βο. svkl-ja-mi. /■ >Δ^ gloc-io. / ν Ay h'mk-iu (Princ. i. 196). '/ - > ,\ main-ye-te (he thinks) \ (cp. the Gk. μαίνεται Piinc. i. 387). mel-jq (I grind). / x^ &Φ Q V -A? £ αλΧομαι Lat. άσπαίρω Lith. δαίω divide Skt. δβι'ρω Lith. Ζζομαι Lith. θυίω Lat. Ιδίω Skt. Lat. cap-io Goth. Skt. kup-ja-mi Lat. κΚώσσω Lat. λβνσσω Lith. Skt. man-ja-te Zd. μνΧλω Ch.-Sl. οζω Lith. πτίσσω Old-La V -hi (smell) jnns-io. 296 204 THE I-CLASS. CH. xi. ρεζω, Ζρδω Zd. verez-ya-mi. ρνζω Lat. rug-io. Skt. (8)pac-ja-mi Zd. ςρας -ya Lat. spec-io, Zd. ukhsh-ya-nt (part.) Goth. vahs-ja. φράσσω Lat. farc-io. Besides these there are a few more instances, some of which are doubtful, while others are of an exceptional character. The Gk. σφάλλω, for instance, and the Lat. faUo can only be compared on the assump- tion that Ij sometimes turns to 11 in Latin also, μνζειν groan can be compared with mug-i-re if the g of the latter is not a weakened k as might be inferred from μνκάο-μαι. Undoubtedly the Lat. mor-io-r is to be compared with the Skt. mri-jd-te he dies=Zd. (fra)-mair-yei-te and the Old-Pers. a-mar-iya-td he died (Joh. Schmidt Voc. 244). But the syllable ja has in the Sanskrit word the' force of the mark of the passive voice, βαίνω, as we remarked on p. 185, is of the same forma- tion as the Lat. ven -ίο, but in Latin, as in Oscan and Umbrian (3rd sing. fat. ex. ben-ust) the nasal sticks fast to the verb-stem all through, while in the Greek verb it appears only in the present. From a Greek point of view then βαίνω is one of the verbs in which the nasal class and the ί -class are united, but ven -ίο belongs exclusively to the i-class. In the cases of τρείω by the side of τρέω, and παίω=ραν-ίο conjectural com- parisons will be given below. Having thus set the antiquity of this class of verbs in the right light we have now to consider what was the origin of the syllable ja. On this point there are practically only two views to choose from. Either the syllable -ja is just as much a noun-suffix as, according to the view argued out on pp. 108 f. and 164, are the syllables -na, -nu, and -ta, which constitute the marks of the nasal class and the ί -class, or else we have to deal with quite another sort of formation, i.e. a compound ; in other words, that is, the syllable ja is of verbal origin and identical with the verbal root ja, Skt. ja. Each of these two views has redoubtable names on its side. Schleicher avows the former (Comp. 3 753), and the latter was first stated by Bopp (Vgl. Gr. ii. 2 357), and adopted, among others, by Benfey and Max Miiller. I have myself repeatedly (especially in my Erlauterungen 3 103, in my 'Zur Chronologie ' 2 57, and in the Intro- duction to this book p. 12) declared for the second of these views. 2 On the side of the former view may be urged the analogy of the above-mentioned suffixes. The suffix -ja moreover is of very frequent occurrence in verbal adjectives, it is used in Sanskrit in the formation of gerundrval adjectives like ja(j-ja-s (rt. ja{j) venerandus=Gk. άγ-ιο-ς, though no definitely established meaning was uniformly attached to it. For instance paft-ja-s from the rt. pa% cook means ripening, and the cor- responding Zend form likewise. Emphasis might even be laid on the fact that in Sanskrit the suffixes an andja are found united in the later fuller gerundive termination -an-ja, -an-tja, and that in the verb likewise 2 Since the above was written an attempt lias been made in Bezzenberger's Jieitrage i. 1 20 ff. by Fick and Fuhrer, to show that the ■ so-called ja-suflfix ' was from the first an element inherent in the verb. I confess I see no reason for this view, and it seems to me that no proper regard has been paid, in making the lists there given, to the period at which the several words occur. I fail to see that, e.g. the altogether late β-ηχία, hoarseness, can be of any use at all in ex- plaining βήσσω or rice versa. CH , χι. ORIGIN OF THE ja. 205 both elements occur not seldom in conjunction, whence comes e.g. the Skt. bhur-an-jd-mi I start, and the Gk. υψ-αίνω i.e. νφ-αν^ω from the rt. v0=the Skt. vap (cp. above pp. 177, 185). Single instances in which to 297 actually occurring adjectives in -ja we can find actual present-stems cor- responding in sound and meaning seem not to be altogether wanting. Eor instance in Justi's Zendworterbuch the adj. verez-ya (rt. varez=Gk. fopy) meaning effectual, and the pres. verez-yd-mi I do, i.e. I am effectual, occur side by side. Elsewhere, it is true, the meaning of the verb does not square so well with that of the adjective. For instance, άζομαί=. ay-jo- μαι I dread does not fit in with either άγ-ω-ς or jag-ja-s, nor the passive or intransitive meaning of the Skt. mri-ja-te he dies and the Lat. mor-i-tur with the Zend mair-ya destructive, deadly, and there is no very close connexion in meaning between πλάγ-ιο-ς knocked out of shape, and hence crooked, and πλήσσειν i.e: πληκ -jeiv. It must be admitted that in the case even of those* present-stem formations which we have seen good reason to regard as of nominal origin, the develop- ment of meaning in the forms which at a later time were used only nominally, took quite a different course from that in the present-stems, and indeed all such introductions of nominal-stems with their various suffixes into the structure of the verbal system must have belonged to such a very early period that all consciousness of any connexion between the nominal and verbal form must very soon have disappeared. All the same the preponderance of probability is on the side of Bopp's view. The element ja evidently plays a more important part in the verbal structure than all those other syllables na, -nu, -ta, or -ska — of which, as component parts of the present-stem, we thought the origin was to be traced to nominal suffixes. The syllable -ja, with the accent on it, and as a rule in connexion with middle terminations, does duty in Sanskrit as the mark of the passive, as does the syllable -ya in Zend and Old-Persian, where it is not uncommonly joined with active personal ter- minations as well (Spiegel Altpersisch p. 169). Accordingly the majority of the Sanskrit verbs of the 4th class with which we are here concerned have also an intransitive meaning. Max Miiller (Skt. Gr. p. 188) is of the opinion that there are traces which show that the verbs of the 4th class were originally accented in the same way as passive verbs. Further the syllable ja appears as an essential part of the suffix -aja by means of which derivative, and, more particularly, decidedly denominative verbs are formed . in Sanskrit and Iranian, 298 and which has become the source of the manifold denominative forma- tions of all the related languages. We shall come back later on to what I hold to be the unmistakable fact that this -aja is nothing else than an a which is the final letter of a noun-stem, and this very ja which is used to form the present. The syllable ja — in the form now of ja and now of i — is also the modal characteristic of the optative, and is to be seen again in the future as the second element of the suffix -sja (Skt. da~s-ja-mi=T)or. ΰω-σ-ίω). It might be urged against this that we have here to deal with elements which, though alike in sound, are of totally . different origin. But there is a probability from the meaning too that the ja in the optative is the same as that in the future. And since . "in the future all rightly agree in referring the first element in the suffix to a verbal root, i.e. as, there is a special probability here that the second is not a mere nominal suffix but a verbal stem. It is not to be 206 THE I-CLASS. CH. XI- denied of course that suffixes which occur elsewhere as nominal suffixes occasionally perform other functions. For instance the suffix -na ■ gets that of expressing the passive in Gothic. But where is a nominal suffix to be found with such manifold ramifications of meaning as this -ja 1 Are we to imagine that in the structure of the verb such essential cate- gories as passivity, modality, and the designation of the futm-e, as good as came out of nothing at all, or, in other words arose from the chance difference in the application of a nominal suffix which has in itself no meaning, or at least none to distinguish it essentially from other nominal suffixes 1 This seems to me an impossibility, and I believe that Schleicher himself would hardly have maintained the pronominal origin of the syllable ja if he had not, with the rigid exclusiveness of attention which was peculiar to him purposely refused consideration to many of the abstruser questions about ' function.' And yet it is only by the conscientious and combined consideration of both sound and -meaning that a satisfactory solution can be reached of the problems set us by the Science t>f Language. If we proceed to ask what is the way which modern languages have taken, in periods that are more open to our observation, to express categories like passivity, modality and futurity there is no doubt about the answer. It is by the application of auxiliary verbs which, in 299 virtue of the meanings which had been already determined in their independent use, carried in themselves the germ of the expression of these relations. It was these clear analogies which Bopp had in view when at the very outset of his vast labours he conjectured that auxiliary verbs had been made use of in earlier periods of linguistic history. And seeing that Bopp's explanation of the sibilant in the verbal structure as being the rt. as is as good as universally accepted, and that of the dh (Gr. Θ) as being the rt. dha place, do, is pretty generally adopted, we shall be justified in inclining to his derivation of the syllable -ja from »the rt. ja (Skt. ja) go. The idea of going contains in itself, as is shown by W. von Humboldt (Ueber die Verschiedenheit des menschlichen Sprachbaues, p. 257 ff.), and as we have already pointed out on ρ 12, the germs of the most various meanings. To begin with, going is a continuous action, and as such is adapted to be used in the stem of the durative present- tense. Take for instances such phrases as the Germ. schwanger gehen ' to go with young,' ' to go walking,' l to go begging,' (' to go shares ') or the Latin exsequias ire. Then going is intransitive and, where it means not the striving after an object but the getting into a state, it can give rise to a passive meaning, as for instance in the German verloren gehen ' to get lost,' nnafeil gehen ' to go for sale,' and in the Lat. venum ire. Bopp (Vgl. Gr. iii. § 739) mentions that in Bengalee Icord yd'f, properly ' I go making,' means ' I get made.' The same verb again can take us further to the notion of striving. How far it is possible that we may derive hence the explanation of the modal use of the syllable ja in the optative will be discussed when we come to deal with this mood. The force of a future is clearly possessed by the verb ire in dejectum ire, amatum iri, with which, besides the analogies from French mentioned on p. 12, we may compare the German 1< gehen * to go to bathe ' (and the English ' I am going to do it '). It is easily conceivable, if this view be taken, that the force of this affixed verbaL root should in many cases become considerably weakened, and ch. χι. PEESENTS IN -ιω. 207 should even be entirely lost sight of, and that consequently, after all suspicion of the origin of the syllable in question had vanished from the consciousness of the speaker, it might degenerate into a purely formal constructive element and be used in transitive or even causative verbs. We may even find it possible on a more detailed investigation to imagine many ways in which the transference of use might have taken place. The apparent absence of meaning in the syllable which we are obliged 300 to acknowledge in many cases is not enough to outweigh its significance in many others. This absence of meaning is in accordance with the general tendency of language — just as in the case of the inchoatives we saw a meaning which at first had been a specific one preserved in but a small circle of verbs — while the significance which it possesses in the future and the optative could hardly be explained without the help of Bopp's assumption. Max Miiller is so thoroughly convinced of the origin of our syllable ja from the root of the verb to go that in his essay * On the Stratification of Language ' p. 31 he actually refers the primary nominal suffix -ja, fern, -ja, to this root. In this way then the second of the two possible explanations would coincide to a certain extent with the first. However, I admit that I still feel considerable doubts as to the soundness of the latter explanation (cp. the note on p. 204 above). With respect to the Greek representatives of the {-class, which we have now to review in their several ramifications, it is certain that there cannot be said to be any special modification of meaning in the present- stem as opposed to the verb-stem, even to the limited extent in which this could be maintained with respect to the inchoative class. The Greek ί -class falls into two main divisions, according as the vowel ι or the consonant j is the basis of the suffix. The first of these divisions is but poorly represented, the second branches out in the most various directions, and has therefore to be divided into several sub- classes. I. PEESENTS IN - ίω . Greek forms constructed like such Latin presents as cap-io f fod-io are rare. Such as there are fall into two subdivisions : A) where -ιω has been preserved pure, B) where -ιω has coalesced with other vowels to form diphthongs. A) εσ-θ-ίω, which occurs from Homer onwards, with the (also Homeric) by-form εβ-θω and the unexpanded εΐω, is the , only present with a 301 movable ι. Since, however, the ι is here preceded by another stem- expansion, i.e. Θ, we have evidently here what we have encountered so often before, e.g. informs like υφλι-σκ-άνω, αισ-θ-άνο-μαι, the conjunction of two elements of stem-expansion. The stem without the ι is just as much a present-stem as that with *the ι, so that properly speaking we cannot call this a present-forming ιω. Delbriick (Verb. 202) discovers an isolated parallel to εσ-θίω in the Vedic gru-clMjd-ti he obeys (rt. $ru hear). All the other verbs in -ιω keep the ι in the other tenses as well as in the present : άλίω ηλισα, κυλίω εκΰΧισα, όίω Horn, οισατο. In the case of denominative verbs like μηνίω, ΰηρίομαι, κονίω this is hardly to be wondered 208 THE I-CLASS. CH, XI. at. All these verbs are like the Latin verbs with a permanent i like audire, lenire, and not like such as cupere, fodere. . It is possible that this per- manency of the t was not a primitive feature and that the extension of the domain of the ι was, like that of the nasal affixes, only gradual. But where we find forms with a permanent ι occurring in the very earliest times we are hardly entitled to make such an assumption. In the case of άίειν hear we must not overlook the post-Homeric ήισα and the σ of the Herodotean verbal adjective επάϊστος, — all the less as in the evidently related ai -σ-θ-άνο-μαι we see the other stem-expansions following the same consonant. Hesychius's αετέ ■ άκούετε does not make against this. Although then the facts here adduced by no means exclude the possibility of the connexion of ά'ίω with the rt. av conjectured at Princ. i. 482, it cannot be said, strictly speaking, that the ι of this verb is instrumental in forming the present-stem. We have more right to maintain this in the case of lB -ίω (Princ. i. 300). For here IB, originally σΛδ, is unmistakably the root, which appears unexpanded in ΙΒ-ος iB -ρό-ς, ΙΒ-ρ-ώ(τ)-ς, and the corresponding Sanskrit root svid forms its present after the fourth class : svid-jd-mi. But we look in vain for a future *'ί-β•ω or an aorist *ί-σα. All that is preserved is έ'ξ-ίΒϊ-σα in Aristophanes Av. 791 and forms of the same kind in Aristotle. The length of the t moreover in Attic (πρϊν av IBtyg Aristoph. Pax 85), as contrasted with the Homeric 'ιΒιον ν 204, is remarkable. 302 B) If, then, we count εσθ-ίω and lB -ίω as two instances we may begin this division with no. 3. Presents with diphthongs before the thematic ω have in some cases no corresponding forms of a shorter stem, as -καίω επαισα επαίσθην, πταίω ϊπταισα, σείω έσεισα σεισμός. In these presents then there is no element to be seen of the kind we seek. Where there are forms of the two kinds, we must distinguish between two classes. On the one hand, a vowel-stem may have directly coalesced with -ιω ; on the other, a consonantal stem ending in f or σ may have become exposed to the same transformation after the loss of its final letter. The cases of the first kind are arranged under 1), those of the second under 2). i) 3) άγα-ίο- μαι by the side of άγα-μαι, άγά-ο-μαι, άγά-ζω } has been already mentioned at p. 118. 4) Ια-ίο-μαι divide, which we gave at p. 203 as one of the verbs which showed the same method of formation in Greek and Sanskrit. Along with Βα-ιό-μεν-ος ρ 332, Βα-ίε-το ο 140 we get forms like Βά-σονται X 354, Βά-σασθαι Σ 511. The forms without an ι might certainly be referred to the stem oar which underlies Βατέομαι. Cp. Leskien, Stud, ii. 122. The double σ of αποτάσσομαι Ρ 231, Βάσσα ντο A 368, in ΒέΒασ- rat A 125 άνάΒαστος (Plato) might be appealed to in support of this. It is anyhow remarkable how the t has made its way into other verbal forms and noun-stems Βε-Βαί-αται (only α 23), Βαί-νυ-μι (fut. οαι'σω), Βαί- (r)-c, cai -τρό-ς, Βαι-τύ-ς, Βαι-τυ-μων, which, however, finds a complete analogy in the Skt. daj {ddj-a-te) divide, confer, allot. The meanings of this verb with the j preserved actually come nearer to those of the ch. χι. PRESENTS IN -ιω. 209 Greek verb than those of d-jd-mi, though the latter, according to the Pet. Diet., in composition with various prepositions means also allot, divide, and thus comes so near to the Greek verb in meaning that it can hardly be doubted that it has the same root. We have here a very clear instance of the vacillation early manifested by language between a merely partial stem-expansion — one limited to the present-stem, that is — and one that goes all through the verb. 303 5) κερα-ίω, by the side of κέρα-μαι, κεράο-μαι, cf. p. 120. 6) θυίω. 7) Aeol. φυίω, which, along with other related formations in which the t is sometimes found and sometimes deducible, have been discussed on p. 147. 2) 8) γα-ίω, only preserved in the Homeric part. γαίων A 405. Still, γαϋ-ρυ-ς and the Lat gau-d-eo (Princ. i. 211) make it probable that it comes from γαβ-ιω. There are no other tenses. 9) ΰα-ί ω kindle, poetical from Homer onwards in the active and middle. Its origin from ΰαΐ-ιω is established by ΰε-ΰαυ-μέ-νο-ς (Princ. i. 285). tiaf corresponds to the intensified Sanskrit root du (du-no-mi) burn, whence comes dav-a-s, a burning. The perfect is ΰέ-ΰη-ε, and there, is an aor. ΰά-η-ται. 10) κα-ίω with the Attic by-form κάω, common to all Greek from Homer onwards. The Attic καν•(Τω, ε-καν-σα, κέ-καυ-μαι, καϋ-μα etc. establish καΡ as the stem of the verb. 11) κλα-ίω in Homer, Attic by-form κλάω. Homer has κ\αΰ-σομαι, κλαν-σε, ά-κΧαυ-το-ς, the tragedians κε-κΧαυ-μένο-ς, so that the stem of the verb must be κλα/, which has perhaps been developed from the rt. κΧν wash (κλν ζω cp. plo-ra-re and the rt. plu), in the same way as laf (no. 9) from du. 12) λι-λα-ίο-μαι an isolated epic present, which we should be able to derive straight from the rt. λα (λή-μα), if it were not that there is no definite evidence of the existence of such a root (Princ. i. 450). The rt. las, on the other hand, is well established, and in Sanskrit it forma a present Idsh-jd-mi (as well as Ids-d-mi) of the same meaning as the Greek word. It is only the reduplication, for which we shall find numerous analogies, especially where the meaning is intensified, that distinguishes λι-λαίο-μαι from lash-jd-mi. The Homeric Χε-Χιη-μέιο-ς ought perhaps to be referred to Χε-ΧιΧη-μένο-ς. 13) μα-ίο-μαι. Leskien (Stud. ii. 88) has made it exceedingly probable that the rt. μας is at the bottom of this present, which occurs in poetry from Homer onwards (E 748 'Ήρη 2έ μάστιγι θοώς επεμαίετ άρ 'ίππους). The forms μάσσεται Ι 394 (Aristarchus), έπιμασσαμένη τ 468, ίπίμαστος ν 304 377, μαστήρ, μάσμα all make for this. This μας we may regard as a sigmatic expansion of the rt. ma which is to be found in ma-nu-s and μά-ρη hand (εύ-μαρής) with the fundamental notion touch, feel. The fundamental meaning is readily distinguishable in some of the Greek forms, while in others the derived meaning i feel after something, long for,' is prominent (Princ. i. 388). As it is probable that the rt. ma measure (Gk. με) is also to be referred to the same fundamental notion we should not be excluded from identifying the Skt. mas measure with the Gk. rt. μας, only this Skt. root is only to be found in lists of roots ρ 210 THE I-CLASS. ch. xi. (Pet. Diet.), though the present form mas-jd-mi, which is completely identical with *μασ-)ο-μαι, is also given. 14) να-ίω used by poets from Homer onwards. The existence of a σ is made probable by. such forms as νάσσα δ 174, άπενάσσατο Β 629, νάσθη Ε 119. On the related rt. νες (νίσσομαι, νόστος), which derives support from the Skt. nas to join oneself to, cp. Princ. i. 391, Leipz. Stud. i. 141. The t, as in no. 4, appears beyond the present-stem in να-ι-ετάω (cp. ναέται, ναετηρες' οικητορες Hesych.). 15) τρε-ίω instead of the usual τρέω is quoted by Veitch from Timon Phliasius fr. ix. (Wachsmuth) ; ήν πλείστοι νποτρείονσι σοφισταί (cp. Oppian Cyneg. i. 417, iv. 117). If we ought to discern in this word the effects left by an early linguistic process, τρείω would correspond to the Skt. trds-ja-mi. But as it occurs in these late poets, it is possible that it is only due to an imitation of the epic forms we are just about to discuss. In conclusion, neglecting the alphabetical arrangement, we may group together the following similar epic presents in ειω : 16) θείω. 17) πλείω. 18) πνείω. 19) χείω (χέω). After all that has been said by others and by myself (Princ. ii. 201 f.), it seems to me most probable, as may be gathered from what has been said on p. 156, that the first three ought, just like δα/ω, καίω, κλαίω, to be referred to primary forms with an ι, i.e. to θεΡ-ιω, πλίϊ-ιω, πνεϊ-ιω. πλί^-ιω — by the side of πλύνω, πλντός, πλνμα and πλόος, i.e. πλόΡος — gains support from the Ch.-Sl. plov-jq, the Lith. plau-ju and the 0. H. G. flew-iu j(Princ. i. 347). — To πνείω (Aeol. πνεΰω) belong 305 the aorists άμπννε (X 222) and αμπνϋτο and the noun πνοή (for πνοβη) and πνοιή (for πνο/ιη). — Of forms belonging to no. 16 Homer has θεί ν Ζ 507 beside θέ ν σι Σ 601, θείειν Κ 437, Π 186 etc. beside θέειν Β 183, Λ 617 etc., of those belonging to no. 17 πλείειν I 418, ο 34, πλείον- τες π 368 beside πλέων Η 88, of those belonging to no. 18 πνείει Ρ 447 beside πνέει ε 469, επιπνείτ^σι Ζ 357, άποπνείων Ν 654, άποπνείονσαι δ 406 and the like, χείω (no. 19) is represented in Homer only by έγχείρ ι 10, and Hes. Theog. 83 has χείονσι. By the side of this come χόο-ς, χους, and the shortest forms εχυτο, χυμένος, κέχνται etc. Hesiod is our only authority for — 20) ρείω : fragm. 237 Gbttl. ποταμω ρείοντι εοικως. We may here add the word discussed by. Usener in Fleckeisen's Jahrb. 1872, p. 741 ff. 21) hi v. Ζεϊν as a neuter participle=^f'ov is there quoted, in con- formity with the testimony of old grammarians from several passages in Attic prose writers, especially Lysias 14, 7, Xenoph. Hell. vii. 4, 39. Since ΰεϊν bears to δέον the same relation as that of πλεϊν more to πλέον, Usener justly concludes that there was a present form *δείω, of which the participle ΰεΐον is the primary form of h~tv, just as πλέων is that of πλεϊν. *ΰείω he rightly refers to ^fZ-Jw, just as at Princ. i. 289 the stem hf is given for Ιέω. There is ground for suspecting other presents with t-diphthongs of having lost consonants in a similar way. But the ι appears to have established itself firmly through all the tenses, as is the case in κ να-ίω, κναίσω, εκναισα by the side of κνά-ω, κνή-θω, which Fick i. 3 49 refers to a rt. knaa, in παίω, παίσω or παιησω, επαισα, έπαίσθην — which at Princ. i. 333 I have compared with the Lat. ραυ -io. The t of the latter verb is ch. χι. PRESENTS IN -λλω. 211 treated as if it were that of the i-conjugation (cp. also pavlmentum), but Paul. Ep. p. 70 quotes from Lucilius the perfect de-puv-i-t from de-puv-io, so that the * of the verb appears to have been a movable one. — πταίω πταίσω, επταισα is of too uncertain etymology to yield us any result for our present purpose. II. PRESENTS SHOWING THE EFFECTS LEFT BY AN EARLIER -jo), 306 A) Yerbs IN -λλω -λ^'ω. 1) αλλο-μαι, ordinary Greek, by the side of the Homeric άλτο, conj. αλε -rciL (cp. above, p. 130), the Att. άλοϋμαι etc. Lat. sal-io (Princ. ii.167). 2) βάλλω, ordinary Greek, with the Arcadian by-form ζέλλω, δε'λλω (Princ. ii. 76), by the side of εβαλον (Arcad. εζελον), εβαλόμην, βάλω, βέλος, βολή. If we are not mistaken in the comparisons given at Princ. ii. 76, and the assumption that the primary meaning was flow, glide, the Ο. H. G. quillu scaturio (pret. qual) is due to the same method of present-formation. — ΙιαΙέΧΚειν' ΰιασπάν Hesych. can be nothing but ΐιαβάλλειν. 3) βΰάλλω. Plato Theaet. 174 d is the earliest passage in which the verb is found. Of forms of other tenses there occurs only εβΰήλατο. 4) βΰέλλων τρέμων η βΐέων Hesych. as also βΰΰλλειν' ΰεΰιεναι, τρέμειν rj βΙεΊ. ν belonging to βίελ-νρό-ς which is an expansion of the root of βίεω which was originally βίες (Princ. i. 284). 5) βάλλει* κακουργεϊ Hesych., if genuine, belongs to δαλ rj' κακονργ^, ΰαλήσασθαι' λνμηνασθαι, άδικήσαι and ΰηλέομαι. To the latter δάλλω bears exactly the same relation as θηλέω to βάλλω. 6) θάλλω, the present-stem not till after Homer (who has θήλεον ε 73, αναθηλησει A 236), later it is found in ordinary Greek, by the side of the Homeric τε-θαλ-υ'ια, τεθηλώς, θάλος. 7) ίλλω cannot be put here with complete certainty, inasmuch as it seldom occurs without the variant ε'ίλω or είλλω, while at p. 179 we referred είλειν press hard to an earlier ΐελ-νω. Still it is not impossible that from the rt. ΐελ twist, turn there should besides ε'ίλω have been formed an ίλλω standing for βελ-βω. Cp. Buttmann Lexil. ii. 150 f. 8) κ•ελλω does not occur in the present, but ό-κελλω is frequent in Attic prose (Princ. ii. 397). 9) μέλλω, ordinary Greek, with λλ all through the verb (Att. μελ- 307 λησω, εμελλησα), but it is certainly from the same root as μίλει (Princ. i. 412). 10) μνλλω only Theocr. 4, 58, mentioned in the scholia on the passage and elsewhere by grammarians, generally sensu ohscoeno (μνλλει* πλησιάζει Hesych.), undoubtedly however related to the Lat. molere (cp. permohre) and μύλο-ς. The same present-formation occurs in the Ch.- Sl. mel-jq,. 11) χάλλω, in use from Homer onwards, especially with poets, by tlje side of άμ-πε-παλ-ών Γ 355, πάλ-το (cp. above, p. 131), πάλο-ς. 12) σκάλλω scrape up earth, dig, by the side of σκαλ-εύω, σκαλ-ίζω, from Herodotus onwards (ii. 14). 13) σκέλλω dry (trans.) κατεσκίλλοντο Aesch. Prom. 481, by the side of σκελεω, σκελε-τό-ς etc., perf. εσκλη-κα, σκλη-ρύ-ς. The aorist ε-σκηλα (σκήλειε Ψ 191) should by rights have a present σκάλλω. There was p2 212 THE I-CLASS. ch. xi. clearly a similar, variation between α and ε before λ here as in the rt. μελ (μέλει) with the perfect μέμηλε (Dor. μέμάλε). 14) σκύλλω, from Aeschylus onwards, later there is an aor. εσκνλα etc. 15) στέλλω, as early as Homer (M 325), by the side of the rut. στελ-έω (/3 287), aor. στε ϊλα (£ 248), later έ-στάλη-ν, ε-σταλ-μαι, ε-σταλ-κα. 16) σφάλλω, from Aeschylus onwards by the side of σφαλώ (also middle), εσφάλην, εσφαλμιιι. Homer has only the aor. σφήλαι Ψ 719, μ 464. 17) τέλλω, common from Homer onwards by the side of ετειλα, τέταλμαι. 18) τίλλω, as early as Homer (X 406), and from Attic writers πλώ, έτϊλα, ετίλθην and other forms. 19) ψάλλω, from Aeschylus onwards, έ'ψί/λα. We may conclude from Hesychius's σήλατο• έσεισε that there was a verb *σάλλω belonging to σάλο-ς. The disyllabic stems are partly, like ποικίλλω (as early as Homer), unmistakably denominative. Of these we shall treat later in connexion with the remaining denominatives of this class. There is however a group of disyllabic stems which we will here give apart from the rest. Reduplicated Present-stems. Some of these remind us of the Sanskrit intensives in which the reduplication syllable has been strengthened, as Schleicher (Comp. 3 758, 308 cp. Bopp Vgl. Gr. § 756) has already remarked. Anyhow the con- junction of reduplication with the mark of thej-class is common to both these formations. The other divisions of the class will furnish us with abundance of analogies to this. Gerland (Intensiva und Iterativa, Leipzig 1869) discusses this kind of Greek intensives at p. 32. 20) αιόλλω ν 27, in Hesiod and Pindar, later αίολέω. The ο has led to the assumption that the verb is derived from αΐόλος. No other tenses occur. 21) δοαδάλλω in Homer and Pindar. The latter also forms ΰεΰαι- ΰαλμένος, ΰαιΰαλθείς and (from an evidently denominative by-stem) haida- λωσέμεν (01. 1, 109). It by no means follows, however, that Ιαί-Ιαλο-ς was earlier than ΙαιΖάλλω. Cp. Princ. i. 286. 22) ΰεν-ΰίλλω I 180, Apoll. Rhod., with no other tenses. The San- skrit d-dar trouble oneself, take thought for, compared by Fick 3 i. 106, which is only used in composition with the preposition a, shows a kindred present-formation in d-dri-jd-te. Cp. also the O. H. G. zil-jan. The reduplication is like that in ΰέν-ΰρε(ΐ)ο-ν beside ΰρϋ-ς. 23) ί-άλλω, from Homer onwards, by the side of the aor. 'ίηλα, in poets. Cp. Princ. ii. 171. The root must be o^. = Skt. ar go, from which likewise there is formed the reduplicated present ij-ar-mi, which besides the intransitive meaning has the transitive meaning * move, bring.' 24) κοι-κνλλω, only in the present-stem in Aristoph. and in gram- marians. The etymology is obscure. 25) μοι-μύλλω, related to μυειν, is explained by Pollux ii. 99 by συ\ά•γειν ra χείλη, but Hesych. renders it θηλάζει?, εσθίειν and it stands, ch. χι. STEMS ENDING IN p. 213 thanks to Meineke's striking conjecture, in the latter sense at Hipponax fragm. 80 Be. 3 26) παιπάλλω only given by lexicographers : παιπάλλειν ' σείειν Hesych. The word, if genuine, is a kind of frequentative to πάλλειν. It is possible that it arose from a nominal stem which underlies the Homeric 7ταί7ταλόείς. We may also mention here the etymologically obscure άτ-ιτ-άλ-λω with its trisyllabic stem, — possibly a denominative related to αΓαλός, άτάΧλω. Β) Epenthesis of the «. 309 1) Stems in p. The Lesbian Aeolic dialect took just the same course in the case of stems in jo as in that of stems in λ, i.e. that of progressive assimilation : ψθέρρω : *depjv ', \ βάλλω Ι *βαλ]ω. Cp. Ahrens Aeol. 53. The other dialects took a different course. It is true that it is almost exclusively from the Ionic dialect that we get instances of the real anticipatory epenthesis of the t. Still the assertion of the grammarians, which Ahrens wrongly calls in question, that the Dorians said ψθαίρω for φθείρω, leaves little doubt that the Dorians agreed in this formation with the Ionians. We have a distinct testimony to this agreement in the Cretan ΣΠΕΙΡΕΝ adduced by Brugman Stud. iv. 99 from C. I. no. 2556, i. 18, a form which is important for the explanation of this present- formation. For since in the Cretan dialect ει can never come from ε by compensatory lengthening, it proves incontestably that the ι was here really introduced by anticipation from the following syllable. The case is different with stems in vp. These show the effect of the j in the following syllable only in the lengthening of the υ : κύρω, <ρόρω. To assume the same process here would be too artificial an hypothesis, and Brugman (Stud. iv. 100, 117) has shown conclusively that in this instance the length of the syllable is due to compensation, and that κύρω came immediately from a *κυρρω which we may assume on the analogy of the actually occurring Lesbian όλοφνρρω. It must be admitted, how- ever, that such presents as have no attested Aeolic counterparts in pp, ought possibly to be put into the lengthening class — may perhaps have been formed, i.e. like τύψω, φρύγω (p. 158). 1) hi ρω, by the side of ΰέρω, in Hdt. (ii. 39, iv. 64) and Attic writers, Lesb. ΰέρρω. ΰαίρω which occurs in some M.SS. at Aristoph. Nub. 442, Av. 365 Dindorf is no doubt right in altering to δε ίρω. For the orthographical rules of the grammarians — e.g. Herodian ii. 490 — know of nothing but Ζείρω, which analogy demands, and the Aeol. Ιέρρω (Ahr. Aeol. 53). Ιείρω i.e. *^ερ^ω has been above (p. 203) compared with the Lith. dir-iu. A Sanskrit dir-jd-mi of the same formation is 310 also mentioned in the Pet. Diet, as given by grammarians. 2) είρω say, from Ρ είρω β 162 τάΐε είρω (λ 137, ν 7), by the side of the fut. ερίω, ερώ from Homer onwards; there is also the present ηρέω, Hes. Theog. 38 είρεϋσαι saying. 3) είρομαι ask, seek ε'ίρεαι γ 80, ε'ιρετο A 513, but like no. 2) with by-forms from a stem in ε : ερίων Η 128, conj. ερείομεν A 62, ερέωμαι ρ 509, ερίοντο Α 332. The stem ερ appears without any expansion in the middle aorist-forms in use from Homer onwards ερωμαι, εροίμην, 214 THE I-CLASS. CH. XI. ερέσθαι (y 243), to which was later added the indicative ήρόμην. On the difficulty of connecting these forms with εφω say cp. Princ. i. 429. '4) εφω set in a row, fasten.. The present from Pindar onwards. The Homeric form ήεφε discussed on p. 81 belongs either to the imperfect or to the aorist, while the unexpanded stem is certainly to be seen in the Homeric εερμένος, εερτο. ΰι-έρ-σαι is quoted from Hippocrates. Cp. Princ. i. 441. The Lat. sero is an unexpanded form from the corre- sponding Latin stem. 5) κείρω from Homer onwards (Λ 560, λ 578) by the side of the fat. κερ-έω, κερώ aor. ε-κερ-σα Ν 546, Princ. i. 181. The Lesbian κερρω is often given. 6) κύρω. κνρον Ψ 821, ε-κνρο-ν Soph. Ο. C. 1159, κύρεται Ω 530. A by-form of the present-stem occurs in κϋρ-έω Aesch. Prom. 330, εκνρονν Soph. El. 1331, whence we find later on κυρήσω etc. The pure stem κυρ occurs in ε-κνρ-σα, κυρ-σω. 7) μείρομαι. μείρεο in the suspicious verse I 616, and άπομείρεται is not quite beyond suspicion at Hes. Theog. 801, Opp. 578 (cp. Kochly) ; the only other passage adduced for it is Aratus 657 (μεφομένη). The forms from the unexpanded stem however are of frequent occurrence : the Homeric εμμορε, εΊμαρται, μέρος. At Princ. i. 412 I have connected with it the Lat. mer-eo. 8) μύρομαι. Homer has μύρονται Τ 213, μνρόμενος τ 119, also later poets, and Hes. Scut. 132 has the active impf. μϋρον. 9) Εύρομαι, a by-form of ζνρέω, not before Plutarch. 10) πείρω, poetical from Homer onwards (v 91, γ 33). The shorter 311 stem occurs as early in πε-παρ-μένος. At Princ. i. 338 I have compared with it the Ch.-Sl. pra-ti scindere, of which the 1st sing, por-jq, — though not supported by quotations in Miklo&ich's Lexicon — is formed in pre- cisely the same way as πείρω. 11) τττνρομαι, only found in Hippocrates, while the aor. επτνρην occurs in Plutarch. 12) σαίρω Eurip. Ion 115 and elsewhere, the shorter stem in σεσηρα (comic poets), Aor. εσηρα (Soph.). 13) σκαίρω only in the present-stem, σκαίροντες Σ 572, σκαίρωσιν κ 412. The pure stem is shown in σκαρίζω and with a thinner vowel in σκιρτάω. 14) σκαίρω (cp. below άσπαίρω), only in Alexandrine poets, in Aristotle and in late prose. There are no forms found except those from the present-stem. 15) σπείρω, in common use from Hesiod onwards (Opp. 463), by the side of the fut. σπερώ, pf. εσπαρται, aor. εσπάοην, σπορά etc. Lesb. σπέρρω. ^ 16) σύρω. Present forms first in Hdt. (ii. 60) and Aristophanes (παρασύρων Equ. 527), εσνρην not till late prose. 17) τείρω (Lesb. τερρω) poetical from Homer onwards (Δ 315), but only in the present-stem. The pure stem must be sought in τέρ-ε-τρο-ν, τέρ-ην and the Lat. ter-o. The latter is related to τείρω exactly as eero is to ε'ίρω (no. 4). 18) φθείρω, Lesb. φθέρρω, from Homer onwards with the fut. φθερσω, later φθερώ, ί-ψθάρ-ην, ε-φθορ-α etc. 19) φόρω, in general use. Ω 162 ϊάκρνσι είματ εψνρον. We cannot put this verb here without reserve, for the short stem φΰρ which we thus CH . χι. STEMS ENDING IN p. 215 are obliged to assume, is nowhere to be found. Even the derivative ώυράω has ν (Aesch. Sept. 48). The present-stem seems therefore to have become completely petrified, unless indeed we ought to explain the long vowel in quite another way. 20) χα/ρω, in general use, beside ε-χάρ-ην — Homeric κεχαρήσω, κεχαρ- ηόπ, κεχάροντο. If at Princ. i. 244 we were right in comparing the Skt. hdr-ja-mi amo, desidero, there is a correspondence even in the formation of the presents, as also in the Umbr. her test volet, and the Osc. heriiad capiat. There is, however, a difficulty in the difference of the meanings. 21) ψα/ρω, a rare verb, denoting vibrating motion, used only in the present-stem. Aesch. Prom. 394. Besides these monosyllabic stems there are also a few disyllabic: viz. 312 a) Reduplicated stems, corresponding exactly to those mentioned on p. 212 f., i.e. 22) γαργαίρειν swarm with, in Cratinus (Meineke ii. 221) : ανδρών αρίστων πάσα γαργαίρει πόλις. It is natural to conjecture a relationship to άγείρω, αγορά, πανήγυρι-ς (no. 27), to which γαργαίρω seems to be an intransitive intensive. 23) καρκαίρειΐ' ring or quake : κάρκαιρε ΰε γαϊα πόΰεσσιν Υ 157. 24) μαρμαίρειν shimmer, poetical from Homer onwards (Ν 22). The unexpanded stem appears in μαρμάρεος, μαρμαρυγή. 25) μορμνρειΐ' boil, bubble, the same. 26) πορφόρειν move in waves (used of the play of colours), the same. A short υ appears in the related πορψΰρέω (late), πορφυρός. b) Stems with prothetic vowels. 27) ά-γείρω, ordinary Greek by the side of the Homeric aor. άγεροντο, άγρόμενοι, ήγέρθην, αγορά, παιήγυρις. Lesb. άγέρρω. The derivation from d copulative and the rt. gar (γηρύω) does not suit the use of the words well, άγείρεσθαι, άγέρεσθαι in Homer denotes the carrying out of the command proclaimed by the herald : Β 52, 444 rot <Γ ήγείροντο μάΧ ύκα. and αγύρτης beggar (i.e. collector), άγυρμός have nothing what- ever to 'do with calling. It is better with Fick 3 i. 73 to compare the Skt. grd-mas troop, mass, with which, however, we must also connect the Lat. grex and γαργαίρω given at no. 22. 28) ά-είμω (Lesb. άίρρω) from Homer onwards (Ψ 366 άεφομένη, Τ 386 impf. άειρε, Hdt. ηειρε) with the plupf. άωρτο, aor. αερθεν. The con- traction begins in Homer (οφοιτας Ρ 724, αρθείς Ν 63), in the tragedians we find 'άρω (e.g. Aesch. Pers. 795, Iph. T. 117), αραι, ήρμαι and αίρω, and when we come to Attic prose only the contracted forms are found. 29) ά-σπαίρω, by the side of σπαίρω no. 14, from Homer onwards (M 203). For its derivation see Princ. i. 358. Since this stem, like no. 28, begins with two consonants, it is possible that the a is here the remnant of a reduplication, in which case ά-σπαίρω : γαρ-γαίρω I * ε-σπαρ- rai : γέ-γραπται. 30) ε-γείρω (Lesb. ιγερρω) from Homer onwards, by the side of the 313 Homeric ε-γρετο, ε-γρή-γορ-α, ήγέρ-βην etc. Cp. Princ. i. 221. Fick 3 i. 72 compares εγείρω with the Skt. causative (ja-garaja-mi I awaken. In this case εγείρω would be a derivative verb. 31) ό-ΰύρομαι, the prevailing form from Homer onwards (X 424), by 216 THE IrCLASS. CH. XI. the side of ϊνρομαι (Aesch. Prom. 271, Eurip. Hec. 740) and πάνΐυρτος (Aesch. Soph. Eurip.). 2) Stems in v. Since the ν in many of the stems which fall under this head is just as movable as the epenthesised ι, many of the following verbs have been already noticed among those of the nasal class, especially βαίνω on p. 185. In cases where we have no -short- vowel forms of other stems than the present it is often impossible to decide whether a present has been formed according to the nasal-class or the i-class. For instance it is by no means impossible that the verbs £tVw, Ιννω, θύνω, πύω, <ρύνω given on p. 178 f. arose from *3ir;w, *$vvj(u etc.; and when on the other hand we assign σίιομαι to the i-class, and so refer it to *σιι$ομαι, the only reason we have for this is that no root *σι is anywhere to be found. It is possible all the same that there was such a root, and that σίνο-μαι came from *σι-ιίο-μαι. Only where, as in θείνω, καίνω an i-diphthong, or where, as in κρίνω with its fut. κρϊνώ, forms with a short vowel and a ν occur, have we clear and certain proof of the original existence of a j in the present-stem. It should be said that out of the following list, besides σίνομαι above mentioned, θείνω, καίνω and σαίνω are the only verbs whose stems show a ν under all circumstances. 1) γείνομαι, poetical, γεινομένω Υ 128, I 208, Hes. Theog. 82, the usual present being γίγνομαι, so the Skt. tfa-je, one of the presents of the rt. (fan, though this has no n. 2) Ιραίνω, belonging to Ιράω do, only at Κ 96; we get the stem without the t in όλιγοΰρανέων, Princ. i. 294. 3) θείνω poetical from Homer onwards (Π 339) ; Attic dramatists have forms with ε, which are now rightly regarded as aorists, and are hence accented θενεϊν, θενών. The rt. fov = the Lat. fen in fen-do (Princ. i. 316). 314 4) καίνω from Aeschylus onwards by the side of κανώ, εκανον, κέκονα; it can hardly be from a different root from that of the fuller and older form κτείνω and the Skt. kshan (Princ. i. 192). 5) κλίνω (Lesb. κλίννω) ordinary Greek, with the fut. κλϊνώ (Ari- stoph. Plut. 621), εκλίνην (Aristoph. Lys. 906), εκλϊνα. The rest of the tenses come from the shortest stem κλι, κλϊιή from the present-stem. 6) κραίνω. In Homer the apparently denominative κραιαίνω is the prevailing form, κραίνονσι τ 567. κραίνω from Pindar onwards in poets. Homeric fut. κρανέίσθαι, aor. εκρηνα etc. 7) κρίνω, ordinary Greek with the fut. κρΧνώ (ΰιακρινεει Β 387), εκρίνα, έκρίνθην, later εκρίθην from the shortest stem, like κέκρικα, κέκρι~ μαι. The Lesb. present-form κρίννω has the testimony of an inscription C. I. 2166, 23, επεκμίννετο. 8) κτείνω (Lesb. κτέννω), cp. no. 4, than which this is an older and commoner form, from Homer onwards with κτενώ, εκτανον. The shorter stem contained in κτά-μεναι etc. was discussed on p. 130. 9) μαίνομαι (rare active εκμαίνω Eurip.), in use from Homer onwards, with μανοϋμαι, εμάνην, μέμηνα. The corresponding Indian and Persian formations, which, however, have a different meaning, are mentioned on p. 203. 10) ϊ,αίνω χ 423, later in common use, ξ(Ίνώ, (ΙάνΒην. Cp. £cw, ζνω. ch. χι. STEMS ENDING IN v. 217 11) ραίι ω, from Homer onwards (ραίνοντο Λ 282). The forms ε-ρράΐ-αται ν 354, έρράΰατο Μ 431, ράσσατε υ 150 point to a root pah which again is identical with apl• (άρΰω). βαίνω therefore perhaps stands for pal-rjb) (Princ. i. 283), and here also we have a threefold formation : pal• pa(l)v pa($)vL. Fut. ράνω. 12) βαίνω, from Homer onwards (je 219). There appear to be no forms with a short vowel. 13) σίιομαι (Lesb. σίννομαι), from Homer onwards (μ 139). No forms but those of the present and weak aorist stems. 14) τείνω. This present-stem is not clearly established in Homer, as τείνη Π 365 may be an aorist, but from Aeschylus onwards it is in common use, by the side of τενώ, έτεινα and the forms from the rt. τα τέταται, ετάθην, τατός. — Cp. τιταίνω. 15) φαίνω bears to φαείνω a relation similar to that of κραίνω to κραιαίνω. The present-stem is in universal use by the side of φανώ, φανοϋμαι, εφάνην, πέφηι>α. The shortest stem φα may be seen most 315 clearly in πεφήσομαι (Ρ 155). 16) χαίνω a late present, first found in the poets of the Anthology, to the St. χαν (χανών Π 350, κεχηνώς Π 409, εγχανοϋνται Aristoph. Lys. 271), for which the present in use is χάσκω (cp. p. 197). There is also χανενειν βυάν Hesych. (Gust. Meyer n. P. 50). 17) χραίνω, from Aeschylus onwards; forms without the t are rare : χράνώ Pseudo-Eiirip. Iph. Aul. 971, late εχράνθην. Reduplicated Forms. 18) αναίνομαι formed apparently from the negative av, in use from Homer onwards, the only form from another stem being άνηνασθαι. 19) βαμβαίνω Κ 375 ό 2* α ρ' ε στη τάρβησέν τε, βαμβαίνων, later also of trembling and hence stammering of the tongue. Only in the present- stem. 20) παμφαίνω epic (Λ 30, Τ 398) with the part, τταμφανόων. Cp. φαίνω. The intensive force comes out clearly. 21) παπταίνω poetical, Ρ 674 πάντοσε παπταίνων ) Soph. Aj. 11. Aor. πάπτηνα as early as Homer. A by-form in Lycophron is παπτα- Χάομαι. 22) τετ^αίνω. The aor. τέτρηιε, which shows that the reduplication stuck to the stem, is as early as Homer, the present not till Herodotus and Aeschylus. Cp. τιτράω. 23) τιταίνω only in Homer (B 390, Ψ 403) and other epic writers. Also τιτηνας (Ν 534). καγκαίνω and τετρεμαίνω, the latter of which is characterised by an inserted av as well, have been already mentioned in the nasal-class (p. 185 f.). — Hesychius's glosses γαγγαίνει ν ' το μετά γέλωτος προςπαίζειν, ΰαοαίνειν, ΰανίαίνειν άτενίζειν, μεριμνάν are of too uncertain a nature (cp. Mor. Schmidt) to be included in the list. 3) An isolated stem in κ. πε'ικω only σ 316, ε'Ίρια πείκετε χερσίν, Hes. Ορρ. 775 οίς πείκειν. Cp. above, pp. 162, 168. Isolated though this verb is in its present-forma- tion no doubt is possible on the subject. Moreover analogies in other 218 THE I-CLASS. ch. xi. classes of formations for the anticipatory epenthesis of the ι before a κ have been collected at Princ. ii. p. 335 f., and it is possible that the 316 effort to distinguish between πείκω and ττεσσω had something to do with the special treatment of the kj in this case. C) Presents in σσ (rr). I have discussed the origin of this σσ in its proper place at Princ. ii. p. 323 ff. It is an undoubted fact that σσ (Att. and Boeot. ττ) bears to ζ the relation of tenuis to media, and consequently for presents in σσω (ττω) we are to expect to find roots ending in κ, χ or r, Θ, and for presents in ζω (Boeot. and Megar. Ζδω) roots ending in γ or L The few exceptions to this, — to be marked with a *, — are mainly referable to older and sometimes actually occurring by-forms with a harder final letter to the root, which hard letter afterwards became softened. But few verbs point to a radical sigma, and many occur only in the present- stem or else show a fluctuation between different stems. 1) Verbs from guttural roots. 1) *ά^σω an unauthenticated by-form of άγννμι only quoted in Steph. Thes. from the Ε. Μ. κατάσσω in Appian and Artemidorus. 2) βήσσω quoted from Hippocrates, comic poets, and-Xenophon, βί/ζω, έβηξα. Cp. the noun-stem βηχ nom. βήζ. 3) βρνττω, only given by lexicographers : Hesych. βρυττειν έσθίειν. Ε. Μ. p. 216, 25 however gives βρΰττοντες as Athenian for πυρέττοντες, πάρα τον βρνχετόν, so that the use of the word arose from the chattering of the teeth in a fever, βρύττω is only another present for βρύκω. It seems though that we ought to assume two stems of the form βρνκ, one with the meaning σννερείΖειν τους όΐόντας μετά ψόφου (Hesych.), to which belong the aspirated βρυχη, βρΰχω, the other with a force differing little from that of the apparently related βιβρώσκειν.* 3 b) γλαύσσω, διαγλαΰσσουσι Apollon. Bhod. 1280, cp. γ\αυκό-ς. 4) φράσσομαι. Homer has only ΰεΰραγμένος, the tragedians only 317 perfect-forms. The present occurs first in Herodotus {ϊρασσόμενος iii. 13); Aristoph. (Ran. 545) has εΰραττόμην, and Plato (Lys. 209) ΰραΐ,άμενοι ; the active first occurs in Pollux. From fy>a£, gen. δρακός, hand, I have inferred (Princ. ii. 98) that the root is ΰρακ, while Fick 3 i. 107 starts from ΰραχ and compares this with the Zd. drazh, the Ch.-Sl. druzati hold fast. 5) θράσσω the shorter form of ταράσσω in Pindar, the tragedians etc. Aor. εθραζα. Perf. τετρηχα. 6) εν-ίσσω, by-form of ένίπτω, X 497, cp. above, p. 164. 7) προ-ίσσομαι Archil, fr. 130 B. 3 along with προ-ίκ-τη-ς beggar (p 352) and προίζ gift, present, apparently belongs to the stem isk ask for mentioned on p. 189. 8) κλώσσω, a rare by-form of κλώζω, only preserved in Suidas s. v. φω\άς, cp. the Lat. glocio. 9) λεύσσω, poetical from Homer onwards (Γ 110), no other tenses. Not till poets of the Anthology, Manetho etc. do forma like λενσω, • δα-ί<5σ<τ«<Γ0αι, iat-5u, οω, the Latin in are, ere, all the weak verbs of the Teutonic languages, among which the Gothic which have kept the j are the most instructive, and a large portion of Slavonic and Lithuanian verbs, among which the j is more or less widely spread. This view is advocated by Schleicher (Comp. 3 340), Leo Meyer (Vergl. Gr. ii. 3), and Scherer (Zur Geschichte der deutschen Sprache p. 183), so that we shall not be far wrong in saying that the above-mentioned derivation of the most extensive class of derived verbs from the single primary form, which has survived in Sanskrit, is the view generally adopted by Comparative Philology. This view has nevertheless not been without its opponents. Pott, in his Et. Forsch. ii. 2 977, brings forward at great length the difficulties which stand in the way of ' making the various weak verbs of the Indo-Germanic languages all spring from one stalk — the Sanskrit 10th class.' All he really does how- ever is to give expression to doubts and objections turning mainly on the various meanings associated with the various derivative formations in the several lan- guages, and rightly calls attention to the fact that many such formations were * afterbirths,' which did not take shape until they had reached the soil of the several languages. That on the other hand the type and starting-point for all the forms under discussion must be sought in these same Sanskrit 332 verbs, seems to me as clearly established after Pott's exhaustive investiga- tion as before. And this surely is all that was held by anyone. For the assertion that the Greeks formed no verb in ea> or οω, the Romans none in are, ere, that had not its exact prototype from that identical stem in primitive Indo- Germanic, is too preposterous to have been made by any reasonable scholar. It was necessary, no doubt, to establish and elucidate more clearly the ramifi- cation maintained by Bopp, and in so doing to bestow fitting attention upon the meaning of these formations. We may say at once that this latter point throws but slight difficulty in the way of Bopp's theory, for the reason that the Sanskrit verbs in -ajami, though set down for the most part as 'causatives' in our grammars and lexicons, in reality often show quite a different meaning, and that of just as various kinds as are shown by the derivative verbs in European languages. For instance, the Skt. dharsMja-mi venture on something, is in no way different from θαρσίω, and rghaja-7ni shake, rage may even in 230 APPENDIX TO THE I-CLASS. meaning be compared without difficulty with ορχίομαι, while the really causative smarajd-mi make to think, call to mind, agrees with the Goth, meijan and the Latin f eduplicated memorare. The best review of the manifold Indian forma- tions is to be found in Delbriick p. 200 ff. The prevailing view of the origin of the derivatives has of late been attacked, in particular by Oorssen, on a point of importance. In his Aussprache ii. 2 733. while admitting the connexion with the Sanskrit verbs in -ajd-mi in the case Oi the Latin verbs of the e- and ί -conjugation, he emphatically denies it in that o' the α -conjugation. On p. 736 he leaves it an open question whether the Greek verbs in -αω are also to be excluded from this connexion. The formal and distinct contradiction of one of the main points of the theory on the part of so distinguished a scholar compels me to examine, however briefly, both the grounds he alleges against Bopp's view, and the theory which he himself propounds. Corssen's main objection to what till then was the prevailing doctrine is this, — that it is improbable that the same formation should have been transformed in so many different ways — that the old ajar-mi should appear 333 now as -ao, now as -eo, now as -io. This objection, if valid, would serve equally to disprove the original identity of the Greek verbs in -αω,*-οω, and -e<» which so often take each other's places, as also the original identity of the three Gothic weak conjugations, and the same principle might even excite doubts as to that of the Latin verbs in -eo and -io. Still, if the same suffix -tar has survived in Greek as -rep (πάτερ) and -το ρ (ρητό ρ), -tar as -τω ρ (μήστωρ, Lat. dator) and -τηρ (δοτή ρ), if the suffix -as appears sometimes as -os (-us) e.g. in decue, decor-is, sometimes as -es (-os) e.g. in genus, gener-is, if Corssen himself • regards not only these suffixes, but also -on and -en (-in), -ont and -ent, -aio, and -eio as in each case essentially the same, this is, I think, a sufficient dis- proof of the applicability of this principle in general, and it even proves unmis- takably on the other hand that what were originally purely phonetic trans- formations have not unfrequently led to the growth from one and the same primary form, of different secondary forms, which have then, in the course of time, won themselves a separate existence. Objection is taken in the second place to the disappearance of the j between the two α -sounds. ' Neither in Latin,' says Oorssen, ' nor in Oscan does I or j ever fall out between two a's without leaving a trace, and allowing the a-a after- wards to run together into a.' This may as a matter of fact be quite true ; still I do not think there is much to be gained by phonetic rules which are of purely theoretical construction, and not deduced from a series of given cases. How often, I should like to know, was an opportunity given in the Italian language for a j to fall out between two a's ? Besides, the assumption that those who identify the Latin doma-t with the Skt. damaja-ti hold that an % fell out between two a's, is a mistaken one. For it is certain that the a of the thematic vowel (cp. p. 145) had split up into e and ο long before the^' fell out. The un- contracted Greek forms make this as clear as can be. We here have actually occurring forms like δαμάα, and can consequently have no hesitation in assum- ing a middle form *damajet between damaja-ti and domat. They, then, has fallen out between a and e. Now of this disappearance aes = Skt. djas furnishes 334 us with an undoubted instance, for in this case the middle form aes has actually survived in aeneue = aes-neu-s (Unibr. ahesnes Oorssen i, 103)•. The 1st sing. domo has evidently been contracted from doma-6. Why the spirant, which (cp. Coresen i. 308) so readily falls out between vowels^-e.g. in Pompeus by the side DENOMINATIVE VERBS. 231 oi.Pompejus, in aureus by the side of χρνσαοε and the Skt. -eja-s, in quadri-ga for quadri-juga — should have been inviolably protected from falling away by the very fact that a long a went before it, I cannot conceive. In the case also of the genitives in as in Old Latin, Umbrian, and Oscan (escds, Osc. moltas) I still think it probable, in spite of what Corssen (i. 2 770) urges in behalf of a different view, that between the a at the end of the stem and that of the genitive termination, there has been lost the j which survives in the Skt. -d-j-ds. Where else can the a be said to manifest its power of defending the j from extinction ? With Mdja, major, djo the case is quite a different one ; here the j has come from gj and was probably pronounced differently, as in pejor and other forms. A further objection is expressed by Corssen as follows : ' Among the super- abundance of Latin verbs of this conjugation, there are hardly any which would correspond in root with a Greek verb in -αω or a Sanskrit verb in -ajdmi.'' In his note he mentions the verbs bovare (reboare), comare, cacare compared by Leo Meyer with βοαν, κομάν, κακκαν, as the only three instances of such a cor- respondence, and tries to get rid of the Latin verbs by showing them to be borrowed from the Greek. Whether he was right in so regarding them I will here leave an open question. But beside these verbs there are a large number of undoubtedly genuine Latin verbs which can without any etymological diffi- culty be compared with Sanskrit verbs in ajd-mi, Greek verbs in -αω, and Teutonic weak verbs. It may suffice here to enumerate the following 11 (cp. below p. 236) : Skt. dhumajd-mi (smoke, steam) Skt. sdddjd-mi (set) Skt. svandjd-mi (sound) Skt. stkdpdjd-mi (fix firmly) wtam (by the side of νζάζω veo, and the Latin in uo like acu-o, statu-o where such a union has apparently taken place, we shall find (cp. p. 246 f. below) good reason to assume the loss of a/. 3 A more thorough-going attack has of late been made upon Bopp's view by 337 Savelsberg, who in theZtschr. xxi. in his ' Umbrische Studien ' not only denies that there is any connexion between Italian derivative verbs and the Sanskrit verbs in -ajami, holding instead, on the ground of a few Old-Italian verbs of equivocal " Corssen has in his last work (p. '493 ff.) made another effort to vindicate his theory. He alleges, as far as I can see, no new arguments, and seems to ignore the most undeniable facts in the history of language. He ignores the loss of the * in the ordinary Greek μ^θύω which is clearly established by the Aeolic μ^θυΐω, and the other traces I have pointed out of an i.in Greek derivative verbs, as also the fact that the contraction of a Latin ae to d, which he denies, occurs in Sdturnus by the side of an older Saeturnvs (on which head an untenable conjecture is advanced at i. 417), in mdlls = ma-reUs, and in amdrunt for amdv-Bruut. (Cp. Corssen himself ' Aussprache,' i. 2 317.) DENOMINATIVE VEEBS. 233 meaning and obscure form, such e.g. as the Osc. tribaraka-v-um, that these have lost a v, but even gives expression on p. 197 to the like denial for one and all of the Greek verbs of the same description. I do not imagine that this view is shared in by many and I therefore content myself here with noticing it in pas- sing, the more so, that I believe that the whole of the following exposition will place in the clearest light the close connexion of the Graeco-Italic derivative verbs with the Sanskrit verbs of the 10th class and all kindred formations in the related languages. We will proceed then, in spite of these objections, to derive the Greek con- tracted verbs from the verbs in -aja-mi. The next question that arises is, what are we to say about the origin of these verbs in -aja-mi ? Two explanations of them have been advanced. According to one, which follows in the steps of the Indian grammarians, the element which is the characteristic of the verbs of the 10th class is properly speaking i, and consequently, e.g. in the case of the Skt. veda-ja-mi I make to know, the stem to be given is ved-i, and the aja is an after- growth from this, due to an addition of sound (Guna) and the affixed thematic vowel. This view has been developed most consistently by. Grassmann Ztschr. xi. 81 fK It compels Grassmann altogether to separate verbs which, like namas-jd- mi I honour, from namas reverence, clearly show an added -ja, from those in -aja-mi, as also to deny that any connexion exists between the first a in -aja-mi and the final a of noun-stems. In Sanskrit this division into two classes is ap- parently favoured by the difference of the accent. The verbs of the tenth class, the so-called causatives, accent the a before the syllable -ja \ veda-ja-ti, while the others accent the -ja itself: deva-ja-ti he serves the gods. But Delbriick (p. 209) shows that ' the partition between the two classes is not free from gaps.' The following verbs e.g. are unquestionably denominative : artha-je strive, desire, from artha-s aim, advantage, mantra-je deliberate, from mantras counsel, mrga-je hunt, which in meaning is to be referred in just the same way 338 to mrga-s gazelle, as θηρ-άω is to βηρ. Such comparisons as these give the greater probability to the second view, which is especially represented by Schleicher Comp. 3 341 and Leo Meyer ii. 19, according to which the first a is identical with the final α of a large number of noun-stems. We cannot, it is true, find a noun-stem in a for every verb in aja-mi. But there is, on the one side, nothing to hinder us from assuming that there were large numbers of stems of this character at an early period, while on the other, it was precisely in the formation of derivatives that the force of analogy procured for a form, when once made, a wide extension of its original domain. The syllable -ja however is in that case clearly the same which we have found discharging on so large a scale the function of a present expansion. It might be objected that there is an important difference here, that, whereas in the fourth class of Sanskrit verbs the syllable -ja is confined to the present stem, in the tenth it extends, along with the preceding a, through all tenses. Greek itself however can show us that this difference has nothing to do with the special nature of the denominative verbs. Numerous denominative verbs such as e.g. ποικίλλω from ποικίλο (for ποικίλλω), θανμα'ινω from θανμαν (for 6avpav-ja>), φυλάσσω from φνλακ (for φυλακ-^'ω) have this addition only in the present-stem, and form the remaining tenses straight from the unexpanded stem. We have repeatedly seen that the general relation between present-stem and verb-stem is not rigidly determined by an impassable barrier, and we may therefore well assume that the habit of regarding the syllables aja, like their late phonetic representatives in Greek 234 APPENDIX TO THE I-CLASS. d, η, ω, as a fixed element in the stem, was only introduced gradually as time •went on. In the Vedas forms like the late future Jcoraj-ishjd-mi and the like (Delbriick p. 184) are rare. Nor is it unheard of in Greek, and still less in Latin, that derived contracted verbs should have part of their stem movable : e.g. γοάω, aor. ε-γο-ο-ν, Lat. domd-re pf. dom-uijfrica-re part, fric-tu-s. This difference then can hardly shake us in the conviction that this syllable -ja was the main element in all derivative formation for all the languages of our stock. <53y if now m the verbs in ajd-mi the vowel which precedes this syllable was from the first the final vowel of the noun-stem which is the basis of the verb, it is easy to see how the bond between the final letters of the stem and the special form assumed by the derived verbs should never have been broken. What is more, we shall see in this very relation, as our more detailed examination will show us later on, the main explanation of what is at first sight so surprising a multiplicity in the forms assumed by this widely-ramifying class. The Greek derived verbs fall into three classes. Thejirst consists of those which come from vowel-stems and keep the vowel, e.g. τιμά-ω, δηλό-ω, μηνί-ω, δακρν-ω, άριστεν-ω. The second comprises the verbs which come from consonantal stems, e.g. μελαίνω, i.e. μελαν^ω, τεκμαίρομαι, i.e. τεκμαρ -jo-pai, κηρύσσω i.e. κηρνκ^ω. The third comprises those whose final vowel has been knocked off before the de- rivative termination, syncopated vowel-stems as they might be called, e.g., καθαιρώ for καθαρ-^'ω from καθαρό, άγγελλω for άγγελλω from άγγελο, φαρμάσσω for φαρμακ^'ω from φάρμακο. The third class is obviously very nearly con- nected with the second, and as the phonetic processes are the same in both, it is not always easy to determine whether the noun-stem which forms the base of the verb had a vowel once, or always ended in a consonant. For these reasons we shall join the second and third classes together and arrange the verbs in two main divisions, We first in which the sound before the derivative suffix -ja is a vowel, the second in which it is a consonant. Briefly the first main division may be called the voealic and the second the consonantal. No doubt each of these classes has been developed beyond its original domain, and hence the grow- ing force of analogy is an element of importance which must be taken into ac- count throughout the whole of this investigation. I. VOCALIC DIVISION. .1. VERBS IN -αω, -αιω, -αζω. It is an important fact, which meets us when dealing with the question of the connexion of the contracted verbs with the Sanskrit verbs in ajd-mi, that 340 we can 8t ^ point to some traces of they at this place in Greek. I have called attention at Stud. iii. 191 to the remains of derived verbs which kept the t. The t has survived after an α in παλαίω (Boeot. παληω) which it can hardly be doubted is a denominative formed from the stem πάλα (ή πάλη wrestling), especially as we find an aorist επάλησα in Hdt. viii. 21 (παλησειε), though with a special modi- fication of meaning. Other present forms, already discussed by Lobeck on Buttm. Ausf. Gr. ii. 2 59, showing an ai in derivative verbs are βιαίω (Hesych. = βιάζω), διχαίω (Hesych. = διχάζω), χαλαίω\)γ the side of χαλάω, σταλαίω by the side of σταλθώ and σταλάζω, ίσαίω (Aratus = ίσάζω), the last of which is confirmed by Hesychius's Boeotian ίσηΊ ίσάζει, for a Boeotian η is without excep- tion the representative of an ordinary Greek at (cp. above p. 60 f.). The Homeric παραφθαί^σι discussed on p. 39 f., would find a place in this list if we were to VERBS IN -αω, -αιω, -αζω, 235 follow Johannes Schmidt in taking it as a conjunctive, as also πταίω, which in a causative meaning, I make to stumble, is aptly compared by Fick i. 3 658 with the Skt. pata-jd-mi the causative from pat fall. A further case in point is that of the Lesbian Aeolic μαχαί-τα-ς = μαχητής (Alcaeus fr. 33 Be. 3 ), in as far as it points to a *μαχαίω. Other words, some of wbich have been mentioned above (άγαίομαι ρ 118, Kcpau p. 120, μιμναίσκω p. 190), we will pass over here. In all these cases then the ι represents the original j after an α in exactly the same way as in the derivative adjectives in -αιο -s e.g. άρχα-7ο-ς, κορνφα-ω -s as compared with the Skt. eja-s i.e. a-ja-s e.g. paurusheja-s coming from man (pwusha). These forms come nearest to the Gothic weak verb-stems in -ai. If, e.g. we are entitled to conjecture that in the first part of poetical compounds like ταλαί- φρων, ταλαί-μοχθο -s there is likewise the remnant of an old verb *τάλαίω as present to e -τλη-ν, i -τάλασσα, τετλάναι, this verb would correspond exactly to the Goth, thulai-th he endures (inf. thulan). Perhaps too Leo Meyer (Goth. Sprache p. 683) is right in comparing the Skt. tulajd-mi weigh, lift, and the Latin tollo is perhaps another parallel, which may be explained to come from *tol-jo for *tola-jo. Savelsberg (Ztschr. xxi. 200), who appeals to Thiersch, recommends that we should regard the -ai in the above-mentioned Greek words as a way of lengthening an α ; but this amounts to abandoning all attempt at a 341 real explanation. The original j is preserved in another form in the verbs in -ζω. That the ζ is to be regarded here as the representative of the / was the view held by Bopp, who sets the verbs in -αζω along with those in -αω, -οω, and -fa in the same class as the Sanskrit verbs in -ajdmi. The representation of an old j by ζ has been discussed by me at length in my Principles ii. 263 if., and I there call special attention to the fact that many verbs have presents in both -αω and -αζω as a strong argument that the two forms had a common origin. I will content myself here with enumerating the verbs in -αω which have presents in -αζω as well, or which have forms from other tense-stems which would naturally accompany such presents. There are the following 18, of which as many as 10 show the twofold present in Homer : αγαπάω (Horn.) άγαπάζω (Horn.) άγοράομαι (II.) αγοράζω (Hdt.) άνιάω (Horn.) άνιάζω (Horn.) αντιόω, άντιάαν (Horn.) άντιάσω, άντιάσαντα (Horn.) άμπώμαι (Hesych.) αρπάζω άτιμάω (Horn.) 4 ατιμάζω (Horn.) βιάω, βφίηκΐ (Horn.) βιάζω (Horn.) yeXcio Aor. fye\a£e (Theocr.) ίπι-δικατό-ς (Stud. iii. 189) δικάζω προς-δοκάω δοκάζω (Sophron) (Ιλνφόων (Horn.) €Ϊλνφάζ€ΐ (Horn.) ΐΰνησα, εννηθήναι (Horn.) ei /νάζςσθαι (Horn.) 4 Nauck Melanges iv. 38 if. will not admit ατιμαν in Homer and calls it an 1 erroneous form.' Cp. Euripid. Stud. ii. 179. It is questionable though whether ατιμάι/ stands on the same footing as the undoubtedly anomalous compounds like δυί-θνήσκ€ΐν which are there discussed. As the negative of τιμαν ατιμαν would no doubt be ' erroneous,' but not as a denominative from άτιμος. Cf . ασχαΚααν from *α-σχα\ο-5, κακοδαιμοναν (Aristoph.) from κακοδαίμων, beside ΰαιμοναν (Aeschylus), and ανομοιουν (Plato) beside δμοιονν. 236 APPENDIX TO THE I-CLASS. Ισάσκίτο (Ω 607) Ισάζω (Horn.) Ρ€αω (veoa>) ρ^άζω οντά(, οϋτησα (Horn.) οντάζω, οϋτασται (Horn.) παράω, παρηθήναι (Horn.) παράζω (Horn.) σχαν (Aristoph.) σχάζω χψάν ' ριγονρ (Hesych.) χιμάζαν ' top χαμωνα dtdyeiv (Hesych.) ο4 Δ "\y e h ave "beside these to consider the whole class of frequentatives in -rda> which have by-forms in -τάζω. These verbs are clearly identical in formation with the Latin frequentatives in -tare (Leo Meyer Vgl. Gr. ii. 10). νακτάω may be compared for its meaning with the Lat. habitare, as also μελετάω with the Lat. meditari, σκφτάν bears to amipeiv the same relation as that of saltare to 8alire, λαμππάν (Horn, λαμπετόωντι) to Χάμπαν that of cantare to canere. We may compare also (ίχ^τάασθαι (Horn.), έρωτάρ, aeprav {άρτάρ) pensare by the side of aeipup pendere. ϊτη-τβορ is to be derived from an *1τάω, identical with the Lat. itd-re, and now supported by the Elic iir-ap -ιτα-κωρ i.e. *έπαριτηκώς (the Damocrates inscription, Archaeol. Zeitung Ιβ76, cp. ςξ-ίτη-λο -s). From the point of view of Greek, the natural thing is to derive the majority of these verbs from masculine stems in -τα — ραιβτά-ω from the stem ρακτα- etc. But the Latin verbs have no such corresponding nouns, and if we want to maintain them to be of a similar origin we must assume that nouns of this kind survived in these derivatives and nowhere else. Both Greek and Latin would admit of the derivation from verbal-adjectives in -το, and most of the Latin verbs are found along with such adjectives. In any case the Graeco-Italic verb-stems in -ta provide us with a fresh argument for the identity which I believe to exist between the Greek and Latin «-conju- gation. Now several of the Greek verbs in -ταω have by-forms in -τάζω, some of which are of great antiquity : e.g. έλκυστάζω, which is related to ίλκνω, ίλκνζω as tractare to trahere, ρυστάζω by the side of έρύω, άλνκτάζω (Hdt.) by the side of the Homeric άλαλνκτημαι, άγυρτάζω beg (Od.) by the side of αγβι'ρω, ριπτάζω (II.) which bears to ρίπτω exactly the same relation as that of jactare to jacere, with the iterative ρίπτασκορ Ο 23, which points to a *ριτττάω, όροτάζω by the side οι ο po pat, οιροποτάζαρ by the side of ποτη-τύ-ς (Hesych.) and the Lat. potare. Hesychius has also δροκτάζίίς' nepfiXtneis, which bears to δίρκομαι the same relation as that of spectare to *speeere, (νκτάζον ' βϋχον which is clearly nothing but a by-form of €νχ€τάασθαί, ίκνιτταζορ ' ίκνπτορ like cubitare by the side of cubare, φ€ρτάζ(ΐ * φίρα. like gestare by the side of gerere, ρ^νσταζω 343 (Horn.) and ρνστάζω (Plato) by the side of ι/εύω like the Lat. nfttarehj the side of nuere, φαρτάζω by the side of φαίρω like ostentare beside ostendere. In one or two cases the Greek -ταω seems to have become weakened to -τ*ω, as in ριπτίω. Verbs in -τιζω too, like παρψίζω, will be found to be related. These formations prove, as clearly as anything can be proved, that the ter- minations -αω and -αζω are identical. AVhat else could have produced this two- fold formation if it was not such identity ? It is obvious that this identity further entitles us, when we find verbs in -αζω, which are not frequentatives, of the same root as Latin verbs in -are, to treat them as identically the same verbs, and thus to identify e.g. τνρβάζ<ιρ with the synonymous turbare, χπμάζαρ, which among other meanings has that of to pass the winter, with hiemwe which means this only, διπλάζαρ with duplare, and even δικάζαρ (Lacon. ίπιδικατός) with dicare, for both come from the same noun-stem dika way, manner, which VERBS IN -αω, -αιω, -αζω. 237 in Greek has settled down to the special meaning of the right way, law, right. How impossible it is to derive the ζ of the forms above mentioned from stems in -δ (-αδ) or from the analogy of such stems, as some have tried to do, is made sufficiently clear by the comparisons made above, but it is put beyond a doubt by the numerous verbs in -ζω denoting sounds which are derived from interjec- tions (Lobeck Rhem. 216) like αλαλάζω (αλαλά, άλαλη), αΐάζω (ami), βανζω or βανζω (βαΐι), γρνζω, ενάζω (eua), Ιύζω, οϊζω, οΐμώζω (οΐμοι), φενζω (Aesch. Ag. 1307 ΚΑ. φ*ν φευ. ΧΟ• τί τοντ ϊφβνξας ;), ωζω. There is not an atom of proba- bility that such verbs have been formed on the analogy of noun-stems in δ or γ. We may also learn from them that the guttural which appears outside the pre- sent tense need not belong to the root, but is often, like the ζ of the present, to be regarded as the representative of the oldy. The phonetic side of this question I pass over here on purpose, as, besides the above-mentioned passage in the Principles, I have given it a special dis- cussion at Studien ii. p. 185 ff. The splitting up of a single original sound into several is, as we saw above on p. 230, a common phenomenon in language. It was in this way quite possible that, in early times a distinction should arise be- tween a j which inclined to a vowel, and hence was easily volatilized between two vowels, and a thicker^" which was almost pronounced like jj, which afterwards, 344 like the initial of the Skt. juga-m, Lat. jugu-m, Gk. ζνγό-ρ produced a d before itself, and from this dj a ζ. It might be imagined that the choice between the two paths depended on the quantity of the preceding «-sound, in the same way as in Sanskrit we find two forms of verbs in aja-mi e.g. aghajami threaten from agha-s bad, and virajate behave oneself like a man from virus man. But on the one hand this parallel would give us no help in the case of the verbs in -ιζω by the side of -<τω formed from o-stems, because we can hardly imagine these to have had a vowel which was always long, while on the other hand it is more probable that the twofold Indian formation in point finds its counterpart rather in the interchange between α as the representative of the long a and an e-sound whic"h became later an «-sound, as the representative of the short a. It would be as hard to find, a definite and consistent explanation of the twofold forms, as for the greater multiplicity of vowel sounds in late linguistic periods as con- trasted with the greater simplicity of earlier times, or for the threefold forms in the Teutonic and particularly Gothic weak conjugation. As regards the relation of the verbs in -αω and -αζω to the stems from which they come, it appears (cp. Leo Meyer VergL Gr. ii. 6) that the number of the verbs in -αω which come from stems in -a, like αϊτιάομαι, άρετάω, αί>δάω, βροντάω, διψάω, ηβάω, νικάω, δρμάω,σιγάω, σκιάω, τολμάω is very considerable. Far less numerous are those from stems in o, like άντιάω, 'άριστάω, ιςράομαι, ν^άω, and still rarer the derivatives from other stems, like Ιχθνάω, ν^μ^σσάω, φνσιάω, in the case of some of which last it is not impossible that feminine substantives like * Ιχθνα (cp. θήρα) *vepeaia (cp. θυσία) were the immediate antecedents of the verbs. In the case of the verbs in -αζω the relation is somewhat different. A considerable proportion, such as αιχμάζο» (cp. αΐχμητή -s), ανγάζομαι, δικάζω, (Ιλαπινάζω, σκοπιάζω (Horn., late poets have σκοπιητψ), point to «-stems, but there are almost as many from o-stems,• such as ίτοιμάζω, λιθάζω, σηκάζω, (saepire), μετριάζω, δοκιμάζω, τοξάζομαι. Along with the latter may probably be classed the verbs in -σκάζω already mentioned among the inchoatives on p. 189, such as άλυσκάζω, ηλασκάζω, πτωσκάζω, in so far as they are based on noun-stems in -σκο. Besides these we have derivatives from stems in μα(τ) zak 238 APPENDIX TO THE I-CLASS. like ονομάζω, θαυμάζω, χαμάζω, in which the α must likewise have had its origin in the stems from which they were derived. Beyond these there are not many verbs in -αζω, and these are mostly to be explained on simple analogies. On the whole then it appears that in this particular class of verbs it is im- possible not to see that the α -sound corresponds to the final letter of the stems from which they are derived. The substitution of α for ο is of course to be at- tributed to original identity of the two vowels. Consequently the α of άντιάω and μετριάζω stands on the same footing as that of Ύάλθυβιά-δη -s, and Ή.€λοποννησια-κό-ς. 2. Verbs γν -οω, -οιω, -οζω. It is possible that this very sense of the connexion between the vowel that distinguishes the conjugation and the final letter of the stem from which the verb is derived, which has been noticeable hitherto, was what led to the forma- tion of an o-conjugation by the side of an «-conjugation. In my essay ( Ueber die Spaltung des A-Lauts' (Ber. der K. sachs. Ges. d. Wissensch. 1864) I be- lieve I have shown that the o-sound took its "place beside the α -sound later than the e-sound. There was therefore probably a time in which the Greeks had verbs in -αω and verbs in -€β,.λ«οω with levare, λοξόω with luxare^ ομαΚόω with simnlare, and to the Lat. novare corresponds, not only νβάω and 34ο ν(άζω which have been compared with it above, but ν€Οω ae well. Herein we may discern a fresh bond of union between Latin and Greek derivative forma- tion. It is possible that it was in the lively sense of the connexion of the derived verbs with the final letters of noun-stems that brought about the desire for verbs in -οω beside those in -αω, and that in a relatively late linguistic period. It accords with this view that the verbs in -οω come to a very large extent from stems in -o, as e.g. βιόω, γνμνόω, &δνόω, κακόω, χόλου μαι, μονόω, όρθόω. There are also, it is true, some which, like κορυφόω, ζημιόω, ριζόω, yc φυρόω, stand by the side of stems in -a, and a still smaller number from more out-of-the-way stems, such as ριγόω, γουνοΰμαι, στομόω, πυρόω, some of which may probably be explained by the fact that the noun had two different steins. . It is worth noticing, however, that, as Leo Meyer Vergl. Gr. ii. p. 34 has shown, the exceptions to the rule are very rare in Homer. Though there are more than forty verbs from o-stems there are only seven from others. There was once a/ here too, and the \erbs in -οω show some traces, though only faint ones, of this letter. The j has survived (Stud. iii. 193) as i in άρμοί-ματα ' άρτυματα in Hesychius and in the Homeric κυΰοι-μό-ς , which can hardly be explained otherwise than as coming from an obsolete *κνδοίω, and VERBS IN -οω, -οιω, -οζω. 239 perhaps also in ετοι-μο -s. There are only two verbs in -οζω, αρμόζω (from Homer onwards) and δεσπόζω. The former can hardly he anything but a de- nominative, either from the stem άρ-μο (άρμό -s), which does not occur before Sophocles, though its adverbialised locative άρμοΐ just, just now (Aesch.), looks as if the stem were an old one, or else from the stem άρμον, which survives only in the derivatives Άρμον-ίδη -s (E 60), άρμον-ία, άρμορ-ικό -s, and in the com- pound βητ-άρμων (θ 250)— in the last word with the original spiritus lenis — and comes very near to the stem of the Homeric άρμα team. There is in any case no reason why we should not assume the same relation between the ζ of αρμόζω and the ι of the above-mentioned άρμοίματα as between the ζ of the verbs in -αζω and the ι of the verbs in -αιω. It should be noticed moreover that outside the present Homer knows only ήρμοσ€ and that the older Attics show a pre- ference for the dental inflexion (ηρμοσται, άρμοστεος). άρμόξαι and the like are Doric (Alcman, Pindar) and are by no means(cp. p. 236) an indication of a guttu- ral stem. In view of the Homeric ηρμοσε on the other .hand we may safely 347 assert that the ξ is the result of the derivative^'. It was probably the analogy of this guttural form which produced the late Attic άρμόττω. The γ of άρμογη — which occurs first in Eupolis — is either the remains of the old j, or else is due to the analogy of ταγή by the side of τάσσω, αλλαγή by theside of άλλάσσω. The second verb in -οζω is δεσπόζω. It has often been maintained that δεσπόζω was derived {τοτη.δεσπότη-ς, — from the stem δέσποτα that is, — but this is both phonetically and historically improbable. There is not a single one of the innumerable stems with r-suinxes which has a derived verb in -ζω corresponding to it. It is no use to cite verbs like ονομάζω by the side of όνόματ -os, for, as is shown by όνομαίνω and the Goth, namn-jan, there was a preliminary όνομαν from which ονομάζω can be derived. But where are we to find a *πο\ίζω formed from πολίτη -s, a *δόζω by the side of δοτός or anything resembling them ? And yet, there was just as much reason why such verbs should be formed, if phonetically possible. We need not attach overmuch weight in this discussion to the fact that δεσπότη -s is as a matter of fact first found somewhat later than δεσπόζω or rather than the future δεσπόσσεις, the former occurring first in Tyrtaeus fragm. 7, 1, the latter in the hymn to Ceres v. 365. Homer knows only δέσποινα, while Tyrtaeus and the hymn to Ceres have δ'εσπόσυνος as well. Inasmuch as I consider that even the identity of -ποινα with the Homeric πότνια, πότνα and the Skt. patni is by no means so certain as seems to be supposed, and that it would be as hard to get δεσπόσυνο -s from the stem δέσποτα as δεσπόζω, it seems to me far more likely that we ought to start from a compound shorter stem *δεσ-πο. The syllable πο would here correspond to the -pa of Sanskrit com- pounds like nr-pa lord of men, go-pa lord of cows, and ruler in general, and other words. In that case δεσ-πό-ζω would have come from *δεσ-πο in just the same way as the Skt. verb go-pa-ja-mi from go -ρά. δέσποινα and δεσπό-σννο -s (cp. ευφρόσυνος, γειτόσυνος) seem to point to a stem *δεσ-πο-ν, the second part of which reminds us of Πάν, and the Lith. ρό -na-s, Ch.-Sl. pa-nu. • 3. VERBS IN -εω, -ειω, -εζω AND -ιζω. 348 A portion of the verbs in -aja-mi seem very early to have weakened the first a into e. The following are the Greek verbs which can be compared with verbs of the like formation in the related languages : άρκεω, Lat. arceo. άρτεομαι, Skt. rta-je, set in order. 240 APPENDIX TO THE I-CLASS. γηθέω, Lat. gaudeo. θαρσίω, Skt. dharshd-ja-mi (venture on), cp. above p- 229. κοίω, Lat. cav-eo, Goth, us-skav-jan, Ch.-Sl. aivaja (sei'vo), 0. H. G. scaiv-on, Lith. kavo-ju (watch, protect). όρχέομαι, Skt. rgha-jd-mi (quake, rage), cp. above p. 229. 6χίω, Skt. vdhd-jd-mi (drive, carry), Goth, vag-jan (move). ροφίω, Lat. sorbeo, 7°*% ] Lat -/o>'? M ™> °• H • G • drdhrjm (turn). φορέω, Skt. bhdrd-jd-mi (let out on hire), Oh.-Sl. su-bira-jq (cottigei'e) , Zd. uz-bdra-ja-t (he carried out or forward). ώνέομαι, Skt. vasnajd-mi (bargain). (Sophron ώΐΌίσβΓται). To these we may add a few others which, though not preserved in Greek, throw light on the related Latin e-conjugation: Lat. cens-eo, Skt. qqsd-jd-mi (give notice, announce). Lat. lub-et, Goth, lubai-th (he hopes). Lat. sil-eo, Goth, ana-silvan. Lat. tac-eo, Goth, thah-an (0. Sax. thag-jan, 0. H. G. dagen). Lat. terreo, Skt. trdsd-jd-mi (terrify), Goth, thlas-jam (?) (terrify). And the following, which are instances of the variation of vowel in the derivative syllable : βαρίω, Lat. gravdre, gravdi'i, Goth, kaurjan (burden). heinvea>, Lat. dapindri. ήγίομαι, Lat. ind-dgd-re (properly to drive into the net), καλεω, Lat. cald-re by the side of Cale-ndae, O. Sax. halon (call). 349 καλητωρ, Lat. caldtor, 0. H. G. holen (fetch). Xoeco, Lat. lavd-7'e. σκοπέω, Lat. (au)-spicdri, 0. H. G. spehon (spy). The converse of this relation holds between οργάω and wr^reo, which meet in the Skt. ur(jajd-mi (part, urtfajant powerful). The verbs in «-ω are by no means so closely related to any particular class of noun-formations as those in -αω and -οω. Some come from o-stems, as αϊνίω, δινεω,κοιρανέω, κοσμίω, οικίω, some from Λ -stems, as αττβίλ/ω, άντίω, φωνίω (Pind. φωνασ(), ψενστϋω, as many from s-stems (cp. Skt. d^djd-mi from otfas strength), as a/ueXeo), άνθέω, ςντνχίω, θαμβίω, κρατίω, some from stems ending in other consonants, as άφρον-έω, ιστορ-4ω, έπιχαρ-έω. This indifference as to noun- stems shown by the verbs in -eo> explains the fact that there is often, as in the case of άσκίω, ανχίω, δηλέομαι, πούω, νλακτέω, either no noun-stem at all or one which occurs only in compounds, e.g. in φων-ασκό -s, λογο-ποιό -s, as also, on the other hand, the fact that the forms of the verbs in -ea> and -ω, as we shall see later on, so often interpenetrate each other. Of presents in -ua> there are seven in Homer (Stud. iii. p. 192) : άκίομαι (also in Pind. Pyth. ix. 104), μαχ^ομαι, νπκ^ίω (also in Hesiod and Theocritus), οϊνοβαρ^ίω, 6κν€ΐω, π*νθ(ίω, τίλβιω. With regard to five of these Leskien has shown at Stud. ii. 95, that they are founded on sigmatic noun-stems, those i.e. of the words άκος, νίϊκος, οϊνοβαρψ, πένθος, τίλος. He conjectures not im- VEKBS IN -εω, -ειω, -εζω AND -νζω. 241 probably that the same is the case with όκνείω, for it is just as conceivable that there should be a to *6κνος, formed like έθνος, ετνος as well as a ό οκνο-ς, as that there should be, as there is, a to o-kotos as well as a ό σκότο -s. Still this is mere conjecture, and as regards μαχείομαι there is absolute lack of grounds for saying that in this instance also the ει owes its existence to a σ which once existed between the e and the t. There is on the other hand good ground, in view of the Lesbian μαχαί-τα -s mentioned on p. 235, for thinking that μαχεί- ομαι {ρ 471 μαχειόμενος) existed along with μαχαίω and was formed from the stem μαχα (μάχη). To this may be added νμνείονσαι in the proem to Hesiod's Opp. 2, which is certainly to be referred to ΰμνο-ς, and οικείων (Theog. 330) by the side of οΐκο -s. It is possible that we ought not to attach much weight to the 35Q forms used by late poets, of which Lobeck Rhemat. 92 gives a list, for these are probably all blind imitations of Homeric prototypes which owe their existence to the delusion, not even yet rooted out, that the Homeric dialect admits of the lengthening of any ε whatever into ει. It is to be noticed moreover, that Herodian (in the Ε. M. p. 620, 44, ed. Lentz ii. 267) regarded όκνείω by the side of όκνεω not as mere πλεονασμός, but as παράγωγη, ωσπερ πάρα το θάλπω γίνεται θάΚπείω, ριγώ ριγείω, όντως όκνώ όκνείω. So too at ii. 462, where he adds the desideratives like πολεμηοείω. Such a view was perhaps based on the fact that in some cases there were only so-called barytone verbs in use by the side of those in ειω, as in this very instance of θάλπε ίω and θάλπω, and that of θερει- όμενος (Nicand. Ther. 124, Al. 567) and θερομαι. A bare mention may here suffice for the following forms : άπείλείω (Nonnus, Musaeus), καπνείων (Nicand. Ther. 36), κελενθείοντες (όδενοντες Hesych., the M.S. has κελενθίο ντες), νδείομεν (Oallim. in Jov. 76). It is only θαλπείω and θερείω which could have anything to do with stems in cr. Considering all this I think it must be admitted that -ειω was an old by-form of -εω, related to it in much the same way as -αιω to -αω. The Boeotians said ιω for εω : ανλίοντο, δοκίει ( = δοκετ)) etc. (Ahrens Aeol. 179). Since 1, in this dialect, when long, is the regular representative of the diphthong ει, and when short, can stand for ε too, and since the forms given above are only known to us from inscriptions, it is impossible to say whether the step before the ιω was ειω or εω. There is, according to Herodian (i. 443, ii. 949), only one present in -εζω of more than two syllables : πιέζω, which is found from Homer onwards (Π 510, δ 419), with the Doric by-form πιάζω (Alcman fragm. 44 Be. 3 επίαζεν) and the doubtful Ionic πιεζεω which Herodian (ii. 140) attributed to Apion, while reject- ing it himself. The Doric πιάζω was, as Herodian saw, άναλογωτερον. Pick, (i. 3 146) is probably right in taking the rt. of the verb to be pis, Skt. pish, which, though its primary meaning, retained in πτίσσω, is * pound,' comes very near to πιέζω in many of its compounds, e.g. in ά -pish press hard, ut-pish crush, ρη-αίί- 35 \ pish rub against anything. I cannot agree with Tick however in regarding the ζ of πιέζω as related to the d of the secondary root pid for *pis-d (to be hard pressed). The *πισ-ε-δ or even *πισ-α-δ which, he assumes, and supposes to have arisen from *pis-d by the introduction of an auxiliary vowel, has no analogy to support it. This explanation, too, will find obstacles in the forms with ξ, γ and χ, and especially in the Dor. πιάξας (Theocr.), in επίεξα, πεπίεγ- μαι, επιεχθην (Hippocr.), which exist on good authority beside επίεσα and 7re7r/e- σμαι. For with very rare exceptions, among which e.g. is καθίξη in Theocr. (1, 51) from the rt. id, such guttural forms are found in conjunction with a present in ζ only when this ζ ia the product of γ +j or a simple^', and not when. 242 APPENDIX TO THE I-CLASS. it has come from b +j. It results from this that the ζ of πιάζω, πιέζω is a derivative ζ of the same kind as that in αΚαπάζω, πτολ€μίζω. The primary form we may take to have been *pisaja-mi. From the rt. pis a noun-stem *pis-a, Gk. πι-σο, πι-ο, must have been formed, and from this the derivative verb πιάζω, πιέζω, of which the latter must bear to the former much the same relation as that of the New-Ionic 6 ρέω to δράω. The further weakening of e to ι, which took place as a rule in the analogous verbs, was prevented in the case of πιέζω by the preceding t. Op. Mangold Stud. vi. 155. — Further traces of an e in this position are the Homeric forms άκηχέδατ' (Ρ 637) and έληλέδατ' (η 86), on which I may refer to Prine. ii. 293. As a companion to the former we find at Μ 179 the specially noticeable άκαχς'ιατο, of which the «, which has been transported into the perfect stem, can hardly be a mistake for η. άκηχέδατ 1 derives addi- tional support from the substantive aiajxeSoves ' λύπαι (Hesych.). The δ is here of just the same kind as in the Herodotean κχωρίδαται, only before it the older e, which in άκαχίζω has sunk before the double consonant to ι, has survived intact. In the case of the reduplicated stem ελ^λβδ we may assume the same relation to *ίληλαδ as that between πιέζω and the Doric πιάζω. As a present to it we should have expected Άφι, and with the fuller vowel *έλάζω, to the latter of which moreover point such forms as ήλασα, έλαστός, ηλάσθην. Perhaps too there is to be seen in Homer's epithet for Notos, apyearrjs (apyearao Noroio 352 A 306) the trace of a verb *άργέζ€ΐν to make clear, from the adjective apyo-s, in its meaning of Xev<09 which is preserved in Hesychius. — The Tarentine aveypa ' αίνιγμα Hesych. points to *αινέζω as an older form for αΐνίζω. We have in this way traced the steps which connect the verbs in -eo> with those in -ιζω. Of the latter it is only a portion which belong to the division now under discussion, while others, in which, e.g. in όπίζω from the stem 6πι, ελπίζω from the stem έλπι, we have no reason to deny the antiquity of the ι, will take rank along with the verbs in -ιω. Those verbs in -ιζω, on the other hand, which in formation and meaning approach nearly to those in -eo>, belong to our present class. The fact itself, that ιζω takes rank beside «o, as αζω beside αω, was first recognised by Grassmann, and by him demonstrated at length at Ztschr. xi. 97 if. The simple explanation thereof is to be found in the double consonant, before which in ρίζα (cp. rad-ix), ιζω ( = «δ-^'ω), χθιζός beside χθες a hard vowel has passed into the corresponding soft one. Further parallels to this weakening are to be seen in ΐσ-θι by the side of έσ-τί, ίσ-τία by the side of εστία and other phenomena discussed at Princ. ii. 379. Lobeck saw that the verbs in -ιζω were closely related to those in -eo>, and he even declares at Rhemat. 227: ' Eorum quae apud Homerum leguntur pleraque metri causa diverse declinata sunt : κομέ^ιν κομίζων, χατέουσι χατίζΐΐ etc.' ; though he has sufficient insight to make him shrink from finding the exclusive ground for the twofold form in the licence of poetry, and prefers instead to conjecture ( hunc paraschematismum cum sermone ipso natum esse.' Homeric Greek presents us with 9 cases of this twofold form : αινησουσι (π 380) αινίζομ' (θ 487) 5 αΐτησων (ρ 365) αιτίζων (ρ 222) άκάχησ -t (Ψ 223) άκάχίζ*ις (π 432) β ' • To this verb belongs αίνιγμα, in which the y is of just the same kind as that of ο,ρμογί), and just as this latter had αρμόσοω formed at a late period, so the former had αΐνίσσομαι formed in Attic Greek. * ακαχίζω is, like γσργαρίζω and a few other reduplicated verbs, an intensive VERBS IN -εω, -ειω χ -εζω AND -ιζω, 243 κανάχησε (r 469) κανάχιζε (Μ 36) κομεειν (ζ 207) κομιζόμενος (θ 451) 353 κονάβησε (Ο 648) κονάβιζ€ (Ν 498) μόχθησαν (Κ 106) μοχθίζοντα (Β 723) όχλ*ί)ι/ται (Φ 261) όχλίσσειαν (Μ 448) (I. Bekker οχλ?7σ«αϊ>) προκαλεσσατο (Η 218) προκαλίζετο (Γ 19) To these may be added άράβησε and apci/3t£e in Hesiod, the Homeric σμαρα- γησαι and Hesiod's εσμαράγιζε, and a large number from late poets and prose- writers, such as τειχεω and τ€ΐχί'£ω, άτρεμεω and άτρεμίζω, νστερίω and νστερίζω, ηρεμεω and ηρεμίζω, άσμενεω and άσμενίζω, σελαγεω and σελαγίζω. With re- spect to the differences which are to be seen in some of these cases in the meaning, e.g. in δειπνεΐν (take a meal) and δειπνίζειν (entertain a guest), both in Homer, and in others in the matter of dialect or style, we may refer to Lobeck. The only point I wish to emphasise here is that several aorists and futures with a short vowel are thus satisfactorily explained, άκήδεσα (S 427), αρκέσω, καλίσσατο, αχνίσω it is quite simple to regard as belonging to *άκφεζω etc. *καλίζω, * αχνίζω were the forerunners of the actually existing καλίζω and αχνίζω. In fact καλίσσαι bears to καλίζω exactly the same relation that «Wat does to ΐζω. The close relationship of the two formations entitles us to compare the verbs in -ιζω as well as those in -εω with Latin verbs of the e-conj ligation. Thus μαδίζω the by-form of μαδάω comes perhaps nearer to the Latin madere than the «-form does, and προκάλίζω as near to the Latin Calendae as καλίω does. φατίζειν seems to have sprung from two sources : on the one hand, in the mean- ing speak, spread a report, it comes very near to φάτι-ς, while in that of promise, appoint (e.g. Eurip. Iph. A. 135 os τω της θεάς σήν παϊδ' αλοχον φατίσας ηγες) it strongly suggests fateri, profiteri. The interchange between e and α is so common that we are entitled also to assume the relation between όμαλίζω (also όμάλόω) and simulare it ορίζω and par are ■γνωρίζω and i-gnorare 1 354 to be none other than that with which we met above in the case of βαρίω and gravare. If we inquire into the sources from which the verbs in -ιζω are derived within the Greek language, we shall find that many, such as εναρίζω, λογίζομαι, οικίζω, οπλίζω, νομίζω, προμαχίζω, come from cisterns, a very small number, such as καναχίζω, πειρητίζω (cp. above p. 236), πληκτίζομαι, from α -stems, several, such as τειχίζω, μερίζω, θερίζω, ονεώίζω, κτερεΐζω, from s-stems. By far the largest portion of verbs in ιζω developed as the language went On, gradually, and tome quite late. Those of them which point to a consonantal stem, such as αγωνίζομαι, ακοντίζω, μακαρίζω etc., will be placed in the other division, in which the ι is the original letter. formation (cp. the Homeric αχίων), and thus takes rank with the intensives, with which we became acquainted at pp. 212, 215, 226 etc., though it is distinguished from them by the fact that -ιζω, like -εω, points to a preliminary noun-form. In the case of μερμηρίζω this form is actually found in μίρμηρα. 7 Both verbs have evidently come from an adjective-stem almost identical with the Lat. gndru-s, i-gndru-s. r 2 244 APPENDIX TO THE I-CLASS. First Excursus. On the Interchange and the Meaning of the Verbs in -αω, -οω, -*ω. Having thus reviewed the origin of the three commonest classes of deriva- tive verbs, we have two things left to do before we proceed to the other classes. We must first say something more on the grounds for the division of what was originally a single class, more particularly in respect of the meaning, and then we shall have to discuss the phonetic treatment which the so-called contracted verbs received in the various Greek dialects. The source of the division of the derivative verbs was, as has been above stated, the splitting up of the old α -sound, which resulted, as we conjectured, first in the development of an e beside the a, and so of an e-conjugation beside the α -conjugation, and subsequently of an o-conjugation as well. It is clear, however, that these fresh growths did not unfold themselves unaided by the operation of analogies not only of sound but of meaning as well. At first all derivative verbs had hardly any other function than that of denoting a state or 355 action standing in some relation or other to the notion of the noun-stem from which they were derived, and even as to the relations in which we find them in the period of Greek of which we have full knowledge, it is, as Lobeck says (on Buttm. Ausf. Gr. ii. 3 384), labour lost to try to get a definite formula for the use of the different kinds of derivative verbs. It is, however, possible, as I have pointed out in my essay ' Ueber die Spuren einer lateinischen o-«onjuga- tion ' (Symbola Philol. Bonnens. i. p. 272), at all events to find some pretty thorough-going analogies. On reviewing the lists given by Leo Meyer of Homeric verbs in -αω, -ea>, -οω, we perceive immediately that those in -οω are formed, in the majority of cases, from adjectival osteins — out of 40 there are 21 such — and that these have clearly a causative or factitive meaning, as e.g. in ά'ΐστόω, άλαήω, άλίόω, γυμνόω, Ισόω, κακόω, κνρτΐω, μονόω, οίόω, ομοιόω, όρθόω, σαόω, χηρόω, all of which we can translate to make something. Along with these go others which come from substantives, and have a similar meaning, — that of ( bring about something, provide with something' — e.g. νπνόω, χολόω, θριγκόω, nrepoa) (cp. the Skt. pattrdja-mi provide with feathers), στεφανόω. On the other hand the verbs in -αω, which come from feminine noun-stems in a, get their meaning from these nouns, and generally denote the exercise of some activity or the existence of some state, e.g. αγαπάω, άγοράομαι, αΐτιάομαι, ανδάω, βοάω, δ\>αω, ήβάω, βΐάομαι, θοινάομαι, μηχανάομαι, νικάω, πειράω, πλανά- ομαι, ποτάομαι, σιγάω, τολμάω. This analogy makes itself felt also in the com- paratively late formations in -ιαω, which denote a bodily or a mental languishing or desire (cp. Chap. XXIII.), such as νανσιάω, ώχριάω, κλαυσιάω, στρατηγιάω, τνρανριάω. Most of these verbs presuppose, in idea at least, feminine nouns in -a. With regard to the verbs in -ea>, however, I would venture to conjecture that they were formed under the influence of a twofold analogy. On the one hand it looks as if an intransitive meaning was developed very early in a large number of them. In view of Latin pairs of verbs, such as albare and albere, clarare and clarere, salvare and salvei-e, as also of the numerous inchoatives in ~esco (cp. above p. 190), and the prevailing intransitive or inchoative meaning of 35ρ the Church-Slavonic verbs in ej'e-ti (e.g. HuU-je-U flavescere) one is tempted to conjecture that in a period preceding the differentiation of the Greek language as such, this difference had developed itself between the a- and e-conjugations, THE MEANING OF THE CONTKACTED VERBS. 245 that the former — from which sprung later the o-conjugation as a separate formation — took more particularly the transitive meaning, the latter the in- transitive. This rule holds good at any rate in Greek in a large number of compounds. In this language -eo>, — and -ιζω too, — was the formation uniformly employed for the numerous and mostly intransitive verbs formed from nomina agentis, of which Leo Meyer (p. 25) finds 21 in Homer, such as άέΚπτίω, άπιστέω, 4πικουρί<ύ, νπερηφανέω. A second analogy is to be found in the extraordinarily large number of masculine substantives in ο of abstract meaning which existed in Greek from the first, such as άθλος, αίνος, γάμος, κόσμος. Out of 66 Homeric verbs in -eco 44, such as άθλέω, αϊνέω, γαμίω, κοσμίω, κτυπέω, μοχθίω, οκνέω, δμαΰεω, 6μιΧ4ω, ποθίω, πονέομαι, στοναχίω (στβναχίζω), φθονέω, come from stems of this kind. In this way it happened that a much less definite contrast arose between verbs in eto and those in αω than between verbs in eco and those in οω, which we assumed, as a rule, to be formed from adjective-stems. It was the coincidence of the most various circumstances which ob- literated the boundaries between the thre». conjugations. In the first place a conflict arose in many instances between the analogy of meaning and that of sound, γςφυρόω probably owes its existence, in spite of γέφυρα, to the prepon- derance of the former, as does Ιεράομαι in spite of Upo -ς, while λωβάομαι and τβλβυτάω, in spite of their somewhat causative meaning, to the phonetic analogy, i.e. to the prevailing sense of their connexion with α -stems. Of course there were other tendencies which helped to shift the verb from one class to another. Possibly it is sometimes only our defective knowledge of the Greek vocabulary which makes us think this. There are, for instance, many instances of α and ο stems existing side by side ; and it is sometimes a mere chance that one has been preserved and not the other. How easily, e.g. might the remarkable κορυφουσθαι have come from a lost *κορνφο-ς existing by the side of κορυφή ? On the other hand there were no doubt cases where phonetic weakenings took place. Many an original -αω, -αζω may in this way have degenerated to -*ω, -ιζω, 35 7 more particularly in cases in which the consciousness of the connexion with an α stem had been lost, as we conjectured to have happened in the case of ριπτέω by the side of the Homeric ρίπτασκ€, ριπτάζω. In the Greek dialects the mutual interchange between all three conjugations is as common as that within the same dialect at different times, and where two or even three forms of the same word were in vogue at one time, it was necessary that the tendency towards differentiation should give rise to small varieties of use which are not in all cases quite consistent with the prevailing analogies. In the chapter ' de confusione terminationum conjugationis circumflexae ' in Lobeck's Rhematikon p. 163 if. and in his note on Buttmann ii. 2 53, is a copious store of material, although all that is purely dialectic is there excluded. It will be enough for us to adduce a series of facts. The numerous Ionian by-forms in -*ω for the verbs in -αω seem clearly to be the result of a weakening. In this case we are inclined to find the special ground in the Ionic preference for e as compared with other vowels. Hence rjvreov, pevoiveov, όμόκλ^ον by the side of forms with the α in Homer, ipia> for ipaa>, διψίω for διψάω in Archilochus, κυκ€υμ€νος for κυκωμενος in Solon, έχρέοντο, olbeovro, 6ρ4ω in Hippocrates (Renner Stud. i. 2, 43), and much of the same sort, such as άρωτςον, έμηχανέοντο, ωρμέοντο, φοιτ^όντ-ων, τολμέω in Herodotus (Bredow dial. Herod. 382). This explanation, however, does not hold for all cases, for while it may be said that ωνασάται by the side of ώνέομαι is an 246 APPENDIX TO THE I-CLASS. antiquated Doric form, on the other hand ορίων, σνλέων, επιτιμίων, ipewiovrcs are Doric as well as Ionic (Ahrens 310, Sitzungsber. der k. sachs. Gesellsch. d. W. 1864, p. 221), and though Plato uses αίολίω, the Ionic Hippocrates has αίολάται. ξνρίω is the older and ξυράω the post- Attic form, and there are other instances in which the Atticists warn their readers against the αω of the κοινή as opposed to the 6 and οω, such as is to be seen in the Ionic forms άντι^νμ^θα, άξιεύμςνος, μισβ(.ννται, ibiKaievvro (Bredow 391). We find κνκλέω and κνκλόω, piyeco and ρι-γόω existing side by side without essential difference of meaning. There are even cases where all three forms exist, as σκηνάω, σκηνίω, σκηνάω, all three good Attic, and with no definite variety of meaning. The unity of all these forms is, I think, con- firmed afresh by the extent to which they were used for each other. Second Excursus. On the Inflexion of the Verba Contractu. I have treated the inflexion of the verba contracta in full at ' Studien ' iii. p. 379 fF. I may therefore be allowed to give here a very brief recapitulation of the results of the investigations there detailed. As distinguished from the other dialects, the Aeolic dialect, it is generally stated, treated the contracted verbs like verbs in -μι. A closer examination, however, reveals the fact, that the use of different terminations was quite an in- significant element in the difference between the dialects — for in reality it is only in the first person singular : Aeol. φίλημι by the side of φιλέω, and in the infinitive active : φιλήμ^ν or φιλήναι by the side of φιλίϊν, that there is a dif- ference in this respect — the main point being that they differ in the mode of treatment of the vowels and diphthongs which appear in the body of the verbal forms : 359 Aeol. φίλημίν Att. φίλονμ^ν „ φίλ^ισι ,, φϊλονσι „ part, φιλί is „ φιλών „ φιλήμ€νος Μ φι\ονμ€νος These vowels are in many forms just as long in Aeolic as in Attic, so that φΐλη-μ(ν and τιθί-μΐν *άσάμ*νος and ιστά-μ€νος THE INFLEXION OF THE CONTRACTED VERBS. 247 are distinguished from each other equally clearly in both dialects. The only exceptions are such forms as either have, or once had, vr after the conjugational vowel : φίλεις, gen. φίλε-ντ-ος, 3rd pi. φιλεισι for *φίλε-ντι, and the vowels preceding the modal characteristic of the optative : φΐλε-ίη-ν, γελα-ίη-ν. We have already seen (on p. 135) the neighbouring sounds exercising the same shortening influence in the primitive aorists, and we referred then to the forma- tions now under discussion. Now the long vowel sound which we thus find to be the rule must without doubt be explained to be the result of contraction. The Aeolic verbs φίλημι, γέλαιμι, δοκίμωμι are just as much contracted verbs as those in the other dialects and in Latin, to the latter of which the Aeolic forms bear the closest resemblance, e.g. δόκη-μεν = Lat. doce-mu» δόκει-σι = „ doce-nt δοκή μένος = „ doce-mini. In the conflict between the two vowels Aeolic like Latin lets the first vowel, which we may call the conjugational vowel, prevail throughout, while the main Ionic dialect was far less consistent in its procedure. In order to comprehend the rule of the Aeolic contraction we must start from -ajd-mi as the primary form. This is only feasible, however, on the three following assumptions : 1) The second a may, at the time when the Greek dialects still formed a single whole, not yet have manifested that regular change of ο and e sounds which is the characteristic of the thematic vowel. There is no way of getting from φιλεομεν to the Aeol. φιλημεν, or from φιλεόμενος to φιλημένος. On the other hand all the vowels can be completely explained as soon as we assume 360 that the -ajd-mi first appeared on Greek ground in the forms -a-je-mi, -o-je-mi, -e-je-mi. At p. 206 we recognised, as we thought, in the syllable -ja the verb to go. In the forms U-vai, ιε-Ίη-ν we have formations which differ from the -je-mi of our search, only in having the vowel ι in the place of the spirant j. Hesychius has moreover preserved the gloss ε'ίη-μι' πορεύομαι. If this is, as Lobeck conjectured, a mistake for ϊη-μι, we have here the 1st sing. ind. to Ιε-ναι. Perhaps too θίη-μι ' ποιώ, θιη-σαι' ποιησαι (Hesych.) with 6ι for θε, is to be referred to *dha-jd-mi, so that it comes quite near to the Ch.-Sl. d£-jq I do. A remarkable form however, which, as I believe, confirms our conjecture of the original presence of the e-sound in this place, is the hitherto unexplained and extraordinary present άχνάσδημι quoted in the E. M. p. 181, 44 from Alcaeus. The authority for it is Herodian περ\ παθών (ed. Lentz ii. 290), and this is enough to give credit to the form. Alcaeus wrote άχνάσδημι κάκως meaning ( I am sorely vexed.' Herodian is no doubt right in connecting the form with αχός, or, as we should say, with the rt. άχ, though in a way of his own : ' εστίν άχω, παραγωγόν άχάζω, άχάζημι κα\ τροπτ} τον ζ εις σ καί δ πλεονασμοί τον ν άχνάσδημι.' Without vouching for the perfect completeness of this account, we may anyhow be sure of the form itself, άχνάσδημι was the Aeolic form of what in Attic was άχνάζω. This form we may suppose to have arisen in the following way. There is nothing to prevent our presupposing a noun-stem *άχ-να, formed from the rt. άχ in. a similar way to that in which τεχ-να came from the rt. τεκ. The first formation from this stem is άχνάζω* which might 8 Hesychius's gloss αχνάζει • άχθεται, μισεί, ψε-γει has been omitted from both of Mor. Schmidt's editions, no doubt by an oversight, as the note to 92 shows The older editions have it. 248 APPENDIX TO THE I-CLASS. have come from it exactly in the same way as δικάζω from the stem δικά. Inasmuch now as we have recognised the ζ of such formations to he the repre- sentative of an original j, we are "brought to a form *άχνα^'ω. In the place of the ω, however, άχνάσδημι, which has incontestahly come from άχνα-^η-μι, shows 361 us the very e sound which we were looking for in verbs of this kind. It is remarkable what confirmation this form gives to both of our hypotheses, first that the e sound is the predecessor of the ο sound, and again that the verba contracta were originally identical with those in ζω. From the primary a-ja-mi then there was developed, as we may now assert more definitely, first a-je-mi. Thi9 again split up, according to the way in which the j was treated, on the one hand to α-ζη-μι, on the other to α-η-μι, and inasmuch as the a can appear in either of three forms a, e, or o, there results all the variety of ending which we have here to try to comprehend. 2) The second assumption which is requisite for the comprehension of the Aeolic formation is that the /, previously to its disappearance from between the two vowels, was in all cases productive of lengthening. The most various traces survive of the original length of the vowels of the verba contracta, as we shall have occasion to notice immediately. We shall therefore have to assume, for an early period of Greek, forms like *γ€λάη-μι, *φΐ\ήη-μι, *δον\ωη-μι, *γ(λά€-μ€νος, *φιλη€-μ€νος, *8ονλα>€-μ€Ρος, from which by contraction came φιλημι, δούλωμι, yeXd/xeiOs (cp. άσάμ^νος), δονλώμενος. 9 3) Our third assumption is that a transition, even among the Aeolians, was gradually made to the conjugation in -ω, though perhaps only to a small extent. The only actual instances we have from Aeolic Greek are καλήω (Herodian ii. 332), ποθήω, άδικψι. ποθήω bears the same relation to πόθημι that δεικνύω does to δείκννμι. Strictly speaking what happened was that the e-sound in ποθή-ψμι, plur. ποθη-€-μ€Ρ, underwent the change to an o-sound which regularly happens in the verbs in ω. The first person ποθήω, later ποθίω finds its c^sest analogy in the Lat. mone-o, while mone-mus, mone-nt, as we saw, correspond to *πόθη-μ€ν, *πόθ€ΐ-σι. The Cyprian καλήζω (Princ. ii. 266) preserved by Herodian (i. 444, 362 ii. 332) is a remarkable form. It proves on the one hand that ζ represents j alone, and shows, on the other, that the lengthening assumed above belongs to a period previous to the origin of the forms with the ζ. % Homeric Greek, as is well known, has retained lengthened forms of this kind in great abundance: from the «-conjugation forms like σνλήτην, προςανδή- την, συναντήτην, φοιτητην, άρήμ^ναι, γοήμςναι, π^ινημςναι, ονήμςνος, which are only distinguished from the Aeolic formations by the Ionic η ; from the e-con- jugation forms like άπΐΐΚητη ι>, όμαρτήτην, καλημζναι, πίνθημ€ναι, φιλήμ^ναι, which may be compared with the Boeotian καρτ€ρημ(ν (probably more correctly καρ- Tcpeipep Ahr. Dor. 523), akiTijpevos which corresponds to the Arcad. αδικημένος, the Lesb. φορήμ^νος and the Boeot. Delph. and Locr. participles, like άδικύ- μΐνος, άφαρύμίνος, /caXet'/xf νος ; from the ©-conjugation the 3rd sing, σάω (Π 363, Φ 238), which could not conceivably have come from *fVaoe, but which falls • The specifically Aeolic epenthesis of the ι in γέλαιμι, and other irregularities, such as those enumerated by Ahrens Aeol. 139 — 3rd sing. ψΙλ*ι, χαύνοι, apparently the result of weakening— need not be mentioned here. VERBS IN -ιω AND -ίζω. 249 into its place quite simply as the 3rd sing, to the 2nd sing, preserved in Alcaeus fr. 73 or' ασφ' άπολλνμένοις σάως, as also does the imperative σάω ν 230, ρ 595 as analogous to the so-called Aeolic μνρω (Ahrens 140), so that we have to recognise in it a contracted form of *σάω€. To these must be added the forms έάλων, αΚο'ιην, άλώναι, άλονς, (βίων, βιοίην, βιώναι, βιούε which survived after Homer's time, and which, notwithstanding their aoristic meaning are, as we saw on p. 133 f., undoubtedly presents in origin, and provide us, in their long vowels, with the clearest evidence in support of our view. It is thus placed beyond a doubt that here, as in other cases, the Aeolians preserved formations of a kind peculiar to the earliest period of the Greek language in general. There is a like plenty of forms from various dialects like the Aeolic ποθήω, άδικψις, forms, that is, in which the thematic vowel has made its appearance, while the conjugational vowel is still long. Such are the Homeric πςινάω, διψάω, μ(νοινητ]σι (Ο 82), the Hesiodic άμάαν (Opp. 392), the Attic neivfjs, διψη, έδίψη, χρηται etc., which can only be conceived as coming from neivrjeis, διψψι etc., the Dtelphic σνληοντ^ς, by the side of σνλίων, avXeovrcs, and contracted forms like σνλήν, συλητω, the Homeric νπνώονπ-ς, ίδρωουσα, just like the forms from the Delphic dialect : άπαλλοτριώονσα, άπαλλοτριωοίη, στ€- φανω€τω, μαστιγώων, by the side of the contracted δουλωη, of which the last may be compared with well-attested Attic forms like the infin. ριγών, conj. 3. s. 353 ριγώ, Opt. ριγωην, part. dat. ριγώντι (Aristoph. Ach. 1146) and Hippocrates's Ιδρωην, Ίδρώσι, ιδρώντ^ς. The view, which in itself is objectionable, that forms like these are in Homer due to metrical license, is completely upset by these facts. What place is to be found, in this history of the verba contracta, for the so- called lengthened, but more properly speaking, assimilated Homeric forms like όρόω, όράας, μνωόμ^νος etc., is a question which I will not again enter on here, as I have on several occasions expressed my views with regard to it, — most recently at Stud. iii. 400, — and as I can now refer the reader to Mangold's paper ' De Diectasi Homerica,' Stud. vi. 139 ff. 4. Verbs in ιω and ιζω. Just as the noun-stems in ι are far less numerous than those which end in α or o, so too the verbs in -ιω and those verbs in -ιζω in which the t may be regarded as native to the stem fall in number far below the three classes hitherto discussed. The only verbs in -ιω from noun-stems in ι are δηρίω (Homeric δ»7ρί- σασθαι from δήρι -s), κληΐω (Horn, κληϊσαι from κλη^ί-ς = Lat. clams, Hdt. κληΐω, Att. κλτ]'ω, later κλβι'ω), κονίω (κόνι -s, later κον'ιζω), μαστ'ιω (μάστιε Ρ Q62 from the St. μαστι ace. μάστι-ν), μηνίω (μήνι -s, μηνϊζ, μηνιααε), μητίομαι (μήτι -s, μητισομαι). We cannot be so sure that the Homeric κηκίω (άνςκήκιε, Soph. κηκΐον) comes from the post-Homeric κηκί -s (gen. κηκϊδος). The first two of these verbs have a long t throughout, and this must undoubtedly be held to be older than the short vowel, and is perhaps the product of the final ι of the stem and the derivative j {κονι-^ω, κονιω). To these correspond, in Sanskrit, verbs like (fani-ja-ti he wants a wife (Jani-s), arati-ja-ti he threatens mischief {ardti-s mischief), in Latin, verbs like finire, grandire, lenire, potire, in-retire, vestire which are evidently founded on noun-stems in i. Where no such nouns can be found for Greek verbs in -ιω, as is the case e.g. with άλίω, κυλίω (by-form κνλίνδω, κυλινδέω), papUiv (οχλίϊσθαι, nvptrruv Hesych.) we ought perhaps to assume that such stems have been lost. For μαρίπν, which has been wrongly 250 APPENDIX TO THE I-CLASS. suspected, such a stem might be deduced from μαρί-λη glowing coal, coal-dust,* and it is related anyhow to the name of an inflammable stone which in Aristotle 364 according to Bonitz's Index is μαριεν-ς {μαρύα v. 1. μαριθάν), in Hesych. is μαριζεύ-ς. The verbs in -ιω which are, at least apparently, primitive, we have discussed on p. 207 f. Along with these verbs in -ιω come a number of forms in -ιζω, which point to primitive stems of the same kind. They bear to the verbs in -ιω the same relation as that of the above-mentioned μαριζενς to papievs. Such are κιθαρίζω, νεμεσίζομαι, νοσφίζω (adv. νόσφι), όμηγνρίζομαι, οπίζω, πολίζω, υβρίζω, χαρίζο- μαι, χατίζω, all Homeric verbs, to which may be added e.g. from later Greek ραχίζω (cp. also ραχίτη -s), φημίζω. None of the noun-stems belonging to these verbs show a δ in their inflexion, as is the case with έλπί-ς, epi-s, πα-ΐ -s (παϊ-ς), ριπί-ς, φροντί -s, to which the verbs ίΚπίζω, ερίζω, παίζω, ριπίζω, φροντίζω correspond. Hence I cannot believe that the δ of the noun-inflexion, the sporadic character and origin of which I have discussed at Princ. ii. 278, has any special connexion with the ζ of the verbs, and hold rather that the latter is in all these cases the representative of a simple^', before which an involuntary δ was developed. It is not quite so easy to answer the question whether verbs like μαστίζω, σαλπίζω, φορμίζω, which show a guttural in the tenses beyond the present stem, have sprung from the noun-stems μαστιγ, σαλπιγγ, φορμιγγ, or from shorter ones, such as actually exists as a by-form for μαστιγ. This is a question which we can here afford to leave unsettled. That it is possible that in deno- minative verbs of this kind, to which we shall return in Chapter XVII. (p. 270 of the marginal paging), gutturals may appear which are not to be found in the noun-stem, is proved by παίζω, παιξονμαι (Laconians in Xenoph.), Έπαιξα, επαίχ- θψ, παίγνιον, and we saw likewise at p. 237 that in the inflexion of the verbs in -ζω formed from interjections there are gutturals which have come from a simple j. To this class of the verbs in -ιζω we must also add, I think, those which, like αγωνίζομαι, ακοντίζω, άνδρίζω, κελητίζω, μακαρίζω, σωφρονίζω, Come from noun-stems ending in a consonant. In this case it seems to me most natural to regard the ι as a vowel produced involuntarily before the j, so that e.g. εμποδίζω and the identical impedio would have to be referred to a denominative *])ad- ja-mi, *pad-i-ja-7ni. For several of these verbs, however, other explanations are possible. 365 5. Verbs in -νω and -νζω. The following verbs in -νω are clearly denominatives : άχλνω, γηρνω, δακρνω, ε'ρητνω, ίθνω, μεθνω, οίζνω, to which we may add φϊτνω m Lat/w£wo, though the corresponding noun-stem does not occur till a much later period than the verb, and also τανω or τανζω which is to be deduced from Hesychius's τανσας ' μεγα- Xvvas (cp. Tavs' μ^γας, πολύς). That in such verbs a j has fallen out between the ν and the thematic vowel is the natural and perfectly unobjectionable con- clusion suggested by the exactly analogous formation of Vedic verbs like r(fu-ja-ti he is upright, from fffti-s upright, kratu-jar-ti he puts forth strength, from kratu-s strength, taken in connexion with all these derivatives, and the same assumption may be made for the corresponding Latin verbs like acuo, metuo, itatuo. An important parallel is observable between the Greek δακρύειν and the synonymous Gothic tagr-jan, which, on the assumption that the Gk. word once was baKpvjttv, coincide exactly, while the loss of the j after the «* in Latin is VERBS IN -ενω AND -ονω. 251 exemplified by the Lat. struere as compared with the Goth, straujan (cp. the Skt. starajd-mi). We have however a superfluity of unmistakable intermediate forms in Greek itself in the Aeolic presents in -νίω, which have been noticed above at p. 147. One of these presents is μ^θνίω (Herodian i. 456), evidently a denominative, while άλνίω and οπνίω are as yet somewhat obscure. The other representative of the/ which we should expect in this place is ζ. We have no instance though of an undoubted denominative in -νζω. It is possible that ίρπνζω comes from a noun-stem *ίρπυ, known to us anyway as the proper name of the Theban "Ερπν-ς. A direct branch of the stem is perhaps to be seen in the name of the creeping plant Ζρπνλλον, epnvXXos, έρπνλλιον. Forms like βρπΰσω, ίίρπνσα, Ζρπνσις, βρπυστίκός bear to ϊρπω exactly the same relation as that of ίΚκϋσω, etX /κνσα, ίϊλκνσται, (ΐΚκύσθην, ελκΰσιε, ίλκνο-τάζω to ίλκω. Not- withstanding the latter are commonly referred to a present ίλκνω which does not occur till the latest Greek, though analogy speaks for ίλκύζω. The remaining verbs in -νζω are almost entirely onomatopoetic verbs with a guttural in the stem, and are widely different from the formations here in question. Some of these we have met with above on p. 237. 6. VEEBS IN -evo> AND -ονω. 366 That these verbs also once had a j before the thematic vowel, though not proved by the existence of by-forms with a ζ — for the awkward combination -(νζω probably does not occur except in the above-mentioned onomatopoetic φ€υζω — is extremely probable on various other grounds. In the first place the already well developed and well established rule for the denominative verbs in general speaks for it. At p. 232 we saw that, though a noun-stem may as such do duty as a verb-stem, e.g. deiKw in δείκννμι, the conversion of a noun-stem into a verb-stem by the addition of the thematic vowel was unheard of. And yet such a conversion must have taken place if it is to be assumed that βασίλβύ-ω came from βασιλεν without the loss of a j. The only imaginable explanation of such a form would be that *βασί\(ν-μι was formed from βασιλεν as was δείκννμ,ι from deiicvv, and that subsequently the thematic vowel came in, as in δεικνύω, on the analogy of the verbs in ω. But I know of no support for such a view. There are moreover some hints at all events that there was once a j here. If the diphthong ev had stood from the first directly before the thematic vowels, it would be surprising to find it so well preserved. If *iepev-as and *iep€v-es became Uprjos and Ιερψς, why should not we get ίερήω, Ιςρψις and the like ? Presents with a ev in them which have come direct from the root show a corresponding loss of sound e.g. oVo -μαι by the side of devo -μαι (cp. above p. 210), πλίω, πνίω, ρβω, beside δβι'ω, πλβι'ω, πνείω of the ^-class. In denominative verbs in βυω no such loss occurs. Again, we have repeatedly remarked, and shall discuss at length in Chap. XXI., p. 369 (marginal), the fact that the σ which appears to have intruded in some tenses e.g. in €ΐλκν-σ-ται, (ΐΚκν-σ-βη-ν, i -μνη- σ-θη-ν, κ€κνλι-σ-μαι, and in a corresponding way in noun-formations e.g. πρι-σ-μα (pres. πρίζω beside πρίω) points to the existence at one period of &j which has in many instances become ζ. And this same σ occurs also in some verbs in βυω, 367 especially in ΜκίΚςνσμαι, ίκίλξνσθην, ςλςνσθην, the latter of which is from the monosyllabic stem Xev for Xafa (Kaas). With this too agree the primitive verbal forms πίπλίνσται, cnvevafyv, for we found by the side of 7rXea>, ττνίω the forms πλ6#"ω, ττ^ω so that we have here another case where this σ confronts a j in 252 APPENDIX TO THE I-CLAS& the present. For these reasons Τ make no doubt that Schleicher Comp. 3 355 and Leo Meyer ii. 43 are right in referring the presents in -ευω to -ev /ω. As to the origin of the verbs in -ευω, we must look for their starting point to the noun-stems in -ευ. It must be admitted however that but a very few of these verbs have noun-stems of this kind corresponding to them. Leo Meyer gives 44 Homeric verbs which have no noun-stems in -ευ and only 10 altogether which have, and even from these 10 we must deduct 2 or 3, for it is not till the latest Greek that ήγεμονεύ-ς (for ηγεμονεύω), and υδρεύ-ς (for υδρεύω) occur, and πορεύ-ς is only found in Hesych., and that with the meaning πορθμούς, so that it is but a poor foundation for πορεύω, αριστεύω though, from the Homeric άριστεύ-ς, has been overlooked. There remain therefore but 8 Homeric verbs in -eva> which show the normal relation to the noun-stem, though there are some verbs of very common use among them ; they are αριστεύω, βασιλεύω, ήνιοχεύω, ηπερο- πεύω, ιερεύω, νομενω, πομπεύω, χαλκεύω. To these we may, it is true, add a few from post-Homeric Greek, such as άλιευω, βραβεύω, γραμματεύω, ιππεύω, σκυτεύω, τορεύω, φονεύω. But for Homeric verbs like άγορεύομαι, άεθλεύω, άρχεύω, βονλεύω, θηρεύω, θητεύω, μαντεύομαι, συλεύω, τοξεύω, and for post-Homeric verbs like παιδεύω, πιστεύω, χορεύω it is vain to look for noun-stems in -ευ. The related languages too give us no help whatever in respect of this whole forma- tion. The solution of the riddle is however probably to be found in the fact that in Homer the nominal-suffix -εν, as I have pointed out at Ztschr. iii. 78, iv. 213 is constantly added to shorter formations to strengthen them, and espe- cially to stems in -o, with the view of giving greater prominence to the person acting. Occasionally, as e.g. in ήνιοχήα, ψιοχηες by the side of ηνίοχο-ς, ούρηων Κ 84 by the side of ovpo-s, πομπηες, πομπηας by the side of πομπός, Αίθιοπηας by the side of Αιθίοπες, this addition is made in certain cases only. It is not sur- 368 Posing therefore that this suffix should have introduced itself in just the same sporadic fashion before derivative terminations, and even before the derivative -ia of feminine epithets, as e.g. on the analogy of δυςαριστοτόκεια {τοκεύ-ς) in ενπατερ-ε(ϊ)-ια, and ευρυόδε(β)ια as also in the verbs with which we have here to do. It was felt pretty much that every nomen agentis in -o might have a by-stem in -ευ as well. On the pattern of such a noun, i.e. *οϊνοχοεύς, which is just as conceivable as ηνιοχεύ-ς, was formed οίνοχοεύω, on that of *ικετεύς, a conceivable by-form of Ίκετη-ς, Ικετεύω. A very large number of the noun- stems in -ευ are designations of persons which are derived from the name of a thing and denote the man who has to do with that thing e.g. ίππεύ-ς, οίκεύ-ς, σκυτεν-ς. Perhaps there were once, on the analogy of these words, such nouns as *θηρεύ-ς, *παιδεύ-ς, *τοξεύ-ς. From these, whether they actually existed or were only present to the mind as types, came verbs like θηρεύω, παώενω, τοξεύω. Hence their decidedly preponderating intransitive meanings (Buttmann ii. 383). The verbs in -ευω have all from the first the meaning to conduct or behave oneself after the fashion of some person or other who was denoted, at least in idea, by the suffix -fv. The need of verbs to discharge this function evidently conduced to swell the list of verbs in -ευω. In use, these verbs come nearest to the verbs in -εω. This sufficiently explains the constant, though not always meaningless interchange between the two classes, for which see Lobeck Rhem. 199 ff. No one will try to deduce a phonetic connexion between these two formations from the fact that we find side by side ζητεύω and ζητεω, οίνοχοεύω and οίνοχοεω, επιδημεύω and επιδημεω. There are only a few verbs in -ονω. These few appear, at least in some VEEBS IN -νω AND -αινώ. 253 cases, to be related to those in -evto. Thus Hesychius's μολυυ€ΐν (έγκόπταν ras παραφυάδας), as was recognised by Lobeck Rhem. 206, is identical with the Attic pokevetv quoted by Pollux vii. 146 from an Attic law, and explained by ras αυτομολίας κόπτειν. μυλήω however stands in the same relation to (αύτό)μολο-ς that άρχευω does to άρχος, κολούω seems to me to be in the main rightly explained by Fick 3 i. 240, when he refers it to *κολοΡό-ς, the form to be expected as the oasis of κυλοβό-ς, so that it may possibly have gone through 369 the intermediate stage KoXop-ja), κολου^'ω, and may bear to κολοβόω the same relation as that of στρογγύλλω to στρογγνλόω. It is not so easy to explain ακούω and ορονω. If this view of these verbs is correct, they have the closest resemblance to the Church-Slavonic verbs in -u-ja inf. -ova-ti e.g. kupujq, inf. kupovati buy, the only difference being that here an α -sound is added in the infinitive. II. CONSONANTAL DIVISION. We have here to deal with derived verbs which originally had a conso- nant before the -ja> of the present, whether that consonant is to be regarded, as in θανμαν-^'ω, the primary form of θαυμαίνω, as the final consonant of the stem, or, as in καθαρ-]'ω, the primary form of καθαιρώ, as part of the suffix {καθ-αρό-ς) by means of which the noun-stem has been formed. 1. Derived Verbs m -νω. These have, beyond a doubt, their starting-point in stems in ν and vo. There are examples of both kinds in Sanskrit of the most transparent forma- tion, such as ukshan-ja-ti he acts like a bull (ukshari), krpan-ja-tihe acts miser- ably (krpana), the latter standing by the side of krpa-ja-ti and krpa-ja-te, just as χαμαίνω, and όνομαίνω have χειμάζω and ονομάζω without v. The Greek verbs of this kind fall into three divisions, two of which are very rich and one poor. Verbs in αινώ and ννω are numerous, those in €ΐνω rare. On phonetic grounds we will put ΐίνω next to αινώ. a) Verbs in -αι,νω. From stems in v, and with a preference for an α -sound, even where there is none in the noun-stem, there are formed verbs like άφραίνω (άφρον), pe- \αίνω (μ(λαν), μζλεδαίνω (μελεδον), πιαίνω (πιον), ποιμαίνω (ποιμςν) and nu- merous derivatives from neuter stems in μαν like ασθμαίνω, δςιμαίνω, κυμαίνω, πημαίνω, χ€ΐμαίνω. To these may be added a small number whose correspond- ing noun-stems seem to point to the suffix -na {vo): βασκαίνω, from βάσκανο-ς, 370 λιταίνω (by the side of λιτάζω and λιτανβύω) from λίτανο-ς (Aeschylus). From stems in -o and -a there come a large number, such as άγριαίνω, δυσκολαίνω, λ€ΐαίνω, λςυκαίνω, ξηραίνω — λυσσαίνω, όρμαίνω, όργαίνω, πικραίνω, τίρσαίνω (cp. the Goth, thaurs-na-ri) . The relation of these to those already . mentioned is just the same as that which subsists between νομ^ύω {νομευ) and πιστεύω (πιστό). It is possible that for some of the nouns from which these verbs come there actually were by-stems in vo. It is as conceivable that there should have been a *λ€υκ-ανο-ς, or *ξηρ-ανο-ς, as that there should be ορφανό-ς by the side of όρφό-ς = orbu-s (Princ. i. 367), or fr -ανό-ν αληθώς, σφόδρα by the side of ereos, eroy. "What is more surprising is, that verba in -αινώ come from not a few sterna 254 APPENDIX TO THE I-CLASS. in -«, and even from adjective-stems in -v, as θαμβαίνω, κυδαίνω, παθαίνω, δυσχεραίνω — γλυκαίνω. It is conceivable that further investigation might re- veal an independent origin for several of these nasals, but even then we should have to assume that the analogy, when once formed, gave the pattern for fresh formations. There can be no doubt that the simple verbs in -αινώ in which the ν is movable had an influence upon these formations (cp. above p. 216 f.). There are moreover derivative verbs in the Teutonic languages, such as the Goth, frauj-inon (dominari), A. S. glis-n-jan (coruscare), mod. Germ, ord-n-m set in order, which have an η in the derivative syllable, while, as Jac. Grimm D. Gr. ji. 174 says, it is only ( & very few of them which presuppose a substan- tive in -in. 1 b) Verbs in -ει,νω. There are only άλ^ίνω, ipctlp» of unknown origin, and φαείνω which must come from the stem αισχννω, κρατννω, μηκννω. It is possible that these may, as is common in the case of sigmatic stems, have modelled themselves on the analogy of o-stems. In one or two cases however other explanations are possible, as in that of καλ- λννω, which is perhaps directly connected with the Skt. kaljana-s, beautiful, a derivative from kal-ja^s = καλός, and with καλλονή beauty. We cannot have 372 too lively a sense, in dealing with these questions, of the possibility of such by- forms. 2. Derived Verbs in -ρω. In Sanskrit we find forms like the Vedic adhvar-ja-nt sacrificing from the st. adhvara sacrifice, while for other verbs of the kind in this language noun-stems with a derivative r can only be supposed to have existed, as is the case with ratha-r-jarti he rides in a carriage (ratha), sapa-r-ja-ti he reverences (rt. sap)'. The Greek formations in ρω are not very numerous. a) Verbs in -αίρω. The only verb from a stem ending in ρ is τεκμαίρω, more commonly in the middle τεκμαίρομαι (τεκμαρ, τεκμηριο-ν) . From stems in -po there are γεραίρω (γεραρό-ς), εναίρω {τα εναρα), καθαιρώ and μεγαίρω, the last being, as Buttmann (Lexil. i. 259) saw, undoubtedly derived from * μέγαρο (cp. μεγαρο-ν a room), the older form for μεγάλο, εχθαίρω clearly belongs to εχθρό-ς, whether we suppose an *εχθαρο-ς formed like γεραρό-ς, or set down the α to anaptyxis, as in the case of the e of the Lat. ag-e-r from *ag-ro-s, Goth, akr-s. ελεαίρω has no noun-stem with an p, and ελεφαίρομαι is an 'obscure word altogether. b) Verbs in -εφω. Ιμείρω from Ίμερος, οϊκτείρω 2 by the side of οίκτρό-ς like εχθαίρω beside εχθρό-ς. εβείρω Φ 347 is obscure. c) Verbs in -υρω. 373 There is no doubt about μαρτνρομαι from the stem μαρ-τνρ with the by-form μαρτνρο, κινυρομαι (by-form κιννρ'ιζω) from κιννρό-ς, μινίιρομαι (by-form μιννρίζω, Lat. minurrio) from μιννρό-ς, while the remaining verbs άθυρω, όλοφύρομαι, πλημίφω (also written πλημμυρώ, with the by-form πλημμνρέω) are of obscure origin. 3 Derived Verbs nr -λω. We may say just the same about these verbs as about those in -ρω. A few verbs in -λω which we might be tempted to expect to find under this head, such as αιολλω, δαιδάλλω we have given on p. 212 f. as primitive intensive formations. From noun-stems in -λο come άγγελλω, αϊκάλλω (αϊκαλος Hesych.), άτασθάλλω, καμπνλλω, κωτίλλω, ναντίλλομαι, ποικίλλω, στρογγνλλω (by the side of στρογγυλόω), 3 The form οικτΙρα; on an old- Attic inscription (Cauer Stud. viii. 253) has led Kirchhoif (Monats-Ber. 1872, p. 237) to take οικτίρω to be the real present, and this does seem to suit οίκτίρμων and the Lesb. οϊκτίρρω ; but it would be the 'only instance of a form of the kind. Since there can be no doubt as to its derivation from οΐκτρό -s we should have to suppose that in this one case an t and not an e developed itself between τ and ρ : *oIkt-i- P o-s, *oIkti P -jw, and I know of no analogy for this. Ought we not rather to regard the t in the same way as in rfaas, τίμη for which we often get the diphthong, i.e. as an early corruption of ει ? 256 APPENDIX TO THE I-CLASS. στωμύλλω. The remainder we shall not be too bold in referring to similar forma- tions. For instance, we have in Homer not only άσχάλλω but άσχάλάω, which brings us to *ά-σχα-λο, from the rt. σεχ, σχε (cp. σχο-λή). For ίνδάλλομαι we may conjecture a *fivd-aXo, formed in the same way as αϊθ-αλο -s, τροχ-αλό-ς. όφίλλω (όφελος) comes most likely from the same stem as the Latin opul-entu-s. Of just the same nature are the Teutonic weak verbs (discussed by Jac. Grimm D. Gr. ii 115) like the Goth, mikil-jan ( = μεγαίρειν magnificare), and the mod. Germ, gangeln to have in leading strings, streicheln to stroke, tandeln to trifle, dally. The diminutive meaning which is here and there discernible in both languages, has its origin in the noun-stems upon which these formations are based. In a certain sense then we may compare Latin verbs like moduldri, ustvlare and the *petvlare which may be deduced from petulans (cp. petid-cu-s) , only that they correspond more to στρογγυλόω than to στρογγύλλω. 4. Derived Verbs m -σσω (-ττω). These verbs fall into two groups, which we may describe, to keep the old nomenclature, as those with the guttural characteristic and those with the dental. There are, however, a number which show neither characteristic. 374 From actually occurring guttural stems we find — from stems in κ or κο: άνάσσω (pavaK and βανακτ Stud. iii. 115), εγγλύσσω (γλυκυ-ς), ελίσσω (ελιξ), θαάσσω (cp. θόωκο-ς), θωρησσω (θωρηξ), κηρύσσω (κήρυξ), μαλάσσω (μαλακός), πτώσσω (πτώξ), υλάσσω (beside υλακτεω, ϋλαζ), φαρμάσσω (φάρμακον), φοινίσσω (φοίνιξ), φυλάσσω (φύλαξ), χαράσσω (χάραξ), — from stems in χ or χο βησσω (βήξ), μειλίσσω (μειλιχος). From stems which show a dental there are far fewer: άγνώσσω (άγνωτος, late), αϊμάσσω (αίματ), βλίττω (μελιτ), Ίμάσσω (ίμαντ), κορυσσω (κορυθ), πινύσσω (πινυτό-ς), πυρεσσω (πυρετό-ς). The verb άηθεσσον stands quite alone, inasmuch as it has come from a sigmatic stem, and accordingly belongs to the list of formations in -as-ja-mi, which we found on p. 240 to be represented in Greek by €ΐω. We have seen above that σσ cannot come from y. For άλλάσσω therefore, and πατάσσω, and πτερύσσω we assume, instead of the actually occurring αλλαγή, ηλλάγην, πάταγος, and the gen. πτίρυγ-ος, older forms with a κ, for άλλάσσω therefore άλλακο, an expansion of άλλο formed in the same way as the Skt. anja-ha-s from the synonymous anja. The O. Ir. ailigim muto corresponds exactly to άλλάσσω. In this language, as I learn from Windisch, there are numerous formations of this kind which all come from adjective-stems with an original -ha. This class is represented in the Teutonic languages by the verbs with a g, discussed by Jac. Grimm D. Gr. ii. 306, some only of which, like the i?. S. ga-halg-jan = to hallow (Germ, heiligen) come from actually existing adjectives, while others, like the O. H. G. chriu- zigon go on a crusade, bimunig&n admonere, either never had, or had lost corresponding adjectives of this kind, as is especially the case with modern German verbs like eteinigen (stone), endigen (finish), petnigen (torment), be- echomgen (gloss over). These common German formations are calculated to throw much light on the formation of derivative verbs in general. Latin has verbs like claudicare, albicare, which exactly correspond to the Teutonic verbs in -igon and what in Greek would be verbs in -κόω. Some of the Greek verbs in σσω show clear signs of a guttural stem in other tenses or in cognate noun-forms, without enabling us to deduce a definite primary noun, as in the case of βδελύσσομαι (βδελυχρό-ς) , σπαράσσω (σπαράξαι) and others. Several of the less comprehensible of these we might perhaps suppose CH. XI. VERBS IN -σσω. 257 to have come from adjectives with a κ analogous to the Latin adjectives in -ax and -ox like vorax, velox. Greek itself is not altogether without such formations, 375 as is shown e.g. by vea£, used, by comic poets with a comical force as a substitute for vtavlas (cp. ν€οχ-μό-ς), and in the same way νέβραξ young animal, from νίβρο -s, Xet'a£ from Xuos smooth (Hesych.),^tXct£ from ψιλός bald. In this way we might suppose an adj. *eyprj-£ for έγρήσσω, and an adj. *άφάξ for άφάσσω feel and the like. For some of the verbs in -νσσω we may deduce forms in -νκο, which, as Ernst Kuhn (Ztschr. xx. 80) has shown, correspond to Sanskrit adjec- tives in -uka, as e.g. by the side of μορμολνσσομαι we find the subst. μορμολύκαον, and by that of the late μαρμαρνσσω, from which άμαρνσσω 5 seems to be not widely different, there is μαρμαρυγή with a γ softened from κ. The verbs in -ωσσω (-ωττω) are especially numerous. They denote a ten- dency, generally of an unhealthy nature, if not an actual defect. Cp. Lobeck Rhem. 248. Among others we may mention άμβλνώσσω, Ικτ^ρώσσω, Χιμώσσω, ν(ώσσ*ι (Hesych. καινίζει he has the reformation-sickness), ονειρωσσω, τνφλωσσω, νπνωσσω. The meaning is not equally prominent in all cases : e.g. νγρώσσων σφόγγος Aesch. Ag. 1329 means simply a moist sponge, and όνειρώσσω dream. These verbs show but scanty traces of gutturals in the other tenses and in derived words, e.g. ονείρωξις (Plato), οναρω-γμός (Aristot.), ωνείρωξε (Butt- mann i. 2 375). On this fact was based the conjecture I expressed in the Symbola philolog. Bonn. i. 281, that these verbs might possibly come from an earlier -ωτ-$ω, and so correspond to Latin verbs of a similar use like balbu-tio, caecu-tio. Guttural inflexion forms no insuperable obstacle to this, as it might very easily have crept in later on the analogy of the numerous presents in σσ (ττ) which come from guttural stems. Anyhow the ω is of the same character as that in τνφλω-τό-ς, νπνω-τι-κό -s. 3 αμαρνσσω seems to bear to μαρμαρύσσω the same relation as that of byc'ipoo to yapyaipv (p. 215). 258 THE Ε -CLASS AND THE 'RELATED FORMATIONS. ch. xn. 376 CHAPTER XII. THE E- CLASS AND THE RELATED FORMATIONS. It was indispensable that we should review the i-class in all its ramifica- tions and in its connexion with the derivative verbs before attempting to explain what looks a far simpler phenomenon, which is connected at all events with the formations desciibed in the last chapter. The alterna- tions, within a single verb, between a shorter stem and one expanded by an e-sound, is one of the commonest of irregularities. We met with a case of such a twofold formation as that of μάθε (μαθή-σομαι) beside μαθ and μαιβανο on p. 178. We may here distinguish between four modes in which this duality manifests itself. 1) The β -stem appears in the present, while some or all of the other tenses come from a shorter stem, e.g. boneco εδοξα. We place this kind first because it bears the most resemblance to the cases of present- expansion already discussed. 2) The shorter stem is peculiar to the present, while the expanded or e-steni appears in other tenses, e.g. μίνω μςμίνη-κα. 3) Both formations occur side by side in the present itself, e.g. αίδομαι αϊδβομαι. 4) Where there is a present expanded in some other way, and also the pure short verb-stem in some other tense, and a stem with no other expansion than an added e-sound occurs in a third. i-ycv -ό-μην yi-yvo -μαι yevrj -σομαι. It is obvious that these four processes, which, to use the old gram- marians' expression, coincide in showing the πλεονασμός of an ε, are very 377 similar, and we must see if we cannot explain them all on a single prin- ciple. The related languages yield us some though not many phenomena which may be compared with these processes. Inasmuch as we have compared the derivatives in -εω with the Sanskrit verbs in -ajdmi, we can adduce something analogous, for processes 1) and 2) at all events, from this language. As a rule the -aja which is the characteristic of derivative verbs pervades all tenses, but the reduplicated aorist forms an exception, being made straight from the root. Hence the aorist a-Jcu-Uura-t he stole bears to the present Uvrafa-mi just the same relation as that, for instance, of the Homeric εστνγε to the present στνγέω. In another case ch. xii, THE Ε -CLASS IN ΤΠΕ RELATED LANGUAGES. 259 we find a form of the 10th class standing side by side with one of the first, though there is no great difference of meaning between the two, e.g. dharsha-ja-mi venture on a thing, beside dhdrshd-mi, the former of which we compared on p. 229 to the Gk. Θαρσέω, while - the latter, in Greek letters, would stand as *θαρσω. Latin shows a still closer resemblance. This resemblance was re- marked by Lobeck, who, on Buttmann's Ausf. Gr. ii. 52 expresses himself in words which are of significance in the history of grammar, when he says : ' The interpolation of an ε, so natural in itself, favoured the purposes of metre, and the propensity to full-sounding forms ; both in the Latin vo- calic words like cieo, clueo, abnueo (conniveo), tueor, and in the Greek ΰαιέω, κυέω, λοέω,θαέομαι. 1 Here is a scholar, as a rule so strict, so disinclined to rationalise, ready at a moment's notice with an explanation which a moment's reflection will explode ! It was going rather too far, to credit Greek, and not only Greek but Latin, with having saddled themselves with a pack of idle vowels all for the sake of greater fulness of sound. Lobeck shows in this an utter want of the historical perception of language as a growth of popular use and convenience. He thought of the ' nominum impositores ' pretty much as learned poets, and was as ready to imagine them adding a sound or two on occasion to their formations, as he might be to imagine a painter adding a few touches to his picture. For all this we need find nothing to object to the par- ticular piece of comparison between the two languages which Lobeck here 378 indulged in. The comparison is a just one, and may even be widely extended. With the exception of the few verbs in -eo which, like neo, fleo, de-leo have come straight from the root, all Latin verbs of the ^-conjugation, as is well known, lose their e in the perfect and in the passive participle, so that ΰοκέω εΐοΐα finds an exact counterpart in doceo, doc-ui. Struve, who in his treatise ' iiber die lateinische Declination und Conjugation ' (1823) was probably the first to recognise this, states the matter thus on p. 188 : ' the larger number ' (of the verbs of the 2nd conjugation) ' borrowed the shorter terminations from the third conjugation.' He was so far right at least in this, that it is really a case of a return to more primitive stems. The phenomena of this kind in Latin may now be found collected in Corssen ii. 2 293 and in JSTeue ii. 2 422 ff. Well attested twofold forms, of just the same kind as έλκω ελκέω are do cieo, duo (κλνω) clueo, fervo ferveo, fulg ο (φλέγω) fulgeo, nuo nueo, olo oleo, scato scateo, tergo tergeo, tuor tueor. There is no instance in Latin of the anomaly given under 3), i.e. the appearance of an e-stem among the other tenses (μέι-ω μψέιη-κα). The appearance of the i-stem however, in peto petlvi petltu-s, quaero (quaeso) quaesl-vi quaesltu-s, is analogous. German too, with its interchange of the so-called 'strong' and ' weak ' forms, presents us with much which is analogous, the only difference being that here the 'weak' forms seem to be, much more decidedly than in Greek, the later growth which gradually forced the older forms into the background. The Slavo-Lithuanian languages are specially characterised by the formation of a so-called second verb-stem. In a large number of verbs this second stem is in Lithuanian characterised by e, and e.g. the Lith. future gelbe-siu bears to the present gelbu (I help), precisely the same relation as that borne by the Gk. ενίψσω to 260 THE Ε -CLASS AND THE RELATED FORMATIONS. ch. xii. εΰΰω. In Church-Slavonic £ is occasionally used to form the second stem, though both languages differ in so far from Greek in the matter, that in them the infinitive, which in Greek stands in the closest connexion with the remaining present-forms, is the main representative of the second stem. Since however this infinitive is of the same formation with the Greek nouns in -τι-ς, -σι-ς, the Ch.-Sl. vMeti know is actually related to viste ye know in exactly the same way that είΰη-σι-ς is to 'ίσ-τε. 379 The similarity of formation thus shown to exist between the related languages, however noteworthy, hardly entitles us to conclude that the Greeks received this peculiarity, like most of the characteristics of stem- formation which we have discussed hitherto, from an older pre-Hellenic period. To begin with, there is a lack of the necessary cases in which the like twofold formation is to be found in different languages in the same stem. We shall not be far wrong, if, instead of this, we regard the e-class, to keep this convenient name for all the formations which belong here, as of no very great antiquity, and suppose that it was not till after Greek became a separate language that its somewhat extensive development took place. It may even be questioned whether all these processes which in their effects are so similar are to be explained on the same principle throughout. For this reason I have thought it safer to give the e-class by itself under this head, and not to be in too great a hurry to incorporate it with the i-class, although the two are obviously related. At the same time I am, on the same account, reluctant to separate the various phenomena here in question, which are in principle at least of a similar kind. "We have first to deal with the first division of the class, i.e. with the presents characterised by the addition of an ε. With respect to forms like δυκέω, γαμέω, and ώθέω I conjectured in my Tempora and Modi pp. 92 and 95 ff, that their ε has come from a j, and that consequently this formation is identical with the fourth or /-class. The fact that the original palatal spirant has occasionally taken the form of an ε in Greek is beyond a doubt, and has been established at some length by me at Princ. ii. 239 ff. κενεό-ς (by the side of κεινό-ς and ^ro-c)=Skt. qunjd-s primary form kvanja-s), and Doric futures like πραίεϊς by the side of the 1st sing. πραζίω and the Skfc. dd-sjd-mi are the clearest proofs of this change of sound. I therefore held to this explanation of the presents in question in the ' Elucidations ' (Eng. Trans, p. 146) as being phonetically unassailable, though I added there that the principles of word-formation would not admit of its application to all the forms, and that on this account the question, whether an ε of this kind in the present tense was to be explained in this or another way, must often remain an open one. 380 In other words the question comes to this j did the -εω come from an older -jd-mi or from -ajd-mi 1 It being shown to be phonetically possible that -εω should have come from -jd-mi, we shall hold this origin the probable one in cases where the related languages, or Greek itself, point to forms with a simple j, and presuppose -ajd-mi on the other hand, where there are decided traces of a denominative formation. Unfor- tunately there are a number of forms about which both sources of in- formation yield us nothing at all, and we can never get beyond a pro- bability one way or the other. The three active presents κϋρέω (from Aeschylus onwards), £ϋρέω, and μαρτνρέω stand beside κύρω, κέρομαι, ζύρομαι (not till Plutarch), and ch. xii. ORIGIN OF THE ε. 261 μαρτνρομαι. It is natural to conjecture, that the primary forms κυρβω, tivpjw, and μαρτνρβω took two different directions, one resulting in an assimilation of the j (*κνρρω) which afterwards, when pp became p and the vowel was lengthened in compensation, disappeared entirely ; the other in a change of j to ε, in which form it survived for a longer time. It is true that forms like κϋρησαι (as early as Hesiod) and ζνράω should make us cautious, as they suggest the possibility that verbs in εω may have existed at an early period by the side of the more primitive forma- tions. — όρέοντο, at Β 398 ανστάντες & όρέοντο and Ψ 212 τοϊ ΰ' όρέοντο, is evidently an imperfect from the rt. op, the aorist of which is ώρτο, and is consequently a by-form of ορ-νυ-μαι of a different present-formation. It is very natural to compare it with or-i-untur, in which the i (cp. ortu-s) is just as movable. On the other hand there is the analogy of the stem στορε (εστόρεσα) beside στορ in στόρ-ννμι. — In the case of 6 -μΐ χ- ε ω by the side of the aor. ωμιζα we might appeal to the Lat. me-jo—meig-jo in sup- port of the view that the -εω comes from -ja-mi. Beside πατ-έυ-μαι Hesychius has (cp. above p. 221) preserved the form πάσσεται, which points to πατ-βε-ται (Princ. i. 335), while the corresponding Gothic fdd- ja-n, inasmuch as it is a weak verb, points to πατ-ε^ο-μαι. — γη-θ-έω beside γέ-γηθ-α and ή-θ-έω (sift) beside ήθ-μό-ς, both from short ar roots ex- panded by a Θ, might be compared with ίσ-θίω in support of the view that the ε and the t are representatives of a j. But by the side of γη-θ-έω stands the Lat. gaud-eo. — αίρέω (Princ. ii. 180), with the Cretan by-form αιλέω must undoubtedly be referred to a rt fap with its by-forms fa\ and /ελ. The diphthong presents some difficulties, but these are removed at 381 once by the supposition that the primary form was fap -βω, in which case we should have to assume the usual epenthesis to have taken place. We cannot be quite sure of this though, as there are no distinct parallels to the word in the related languages. In distinction to these there is another class of these forms which we shall with more or less certainty assume to have come from noun-stems, and consequently to contain the regular derivative εω. This is above all the case with χραισμέω, which undoubtedly presupposes a stem χραισμο for χρα-σι-μο (cp. χρήσιμο- ς), with πεκτέω (cp. above pp. 162 and 168), ριπτέω, which has been discussed on p. 162, and probably for <ριλ-έω in spite of the Homeric έ<ρίλατο, to which a present *φι\Χω would correspond, and this could as well be a denominative from φίλο-ς as ποικίλλω from ποιύλο-ς. It may be doubted whether δοκε'ω can find a sufficient support in the Ιόκο-ς which is only used by philosophers, or in the glossematic c?on/, although οόκιΐ-μο-ς presupposes a noun-stem just as much as πόρι-μο-ς or τρόφι-μο-ς. In the case of γαμέω however, as of Ιουπέω and κτυπεω it can hardly be doubted that they came from γάμο-ς, <)υϋ7Γ0-ς, κτνπο -c, as also ριγέω from piyoc. Ιατέομαι (cp. Ch. XXI. p. 369 marg.) bears precisely the same relation to αποτάσσεται, ΰάσομαι, δάσσασθαι that πατέομαι does to πάσσεται. The ε of κοέω with the aor. ε-κο-μεν (Princ. i. 186) is anyhow, as we saw at p. 240, of the same nature as that of the Lat. cav-eo, and that of λοέω beside λούω as that of the a in the Lat. lava-re. We must take care to notice, on the other hand, that in some presents in ε ω by the side of ω the vowels are by no means such as suit the supposition that the preceding stage was a noun, γεμέω can have nothing to do with γόμος, nor ελκέω with ολκή. Nor can the like assumption be made with any probability for μεΐέων by the side of 262 THE Ε -CLASS AND THE RELATED FORMATIONS. ch. xii. μέΰων. In general it can hardly be denied, that within certain limits a kind of analogy arose for by-forms in -εω by the side of forms in -ω, without the makers and users of such forms being conscious of definite noun-stems to correspond. We will now proceed to review the individual verbs of the first division. It is to be noted that the language showed an extraordinary vacillation between the shorter and the longer stem, and that the num- 382 ber of twofold forms is very great, and that they sometimes both occur in the same periods, and are even used by the same authors. 1)" Presents in -εω with Forms from a shorter Stem in other Tenses. 1) αίρέω common to all Greek from Homer onwards. The stem αφε also occurs with the ε short in ήρε-βη-ν (since Aeschylus), αιρε-θη- σομαι, αιρε-τό-ς (Hdt. Plato), with the long η in αίρησω (II.), άρ-αίρη-κα (Hdt.), ήρη-μαι (Aesch.), while a shorter stem underlies the aorist-forms ειλο-Γ, ελ-ί-σθαι and the late fut. ελοϋμαι. 1 lb) άλιτήμενος, clearly an Aeolising participle like φιλημένος, only 2 807 and Hes. Scut. 91, by the side of the forms ήλιτον, aXir εσθαι which are commoner in poets. 2) γαμέω Homeric both in the active and the middle. In this case the e-stem with a short vowel is only represented by γαμετεον (by the side of γαμητέον) unless we count the futures γαμώ, γαμοϋμαι ; that with the long vowel in earlier times, besides the above-mentioned γαμητέον, only by γεγάμη-κα (Hdt. Aristoph.), which is joined in later times by γαμήσομαι and the like. — On the other hand we have the aor. ε-γάμε (Pind.), εγημε, εγήματο from Homer onwards from the stem γαμ. 3) γεγωνεω almost exclusively poetical from Homer onwards (γε- γωνεΊν Μ 337) with the by-form γεγωνίσκω, mentioned on page 196, which is obviously derived from the perfect γέγωνα of present meaning, the only form from the shorter stem. The e-stem occurs also in γεγωνήσω (Eurip.), γεγωνήσαι (Aesch.). 4) γηθέω poetical from Homer onwards. The e-stem occurs also in γηθησω, γηβησε, by the side of γέγηβα (Pind. γέγάθε) from the shorter stem in the same poets, and in isolated instances in prose. In quite late authors (Sext. Empir., Anth., Qu. Smyrn.) there is also γήθομαι. 5) ΰοκέω in general use from Homer onwards, with ΰόκησε (υ 93). Post-Homeric but good Attic are the forms Ιόζω, έ'£ο£«, Ιέΰοκται, εΖόχθην from the shorter stem, alternating in poets and Herodotus with such forms as ΰοκήσω (Aesch. Pr. 386), εΰόκησε (Pind.), ΰεΰύκηιπ (Aesch. Eumen. 309), δοκηθείς (Eurip. Bacon. 1390). 383 6) (y)loviriu). The present-stem occurs only at Eurip. Ale. 104 and in late poets. The aorist ε(γ)ΰούπη-σα is common in Homer, with ΰιΰουπώς (ΰεΰονπότος Οιοιπόΰαο Ψ 679) from the shorter stem. At Anth. vii 637 there is κατέΰουπε. 7) τ/θεω sift. The present occurs in Plato, and there is the aor. $σα from the st. ήθ, which is quoted in Steph. Thes. from Hippocrates, and the common substantive ι)θ-μό-ς. Still ηγημένος and ήθήσατο occur from Plato onwards. 8) θηλέω a Homeric present for the ordinary Greek θάλλω mentioned 1 itt -ανρίω has been mentioned on p. 195. ch. xii. VERBS WITH E-EORMS OUTSIDE THE PRESENT ONLY. 263 on p. 211, with the perf. τέθηλα (Pincl. τέθάλε), θηλησει (A 236), θάλησα (Pind.). 9) κεντάω makes all its forms (κεντήσω, εκέντησα, κεκεντημαι etc.) regularly from the st. κεντε. The shorter stem is only represented by the άπαξ λεγόμενον κέν-σαι Ψ 337 and by noun-forms like κόντ-ο-ς, κέν- τωρ (for κεντ-τωρ), κεν-τρο-ν. 10) κτνπίω. κτνπέει Ν 140 and also in Attic writers, aor, κτύπε Ρ 595, εκτυπεν Soph. Ο. C. 1456 by the side of κτύπησε ib. 1606. 11) όμϊχέω Hes. Opp. 727, with ώμιξεν Hipponax fr. 55 B. 3 and ομιχμα. Cp. above p. 261. 12) όρέοιτο only in the two passages in Homer mentioned on p. 26 L 13) πατ~έο~μαι. The present is first found in Herodotus, and the aorist επασάμην occurs in Homer, Herodotus, and Attic poets r πάσαντο A 464, έπασσάμεθ* ι 87, πάσασβαι Aristoph. Pax 1281 by the side of the plupf. πεπάσμην il 642-and άπαστος δ 788. Cp. above p. 261. 14) The inrperat. πίει, common on cups by the side of ε-πω-ν, πΊ-θι, πί-νω cp. Roscher Stud. iv. 194. 15) ρϊγεω. The present is only found in Hipponax and Pindar. In Homer (e.g. Π 119) and Sophocles occur ριγήσω, ερρίγησα, ρίγησεν, while the shorter stem is represented by the perf. Ipply Η 114, conj. ερρίγησί Γ 353 ; ερρίγει \p 216, may be either imperf. or pluperf. 16) στυγεω. στνγέουσι Η 112, also in Hdt. and Attic poets, while from the stem στνγ are formed the aorists εστνγον κ 113> Ρ 694 and εστυΐ,α (στύζαιμι λ 502), the perf. ϊστυγμαι (Hesych.), the passive future στνγήσεται. Soph. O. R. 672, and noun-forms like στνγος, στυγ-νό-ς. Still, after Homer, there occur also άπεστύγησα (Soph. O. C. 692)3 84 άπεστνγήκασι (Hdt. ii. 47), στνγηθίν (Aesch. Sept. 692) and the like. 17) ψιλέω. The regular forms of the e-conjugation are too common to need confirmation. From the shorter stem are formed only εφίλατο Ε 61, φ{λατο Υ 304, Imperat. ψΤλαι Ε 117, φίλωνται hymn, in Cerer. 117. Forms of the kind occur in imitators of the old Epos. The Byzantine πέψιλμαι. hardly comes into consideration. 18) χραισμέω. As we ought perhaps to regard χραισμείν in Homer (e.g. A 589) as an aorist, we must go to Nicancler (Ther. 914 χραισμεϊ) for the present. Along with the regular e-forms χραισμησέμεν (Φ 316), χραίσμησεν (Π 837) we find the aorist εχραισμε & 66, γραϊσμε Η 144, ■χραίσμη Ο 32 from the shorter stem. 19) ώθεω, from Homer onwards (γ 295) very frequent in both active and middle. From the shorter stem come άπώσε ι (A 97), the Att. ωσω, ώσομαι, εωσε (Π 410), ώσε (Ε 19), ώσαντο (Π 592), the Attic εωσα, εωσά- μην, εωσμαι, εωσθην, ώσθησομαι, ώστός, ώστϊος, while Attic poets also have ώθησω (ίξωθήσομεν Soph. Aj, 1248), and in late prose we find ώθησα. 2) Presents without an ε by the side of other Forms with ε OR η. Out of the considerable number of the verbs which belong to this division, we will reserve such as show in the present an expansion of another kind, which does not appear in the e-forms as well, e.g. ρεω ρνησομαι, μανβάιω μαθήσομαι, for the fourth division. The forms with which we have here to deal are of the most various kinds. The simplest 264 THE Ε -CLASS AND THE RELATED FORMATIONS. ch. xii. case is that in which the present-forms have a simple stem and the other tenses a stem expanded by the addition of an ε e.g. μέλει μελήσει, νέμω νενέμηκα, εβέλω έθελησω. Here and there comparison can be made with verbs in the related languages. With μεμένηκα we may compare the Lat. mane-o. The forms εΰ-ήδε-ται (Chap XVI. p. 384 marg.), ίΰ-ηδυ-κα, ηΐέσ-θη-ν belonging to εΐω may be compared with the Skt. ada-ja-mi, the causative of the rt. ad, with the meaning ' give food to.' In other cases the present shows a less primitive character. It is possible that εΰδω, κήΰω, ο'ιχομαι are petrified presents of the lengthening class ; anyhow their diphthongs extend into the tenses with an e-stem : ενΰησω, κηΰησω, οϊχησομαι. One of these presents can be connected with an 385 analogous formation in the related languages : from the intensified stem fnfo is formed the fat. ειίησω, and the stem ϊειϊε also forms the basis of the moods of the perf. olla of present meaning (ειΰέω, είΰείην) and is there restricted to the meaning know. The Skt. vedaja-mi is not merely causative, but also means know, the Goth, fair-veit-jan preserves the physical meaning see, and the Ch.-Sl. νΗέ -ti which has been referred to on p. 260, agrees in form and meaning with the Greek ειοέναι. ειίησω is therefore properly the future of an *ειΰέω which corresponds to the above-mentioned vedaja-mi. From a present of the ί -class we get τνπτήσω, from what we conjectured to be a present of the η -class βονλή- σομαι, and from a present of the inchoative class βοσκήσω. The i-class in its various ramifications furnishes a larger contingent : κλαίω κλαιησω, μέλλω μελλησω, οφείλω όψειλήσω, χαίρω χαιρήσω, ϊζομαι καθιζήσομαι, μνζω μνζήσω, οζω όζήσω*. It is obvious that all these β -forms have not come from the verb-stem but from its expanded form, the present-stem. So far they bear the stamp of a not very primitive character, and have in many cases clearly become extended gradually as the language developed by a convenient imita- tion of older types. Two causes were clearly at work in the multiplication of these forms. First, phonetic necessity or convenience (cp. Lobeck on Buttmann ii. 44). From stems like άλ* ζ, αί/£, άχθ, εψ, άλβ there was no possibility of making a form which should be recognisable as a future or an aorist without the help of this convenient ε. From forms like alh, κηΰ, ττετ-, while it is conceivable that there should have been made forms like *ά'ίσομαι, *εκησα, *πέσομαι, it is clear that they would not have been so recognisable as αΐΰέσομαι, έκήΰησα, πετήσομαι. The other cause was the effort to distinguish one form from some other. It was only by the help of the ε added to the stem that οίήσομαι (οίομαι) could be distinguished from οίσομαι (φέρω), ερρησω (έρρω) from ερώ (εΐ/οω), Ιέησω (Ιέω lack) from ΰήσω (ΰέω bind), άχθέσομαι from άζυμαι and άχθήσομαι, έμαχεαάμην from εμαζάμην (μάσσω), παιησω (παίω) from παίσω (παίζω). Pott in his Et. Forsch. ii. 2 957 has collected instances of this kind. Of the variation in the quantity of this e-sound it will be time to treat when we come to the tenses in question. All we need notice here is that the short ε is very rare. The only forms among those which are to be given imme- 386 diately which have the vowel short throughout are αχβέσομαι, άχθεσβήναι. A vacillation between a short and long vowel is to be seen in άκηχε μένος and άκαχημένος, αλθομαι άλθι)σομηι άλβεσθή^αι, μένω μενετός μεμέι-ηκα, μάχομαι μαχησατο, μαχέσασθαι and a few others. Where the vowel is short it is tempting to suppose that it was an auxiliary vowel, and if so, to compare the Gk. ε with the % which meets us sporadically to so large ch. xii. VERBS WITH E-EORMS OUTSIDE THE PRESENT ONLY. 265 an extent in the tense-formation in Sanskrit e.g. a-ved-i-sha-m, the aorist of the rt. vid, tar-i-shjd-mi the future of the rt. tar. But this i itself needs to be examined more closely and ought scarcely to be so decidedly- pronounced to be a purely phonetic element as is usually done. In the case of the Greek e-sound, prevailingly long as it is, it seems to me now far more probable that the whole phenomenon is due to a confusion of the derivative with the primitive formation, and that the variation in quantity is connected with the variation between forms in εω and εζω, ιζω discussed on p. 241 ff. We will now give without further subdivision the individual forms which fall under this head. 1) άλέΕω. In Homer we find active and middle forms like άλεξέμεν (Γ 9), άλεζωμεσθα (Λ 348) by the side of the fut. άλείήσεις (I 251), the aor. ΙιλεΙήσειε (γ 346), and Hdt. has άλείήσομαι. άλέίομαι as a future (Soph. Ο. K. 171) and άλέξασθαι (() 565) belong to the stem άλκ with an unconsciously developed ε (cp. άλ-αλκο-ν). 2) άλθομαι. άλθετο Ε 417, άπαλθήσασβαι θ 405. Cp. άλθήσκω, άλθίσκω ρ. 194. 3) ανζω (Horn, «εξω only in the present) by the side of αυξάνω, fut. αυίήσω, αυίήσομαι, aor. ηυζησα, pf. ηυίηκα, ηυΖημαι, pass. aor. ηυξήθην, verbal adj. αυζητό -c, all well attested in Herodotus and Attic writers. In late prose (Plutarch, Dio Cassius) there occur isolated and hence questionable traces of a present αυζέω, and αυΐοΰ μένος is found in the iambic inscription from Megara C. I. 1066. The e-forms correspond to the Skt. causative vaksha-jd-mi (let grow). Cp. p. 181. 4) άχθομαι. The present-forms are the only ones in use in Homer, but in Attic times we find (ουκ άχβέσει Aristoph. Nub. 1441) the fut. άχβέσομαι and the passive forms ήχβέσθην (Thuc.) and άχθεσθήσομαι 387 (Xen. Plato), άχθήσας' γομώσας (Hesych.) shows a variation in meaning as well. Veitch notices isolated traces of άχθέω in late authors. 5) βόσκω in Homer (O 548) by the side of βοσκήσεις p 559. Other forms with η are rare and mostly late, βοσκητέον Aristoph. Av. 1359. 6) βονλομαι. Homer has only present-forms (among which are βόλεσθε, εβόλοιτο) and the pf. προβέβονλα. These are gradually joined by one or another e-form. βουλήσομαι (first in the hymn, in Apoll. Pyth. 86), εβυνλήθηΐ' (from Sophocles onwards), βεβοΰ\ημα>, βονλητός, all in use with Attic writers, later βονληθήσομαι. Cp. also βούλησις, βονλημα. 7) γράφω only belongs here in virtue of the perf. γεγράφηκα which the Atticiste reject, but which Lobeck (ad Phryn. p. 764) defends. Kiihner (s. v.) points out that the best M.SS. of Xenophon Anab. vii. 8, 1 have γεγραφηκότος. 8) ίί'ω. On p. 210 we met with the present form faiia for *fajj*t 4 We presupposed however the existence of Είΰω (Homeric Ζεύομαι, εδενετο A 602) and *^ε/ω as early by-forms of another present-formation. Homer has also the aor. εΐεϋησεν ι 540 by the side of εοησεν (εμε'ιο On the whole though there are fully 100 verbs which belong to this class. 273 APPENDIX TO THE E-CLASS. It is far less often that we find other movable vowels, but to a limited extent there occur stems in α which alternate, just as the e-stems do, with shorter stems. These are precisely analogous to the far more numerous Latin verbs of the «-conjugation whose a, as in doma-re, dom-ui, domi-tu-m, does not extend beyond the present. The following are probably the only Greek verbs of the kind. ■γοάω, common from Homer onwards, with the isolated aorist eyoov from the rt. yo(f) Ζ 500. μηκάομαι, the present only occurs in grammarians : Bekk. Anecd. p. 33, 8. Byform μηκάζω Nic. Alex. 214, which the scholiast interprets μηκαται ως πφόβα- tov. Pf. μψηκώς Κ 362, μεμακυΊαι Δ 435, plpf. μψηκον ι 439, aor. μάκών II 469. μνκάομαι, μνκωμεναι κ 413, with άμφιμψνκε κ 227, μψνκ* lies. Opp. 508, aor. μνκβ Υ 260. In Attic writers there are found only present-forms, which are joined in late poets by μυκήσω and μνκησας. As a parallel to the numerous verbs with both eo and ω in the present we may notice μνζάω (suck) by the side of μνζω, especially as μνζίω occurs in Hip- pocrates. Veitch, it it true, gives no earlier authority than Aelian H. A. iii. 39 for the α -formatioh. It is impossible to say whether Homer's μνζψϊας (cp. above p. 269) comes from the e- or the α -stem. We may further notice here a few verbs which take an α in the other tenses when their present is expanded in a different way. We may say that επέράσα : περνη-μι. : : εστόρεσα : στόρννμι, while there is no more a *περάω or *π*ράζω as a present in this meaning than there is a *στορίω. It is probable that περάαν 398 Φ 454 is the fut. to the aor. πέρασαν (ο 428), επερασσεν Φ 40. The perf. pass, partic. πεπερημένος Φ 58 bears to it the same relation as that of βφίηκς to έβιασάμην and other examples mentioned on p. 235 of this alternate adoption of the analogy of verbs in αω and of those in αζω. The traces of a movable ο are rarer still, but not altogether wanting. On p. 267 we met with the Homeric perf. ωχω-κα to ο'ίχομαι, in which the ω holds just the same position as that of the η in οϊχη-μαι (Hdt.). — There are besides a few Doric forms, the perfects W -ω-κα or e -νίθω-κα (Ahrens Dor. 340) from the_ stem σ^θ (cp. εθίζω), which occurs in the ordinary εΐωθα without the added vowel, εθωκα is only distinguished from it by the loss of the reduplication. 2 The stems άλω, άναλω, and άμβλω, which appear in ίαΚων, άλώσομαι, 4ά\ωκα, αναλώσω, ή'μβλωσε (cp. pp. 133 and 195), bear the same relation to the shorter stems of the presents άλίσκω, άμβλίσκω that the stem evpe (p. 271) does to 2 άφέωκα, αφίωνται, ανεώσθαι (tabb. Heracl.) are similar in appearance only, for in them the ω is, as Herodian (ii. 236) saw, the representative of the η which appears in ϊ-ψμι. The same may be said of πίπτωκα, εδήδοκα. 274 APPENDIX TO THE E-CLASS. ευρίσκω, and the occurrence of the presents άραλόω and άμβλόω as well is to be regarded in the same way as that of ytyvwt* by the side of γεγωνίσκω. Lastly the ο in ωμόσσαμςν Υ 813, ομοσ^ν Τ 113, which was found in all periods in this aorist, as also in the perfect όμώμοκα, ομώμοται and ομώμοσται, ομωμοσμίνος, the aor. pass, ώμόθη or ωμόσθη, and άνωμοτος, can hardly be of a different character. For ωμοσα : ομ-νν-μι '. '. €-στόρ€-σα : στόρ-νν-μι '. I enepa -σα Ζ πέρ-νψμι. The short vowel is here due to the same analogy which'we discussed above with reference to the α. ωμοσα is in a way the aorist to a lost *6μόζω (cp. p. 239 αρμόζω, δεσπόζω). All these facts go, I think, to prove that the duality of stem, which we have been discussing, is due substantially to the mutual interchange between more primitive and derivative verbs, and this fact justifies us in investigating the phenomena last discussed in an appendix to the e-class. THEMATIC AORISTS, 275 CHAPTEE XIII. THEMATIC AORISTS. We have previously had occasion to notice that there are aorist forms Vol. not distinguished as such by any definite formative elements. In their II. origin they are exactly like corresponding formations from the present p. 1 stem, and they are only distinguished by not having any unexpanded indicative present formed from the same stem. We saw for instance (p. 125) that in formation the aorist ϊ-βη-ν and the Homeric βά-την do not at all differ from ε-φη-ν and φά-την. There is exactly the same rela- tion between thematic forms like ε-τραφο-ν and ε-γραφο-ν, ε-τεμε and έ-νεμε, τέκοι and πΧέκοι, γενέ-σθαι and πένε-σθαι. The forms εγραφον, ένεμε, πλέκοι, πενεσΰαι are presents or imperfects solely because they are accompanied by the presents indicative γράφω, νέμω, πλέκω, πένομαι. The others are aorists, because this is not the case. It follows from this fact, which is of the highest importance in rela- tion to the structure of the verb, and which has found the fullest confirmation in Delbriick's Researches on the Verb in the Veda (p. 16), . that the distinction between aorist and present or imperfect forms is by no means invariably sharp and unmistakeable. We have the following cases : 1) The present indicative, which by its occurrence makes the cor- responding past tense into an imperfect, is wanting in certain dialects and at certain periods of the Greek language, but actually occurs else- where. Thus in ψ 90 Έτραφέ τ ivbvKtois και σον θεράποντ 1 ονόμηνεν. 2 ετραφε is certainly an aorist; for in X 421 we have the unmistakeable imperfect έτρεφε. But in the Doric dialect the present is τράφω (ep. Pind. Isthm. viii. 40 Ίωλκοΰ τράφειν πεδίον), and hence in Theocr. iii. 16 Βρνμώ re viv Ζτραφε μάτηρ the very form, which in Homer passes for an aorist, is taken as an imperfect. In Herodotus the ind. pres. τράπω is common : (e.g. i. 63 τράπουσι)) hence ετράποντο (e.g. i. 80) can hardly be anything but an imperfect, while the same form in Attic writers, occurring by the side of ετρέποντο, is undoubtedly to be considered as an aorist. The same form can therefore at different periods and in different dialects have a different force. And it is not always easy to determine what this is. For instance, does the exclusively Homeric βλάβεται justify us in regarding the form εβλαβε in Quintus Smyrnaeus (v. 309) as an imperfect ί The context seems to be rather in favour of the aorist. Can we regard the Homeric ήλΰανε, to which we shall return, as an aorist, on the ground that in Aeschylus we find a present άλΰαίνω 1 τ 2 276 . THEMATIC AORISTS. ch. xiii. 2) We have, as a rule, a serviceable criterion in the different ac- • centuation of the infinitive and the participle. But frequently the tradition is untrustworthy in this respect. Thus the accentuation άγέρεσθαι, εγρεσθαι (e.g. ν 124) is supported by good authorities. Herodian (i. 452, 26) assumes a present εγρω ; cp. ii. 254, 783. On the other hand κιών κιεϊν is the common accentuation, although in Aeschylus- (Choeph. 680) we find the unmistakeable present form κίεις (cp. p. 146). εχθεσθαι is generally regarded as a present, άπήχθετο as an aorist to απεχθάνομαι. For the accentuation α'ίσθεσθαι, e.g. Thuc. v. 2Q T it may be urged that the present form α'ίσθονται has strong support in Isocr. 3, 5, and is recognised by Herodian i. 441, 2. Yet no one would deny that ησθετο by the side of ήσθάνετο is an aorist. We must therefore admit that the old grammarians are not consistent, and often follow for the accentuation of forms which were not living to them, 3 purely external analogies, borrowed from the spelling. In fact it almost seems as if the usage of language was itself not free from confusion. A man would be much deceived if he fancied that the meaning was in every case sufficient to decide the. question. For the difference between present and aorist-stem is often so slight that in many places both are possible. Under these circumstances we cannot be at all surprised that late writers not seldom constructed present forms after old forms which we have good reason to treat as aorists : e.g. Apollonius Rhodius iii, 895 άγέρονται after the Homeric aor. άγέροντο, by the side of the impf. άγείροντο, Dio Chrysostom 'όφλω — recognised also by Herodian i. 448 — after the Attic aorist ώφλον, Apollonius Rhodius κέκλεται (cp. Herodian ' u. s.) after the Homeric εκέκλετο; and others of the kind, which we shall severally point out. It would be an anachronism to argue from such stragglers to the present character of much older forms. 3) Greek aorists sometimes correspond letter for letter to Sanskrit imperfects, e.g. e-Spa/ie m Skt. a-drama-t €-γω, *γένομαι. 4) Participles which have become substantives may come just as well from the stems of thematic aorists, or from stems like them, as from present stems : τίνων, Ιράκων, Έυϊράμων, in which it is not easy to detect any other relation of time than in μέΐων, Μελπομένη, "Παρμένων, while Ίίνελθων can be taken as Benvenuto. Buttmann first clearly determined the essential characteristics of this aorist formation. From the discussions in the Ausf. Gr. i. 2 399 if., which are still well worth reading, we can see how the thoughtlessness of Buttmann's predecessors had gone so far as actually to assume an aorist *t.(f>i\ov from derivative verbs like φιλέω, and generally to give this formation as one to be expected in every verb. In the face of such 4 absurdities, Buttmann was quite right in laying down the rule that only 4 primitivft ' or verbs which are to be regarded as equivalent to ' primi- tiva/ admit of a 2 aor. act. But the addition ' which are to be regarded as equivalent to primitiva ' is quite sufficient to show that the line cannot «η. χιπ. THE ORIGIN AND CHARACTER OF THEMATIC AORISTS. 277 be very sharply drawn. Verbs like άγγέλλω, έναίμω, οφείλω with their disyllabic stems must certainly be regarded as denominatives ; and yet we cannot deny to them aorists of this form. Even the invented Ηφιλον is not invented against all analogy, inasmuch as it might be supported by the actually occurring 'έ-χραισμο-ν, which certainly goes back to the adjective stem * χααισιμο for *χρασιμο. But it is the most important task of the grammarian to determine the prevailing character of a forma- tion, and to distinguish rigorously what is isolated from what properly forms the rule. And so far Buttmann was quite right in his remark, which was protected from misconception by the addition of ' or.' He was also before his time in comparing (p. 404) the twofold character of the Greek aorist with the double formation of the German preterite, comparing e.g. τρέπω ετραπον with gebe gab, and βλέπω ε/3λεψα with lebe lebte, and in calling attention to the changing nature of the vocalism. No one will certainly reproach him with not having already recognised the German preterite as an original perfect, and the Ablaut as distinct from the Umlaut. Only, with all his acuteness, he was misled by Hebrew analogies, and went wrong in regarding (at p. 368) the 3 sing, of the aorist, without its augment, and with the elision of the thematic vowel, e.g. λά/3', βάλ\ as ' the simplest form of the verb,' and adding the assertion that the Greek language * started with the aorist : ' he went still further astray in assuming for forms without the augment an original preterite force. But it is only by degrees that even comparative philology, with the richer means at her command, has overcome many similar perverse views, and above all has learnt to distinguish between forms that are really primitive, and such as have merely the appearance of primitive forms, owing to phonetic losses. Thematic aorists, or more strictly speaking shorter preterites pro- vided with a thematic vowel, and distinguished from longer preterites, 5 called imperfects, by the absence of those additions, which we have called present expansions, occur not only in Greek, but also in Sanskrit, Zend, and Slavonic. But the traces of snorter moods, infinitives and participles are extremely scanty except in Greek. As a system of connected forms the thematic aorist is properly found in Greek alone, and it is doubtless a result of this fact that it is only in Greek that we can prove a clear distinction of meaning between the shorter forms, and the longer ones which correspond to them. Eor to mark or to preserve any distinction, language needs a somewhat large number of cases, in which it can be brought into exercise. The instinct of the Greek language in the historic time had a safe criterion between forms of the aorist and forms of the present stem in the accent of the infinitive and participle active and of the infinitive middle, by which τεκείν, κραγών, γενέσθαι were sharply separated from κρέκειν, άγων, πένεσθαι. But probably this distinction had itself been imprinted at an earlier period only under the influence of the intellectual tendency towards differentiation. We shall come back afterwards to this question. Under this head we have in Sanskrit that formation of the aorist, which Bopp gives as the 6th, Benfey as the 2nd, Max Muller as the 1st form of the 2 aorist ; but also the 7th of the 2 aorist according to Bopp, the 3rd according to Benfey, the 2nd according to Max Muller, that which includes reduplicated forms. But in Sanskrit grammar the redu- plicated forms are by no means sharply distinguished from those which 278 THEMATIC AOEISTS. ch. xiii. are not reduplicated. Excluding for the present all reduplicating forma- tions, we will content ourselves with quoting here such forms as come from the same stems in Sanskrit and Greek. They are the following : ά -darqa-m = ΐ-ΰρακο-ν Impf. ά -drqa-m = t-bepno-v a-bud/ia-nta = e -πύθο-ντο „ a-bodha-nta = έ-πενθο-ντο a-riUa-t = €-Xt7T6 (according to Benfey Ausf. Gr. p. 394) a-mda-t = e-/iSe (with pres. vtnda-mi) a-gama-t = e -κα/χβ (with pres. qam-na-mi, cp. p. 171) In Zend the only trace of this formation seems to be the 3 sing, bva-t (Schleicher Comp. 743, Justi Handbuch p. 400). bva-t is related to the 6 3 plur. bu -οι, just as an aoristic *'έψνε, quite conceivable by the side of the original present φνίω on the analogy of the Homeric 3 sing, άμ-πννε, is to the actually existing ε-φνν or φύν. The Church-Slavonic aorists of this formation have been discussed by Schleicher Comp. 745 Ksl. 358 ff. With the exception of nesu (from *naka-m)> I bore, which agrees at any rate in root with ήνεγκο-ν, I do not find any Slavonic aorist, which may be placed by the side of a Greek aorist of like formation. Traces of a Latin aorist of this kind I have attempted to establish in my paper ' de aoristi latini reliquiis' (reprinted in Stud. v. 431). Of the forms which appear to me to fall under this category, there are only two which agree in root with any Greek aorist, i.e. taga-on (attiga-m) by the side of the present tanga-m, with the Homeric reduplicated τέταγο-ν and genitur geni with γενέσθαι. A third, the petrified participle parentes corresponds in form to the Greek παρόντες. But the meaning has become modified in a special way in both languages, and we cannot place by the side of πορόντες any present formation comparable with parientes. - As the number of Greek aorists of this formation, excluding those which are reduplicated, is much above a hundred (as we shall see immediately), we must admit that the points of connexion with other languages are but scanty : still they are quite sufficient to prove the existence of this manner of formation in the original stock of the Indo-Germanic verb. We go on now to the special characteristics of the Greek thematic aorist. All the forms, which belong here, naturally divide themselves into two classes : i.e. I. Aorists without reduplication, and II. Aorists with reduplication. The first class is by far the more numerous ; the second class is found largely only in the Homeric dialect, and in later times it is extremely limited. I. Aorists without Reduplication. The first thing which we have to consider here is the relations of the 7 vowels. In the present formations, which are only characterised by the thematic vowel, it appeared (p. 145) that by far the most common vowel is c, while α is especially rare. Here on the other hand α is the most usual of all stem- vowels. On p. 150 we saw that presents with a short t and ν are rare. Aorists with these vowels are on the contrary tolerably numerous. The one observation evidently gives a welcome confirmation of the other, inasmuch as the distinction of the two formations was evidently intended by language. Differentiation by means of vowels is ch. χιπ. THE STEM-VOWELS. 279 one of the favourite phenomena in the structure of the verb. In forma- tions so simple as these the differentiation can only affect the vowel of the stem. Of 116 aorists of this class 54 have an α in the stem-syllable, e.g. ahlv, άμαρτεΊν, άρέσθαι, βαΧεϊν, ΰακε~ιν, Ζραμε'ιν, θανεΊν, λαβείν, μαθεϊν, χαΰεΊν. It is only by means of this difference of vowel that the verb τρέπω gets a simple aorist : ηοαττεΐ»'. ε appears only in 17 instances, e.g. γενέσθαι, ελεϋν, έρέσθαι, όψελείν, πεσεΊν, τεκεΊν. We notice a dialectic variation between α and ε in ταμεΊν by the side of the later τεμεΊν, in βαλεΊν — Arcadian εζελεν with present ζέλλω or Ζέλλω (έςΰέλλοντες In- scription of Tegea 1. 51 [Cauer Delectus Inscr. Gr. p. 138], Michaelis [and Curtius] in Fleckeisen's Jahrb. 1861, p. 587)— and in the Locrian άρέσται (Inscription of Naupactus 1. 32) supposing this to be for ελέσθαι. By the side of the Attic ήμαρτον we have the Homeric ήμβροτον with o. Evidently the oldest of the three hard vowels has an ' elective affinity ' with this old aorist formation, just as the later ε has with the present. We find ο only in 8 cases ί θορεϊν, μολεϊν, όλέσθαι, πορεϊν etc., ι on the other hand in 19, e.g. άλιτεϊν, Ζικε~ιν, θιγεϊν, 1%ε~ιν, λιπείν, υ in 13, e.g. κρνβείν, πνθέσθαι, τνχεΊν, φυγείν. Five aorists with diphthongal stems are quite isolated instances, i.e. αίσθεσθαι, εχαυρείν, ενρεΊν, ΰονπέϊν, χραισμε'ιν. In 9 forms the vocalism is irregular. In three of them there can hardly be a doubt that syncope has taken place, i.e. the participle άγρό- μενοι (e.g. Υ 166) by the side of άγέροντο (e.g. Σ 245) and άγερέσθαι (β 385), and εγρετο, with the imper. εγρεο, the infin. έγρέσθαι and the part, έγρόμενος (κ 50) ; and also ώψλον, which differs indeed in usage from ώψελον, but which is certainly identical with it in origin. The same phonetic process may be noticed in the Skt. a-hsha-n, which occurs by the side of a-ghasa-n as a 3 plur. from rt. ghas eat, devour, ε-γεν-το, which has been wrongly put here, found its proper place on p. 130 by the side of the active εγαν. Metathesis meets us very plainly in ε-ορακο-ν by the side of Ζέρκομαι (as 1 sing, κ 197, άνέΖρακεν Ξ 436), δρακέΐν Aesch. Ag. 602, ΰρακών Eurip. Here. F. 951, with the Skt. a-darga-m and in the quite analogous ε-πραθο-ν beside πέρβ-ω (3 pi. επραθον 2 454, ΰιαπραθέειν Η 32). εΰρακον has only become an aorist by means of this metathesis and the retention of the ο as compared with the ε of the present stem, while in Sanskrit the fuller ddarga^m is regarded as an aorist in contrast with the weakened ddrga-m. We find metathesis appearing as a phonetic affection of no importance for the tense-system in ε-Ιαρθο-ν (post-Homeric) by the side of ε-ΰραθο-ν, ΰραθέειν(υ 143, % 163). In the case of 4 forms it may be doubted at first sight whether they are to be explained by syncope or metathesis, i.e. ε-σχο-ν, ha -σπο-ν, ε-σπυ-ν and ε-πτό-μην. We might be inclined from a fundamental form ε-σεχο-ν (=Skt. a-saha-m) to derive on the one hand by the loss of the σ ε-εχον είχον, that is, the imperfect ; and on the other by the rejection of the ε ε-σχο-ν, the aorist, so that the distinction of meaning here, as in the cases of metathesis, would only have been brought about by what may be called casual phonetic affections. But this is opposed by the imper. σχέ-ς, which shows by its formation, corresponding as this does to that of the conjugation in -μι, that the ε was regarded as a stem- vowel, and that hence σχε was regarded as the root. In this sense we discussed the form on p. 132 and shortly before the completely analogous ενί-σπε-ς. We 280 . THEMATIC AOKISTS. ch. xih. decide therefore for metathesis, which is also confirmed by ε-σχε-θο-ν, σχή- σω, ε-σχη-κα, σχε-τό-ς, σχε-σι-ς, σχή-μα, and α-σπε-το-ς, θε-σπέ-σιο-ς. Hence εσγρν and ενισπον are properly primitive aorists, in which we should, naturally have expected as 1 sing. ind. *ε-σχι/-ν, *ε-σπψν. But the final vowel was in most verbal forms affected by the interchange between ο and ε, so that finally even in the imperative ε ιύσπε, παράσχε and other forms of the kind found a place. The optative σχοίμην is to be looked upon just as θοίμην, only that in the case of the latter the earlier form ϋείμην was also preserved. We meet with a similar change of 9 the radical into the thematic vowel in Sanskrit aorists, only that here, with the firm nature of the a, the change appears not as qualitative, but purely as quantitative, e.g« in a-khja-m ' I told ' from the root khj a. After these two certain instances it is very probable that we have to come to the same conclusion with regard to ε-σπο-ν (επ-έσπον 3 plur. Τ 294), ε7Γί-σ7ττ;ς, επί-σποι, επί-σπέΐν, μετα-σπών. We shall have occasion to return to the middle forms σπέσθαι etc. when dealing with the redu- plicated aorists. It is more difficult to form a judgment as to the aorist ε-πτό-μην, πτέ-σθαι. Here we find α by the side of the thematic vowel : ε-πτα-το, άπο-πτά-μενος as well as άπ-έ-πτη-ν, the relation of which to πεταμαι was discussed on p. 120. It is true that επι-πτέ-σθαι and the like occur even in Homer (e.g. Δ 126), but we have some reasons for regarding the forms with α as the older. Now as we find side by side in the present-stem πέτα and πέτο (πέτομαι) and in the aorist πτα and π™, the balance inclines in favour of syncope. With these forms I compared (1. c.) πελο (πέλομαι) and πλο (ε-7τλε-Γο). This is not however to be understood as implying that ε-ττλε-το (also in the active επλε Mil) is an aorist. The aoristic force does not suit various forms, e.g. επιπλό- μενον έτος η 261 : επλετο is therefore just as much an imperfect as επέλετο. The phonetic difference was not in this case used to distinguish an aorist. As to the origin of the stems appearing in these aorists, the majority correspond to the expectation that we should find in them nothing but an unexpanded root together with the thematic vowel : e.g. ά£ε<ι•, άλέσθαι, /3αλεπ', ΖακεΊν, θανεΊν, ίδεϊΐ', ικέσθαι, ολε'σθαι, αμ-πνυε^ πορεΊν, πυθέσθαι, τεκεΊ,ν. But it is unmistakeably otherwise with a not insigni- ficant number of by no means late and in part very common forms : e.g. βλαστε'ιν, α/σθε'σ(ίαί, αμπλακεϊν, ενρεΊν. These have evidently arisen from stems of various, and, in part, certainly nominal origin. We might call them secondary aorists, and may. divide them into the following five groups : 1) Aorists with an accessory Θ. These fall into two subdivisions, according as the θ is attached to the whole verbal stem, or only occurs sporadically : 10 a) Aorists with a θ attached throughout : αίσθέσθαι, ΰαρθεϊν, μαθεϊν, παθεϊ»', όλισθεϊν. b) Aorists with a sporadic θ : ήλν-θο-ν and iJ\0o-*>, ε-σχε-θο-ν (by the side of ε-σχο-ν). We must return in a later section to a general consideration of the forms with Θ. It is sufficient to refer here to what we have put together in the Principles, vol. i. p. 81 f. If the θ appealing in such forms comes, as is generally thought probable, from the rt. θε, Skt. dha * to place, do,' we must recognise in such aorists compound forms, to be compared with the CH. ΧΙΠ. SECONDARY FORMATIONS. 281 German weak perfects. The case of the thematic vowel is then here just as it was with l-ayo-v : it must have come from the stem- vowel of the verb. 2) Aorists with an accessory τ. There are scarcely more- than the following five, which belong here : i\\tTo-v (gen. mid.), which, as has been shown in Principles, vol. ii. p. 179, is connected with the rt. αλ (άλη, άλασθαι) and especially with ήλί-θιο-ς, ήμαρτο-ν, which has-been already mentioned on p. 163 under the present formations with r. The Epic forms ημβροτον, άβροτάζειν show that the rough breathing is of later origin. Certainly ημβροτο-ν or ημαρτο-ν is nothing but a verbal form directly derived from the adjective *ά-μαρ-τυ * not sharing ' (cp. μίμος, μοίρα, μόρο-ς), as has been already shown in Principles, vol. ii. p. 350. Further confirmation may now be supplied. Hesychius has the gloss άμαρείν with the explanations άκολονθεΊν, πείθεσθαι, άμαρτάιειν. The first two meanings evidently suit only a άμαρείν equivalent in meaning to ό-μαρτεΊν, the third justifies us in assuming a shorter άμαρέω formed from ά-μαρο, and related to *ά-μαρτέΐν as άπραγεϊν is to άπρακτε! v. In the same storehouse of facts we find the gloss επίμορτος ' σπόριμος γή, in the explanation of which the word μόρτη is quoted and explained by μέρος, επίμορτος is evidently the opposite of the ά-μορτυ-ς which is at the bottom of αμβροτεΊν. Other traces of this negative adjective are pointed out by Lobeck El. i. 37. — It is just the same with ε-β\αστο•ν. Here the noun βλαστό-ς actually occurs: its derivation from the rt. vardh Gr. ϊαλθ and by metathesis ΪΧαθ, βΧαθ is discussed in Principles, vol. ii. p. 168. These formations evidently became aorists only by the existence of άμαρτάνω, βλαστάνω, which are II expanded by present-strengthening. — We have further ε-μορτε-ν, known to us only from Hesychius. Lobeck 1. c. regards it as a mistake for ημορτεν, so that it would be equivalent to ημαρτεν. But I doubt whether the explanation of Hesychius άπέθανεν suits this view. It is true that the gloss Ισημορτεν ' άπέθανεν remains obscure. I conjecture that the syllable Ισ- is here as in , ίσ-(()ωρ-ες=ώώρες a dialectic form of εξ, but the η is strange : probably it is a mistake for ε. ε-μορτε-ν, if correctly recorded, is to μορττό-ς, explained in Hesychius by θνητός, just as εβΧαστεν is to βλαστός. — From a much later time we have the last of these forms ομαρτον, i.e. in Orph. Argon. 511 : και pa πανημ€ρίτ)ο-ιν cv εϊλαπίνησιν ομαρτεν. We may conjecture that we may see in this word only an imitation of αμαρτον. There is one more doubtful form in Hesychius, to which my attention has been called by Brugman, Sprachwissenschaftliche Abhandlungen p. 160. The gloss άΰρακτον • Uiov is written by Mor. Schmidt άΰρακτον • εϊΰον, a very happy thought. The frequentative ΰροκτάζεις ' περιβλέπεις suits this very well, a as an augment was dis- cussed on p. 77. We may find in the kindred languages an analogy even for these isolated forms. In Sanskrit from the rt. as i throw ' there is formed the very remarkable aorist as-tha-m. None of the attempts to explain this hitherto, so far as they are known to me, seems at all satisfactory. Perhaps the form belongs to the present group. The aspiration of the t after s is not without analogies. *as-ta-m would be formed in exactly the same way as ε-βλασ-το-ν. Besides this, Old Erse has a widely- extended preterite in t, e.g. 3 sing, bir-t tulit (Zeuss-Ebel p. 454), trans- 282 THEMATIC AORISTS. ch. xiii. lated into Greek letters *ε-ή>ερ'τε=ε-<ρερ-ε. Perhaps the remarkable Oscan preterites (3 sing.) in -ted, which have long been compared with the Keltic forms, e.g. j9rw/a^ed==probavit, belong to this group : Schleicher Comp. 3 823 treats them as perfects. As these languages have lost the augment, the distinctive criterion of the preterite, the question is hard to decide. 12 3) Aorists with accessory nasals. Here we may first mention ήλΰανε (σ 70, ω 368), with the present άλΰαίνω, in face of which the form is regarded as an aorist. But as άλΰαίνω occurs first in Aeschylus, while ήλΰανε is only Homeric, and as the sense in the two passages hardly decidedly requires the aorist, the possibility that ήλΰανε is the imperfect to a subsequently obsolete *άλΰάνω is by no means excluded. On ε-πιτνο-ν and πιτνέω I have stated my views on p. 184. The occurrence of the longer form πιτνέω by the side of πίτνω was. sufficient to mark επιτνον as an aorist. But it is worth noticing that the use of this verb is limited to the poets. But perhaps there are still some other forms to be placed here. If in the case of γίγνο-μαι we start, not from the fundamental form γεν which appears in γενέσθαι, γέγονα, γένος, αίειγενέτη-ς, but from the root γα which occurs in γε-γα-ώς, γε-γά-άσι, then in the stem γε-νο, just as in the Skt. present tfa-na-mi, the second syllable must be regarded as an expansion (p. 272). This is the explanation of the fact that, as we saw on p. 276, a-gana-nta is regarded as an imperfect, έ-γένο-ντο as an aorist. Of the same nature is ε-χανο-ν, undoubtedly an aorist, by the side of χάσκω. But if Gustav Meyer (Die mit Nasalen gebildeten Prasensstamme, p. 50) is right in starting with the root χα for both presents, χα-νο is an extended form just as much as χα-σκο (cp. above p. 197, Principles, i. p. 241). It is much the same with κανεΊν and κτανεΊν. Here too we are brought to a root in a, with which however the nasal coming from the extending syllable -na soon became incor- porated. Cp. pp. 130, 216, Gustav Meyer op. cit. p. 33. We ought not perhaps to attach much importance to the gloss of Hesych. κτάνειν ' κτείνειν, for this may well be an error for κτανε'ιν. 4) Other aorists of secondary formation. There are still a number of isolated forms remaining. A suflix -κο is clearly shown in ε-πτα-κο-ν (καταπτακών Aesch. Eum. 532), by the side of the Homeric κατα-πτή-την. The expanded stem also underlies the present πτήσσω. With these formations we may connect ε-θηκα, 13 ε-ηκα, ε-ΰωκα with their α which reminds us of the sigmatic aorist: to these we shall return below. — The γ in ε-τμαγο-ν may probably be regarded as a weakened κ (cp. Principles, i. 273). ήμπλακον too does not look very primitive, but the etymology presents difficulties. The relation of the Homeric ε-χραισμε etc. with the much later χραισμέω was discussed on p. 259. χραισμο is an adjective stem from χρα-σψο (later χρη-σιμο-ς) formed like φνζφο-ς, hence the verbal form ε-χραισμε is just as closely related to the noun-stem χραισμο as the Homeric θέρμε-το to the stem of θερμό-ς. In the case of εχραισμε the absence of a similarly formed indicative present gives occasion for its aoristic force. — A suffix -po is possibly to be found in the aorist ήναμο-ν, though this does not occur before Pindar : the origin of the word is obscure. But the derivation from εναρα ' arms ' is rightly rejected : Pindar's phrase (Nem. x. 15) iv οπλοις εναρε goes especially against it. Perhaps the word ch. χιπ. NON-KEDUPLICATED THEMATIC AOEISTS. 283 comes from the rt. san (Skt. pres. sanomi) * to win/ * to acquire/ which ■ on p. 122 we thought we discovered in άνΰω, and to which probably -εν-τη-ς in αντο-έντψς, ανθενν-ψς belongs. From the rt. san comes the Skt. san-ara gain, booty, from which we easily arrive on the one hand at τα εναρα (spolia), on the other at εναίρω i.e. εναρ-βω, Ho make booty of/ ' to acquire.' An ί -suffix possibly occurs in ώφ-ελο-ν and ώφλο-ν, but I do not know any satisfactory etymology. The aorists ε-Ιουπο-ν (first in the Anthology), επ-ανρ-εϊν, and ενρ-εΊν show by their diphthongs that they can hardly be radical, but their etymology is unexplained. As 'ίσχω is probably rightly regarded as a reduplicated present, we have in ήμπισχό-μην (Aristoph.) an aorist with present reduplication, which only derives this force from contrast with αμτηαχ-νεο-μαι. Finally the aorist ώσψρό-μην, found first in Aristoph. and for which Herodotus has ώσψράμην, has come from a compound stem. For as we cannot fail to see here the rt. o^ (οζω), the second element of the word must contain a second root, όσ-ψρο for όΰ-φρο appears to be a compound of the nature of καρπο-φόρο-ς, ψως-ψόρο-ς, but to be compared because of the rejection of the vowel with ΰί-φρο-ς. We have actually authority in 14 later time for the substantive οσ^ρα•=.61μη. όσ-ψρο has its counterpart in ol-facio : υσφραίνομαι is a later expansion. So various are the forms which may be united under one common category. I now give a list of the thematic aorists, which are not reduplicated, arranged according to the initial letter of the stems : and hence I gene- rally quote the infinitive, in which the initial comes clearly into promi- nence. The only exception is in the case of isolated forms, not existing in the infinitive. 1) άγγελεΊν. Certainly established in Anthol. vii. 614, 9 Ιάμω Ι άγγελέτην, elsewhere usually with various readings : e.g. Herod, iv. 153 άπήγγελον (v. 1. λλ), Lycurg. § 85. Mid. still less authenticated. 2) άγερέσθαι β 385, also accented άγέρεσθαι (Lobeck Rhemat. 132), άγέροντο Β 94, part, άγρόμενος cp. p. 279. 3) άΰεϊν. Established from Homer, Pindar, Herodotus, and Sopho- cles (Antig. 89). For εαΰον, ευαΰον p. 79. Mid. isolated and late. 4) αίσθέσθαι common from Aeschylus (Prom. Y. 957) onwards. 5) ήλΰανε discussed on p. 282. 6) άλέσθαι. αληται Φ 536, αΧομενα Aesch. Eum. 368 ch., είςαλοίμην Soph. Fr. 695 Dind. ενηλυν M.SS. Aesch. Pers. 516, altered by some editors into ενήλλον, because this aorist is not regarded as good Attic : * forma barbara ' Cobet. N. L. 454. For αλτο see p. 90. 7) άλιτεΐν. ήλιτεν p. 262, άλιτών Aesch. Eum. 316. Άθηναίην άλίτοντο ε 108, άλιτέσΟαί Β 378. 8) ήλφον Φ 79, αλφοι ο 453 ; these Homeric forms are regarded as aoristic only because the present άλφάνω occurs (according to Yeitch) three times in Eur. and Aristoph. 9) αμαρτεϊν from Homer onwards : αμαρτ Δ 491 (with impf. ημάρτανε Κ 372), άμαρτών Ψ 857. Also ήμβρϋτον φ 425 etc. 10) άμπλακεΊν poetic aorist with the bye-form ήμβλακο -r, established from Archil. Fr. 73 B 3 . Cp. p. 195. 284 THEMATIC AORISTS. ch. xhi. 11) άρε'ιν, only apot from quite late prose; but very commonly \ρέσθαι in poets from Homer onwards : την αρετ εκ Ύειέΰοιο Λ 625, 15 μισθον αρηται Μ 435, κλέος έσθλόν άροίμην Σ 121, κνΰος άρέσθαι Π 88, ποΖοϊν κλοπάν άρέσθαι Soph. Aj. 248, never with any other meaning than that of acquiring, gaining, and hence belonging to the present άρνυμαι (p. 110). As αίρω is contracted from άείρω (poet. : to this belong άερθεν, ήερέθονται) the aorist forms with a short α cannot possibly have anything in common with άείρω or αίρω. Cp. Principles, i. 425, 442. Buttmann, ii 2 . 100 was led astray by the parallelism of άείρω αφω with φαείνω φαίνω. But while άείρω (p. 215) is the form invariably used in Homer, it is the opposite with φαίνω, φαε'ινω occurs only in a very limited application, so that the shorter form has certainly not come from the longer, derived from φάος. 12) άρέσθαι * take,' in the Locrian inscription from Naupactus [Cauer p. Ill, B. 18] 1. 32 rav ΰίκαν πρόδιΟ ov άρέσται πότους ΰικαστήρας, άρέσται και ΰόμεν, (cp. Stud. ii. 448), either from the rt. ap, which underlies the present αιρέω, or with a change of breathing (cp. αγεν=.άγειν) for άρέσθαι, and hence akin to the preceding word. The former is the more probable because of the common formula δίκη ν λαβείν. 13) επανρέϊν, almost exclusively poetical : μη τις χρόα χαλκω έπαύρη Ν 649, έπαυρέμεν Σ 302, middle έπαυρέσθαι Α 410 (έπαύρωνται), επαύροιτο Herod, vii. 180. 14) βαλεϊν, βαλέσθαι common in all Greek from Homer (Π 618, Β 45 etc.) downwards. Arcad. εζελεν 'έβαλε ν Hesych. 15) ε-βλαβε-ν only in Quint. Smyrn. v. 509, τίη νύ τοι έβλαβεν ήτορ ; cp. p. 275. 16) βλαστεϊν in Pindar (01. vii. 69), Herod, (vii. 156), and the dramatic poets (Soph. El. 238). 17) βραχεϊν poetical from Homer (E. 863) downwards. 18) βρνχεϊν 'bite,' only in Anth. ix. 252 (εβρυχε). 19) γενέσθαι common in all writers from Homer downwards. 20) γόον 3 pi. probably only in Ζ 500 α? μεν ετι ζωον γόον'Έκτορα. 21) έ-ΰαε-ν only in Apollon. B-hod. iv. 989 etc. with the meaning ' taught' (hence =ΰέΰαε-ν). 22) ΰάηται only in Ύ 316, Φ 375 6πότ αν Ύροίη μαλερω πυρι πάσα ΰάηται. Bt. ^αυ, ΰαΡ, hence" for *caF ηται. 23) ΰακεΊν, common especially in the poets from Homer (E 493) downwards. 16 24) ΰαρθεϊν, common in poetry and prose in composition with κατά, κατέΰραθον ψ 18, καΰΰραθέτην ο 494, κατέδαρθον Thuc. vi. 61. 25) ΰικείν an isolated post-Homeric poetic aorist : Pind. 01. x. 72, 'έδικε πέτρω, Aesch. Choeph. 99 ΰικοϋσα. 26) έ-οουπε, κατ-έ-δουπε only in Anth. vii. 637. 27) ΐρακεϊν in poetry from Homer onwards (έΰρακον κ 197), ε-ΰρακό-μην Anth. vii. 224. 28) ΰραμεΊν common from Homer (Σ 30) downwards. 29) Ιρανών only in Pind. Pyth. iv. 130. 30) έγρέσθαι. εγρετο δ' εξ υπ νου Β 41, καν εγρ*} μεσημβρινός Αϊ*. Vesp. 774, εγρεο Κ 159, έγρέσθαι ν 124, εζεγρέσθαι, εζεγρόμενος Plat. %mp. p. 223 C. 31) ελεϊν, ελέσθαι common from Homer downwards. 32) ήλυθο-ν, ελθεΐν. The trisyllabic form is limited to the indicative, ch. χιπ. NON-KEDUPLICATED THEMATIC AOK1STS. 285 and occurs only in Homer, Pindar, and, especially in melic passages, in * the tragedians : R 28, Soph. Aj. 234 : the disyllabic from Homer down- wards (ήλθον δ 82) as the usual form in the indicative, and the exclusive- form in the conjunctive, optative etc.: ελθοι Ε 301 occurs at all date» and in all dialects. Hence Dor. ήνβον (Epicharmus, Theocritus), Lacom ήλσον, ελσων (Aristoph. Lys. 105). 33) εναρείν, εξεναρε'ιν Hes. Sc. 329, εναρον from Pindar (Nem. x. 15) downwards. Cp. p. 282. 34) ερέσθαι from Homer onwards commonly used in all moods, the infin. and the participle. 35)' ερικεΊι-, ήρικε δ' ΙπποΙό σεια κόρνς περί ΰονρος άκωκ^ Ρ 295, Soph. Fr. 164 Dind. 36) εριπείν, ήριπε 3' εί, όχέων Ε 47, έριπών θ 329, isolated in other poets and in Hippocrates, ήριπόμαν Agathias Anth. P. ix. 152. 37) ερυγεΊν in Homer, perhaps also in Callimachus Fr. 246 (ed. 0. Schneider ii. p. 485) with the meaning ' roar ' (ήρνγε Υ 403), later (also- in Aristot.) ' vomere ' Ar. Yesp. 913 έιήρνγεν, cp. p. 155. 38) ενρεϊν, ενρέσθαί from Homer downwards in ordinary Greek. 39) άπ-εχθέ-σθαι, from Homer onwards; in Φ 83 I. Bekker writes άπέχθεσθαι, but cp. Eur. Med. 290 Elmsley, Dind. κρεϊσσον δε μοι νυν προς σ άπέχθεσθαι, γνναι, • 17^ η μαλθακισθενθ* ύστερον μέγα στενειν where the aorist force is much the more suitable. 40) βάλε, a very doubtful reading in Hymn. Homer, xix. 33, for which Ruhnken conjectures λάθε : άναϋαλέϊν often occurs in the New Test, and LXX. 41) θανεΊι> from Homer onwards very common, especially in com- position with από and κατά. 42) θενεΊν, established from Pindar (01. 7, 28 θενών), Eurip. (HeracL 271) and Aristoph. (Lys. 821). 43) θιγεϊν occurs in Pindar (θίγον Isthm. i. 18), in the tragedians,. Hippocrates, Xenophon, and in later poetry and prose (Aristot.). aiyrjy is Laconian — Aristoph. Lys. 1004. βιγέσθαι Themistius. 44) θορεΊν from Homer onwards (θόρε Ψ 509) especially in the poets ;. νπερθορείν Herod, vi. 134. 45) ΙΰεΊν, Ιϊέσθαι in all Greek. 46) ίκέσθαι common from Homer onwards, in prose for the most part only in composition with άπό, επί. 47) άμπ-ισχεΊν, άμπ-ισχέσθαι, both established from Aristoph. (Eccl, 540), the former from Eurip. (Ion 1159 ήμπισχεν) and Plato Protag. 320. Cp. p. 283. 48) καμείν from Homer onwards, in poetry and prose, καμέσθαι Σ 341 y εκάμοντο ι 130. 49) κανεΊ,ν in the tragedians (Aesch. Sept. 630) and Theocritus. 50) kuiv cp. p. 276. 51) /αχεΐν cp. p. 121 j in poetry from Homer onwards: γ 169 I ν Αέσβω δ' εκιγεν. 52) ε-κλαγο-ν a rare poetical aorist of κλάζω Hymn. Horn. xix. 14, άνέκλαγον Eur. Iph. A. 1062. 53) κ pay ε! ν from Homer onwards (£, 467) frequently in poetry and prose. 54) κρίκε only in Π 470 κρίκε οε ζνγόν. 70) 71) 286 THEMATIC AORISTS. ch. xin. 55) κρυβεϊν first in Apollodorus, in New Test., and in Plutarch. 56) κτανεϊν poetical from Homer onwards (B 701) and in late prose. 57) ε-κτνπε poetical, μεγάλα κτύπε θ 75, εκτνπεν αίθήρ Soph. Ο. C. 1456. 58) κύθε perhaps only in γ 16 οπού κύθε γαία. 59) λαβείν, λαβεσθαι in all Greek. 1 8 60) λαβείν, λαθέσθαι in all Greek. 61) λακε'ιν poetical (λάκε & 25, λακεΊν Soph. Ant. 1094). 62) λαχεία in all Greek. 63) λιττεί^, λιπέσθαι universally used. 64) λιτέσθαι Π 47, λιτοίμην ξ 406. 65) μαβέϊν common from Homer onwards (μάθον Ζ 444). 66) μακών, Π 469, κ 163, σ 98, τ 454. 67) μαπέειν Hes. Scut. 231, 304. 68) μολείν poetical from Homer onwards (μόλτι Ω 781, μολοΰσα Ζ 286), also in Xenophon and in late prose. 69) εμμορο-ν first in Alexandrine poets : Apollon. Rhod. iii. 4 εμμορες, iv. 62, probably from a misunderstanding of the Homeric εμμορε A 278, which will have to be discussed under the head of the perfect [below p. 131 marg.]. μνκο-ν only in Ε 749 πύλαι μύκον and γ 260 σάκος μύκε. όλεσθαι frequent from Homer onwards, especially in composition with από. 71 b) όλισθείν ολισθε Υ 470, Φ 774, ώλισθε Attic poets and later writers. 72) ομαρτε -v a late-formed aorist from όμαρτέω only in Orph. Arg. 511, cp. p. 281. 73) ώρε-το Μ 279, Χ 102, οροιτο ξ 522, οππότ ερις — ορηται Hes. Theog. 782, όρόμενο-ς in the tragedians (Aesch. Sept. 115). Cp. ώρτο p. 131. On the other hand οροντο γ 471 is the impf. from rt. fop ' see, watch.' Cp. p. 144. 74) όσφρέ-σθαι established from the comedians (Ar. Ach. 179) and late prose. 75) ώφελο-ν from Homer (ώς ώ^ελες αντό0' όλεσθαι Γ 428) onwards commonly used in all Greek for wishes, resembling the usage of οφείλω. — ώφλο-ν differing from this only by the syncope, in a special meaning, i I was indebted,' hence more agreeing in use with όψλισκάνω, in Herodotus (ΰειλίην ώφλε viii. 26) and Attic writers. The present οψλω quoted from Dio Chrysostom and Appian seems to be a later formation. παθεϊν in all Greek. παρΰεΊν abundantly established from the comedians, in composi- tion with από, κατά etc. (Aristoph. Pax. 547). 78) πετε'ιν Aeolo-Doric, established from Alcaeus (επετον Fr. 60 B 3 ) and Pindar (εμπετες Pyth. viii. 81). The other form commonly in use from Homer (πέσεν Ε 460) downwards will have to be discussed in Chapter XVII. [p. 284 marg.]. 19 79V πιείν in all Greek. Cp. π~ι-θι etc. p. 129. 80) πιθεΊν from Pindar onwards (πίθων Pyth. iii. 28) here and there in poets, πιθέσθαι in poetry from Homer onwards (επίθοντο Γ 260, cp. C. I. A. ii. 38, 4). 81) πιτνεϊν in Pindar and the tragedians; cp. pp. 184 and 282. 82) άμ-πνυε. άλλα συ μεν νυν στήθι και άμπνυε Χ 222, 3 sing, in Quint. Smyrn. Cp. άμπνϋτο p. 129. 76) 77) CH. XIII. NON-REDUPLICATED THEMATIC AORISTS. 287 83) πορεΊν in poets from Homer onwards (imper. πόρε I 513, πόρων Π 178). 84) πραθεΊν in Homer (ΰιαπραθέειν τόΰε άστυ Η 32) and Pindar (επραθε Pyth. ix. 81). 85) κατα-πτακων only in Aesch. Eum. 252. 86) πταρέϊν common from Homer onwards (p 541 Τηλέμαχος δε μέγ 1 επταρεν): from Hippocrates a conjunctive middle πτάρηται is also quoted. 87) πτέ-σθαι cp. above p. 280, Veitch p. 468. 88) πυθέ-σθαι in all Greek. 89) συν-έρραφε-ν only in Nonnus Dion. vii. 152 μηρω δε συνέρραφεν. 90) ερριφε-ν only in Oppian Cyneget. iv. 350. 91) σπαρέ-σθαι a doubtful reading in Polyaen. viii. 26. Others σπείρεσθαι. Cp. Veitch, p. 529. 92) σπείν belonging to ε7τωίη Homer (οι πάντες ολέθρων ήμαρ έπέσπον Τ 294, οφρα — έπίσπη Β 359, μετασπών Ρ 190), Aeschylus (έπέσπε Pers. 552) and Herodotus' (περιέσπε vi. 44). — σπέ-σθαι (with the reduplicated ε-σπέσθαι) in Homer, Herodotus, and Attic writers (Ύρωσϊν άμα σπέσθαι Ε 423 (v. 1. εσπέσθαι), σπείο Κ 285, έπίσπη Soph. ΕΙ. 967, έπισπόμενοι Thuc. ν. 11). 93) σπείν 'say,' ενι-σπο-ν, from Homer onwards (Β 80, ivL -σποι & 107). The imperative ε-σπε-τε perhaps for εν-σπε-τε. Cp. above p. 280. 94) ε-στιβε-ν ' επαλννεν Hesych. 95) στιχείν Π 258 έ'σπχον, then in Alexandrine poets : for the traces of a present στίχω cp. p. 155. 96) στυγέϊν rare in Homer and later poets (/caret δ* εστυγον αυτήν κ 113, Apollon Ehod. ii. 1196). 97) σχεϊν, σχέ-σθαι in all Greek.. For σχε'-ς cp. pp. 132, 279. 98) ταμεΊν, ταμέ-σθαι are the forms usual in Homer, Herodotus and 20 Pindar, instead of which we have in Attic τεμεϊν, τεμέσθαι (τέμενος τάμον Υ 184, ταμέσθαι Herod, ν. 82. — τεμοΰσα Soph. ΕΙ. 449, ετεμον Isocr. 8, 100). 99) ταρπώμεθα only in the phrase ταρπώμεθα κοιμηθέντες Ω, 636, ί 295, ψ 255. 100) ταφεϊν here and there in poets from Homer onwards (ταφών δ' άνόρουσεν Άχιλλεύς Ψ 101). 101) τεκεϊν in all Greek. 102) ΰι-έ-τμαγο-ν only in η 276 τοδε λαΐτμα ΰίέτμαγον. 103) ε-τορε only in Λ 236 ονδ* ετορε ζωστήρα. 104) τραγεϊν established from Attic comedians (Aristoph. Ach. 809), Hippocrates and late prose. 105)• τραπέίν ) τραπέ-σθαι common from Homer (Y 439, Π 594) onwards ; the middle not unknown even to Attic prose. 106) τραφεΊν, transitive ετραφέ τ ένΰυκέως (v. 1. έτρεφε) Ψ 90, τράφε Pind. Nem. iii. 53, intransitive 'grow up' Ε 555 and elsewhere : so in later poets. 107) τυπεΐν, the only evidence in early times is Eur. Ion 767 (ετυπεν), then first in Achilles Tatius. 108) τυχεΊν in all Greek. 109) φαγείν the same. 110) φανέσθαι a doubtful reading in Xen. Cyr. iii. 1, 34 (φανοίμην), imper. φάνευ in a Laconian saying in Stobaeus Floril. 108, 83. 111) ε-φλαΐο -y. λακίΐες εφλαΐον υπ' άλγεσιν Aesch. Choeph. 28. 288 THEMATIC AOEISTS. ch. xiii. 112) φράΰεν" έλεγε ν Hesych., ϊφραδεν* εΰήλωσεν ib. 113) φυγείν in all Greek. 114) χαδεΰ established from Homer (Δ 24), the Anthology and Hippocrates. 115) χανεϊν from Homer (τότε μοι χάνοι ευρεϊα χθων θ 150) onwards,, established mainly from poets, but also from Herodotus and Hip- pocrates. 116) χάρο-ντο only quoted from Quintus Smyrn. vi. 315 Τρώες b* επι μακρά χάροντο. 117) έ-χραισμο-ν, τεΊχος δ' ουκ εχραισμε τετνγμένον & 66, χραίημ -rf Ο 32 etc., then in Apollon. Rhod. (ii. 218 χραίσμετέ μοι). 21 II. Aorists with Reduplication. The aorists to be discussed here come from the reduplicated stem in just the same way as those hitherto discussed from the non-reduplicated stem. The reduplication therefore serves to strengthen and bring into prominence the verbal stem. After my elaborate attack in Tempora und Modi pp. 150 ff. on the view till then generally accepted, that re- duplication served to mark the tense, and was akin to the augment, it is hardly necessary at the present time to return to this question. For the position which I then established is now universally admitted for Greek, e.g. by Kuhner Ausf. Gr. ii. 513, and has been laid down by Schleicher Comp. 3 p. 739 for the corresponding forms in Sanskrit and Zend. Reduplication, that is to say the repetition — actual or suggested-^• of the stem, can have had no other purpose than to bring the stem into . prominence. We have already (p. 8) had occasion to notice this primitive linguistic resource employed in this way, and we discussed on p. 105 its occurrence in the present, aorist and perfect- stem indifferently. Hence it admits of no doubt that reduplication, like the thematic vowel, was in no way intended originally to denote the nature of the tense. For it is excluded from no one of the three kinds of tenses. We can look back to a time when language formed both a present and a past on the one hand from the pure, on the other from the reduplicated root : thus from da da-mi dada-mi a-da-m a-dada-m . tarp tarpa-mi tatarpa-mi a-tarpa-m a-tatarpa-m. The only reason why in the first instance the form without redu- plication became an aorist, was that the present indicative without re- duplication fell out of use. On the other hand the reduplicated a-tatarpa-m (cp. τετάρπεπ,) gained an aorist force, because the non- reduplicated form here serves as the present indicative. Hence so far we recognise the closest analogy with the relations already discussed. 22 But a new influence steps in from the fact that we have a third com- peting tense, the perfect, which we shall learn to regard as a special ramification from the reduplicated present, and this competition is of the greatest importance for the Greek reduplicated aorists. For the aorist, like the perfect, is distinguished from the present-stem in Greek by that delicate phonetic law, which we toucHbd upon on. pp. 135 and 189. The vowel ι is used without exception for the reduplication of the present-stem ; but never for the aorist and the peifect : here we find for the most part ch. χιπ. KEDUPLICATION IN THE AORIST. 289 the vowel ε. This remarkable differentiation Delbruck thinks he can show as beginning even in the Indo-Germanic time, inasmuch as in Sanskrit the i similarly appears, though only sporadically, in the present- stem, while it is absolutely unknown to the perfect. Hence ti-shtha-mi is to the Doric Ι- στ ά- μι and Lat. sisto precisely as ta-shthdu is to Gr. ε-στα-μεν and Lat. ste-ti. * But Sanskrit treats the aorist in two ways. The a is predominant, and to this again a Greek ε corresponds, e g. a-pa-pta-m (I flew) like e -τε-τμο-ν, but in many cases, especially in the forms with a causative meaning, I appears, though the length is not always preserved, and it is therefore regarded by Delbruck (p. 109 ff.) as later than the short vowel; e.g. a-pl-pata-t (he made to fall) differing from ε-πε-φνε and of like formation with the impf. ε-πϊ-πτε, which in meaning attaches itself to the intransitive πίπτω. By means of this difference in the vowel Greek avoids any confusion of reduplicated aorists with present-stems. But the aorists come thereby all the nearer to the perfects. As in Sanskrit there are past tenses, with regard to which there may be a doubt whether they have come from the perfect-stem by prefixing the augment, and hence whether according to the usual terminology, they are pluperfects or reduplicated aorists, so also in Greek. For Sanskrit Delbruck has thoroughly dis- cussed these questions, and has assigned to some verbs, partly on the ground of their form, and partly from regard to their special force, a place in close connexion with the perfect, while he has made a greater 2 3 separation in the case of others, though on the whole he has recognised the common origin of all reduplicated forms. I entirely agree with him in this, and believe that we must deal in precisely the same way with Greek. Everything leads us to the conclusion that the reduplicated form was at first only distinguished from the non-reduplicated by the fact that the former represented the action as more intensive. From this common stem on the one hand the perfect was developed, originally merely an intensive present, though gradually by special modifications of the endings it grew into a special form of language. We shall have to discuss this fully in Chap. XVI. But on the other side from the same stem the reduplicated aorist grew, and this again separated itself from the forms of the perfect-stem, mainly by the identity of its endings with those of the non-reduplicated aorist. In the past tense, that is, according to the usual terminology, * in the aorist indicative, a confusion with the pluperfect was generally the less possible, that for this tense a compound formation (1 sing, -εα, -η, -ειν) became almost universal in the active, while the forms of the perfect-stem in the middle were sufficiently dis- tinguished from the middle reduplicated aorists by the entire lack of the thematic vowel. The perfect- stem besides has in Greek a preference for the long stem-syllable, the aorist, including the reduplicated, for the short, and in fact for syncope : so that care is thus taken that even the unaugmented form of the 3 sing. aor. πέπιθε is distinguished from the perfect πίποίθε. In spite of this delicate means of discrimination, there are some cases υ 290 THEMATIC AORISTS. ch. xiii. in which a decision is not immediately possible, ίπέφϋκον (Hes. Theog. 152) is shown by its κ and by its meaning to be a pluperfect. But both these proofs are wanting in ε-μέμηκο-ν, for the only passage in which the 3 plur. occurs (ι 439), θηλ€ΐαι δ' έμψηκον άνημελκτοι περί σηκονς, gives no safe criteria. But we have a word of exactly the same forma- tion in έπέπληγον, which occurs more frequently, and which is accom- panied also by middle forms : Ε 504 6v pa (i.e. κονίσαΚον) "* ουρανον es πολνχάΚκον ininkqyov πόΰςς Ίππων : cp. Ψ 363 πέπληγον V ίμάσιν, and further Π 728 πεπληγέμεν, which recurs Ψ 660, and θ 264 πέπληγον δε χορον θεϊον ποσίν. As this form is accompanied by the equivalent present perfect πέπληγα (part, πεπληγώς), and as we find here the length usual in the perfect, we shall be inclined to take επέπληγον as a pluperfect, and to form the same judgment as to έμέμηκον, which is accompanied by the synonymous μεμηκώς fern. μεμακνϊα. The objection may be made, that the infin. πεπληγέμεν and the middle forms πεπλήγίτο (Μ 162, ν 198) πεπληγοντο (Σ 51) prove by their thematic vowel that they have nothing in common with forms like πεπληγώς and the post-Homerio πέπληγμαι. But this objec- tion does not hold good, for in the first place late poets have the by-form of the participle πεπληγόντες in a present sense (Callim. H. in Jov. 53, Nonnus), and secondly other instances are not wanting in which this vowel makes its way into unmistakeably perfect forms after the analogy of the present. The Homeric κεκλήγοντες, well established in Μ 125 (cp. La Roche Textkritik p. 296, 0. Schneider ad Callim. i. 151) by the side of the singular κεκληγώς Β 222, is sufficient to show that no absolute distinction is possible. — We might add to these forms εκέκραγον, if its case . was more certain. In Anth. Pal. v. 87, 2 we find κεκράγεν, but the context requires rather the present, and we are tempted for κέκραγεν ώς to read κεκραγ ως. There is only one testimony for εκέκραγον left, and that from the LXX. — λελάιΜίτο, only in Hymn, in Merc. 145, though it has a short a, shows by its mean- ing * they bellowed ' the closest connexion with the fern. part, λελα- κνϊα μ 85, which also has a short a, so that we may regard it as a pluperfect just as well as έμέμηκον.— The same holds good of some im- peratives formed without a thematic vowel, which properly do not be- long to this group, but which may be discussed here because of their isolated position. The Homeric κέ-κλυ-θι (plur. κέ-'κλν-τε, occurring also 25 once in Pind. Pyth. iv. 13) is to the rt. κλν precisely as τέ-τλά-θι is to rt. τλα. It is of no consequence that the latter comes by metathesis from ra\. Of the same nature is κέ-κγαχ-θί. The reason why τετλαθι and κέκραχβι are regarded as perfects is that they are accompanied by unmistakeable perfect forms like τέτλαμεν (υ 311), τετληυΐα, τετληότος, κέκράγα, while these are entirely wanting to κέκλνβι. But we can no more talk of any specifically perfect force in κέκ-ραχΟι than in κέκλνθί. The posi- tion of the reduplicated forms in the verbal system is therefore just as clearly defined as that of the non-reduplicated. In the case of the latter the occurrence of a present indicative is decisive ; in the case of the former the occurrence of a perfect indicative of similar formation. Iso- lated forms are to be regarded as aorists. — πέπεισθι (or πέπισθι) Aesch. ch. χπι. VOWELS OF THE KEDUPLICATION AND THE STEM. 291 Eum. 599 must be regarded as a perfect imperative because of its mean- ing, which belongs entirely to νέποιθα, Besides Sanskrit and Greek, it is probably only in Zend that we have traces of a reduplicated aorist : these are pointed out by Schleicher Comp. 3 742. The only thematic formation mentioned by him is ta- tasha-t, he wrought, fromrt. tos/i=Skt. tahsh. The 3 sing. perf. of the. same rt. is in Zend ta-tasha (cp. Justi Handbuch, p. 133). Thus here too the two tenses strongly resemble each other. The reduplication- syllable in Greek has always ε. The exceptions are very few. In the first place there is a quite isolated form, from a stem beginning with a consonant, in the participle quoted by Hesych. πα-φω-ν κτείνας (cp. Lobeck, Rhemat. 9) which evidently belongs to the perf. πέ-φα-νται, πε-φά-σθαι (cp. φό-νο-ς). We should have expected πε- φών (cp. ε-πε-φνο-ν), just as the rt. φα i shine ' gives us the non- thematic aorist πέ-φη' εφάνη Hes. We have to mention besides a number of stems beginning with a vowel, in which we find the so-called Attic re- duplication, familiar from the perfect formation. This occurs in six aorists, five of which, i.e. άγ-αγεϊν, ακ -ay/iv, άλ-αλκεϊν, άπ-αφεΊν, άρ-αρεΊν have α, one, i.e. όρ-ορέϊν has ο as the stem- vowel. Delbriick, p. Ill, points out some Vedic forms completely corresponding : dm-ama-t from rt. am ' damage,' and with a weakened stem- vowel dn-ina-t from rt. an * breathe,' ard-ida-t and drp-ipa-t with a loss of the r also, from the rts. ard ' press ' and ar-p, an expansion of ar ' to fasten in.' This agreement 26 is all the more remarkable that this kind of reduplication is unknown to Sanskrit in the perfect. We shall return to it under the head of the perfect. — There is a case of Attic reduplication with ε in the stem and reduplication syllables in ην-ε-γκο-ν (Principles i. p. 384) from a rt. εγκ, which is regularly interchanged with nank (Lat. nanc-i-sco-r). In the Homeric ενεικέμεν Τ 194 the nasal has disappeared with compensatory lengthening (Joh. Schmidt, Yocalismus i. 122). In the stem-syllable, which, as we saw above, rejects a long vowel, we have just the same phenomena as in the non-reduplicated forms, i.e. the preference for the vowel a, which appears in at least 20 cases out of 41, and the metathesis and loss of vowels : with regard to the choice between the former and the latter of these the same doubt presents itself as in the non-reduplicated forms, ε-κέ-κλε-το (rt. κελ), ε-σπε-το, ε-τε- τμο-ν, ε-πε-φΐΌ-ρ are to be judged just in the same way as ε-σχο-ν, and the other words discussed on p. 279. The rejection of a ρ appears in με-μάπο-ιεν Hes. Sc. 252 (μαπέειν 231, 304), a process which finds its analogy in the Sanskrit forms drd-ida-m, drp-ipa-m mentioned above. — Contraction after the loss of a } appears in ε-Ρειπο-ν, which Sonne first in his Epilegomena to Benfey's Wurzellexikon (1847) p. 39, and then independently Ebel Ztschr. ii. 46 (1853) compared with the Skt. a-voJca-m for *a-va-vaKa-m. There is, however, one objection to this attractive comparison. The diphthong ει in ειπείν occurs even in dialects which contract εε into η : thus in Alcaeus Fr. 55 θε'λω τι ϊειπήν, though here the reading is not quite certain. We should have expected ϊηπήν, after the analogy of ήχες=είχες (Sappho 29). And Priscian i. § 54 does actually quote an Aeolic ήττον. But in Old Attic also the common είπεν is always written with a diphthong (Cauer Stud. viiL 257). Perhaps we must presuppose *ϊε-Ριπο-ν formed after the analogy of υ 2 292 THEMATIC AORISTS. en. xi Sanskrit forms like dn-ina-t. The ι of όν-ίνη-μι and όπ-ιπ-ενω arises from a similar weakening. — ένένϊπο-ν with the by-form ήνίπαπον, and ερυκακο-ν are quite irregular. εν-έν-Ιπε-ν, for which εν-ένιπτε-ν is a bad variant, may be simply explained from a mistake as to the preposition, 27 just as ήνεπε, προςήνεπε are forms occurring in the M.SS., which Bergk has adopted in Pind. Nem. x. 79, Pyth. iv. 97. Cp. Rich. Fritzsche Stud. vi. 332. ήνίπαπον and ερνκακον are so far parallel forms that both have apparently suffered internal reduplication, and in spite of the difference in the vowels of the second syllable have α in the third. For ήνίπαπον Ebel Ztschr. ii. 48 following Pott has suggested the explana- tion from the rt. 'ix, interchanging with law, and in Principles ii. p. 59. I have expressed my concurrence. Here the preposition has received the temporal augment. We must regard εν-ϊπ^απε as the original form. The syllable ;'απ is reduplicated by Ίπ. The latter form also is preserved in the verb 'ίπτεσθαι * to damage.' To understand έρύκ-ακο-ν we must start from the rt. hp, with the meaning ' to watch, guard/ which I have discussed in its relation to the Homeric ερνσθαι, εφυσθια in Stud. vi. 272 ff. Expanded by a u, and extended by a κ (cp. όλέ-κω, ε-πτα-κο-ν) we find ΐρύκω with the exclusively negative meaning (so to say) ' to ward off' : this is related to the meaning of ε'ίρυσθαι, as έίργειν ' to shut out ' is to είργειν Ho shut in.' ερ-κ-ος contains the expansion without the v. We must therefore, if ερύκακο-ν has really been formed by reduplication, assume that there were originally two expanded forms varh and varuk, and that the two are here united. I regard *varuk-varka-m as the original form. From this came *varuk-vaka-m, through the same loss of the r as that which we saw in the case of rt. μαρπ, and in the Yedic forms ard-ida-m, arp-ipa-m, and then with a rejection of the / *varuh- aka-m, in Greek letters ερνκακο-ν. We may compare to some extent the perfect νφ-ήφασμαι, though this only rests on the evidence of grammarians (cp. Principles i. 369). It is noteworthy that all reduplicated aorists, with the exception of άγαγεΊν, and also of ειπείν and ένεγκείν which have become unrecognis- able as such, owing to phonetic affections, are limited to the language of poetry, and for the most part find their place among the archaisms of Homer. We will now give a list of all such forms. 1) αγ-αγείν from Homer (ήγαγε Ζ 291, άγαγεν Λ 112, ήγάγεθ' "Εκτωρ Χ 471) onwards, common in the active and middle. 2) άκ-αχεϊν, άκαχέσθαι, quoted from Homer, Hesiod, and Alexan- drine poets : μέγα δ' ηκαχε Χαόν 'Αχαιών Π 822, θυμφ ακα-χων Hes. 28 Theog. 868, μνηστήρες c άκάχοντο π 342. The reduplicated stem ex- tends through all tenses, άκαχίζω, άκάχησα. Cp. αχός. 3) άλ-αλκεϊν from Homer onwards in non- Attic poets : όπως τί μοι άλγος άλάλκοις ν 319, πόΧεμόν περ αλαλκών Ι 605, άλαλκε ί)ε Χειρών Pind. Nem. iv. 60. We find also άλαλκήσω in Apollon. Rhod. 4) άπ-αφεϊν. The active occurs frequently in poets : παρήπαφε S 360, εϊ,απα^ών Eur. Ion. 704 ch. ; the middle only I 376, ψ 216 (άπάφοιτο). Reduplication occurs also in the present άπαφίσκω. 5) αρ-αρεϊν poetic, τους τε κλντός ηραρε τέκτων Ψ 712, αλλ' Ιμε γ' ά στονόεσσ άραρεν ψρένας Soph. ΕΙ. 147, άραροίατο in Apoll. Rhod. 6^ ϊέ-ΰαε 'taught' only in the Odyssey : θ 448, ψ 160. 7^ ΰέ-ΰακε only in Anth. xii. 15. 8) ιίπεΊν in all Greek : Homer, εειπον cp. p. 79. ch. χιπ. KEDUPLICATED AOEISTS. 293 9) ενενιπε only in Homer and Quintus Smyrn. ΊκεταονίΙην ενενιπεν Ο 546, Ψ 473. 10) ήνίπαπε, exclusively Homeric Γ 427, υ 17. 11) τέ-θιγεν ήψατο, probable reading in Hesychius for τέθειγεν. Lobeck El. i. 155. Cp. θιγεϊν, p. 285. 12) τε-θορ-εϊν ΰιαπηδήσαι Hesych., though out of the alphabetical order. Cp. θορείν. 13) κε-καΰεϊν, κεκαΰεσθαι, exclusively Homeric, θυμού- «cat φνχτ}ς κεκαΰών Λ 334, υπό δε Τρώες κεκάΰοντο Δ 497, Ο 574 : from the same stem κεκαΰήσαι ' βλάψαι Hesych. For the root cp. Principles i. 300. * 14). ε-κέ-κλε-το poetic Ζ 66, κέκΧετ' Π 421 etc. κέκλεν Pind. Isthm. vi. 53, κεκλοίμαν Aesch. Suppl. 591 ch., κεκΧόμενος Soph. Ο. T. 159 "ch., κέκλεο • κάΧεσον Hesych. Later poets form besides a present κεκλομαι Apollon. Phod. i. 716 (κ-έκΧεται). In Hesych. we actually find the active κεκΧεϊ, or, as emended by Lobeck Phem. 112, κέκΧει. This is the same phenomenon which we met several times in the case of the non-redupli- cated aorist. 15) κε-κύθω-σι only in ζ 303 όπότ αν σε δόμοι κεκύθωσι και αυλή. 29 Cp. ε-κυθο-ν ρ. 285. 16) Χε-Χαβέ-σθαι only in δ 388 τον γ' ε'ί πως συ ΰύναιο Χοχησάμενος λελαβέσθαι, with λαβέσθαι ρ. 285. 17) Χέ-Χαθο-ν, εκλέΧαθον κιθαριστύν (3 ρΐ.) Β 600, ΧεΧάθη δ' όΰυνάων Ο 60, on the other hand ΧεΧάθοιμι in the sense of Χάθοιμι Apoll. Phod. iu. 779. — ΧεΧάθοντο, οί/δε σέθεν θεοί ΧεΧάθυντο Δ 127, μη τις μοι άπει- Χάων ΧεΧαθέσθω Π 200, ΧεΧάθοντο δε μαΧοΰροπήες, ον μαν έκΧεΧάθοντ' Sappho Fr. 93 Β 3 . The middle always means Ho forget'; only in Hes. Theog. 471 Ho conceal.' — There is a present εκΧεΧάθων, if this is the right reading in Theocr. i. 63 as an epithet of Hades. 18) Χε-Χάκο -i'ro only Hymn, in Merc. 145, cp. above p. 290, for ε-Χακο-ν p. 286. 19) λέ-λαχο-ν, υφρα πυρός με Τρώες και Ύρώων άΧοχοι Χελάχωσι θανόντα Η 80 (cp. Ο 350, Χ Μ3).—ΧεΧάχοι=Χάχοι Anth. Pal. vii. 341. 20) με-μάπο-ιε-ν Hes. Scut. 252 ; the reading γήρας τε μέμαρπον ib. 245 is by no means certain. 21) ώρ-ορε transitive ' aroused' τα μεν r Έυρός τε ΝοΓος τε ώρορ έπαιξας Β 146 (cp. Ν 78, δ 712), intransitive Hose' ώραρε θε~ιος αοιδός θ 539. 22) πε-παγο-ίη-ν is quoted from Eupolis by the Scholiasts on St 241 among aorist forms. There seems to me to be no good ground for doubt- ing such a form, as is done by Buttmann A. Gr. II 2 273 and Ahrens Dor. 330. 23) πε-παΧών. άμπεπαΧών προίει ΰοΧιχόσκιον εγχρς Γ 355. 24) πε-παρεΊν, a reading well supported and accepted by Boeckh, Bergk, and T. Mommsen in Pind. Pyth. ii. 57 εΧευθερα φρενι πεπαρεϊν, with the meaning of άποΰεΊξαι : cp. Hesych. πεπαρε~ιν • ένδεΊζαι, σημήναι, πεπαρεύσιμον, εΰφραστον. Boeckh in his notae criticae rightly recognised in this word a stem akin to the Lat. apparere, but having nothing to do with πορεΊν. 25) παφών, occurring only in Hesych., and discussed on p. 291. 26) πε-πιθεΊν. πε-πιθέ-σθαι, quoted from Homer, Pindar and later 1 ice -κάμω a reading, rejected by Aristarchus and Herodian, for itrei /ce κάμω A 168. Cp. Η 5 eW /ce κάμωσιν, Ρ 658, and La Koche Textkritik, p. 295. 294 THEMATIC AORISTS. ch. xin. poets : πέπιθεν φρένας Hymn, in Apoll. Pyth. 97, πεπίθωμεν I 112, 30 πεπιθεΊν ib. 184, πεπιθών Pind. Isthm. iv. 90 πεπιθονσα θύελλας Ο 26. — ουκ αν h) τις άνηρ ττεπίθοιθ' εφ αυτοϋ θυμό) Κ 204. 27) ε-πέ-πληγο-ν ποίες Ίππων Ε 504 and elsewhere in Homer. Other forms, some middle, have been mentioned above, p. 290. 28) πε-πορεΊν ' ΰοΰναι Hesych. cp. πορε~ιν p. 286. 29) πε-πυθέ-σθαι, ει κεν έμέ ζωον πεπυθοιτ έπι νηνσϊν Αχαιών Ζ 50, Κ 381 (cp. Λ 135). πεπύθωνται' ακοΰσωσιν Άττικώς (?) Hesych. Ari- starchus recognises the reduplicated form in Ζ 50. 30) ε-σπέ-σθαι has so completely coalesced with the non-reduplicated σπέ-σθαι, that it can only be clearly recognised in forms which do not admit the augment, like άμ' εσποίμην τ 579, άμ εσπέσθαι Ε 423 (accord- ing to the cod. Ambros. ; other M.SS. άμα σπέσθαι), and still more indu- bitably in Pind. 01. viii. 11 ψτινι συν γέρας εσπητ άγλαόν. 31) τε-ταγών, ποΰος τεταγών Α 591 (cp. Ο 23), with Lat. tango, old Lat. conj. tag am [Stud. v. 431]. 32) τε-τάρπετο, αυτάρ έπεϊ φρεσιν 1\σι τετάρπετο ΰαίΰαλα λενσσων Τ 19, τεταρπώμεσθα γόοιο Ψ 10, τεταρπόμενος α 310. 33) ε-τε-τμο-ν in Homer, Hesiod and later poets always in the meaning £ come upon,' ετετμε Δ 293, οψρ ετι οίκοι άμύμονα μητέρα τέτμης ο 15, τέτμοιμεν Theocr. XXV. 61. — ε-τέτμετο άσπετος άλμη Orph. Arg. 366, where on the ground of the difference of meaning Ruhnken and G. Hermann read ετέμνετο.' 34) τέ-τορε -V ετρωσεν, έτέτορεν' έτρνπησεν, τετόρη' τρώση Hesych. Cp. ε-τορε. 35) τε-τνκείν only in ο 77, 94 in the phrase ΰεϊπνον ένι μεγάροις τετνκεΊν : the middle is common in Homer : τετύκοντο ΰε ΰαΊτα A 467 etc. Cp. Callim. Dian. 50, τετυκοίμεθα ^όρπον μ 283, τετνκέσθαι Φ 428. 36) τε-τνπόντες only Callim. Dian. 60 ραιστήρες η χαλκον — ηε σίΰηρον άμβολάΰις τετυπόντες. Ο. Schneider (ad Callim. i. p. 150) accents τετΰποντες, and takes the form as a present-like perfect like κεκλήγοντες, but a perfect τέτυπα is quite unknown. 37) τετύχησι, τετνχοιεν quoted only from very late poets (Maximus, Manetho). 6ut τετύχησι was recognised even by Aristarchus and his commentator Aristonicus on Λ 116 as a variant for (ε'ίπερ) τ ε τΰχησι. 38) πε-φιΰέ-σθαι, Homeric : πεφιΐοίμην ι 277, πεψίΰοιτο Ύ 464, πεψι- ΰέσθαι Φ 101. 31 39) ε-πε-φνο-ν poetical : ον επέφνομεν ημείς Κ 478, πέφνεν ματέρα Pind. Pyth. xi. 37, Soph. Ο. Τ. 1497. Other forms only Homeric : ήν τίνα πίφνη Υ 172, πεφνέμεν Ζ 180, καταπεφνών Ρ 539, πεφνόντα Π 827 : Aristarchus and Herodian accentuated πέφνων : cp. Heroclian on Π 827, Herodian ed. Lentz i. 470. In Oppian Hal. ii. 133, v. 390, πέφνρυσι as 3 pi. pres. 40) ί-πέ-φραΰο-ν only in Homer and Hesiod : επέφραϊε Π 51, πιφραΰίτην Hes. Theog. 475, πεφράίοι Ά 335, πεφραΐέμεν η 49. Cp. φράΖεν. 41) κε-χάρο-ντο Π 600, κεχαροίατο Α 256, κεχαροίμεθα Philox. Fr. 2, 24 Be. 3 , cp. χάροντο and κεχαρήσω. At the close of this survey we may point out how large a number of these reduplicated aorists have by-forms without reduplication, not less than 19 out of 41, i.e. ΰέΰακε (7) and Ζαχεϊν, τέθιγε (11) and θιγΰν, τέθορε (12) and θορεΊν, κεκνθωσι (15) and κυθεΊν, λελαβέσθαι (16) and ch. χπι. KEDUPLICATED AOEISTS. 295 Χηβέσθαι, Χέλαθον (17) ΧέΧακον (18), Χέλαχον (19), and λαβείν, λαλεϊ>>, λαχεΐϊ', μεμάποιεν (20) and μαπέειν, ωρορε (21) and ώρετο, πεπνθέσθαι (29) and πνθέσθαι, εσπέσθαί (30) and σπέσθαι, τετάρπετο (32) and ταρ- πωμεθα, ετετμον (33) and ετεμον, τέτορεν (34) and ετορε, επέφραΰε (40) and φράΰεν, κεχάροντο (41) and χάρονΓο : and to these we may add κέκΧυθι, κέκλντε formed directly from the root, as mentioned on p. 290, by the side of κλνθί and κΧντε. We have often pointed out under the several heads the prominent peculiarities of usage. We expect to find always in the stronger form a more forcible meaning. An intensive force may be detected most clearly in the Homeric use of εκέκΧετο by the side of κέΧετο, in κέκλνθι as compared with κΧϋθι, and in the two 'aorists of rebuke' ένένιπε and ήνίπαπε. The isolated aorist τεταγών too has certainly derived its forcible meaning of c seizing ' as compared with Lat. tangere by the help of reduplication. We might conjecture the same for αμπεπαλων. In the case of ΧεΧάκοντο, as we have already intimated, the meaning { howl ' as compared with ΧακεΊν ' to sound ' depends upon the same. The force of ακαχεΊν, άραρεϊν, ΰεόαεΊν, κεκαΰεϊν is decidedly causative, and the same is especially clear in όρορεΊν, in ΧέΧαχον and ΧέΧαθον. In the 32 case of ΧεΧάχωσι Aristarchus remarked on Η 80 ( αντί τον ΧαχεΊν πυιή- σωσι.' The isolated πεπαρεΊ.ν is to apparere as ΧεΧαχεΙν is to ΧαχεΊν. There were often departures from this usage in particular instances. But if we compare the causative force of this device of language in the Sanskrit aorists, and in presents like 'ίστη-μί=8Ϊ8ίο its use can hardly be a matter of chance. — In ετετμον the usage has been otherwise differen- tiated, for the reduplicated form is limited to the meaning ' hit upon.' An anomaly of tolerably wide extent, which the reduplicated aorist shares with the thematic, is found in the occurrence of an α instead of the usually alternating vowels ο and ε. This appears partly only in the post-classical language, but partly also in the language of earlier times, and even in that of Homer. I mean forms like είπα, ήνέγκαμεν for είπον, ήνέγκομεν, and others which made their appearance in Alexandrine or even later times (e.g. εΧαβαν). The right view of ε-ΰωκα, ε-θηκα, ήκα is also connected with this question. But as this whole process rests upon a confusion of the analogies of the thematic and the sigmatic aorists, it cannot be more fully discussed, except in connexion with the sigmatic formation (Chap. XVII.). 296 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiv. CHAPTER XIV. THE MOODS OF THE PRESENT AND SIMPLE AORIST STEM. After our survey of the formation of the present stem and the simple aorist stem, each in its twofold form, with or without the thematic vowel, it remains for us to discuss the signs of the moods and of the verbal nouns. In this chapter we have to do with the moods ; and I place first the Imperative, as being that mood which shows the least difference from what is conjecturally the most primitive form of the indicative. 33 I. IMPERATIVE. According to the usual logical scheme, the imperative is here placed on the same line with the conjunctive and optative. But in formation the imperative is completely different from the two other moods. In the latter the modal element comes in between the stem and the ending, and hence in the same place in which we found the numerous expansions of the stem in the case of the formation of the present : indie. X~pev conj. Χ-ο-μ^ν ,, φερο-μςν opt. φέρο-ι-μεν. We see at once that there is an analogy between the expansion of the verbal stem to the present stem, and the formation of conjunctive and optative forms. On the other hand the imperative is either not at all distinguished from the indicative, e.g. in φέρε-τε, φέρε-σθε, or it is dis- tinguished only by the different form of the personal endings. The place of the imperative formation is to be sought here, i.e. solely in the last syllable of the verbal forms : ' ind. ΐ-στη -s imper. ΐ-στα-θι „ *φίρε-σι „ 0epe „ φ€ρ€-σαι „ φί pe- n -η to use here the most common and indubitable original forms by the side of those actually in use. The second persons of the plural and dual in the active and middle : φέρετε, φέρετοί — φέρεσθε, φερεσθον are not at all to be distinguished from those of the indicative. In the case of the 2 plur. we might indeed conjecture, after the analogy of the distinction in Latin between fer-tis smafer-te, that the Greeks too were not unacquainted in earlier times with this distinction between the indicative and imperative. We may refer for a conjectural *φέρε-τες to p. 45. But there is little probability in this conjecture, for it is only from Latin that we can argue to such a distinction ; and fer-tis, legi-tis show high antiquity. The loss of a final 8, in Latin limited to the imperative, extended in Greek also to the indicative. Sanskrit has in the 2 pi. of the indicative -tha, in the same Ch. xiv. SECOND SING. IMPERATIVE IN -θι. .297 form of the imperative and of the historical tenses -ta ; so that here a 34 distinction arises between bhara-tha 'fertis' and bhara-ta 'ferie'; bnt this distinction is not one which anyone would be inclined to maintain had been created for the special meaning of these two forms. The same is the case with the middle. Greek knows no distinction between φερε-σθε as an indicative and as an imperative, just as little as in this case Latin knows with its ferimini. In Sanskrit the imperative has the secondary ending -dhvam, so that here too the indicative bhara-dhve is distinguished from the imperative bhara-dhvam. The weaker form in Greek made its way into the indicative, as it did also in the active. The second persons of the dual e.g. φερε-τον, ψέρε-σθον also have to serve for indicative and imperative alike, while Sanskrit, just as in the plural, assigns weaker endings to the latter mood, and so arrives at a separation between indicative and imperative. Now as the separation of imperative from indicative forms extends much farther in the 2 sing., it seems to me probable that the stamp given to the mood began here, and that starting with this it was by degrees attempted rather than carried out in the case of other imperative forms. Even in the 2 sing, the Yedic dialect often uses the indicative in place of the imperative form, e.g. ma-si as well as md-hi and the like, as Delbriick points out p. 34. Hence our investigation of the imperative forms limits itself essen- tially to three points, the formation of 2 sing, in the active, that of the same person in the middle, and the forms in -τω, -τω-ν, -σθω, -σθων (which cannot be separated one from the other), together with anything which may be connected with them. A) Formation of the 2 Sing. Act. 1) Termination -θι. This ending is entirely limited to the primitive conjugation, and hence never appears after a thematic vowel. The same is true of the corresponding Sanskrit termination -did and its weaker by-form -hi. The agreement here is therefore complete, and we may put a number of Greek imperatives in -θι side by side with Sanskrit ones, without finding any other differences than those which proceed from the phonetic laws 35 and tendencies of the two languages : e.g. i-hi m 'ί-θι pi-pr-hi qru-dhi pa-hi = (€μ)πί κλν-θι Aeol. -πλη-θι πώ-θι ' drink, If there were active forms corresponding to τά-ιν-ται (p. 113), the im- perative would necessarily be *τά-ιν-θι (cp. ομιν-θι), which would correspond exactly to the Yedic tanu-hi. e-dhi ' be,' coming with an unusual change of sound from as-dhi, corresponds to the Gr. ίσ-θι, for which Hecataeus used the regular form εσ-θι (Herodian ii. 355). Quite in the same way Zend, in which the aspirate dh is regularly represented by d, forms from rt. i the imper. i-di='i-Qi, from gd gai-di (with an epenthesis of ί)=βή-θι, from zd 'know' (by the side of zan=Qr. γνω) ζ-όι=γνώ-θι. It is seen at once from these comparisons that the treatment of the 298 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiv. root- vowel is not always the same. Sometimes it appears long, some- times short. There is the same variation as was discussed on p. 96 and in the case of the aorists on p. 135. In the aorists, where the form is retained complete, the vowel is as a rule long : βή-θι : θ 492 (Lacon. κά-βα-σι Hesych.), γνώ-θι common, ΰϋ-θιΤΙ 64, κλϋ-θι Homeric, also in the tragedians, ττΊ-θι (Arist. Vesp. 1489), άπό-πτη-θι 'fly off' (Philostr. Epist. 11), στή-θι Ψ 97 (Lacon. άττασι ' ανάστηθι Hesych.), τλη -di (Soph. Philoct. 475). Presents with a long vowel, and hence after the analogy of the indicative, are furnished in the singular by ΰίΰω-θι and 'ίλη-θι γ 380 αλλά ανασσ' ΐληθι, δίδωθι δε μοι κΚίος έσθλόν, cp. ϊληθ' 7τ 184. For the latter later poets, e.g. Theocr. xv. 143, Calli- machus Hymn. vi. 139 have 'ίλάθι. We may add εμπίπληθι ψ 311, and ζηθι, quoted from the γνώμαι of Menander v. 191, a line which can hardly have come down to us correctly. A short vowel appears in the ordinary ϊθι, in the Attic φά-θί, which is used from Aristophanes downwards, in ελλαθι discussed on p. 119, in ομννθι (Ψ 585) and in ύρννβι (Ζ 363). 2) s for θι. Six aorist imperatives in ς, i.e. Ιός, θες, ες, σχε'ς, φρες (εκφρες Aristoph. 36 Yesp. 162) and the Homeric ενί-σπες (Λ 186 and elsewhere) are generally explained as arising from the primitive forms *ΰό-θι, *σχέ-θι etc. by the loss of ι and the change of the Θ, which cannot stand as a final letter, into c. But in no language are there primitive forms of the kind assumed, and it deserves to be carefully noticed that by the side of the present φά-θι, which may be to a certain extent compared, there is no trace of any form *φά-ς. And the only certain instances of the change of a final r into ς are προτί προς, *βεβηκότ βεβηκός. The origin of the local adverbs ενΰος and εζος, quoted by grammarians sometimes generally as Dorian, sometimes specifically as Syracusan, from hdodi and *ε£,οθι, is as Ahrens Dor. 366 justly notices, by no means clear. Hence I do not regard the ordinary explanation of these imperatives as absolutely certain. Since in the dialect of the Yedas the termination -si in the case of primitive verbs is not quite unknown to the imperative : e.g. tfe-shi * conquer,' ma-si ' measure,' jd-si l go,' it would not be impossible that ΰός should be shortened from *£-c from *didw-ni = $kt. da-dd-si. This view might also be supported by a Latin analogy in the 2 sing, imper. es from the root ed, which as Neue Formenl. ii. 2 603 shows, is certainly established by one instance in an inscription and several in Plautus. Merguet in ' Die Entwicklung der lat. Formenbildung ' p. 244 suggests doubtfully that this form ' descends from a time when this imperative still ended with *a dental, before which d then passed into s\ that is, that es originated in *ed-di, *es-c?i=Skt. ad-dhi. But as there is no trace of any termination -di on Italian soil, and as there is little probability in the group sd, it is perhaps simpler to think of an indicative *essi es which was used at the same time as an imperative, and to assume the like in the case of es also. Still we cannot consider the older view of the Greek imperative forms in -ς as one to be at once rejected. ch.xiv. SECOND SING. IMPERAT. WITHOUT ANY TERMINATION. 299 3) Imperatives without any termination from verbs without a thematic vowel. Some of these imperatives have shorter by-forms without any personal ending, as ενι-σπε (t> 642) by the side of ε j •ί-σπες, which was mentioned above p. 132, κάθ-ε • with the strange explanation επίΰος Hesych., σχέ established 37 by the metre in the oracle quoted by Schol. on Eurip. Phoeniss. v. 638 (ed. Dindorf ill. 178 1. 1) τήνΰε συ ηγεμόνα σχέ περιτρίπτοιο κεΧενθον, επί-σχε, the reading of the M.SS. Hes. Scut. 446, where there is no occasion with Goettling to read the present επ-ισχε, inasmuch as the aorist is better Suited to the passage (' Αρες, έπίσχε μένος κρατερόν καϊ χείρας άάπτονς), κατά-σχε the reading of the M.SS. in Eurip. Here. Fur. 1210, for which Elmsley and after him Dindorf and Kirchhoff read κατά-σχεθε,παρά-σχε the traditional reading in Eurip. Hec. 842 (πιθοϋ, παράσχε χείρα τφπρεσβντιΰι) for which παράσχες is now read. Porson justly defends παράσχε, though he judges otherwise on Orest. 1330. Finally the Bodleian with other M.SS. has the same form in Plato's Protagoras p. 348 a. Goettling (Allg. Lehre vom Accent p. 45) recognised that in these forms the accent properly falls on the penultimate. Recent editors have been reluctant to accept any of them, and in Attic prose there are difficulties in admitting anything so iso- lated, contrary to the prevalent usage, so long as it is not abundantly esta- blished. But the case is different with the language of poetry. The question is whether in such forms the termination is to be supposed to have dropped off, in which case they would serve to confirm the assumption of the origin of the ς in θι : σχέ would then be to σχέ-ς and the original *σχέ-θι as ούτω to ούτως and the hypothetical *οντωτ. But if we take into consideration forms like παρέσχυ-ν, παράσχω, παράσχοιμι, to which παράσχε bears just the same relation as πάσχε to επασχον etc., and reflect how strong the tendency is universally to confuse the primi- tive formation with the thematic, it seems to me incomparably more probable that we have to admit this explanation also in the present instance, that is, to regard παράσχε etc. just like παρασχόντων etc. as thematic formations. We may mention here two other imperatives without terminations, which present themselves, apparently very anomalously, by the side of the regular βή-θι, στή-θι, — the well-established forms */3d and *στα, occurring however actually only in composition : εσβά Eurip. Phoen. 193, εμβα Eur. El. 113, επίβα Theogn. 847, κατάβά Yesp. 979, πρόβα Ach. 262 — άνστά Theocr. xxiv. 36, παράστά Menander (Comici iv. 105 no. II.). 38 As in Attic writers α in place of η would be unprecedented, we cannot suppose them to have originated from βά-θι, στα-θι by a loss of -θι, and must rather assume here too a transition to the thematic conjugation. The forms quoted are not, so far as their formation goes, primitive aorist imperatives, but thematic present imperatives from βάω (discussed on p. 148) from which προβώντες inter alia is quoted from Cratinus (Com. ii. p. 88). Certainly we have no authority for *στάω, but it corresponds exactly to the Latin sto, so that the imper. sta is* completely identical with Gk. στα. This explanation, suggested already by Lobeck on Butt- mann ii. 2 125, of course does not exclude the assumption that these forms were syntactically accounted as aorists, because the corresponding indicative presents were obsolete. This view is confirmed by the form 300 MOODS OF THE PKESENT AND SIMPLE AOEIST STEM. ch. xiv. εμβη which appears in Ar. Lys. 1303, though not without the variant εμβα. εμβη occurs in the Laconian final song ; it can only be explained as a Doric form by supposing it is contracted from *'έμβαε. Whether the form πώ by the side of πώθι, quoted in Et. Magn. p. 698, 52 from an Aeolic poet (χαίρε και πώ) originated in the same way, depends upon the judgment we form upon present forms with a like ending. We now turn therefore to those present imperatives of the conjuga- tion in -μι, which have altogether lost the syllable θι. Such forms are certainly not less numerous than those which retain 0t, especially if we include the Aeolic and Doric dialects. After what we have just said with regard to the transition to the thematic conjugation, it is natural to suppose that all these shorter imperatives could be explained from the analogy of this conjugation. But we shall not find this sufficient. A form like e.g. ϊστη in Homer, κρημνη in Euripides cannot possibly be explained, in accordance with the laws of Ionic contraction, as from *'ίσταε *κρήμνα€, any more than the Dotic εγκίκρα from *εγκίκραε, for here αε is contracted to η. Hence we divide all the forms which belong here into two classes ; i.e. into apocopated and thematic forms. a) Apocopated forms. The loss of the syllable θ ι finds its analogies in various phenomena of 39 the 1 and 3 sing. ind. and conj., which were mentioned on pp. 28 and 41. The Aeolians of Lesbos formed the 3 sing, by rejecting the per- sonal ending : γελαί, and the 2 sing, imper. in the same way : κέντη. The Aeolic forms of this kind are discussed by Ahrens Dial. Aeol. 140. The grammarians quote as Aeolic ίστα and 'ίστη, εμπίπλη, κεντη, μνρω, ΰίϊω. Ahrens's wish to reject ίστη, which is described as Aeolic in Et. Gud. 283, 40 and in Et. M. 348, 9, arises from a mistake. He is correct only so far, that Ίστη cannot be explained like ϊ στα from a mere rejection of the ending. By its η ϊστη in the case of the Dorians and Aeolians — for it is called Dorian too by Herodian ii. 209 — is shown to be contracted. According to the Dorian and Aeolian contraction it might come from ΐσταε. Hence the identical forms in the different dialects are, remarkably enough, to be explained differently : the Dor. and Aeol. '/στα, and the Ion. ϊστη are apocopated, the Dor. and Aeol. Ίστη and Ion. 'ίστα are contracted. — From Aeolising poets we may quote Ιάμνα Sappho i. 3, κίνη Sappho Frag. 114 Β 3 , νμάρτη Theocr. 28, 3, φίλη 29, 20. — A Homeric form of this kind is presented by Ίστη Φ 313, which reappears in Eur. Suppl. 1230 and Ar. Eccl. 743 (καθίστη). We mayaddf;p///iy7/Eurip. (or rather, as Nauck conjectures, Eupolis) Frag. 918 {κρημνη σεαντήν εκ μέσης άντηρίΰος), πίμπρη Eur. Ion. 974, and εγκίκρα Sophron Fr. 2 (Ahrens Dor. 464). The isolated Attic imperative έ'ξει (Ar. Nub. 633), omitted in our grammars, with which the Scholiast on this passage compares ΰίει and μέτει, is perhaps to be taken in the same way. For it seems to me better to assume an *είθί formed after the analogy of ΰίΰωθι as the original form rather than an ε'ίω, unknown at any rate on Attic soil, and with which we could only compare the conjunctive ε'ίω in Sophron (Frag. 2 Ahrens). If εί as an imperative really originated from the thematic form, it would necessarily have been contracted from *ε7ε, like ΰαϊ = £αΐε in Hesychius, and would completely correspond to the Lat. i (for ei : cp. Imvs). ch. xiv. EEGULAR THEMATIC IMPERATIVES. 301 b) Thematic Forms. Considering the general tendency of the primitive forms to pass into thematic forms, imperatives like δείκνυε (Hes. Opp. 502, Plato), ολλυέ Archiloch. 27 (και σφεας ολλυ ώσπερ όλλυεις,) ομννε (Theocr. 27, 34) 40 present nothing surprising. It is otherwise with ζνν-ιε, which we find in Theognis 1240 at the end of a pentameter. If we are not to correct this into ξννίει, as Buttmann (Ausf. Gr. i. 2 523) suggested, we must recognise here, not indeed the addition of a thematic vowel, but probably the transition of a radical ε into such a vowel. We have clear cases of contraction from forms with an added vowel in τίθει, which is in general use from Homer (τίθει κράτος A 509) onwards, tit (Φ 338, Eurip. El. 592, occurring also in compounds in Attic prose), ΰίδον, which is found from Herodotus iii. 140 onwards. We must add also those forms in which the simple vowel is only thus intelligible, as καθίστα (i. 202), πίμπλα (πίμπλα συ μεν εμοί Xenarch. Meineke Com. iii. 616), Dor. πίμπλη, Herodian i. 464, ΰαίνϋ i. 70, ομνν (Soph. Trach. 1-185, Eurip.), στόρνϋ (Aristoph. Pax 844), στρώννϋ (Com. anonym. Meineke iv. 605). The most difficult form of all is the Pindaric ΰίΰοι (Pind. 01. i. -85; vi. 104; vii. 89; Nem. v. 50). It can only be understood in connexion with the Aeolic ΰίΰοις and the Homeric ΰιΐοϊσθα, and points by its diphthong to a *ΰιΰοίω as an older by-form of *ΰίΰόω (cp. p. 238), which has arisen from a transference into the deriva- tive conjugation. We might assume as the primitive form a * da-da- jd-mi, though this hardly has a parallel. Bopp's notion that ΰίΰοι might have arisen by the loss of θ from ΰίΰο-θι (Ygl. Gr. ii. 2 290) breaks down upon the improbability of such a loss, for which the Greek language offers no analogy. Still less can we follow Bopp in explaining the origin of ΰείκνϋ in this way, and in actually tracing back v.tovt. Nor can we say, as I supposed in Tempora und Modi p. 21 (cp. Kiihner Ausf. Gr. i. 2 p. 524) that ' in the lengthening of the vowel of 'ίστη, ΰίΰον, ΰαίνϋ we must recognise a trace of the original ending 0t.' We do not now look upon ' compensatory lengthening ' in this superficial way, and cannot allow that a lost syllable ever had the power of lengthening the pre- ceding syllable. This brief mention of differing views will suffice. — With regard to the occurrence of many forms here noticed as contracted, we 41 may refer also to Cobet Mnemos. ix. p. 373 and von Bamberg in the Zeitschr. f. Gymnasialwesen xxviii. p. 27. 4) Eegular forms of the thematic conjugation. The ordinary second person of the imperative of the thematic conju- gation has no personal ending. All languages, which have an imperative at all, agree in this. Compare Skt. bhara Zd. bara Gr. φέρε Lat. fer „ gaKha „ βάσκε In spite of this Bopp (Ygl. Gr. ii. 2 291) and Schleicher (Compend. 3 654) assumed that here too there was originally the ending -dhi (Gr•. θι), and that it was afterwards dropped, as in a part of the verbs in -μι. Delbruck pronounces against this view (Yerbum p. 33) as follows : ' we never find a Hharadhi, but only bhara. We cannot, I think, doubt 302 MOODS OF THE PEESENT AND SIMPLE AOEIST STEM. ch. xit. that in these words the simple present stem was used from the first with an imperative force.' And we can well imagine this, for just as an exclamation (Ausruf) even without any special sign in the vocative of the noun may become a call (Anruf), so the bare stem used as an address (Zuruf) may in the verb become a command. Inasmuch as the stem of thematic verbs, as we saw, does not at all differ originally from a noun-stem, sometimes, so far as the sounds are concerned, an imperative and a vocative do completely coincide : e.g. aye=Skt. aga is only distinguished from the vocative άγε by the accent, a distinction which vanishes in Sanskrit, where the vocative rejects an accent on the final syllable. As the present stems in ww=Gr. vv are also in origin noun-stems, we might regard in the same way forms like the Skt. cr-nu hear (by the side of cr-nu-hi). However, for Greek and Latin we are not wholly without traces of a termination, afterwards lost, for the thematic forms here under consideration. We have, it is true, only a single instance of the kind well established for each of the two languages, for Greek άγες * άγε, φέρε in Hesych., for Latin prospices, prospice in Fest. p. 205. There 42 are also one or two doubtful instances. For the senseless ζατές • ζητέι in Hesych. Guyet, followed by Koen on Gregorius Corinth, p. 620 read ζατές • ζητεί ; there is also ζόες ' ζή for which perhaps we should write ζοές • ζή. From Latin we must also mention perfines, perfringas, quoted by Festus on the same page. Bergk in the Index Lectionum Marburg 1847-48 most decidedly regards the Latin glosses, which he looks upon as taken from the Carmen Saliare, as imperatives with personal endings. Corssen, who was formerly (i. 2 286) of the same opinion, afterwards (ii. 2 474) explains prospices and perfines as optative forms. But we cannot see from what sort of stems they are to come. For as forms of prospicere and perfinere, which is probably only another form of perfindere, they can only be futures. Besides, with this explanation we should have again to assume for prospices a by-form of the present *prospico as well as prospicio, and we cannot see why the word, if that was its force, should not have been explained by prospicias. It seems to me that the Latin prospices and the Greek άγες lend each other mutual support. The Greek form is derived by Koen 1. c. from *άγημι (cp. ήγέομαι), and he supports by this his view of the gloss ζατές, to which ζοές must be added, if it is really to be taken as an imperative. However, we cannot see what is to be gained by the analogy of the conjugation in -μι. For an Aeolic present imperative to *άγημι would be *αγη not άγες; an imperative present in ς is just as unheard-of from verbs in -μι as from verbs in -ω. It seems hence much more probable to me that we have in both languages isolated relics of a formation of thematic verbs, which early became extinct. From a purely Greek standpoint we might* be tempted to refer άγες to *άγε-0«, and to find herein a confirmation for the derivation of Ιό-ς from Ιό-θι, and for the assumption that thematic imperatives also were not originally without the termination. But άγε-ς can hardly be separated from prospice-s, and as in Latin the s cannot have come from an original dh (=Gr. Θ), and hence the Latin s can only be regarded as representing the termination -si, we shall apply the same explanation to άγε-ς. The form belongs to the same category as the Vedic forms like ma-si measure, and others mentioned on p. 298, 43 only that the i has here been dropped without leaving any traces, just as in φής, 'ίστης etc. and in the Doric λε'γες=λε'γεις. For we can hardly ch. xiv. IMPERATIVES IN -τω 9. 303 be satisfied with the so-called ' false conjunctives/ as Delbriick calls the unaugmented preterites of the Yedic dialect, used with a modal force, in the entire absence of similar formations on European soil, ζατες, ζοές, if correctly reported, are Doric forms, which are related to the imperative άγες precisely as the Doric indicative ποιες is to λέγες (Ahrens Dor. 176). Perhaps άγες was also limited to the same dialect, and was thus an indicative used as an imperative. Grammarians (especially Ε. M. 302, 36) mention as Syracusan the forms θίγον, Χάβον, άνεΧον for θίγε etc. cp. Ahrens Dor. 304. The ν is probably here, as in the imperative of the sigmatic aorist only a mean- ingless nasal after-sound which had the duller vowel as its effect. We took a similar view on p. 67 of the personal endings -μεθεν and -μεθον by the side of -μέθα. We are excluded from supposing a confusion of the thematic with the sigmatic aorist by the fact that this confusion, of which we shall have to treat below, is not regarded as Syracusan -at all. Finally we may mention here a few more forms in which the thematic vowel suffers contraction with a preceding long vowel or diphthong. Here belongs Xov * Χονσαι 'Αττικοί Hesych., where Mor. Schmidt alters λοϋσαι into λοϋε, for outside the imperative contracted forms like ελου, Xovrai, λοϋσθαι are common. We find also δαϊ, £α<ε, for which see Mor. Schmidt on Hesych. i. 453. He also conjectures that for χρ~ι • χρίει we should read χρϊ '• χυϊε. In Photius Lexicon, p. 348, 1 (ed. Porson) we read : παν • to παϋσαι μονοσνΧΧάβως, from which W. Dindorf, Meineke, and Bergk in Aristoph. Equ. 821, for παν οντοσί of the M.SS. read with Elmsley παν παν οντος. Mor. Schmidt on Hesych. iv. 297 mentions also the form παΊ for παίε, but I cannot find any authority for this. These forms,. in which the thematic vowel coalesces with the preceding one are really not much more surprising than τιμά or ζή (for *ζή-ε). To these belongs also the isolated Homeric imperative τή take,there ! {Ε 219, Ψ 618, κ 287), the plural of which τητε is given by Sophron (fr. 100 Ahrens). W. Dindorf Annot. ad Aeschylum (Oxon. 1841) p. 318 is bold enough 44 on the strength of these isolated instances actually to regard Ziaih• for ΰίαινε as possible in Aeschylus, although this process is not an apocope so much as a rare kind of contraction, as we may see from the forms of λονω. The same scholar is still less justified in adding the dictum * sponte intelligitur apocopen hanc ibi tantum locum habere, ubi idem verbum repetatur.' No authority says a word of this. Is the Latin apocope in die, due, fac limited to repetitions 1 From the established forms of this kind light is thrown back upon the contraction in ϊστά 6Χλϋ etc. 5) Imperatives in τως. Latin possesses two forms for the 2 sing, imper. ; by the side of the usual form without any ending, it has that in -to, which is distinguished by a slight shade of meaning from the former. This -to is identical in form with the only ending of the 3 sing, imper. in Latin and Greek, and the same holds good of the Vedic ending -tat. We have every reason to regard this tat as the primary form, and its final t is still preserved in the third person in forms like the Oscan likitud and O. Lat. estod (Fest. p. 230). We may look upon the fact that the same ending appears in both persons as analogous to the case of the dual -top (cp. p. 51 f.). Probably td-t for the second person may be referred to an emphatically 304 MOODS OF THE PBESENT AND SIMPLE AORIST STEM. ch. xiv. reduplicated -tva. Forms of this kind are abundantly established by Delbruck (Verbum 38), especially for the second person, e.g. voftatat call, vahatdt carry, bhavatdt be. In Zeitschr. viii. 297 I called attention to the fact that the Greeks were not without a corresponding form, and Bergk de titulo Arcadico p. ix. tells us that the same idea had also occurred to him. The gloss of Hesychius ελθέΥως • αντί τοϋ ελθέ Σαλά- μι νο'ι (read Σαλαμίνιοι) is at once explained, if we suppose that the final t here, just as in the ablative sing., changed into ς. I think I have since discovered a second instance of this formation, which had disap- peared from literary usage, in the gloss φατώσαν * γνώθι, proposing in Studieniii. 188 to read without the alteration of a letter φα τώς • άνάγιωθι, 45 and supposing that φατώς, in somewhat the same way as λέγε, was especially applied to the utterance of what was read. Possibly in the dialects which possessed these forms a distinction was made possible between the second and the third person, by the retention of the old dental in the form of ς in the former and its loss in the latter, a distinc- tion maintained neither in Sanskrit nor in Latin. The somewhat sur- prising accent, if we may regard it as recorded accurately, may be com- pared with the tendency of single imperatives like έλθέ, είπε, Χαβά, ίΰέ to become oxytone. Β) The Second Person of the Middle. The termination of the second person of the middle and its conjectural origin were discussed on p. 59. The earliest demonstrable Greek ending is -σο, which here corresponds entirely to the Skt. -sva, whilst this ending is foreign to Sanskrit in the preterite and optative. Hence bhara-sva and φέρε-ο, φερον saJca-sva „ eWe-o, έπου saha-sva „ εχε-ο, έχον correspond exactly in form and meaning. "We also discussed above (1. c.) the conditions under which -σο has been preserved unaltered. Instances of σο retained in the case of verbs in μι are άρνυ-σο (frag. trag. adesp. 3 ed. Nauck), έπίστα-σο Aesch. Soph. Herod., ϊστα-σο from Hesiod onwards, and in compounds not unknown to prose, εσσο ' be ' α 302, y 200, and (quite late) εσο thence derived (cp. πάρεσο ' παραγενον Hesych.), ησο Γ 406, κε'ισο Φ 122, ονησο τ 68, εμ-ττλησο Aristoph. Vesp. 603, πρίασο Ach. 870, τίθεσο Pax 1039.— σοϋσο ' ψχον, πορενυυ Hesych. (cp. Ahrens Dor> 352) Forms, with the σ omitted, but without contraction, are common in Homer, as ΰειΰίσσεο Δ .184, ε'ιρεο a 284, εΚπεο Τ 201, ερχεο Ο 54, θέο * 333, λίσσεο, ψε'ιίεο, χώεο, μάρναο Ο 475, ψάο σ 171. Similar instances from elegiac and iambic poets are pointed out by Kenner Stud. i. 2, 30. Pindar has πίθεο Pyth. i. 59, άναβάλεο Nem. vii. 77. There is abundant evidence in Herodotus for forms like προσΰέκεο iii. 62, άναπανεο v. 19. 46 The contraction of εο into εν often occurs in Homer as in other Ionising poets and in Herodotus, by the side of the uncontracted forms : βάλλεν μ 218, νπόθεν ο 310, επευ Ν 381, ερχεν Ζ 280, ενχεν Ω 290, ΐζεν Γ 162, πείθεν Ε 235— άλέ£ευ Archil. 66, 2, εντρέττευ Theogn. 400, βάλεν Theogn. 1050 — Ηκεν Pind. ΟΙ. ίν. 8 — τέρπεν Herod, ϋ. 78, άνέχευ i. 206. — As the contraction of to into t ν is not unknown to some branches of Dorism, there is nothing surprising in κάθεν in Callimachus h. V. 140. ch. xiv. IMPERATIVE. — THIRD PERSON SINGULAR. . 305 We need give no instances of the Attic custom of contracting εο into ου in thematic verbs. As for the verbs without a thematic vowel, contrac- tion made its way completely through the short aorist forms, e.g. άφον Soph. 0. R. 1521, θοϋ Soph. 0. C. 466, άνάσχον Eurip. Ion 947, but it also very soon attacked the presents : τίθον Aesch. Eum. 226, though without consistency, so that in the comic poets even in the place of κάθησυ — apparently through an intermediate form *καθεσο — κάθον in- truded itself (Veitch, p. 307), and similarly σον 'hurry' from the indie. σοϋμαι (Aristoph. Vesp. 209). — In non-thematic stems in α contraction also came to be usual : 'ίστω Soph. Philoct. 893. Another rare contraction within the word brings about ft from the union of an ε belonging to the stem with the thematic e. Thus αιΰε~ιο from αΐΰεεο Ω 503, ι 269, both times almost in the same for- mula (cp. μνθεΊαι, νεΊαι), and νεϊο, quoted by Veitch, p. 409, from Leonidas of Tarentum 70. In the same way from the primary form άλάεο by interior contraction came first *άλάο, then άλύω (ε 377), for as the contraction of εο to ου is un-Homeric, άλάου as the intermediate form is out of the question. A lightening of the word was more commonly effected by the rejection of an ε, that is, by means of hyphaeresis, for which I may now refer to the thorough investigation of Fritsch Stud. vi. p. .128. Forms like άποαίρευ A 275, μίμεο Simon. C. frag. 29 Β 3 , νψάγεο Theocr. ii. 101, ακίο Herod, iii. 40, φοβέο vii. 50, will suffice as examples. Finally there remain two Homeric forms in which, according to the prevailing view, -εο is ' lengthened ' to -εω. It is needless to point out how utterly incredible it is that among a number of instances of the second person of the imperative formed in the same way, only two should suffer such an affection. Besides the more recent science of language shows 47 a lengthening of this kind to be extremely dubious. Nor is there any probability whatever that in these two isolated forms some very archaic character has been preserved, as Christ has conjectured (Griech. Lautlehre p. 195). He believes that the ει of ερεω (only in Λ 611) and σπεϊο (only in Κ 285) may be explained by compensatory lengthening from the primitive form in a-sva Gr. ε-σΡο. For the phonetic process we might quote the analogy of έίωθα, which is for ε-σΐωθα. Only in this case the course probably was that from ε-σΡωβα came in the first place i-FFwdu, and then έιωθα, while the personal ending sva in Greek at once passed into -σο, from which form we cannot arrive at an explanation of the εί. But ερεω, if we accent it ερεϊο, can be very well explained upon the analogy of alhlo. For taken as an imper. present, it is easily comiected with ερέωμαι ρ 509, ερέεσθαι ζ 298, ερευντο θ 445. The second instance σπεϊο in Κ 285 σπεϊο μοι ως ore πατρϊ αμ eaneo TuSei' διω resists all further explanation. But I leave it undecided whether we are to assume a corruption of the reading (say from εσπε άμ) or that the poet of the Doloneia here, following a false analogy, coined this form. C) Third Person Singular, Active and Middle. It is justly regarded as generally admitted that the ending -ω, common from Homer onwards, hat come from τω-τ and correspondsrto χ 306 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiy. the Vedic ending td-t. For the latter, which has also the evidence of Panini, Delbriick (Verbum, p. 59) points out at least one certain instance, gaWia-tdt let him go, which would correspond to a Greek *βαπχέ-τω. From the Italian languages we have faci-tud C. I. L. no. 813 (=Osc. fac-tud Tab. Bant. 9) with esto-d in Festus-p. 230, 13 (=Osc. es-tud) and also the Oscan forms UM-tud^haX liceto (cippus Abellanus 36), deivatud iurato (Tab. Bant, 5), ac-2wd=agito (ib. 15). Umbrian, like the later Latin agrees so far with the Greek, that it has also dropped the' final consonant : sub-ah-tu (=sub-igi-to), e-tu (=i-to),fer-tu (—fer-to). 48 Cp. Aufrecht and Kirchhoff Umbr. Sprachd. i. 142. With regard to the origin of this termination, the simplest course is to assume an em- phatic repetition of the pronominal stem ta, which in the first instance was pronounced long : hence the primary form was td-ta. In the same way the middle ending -σθω doubtless arose from -σθω-τ, so that σθω-τ and τω-τ correspond, just as in the 2 plur. σθε and τ ε, as in the dual σθον and τον, στων and των. The sign of the middle voice is con- cealed in the σθ, for which we may refer to p. 68 f. I have discussed there 'the few dialectic differences with regard to the combination σθ (Locrian χρήστω, Cretan αποϊει.πάθθω). D) Third Person Plural, Active and Middle. Whilst for the singular of the third person all Greek dialects essen- tially agree, the variety of formation for the plural is tolerably great. For the active we can distinguish four, for the middle three different formations. There was in fact no little difficulty in denoting number and mood, and in the middle a reflex relation to the subject besides, and all this in the way of the imperative, in the termination alone. We may properly place first 1) The forms in -ντω. These are abundantly established from Doric dialects (Ahrens 296) of different regions and times : Lacon. Ιόντω, Heracl. άγγρα-φάντω (Meister Stud. iv. 423), Delphic παρεχόντω, εόντω, άποΰόντω, παρα- μεινάντω etc. (Berichte der k. s'achs. Ges. der Wissensch. 1864 p. 227), and also from the inscription from Tegea (which Kirchhoff. Monatsb. Jan. 1870, has shown to be probably Laconian) οιαγνόντω, and from the Arcadian inscription of Tegea : ποέντω, ζαμιόντω, άγκαρνσσόντω, Ιναγόντω, ZiayvavTO) (Gelbke, Stud. ii. 389). Then follow 2) The forms in -ντων. These are the most widely extended. For they are a) The only forms in use in Homer : αγγελλόντων θ 517, άγεφόντων Β 438, ώενγόντων Ι 47, πινόντων α 340, αντιοωντων Ψ 643 etc. b) New Ionic : λεγόντων, εχόντων, πασχόντων quoted by Kiihner i. 528 from Herodotus (i. 89 etc.). 49 c) Attic, much more commonly than those in -ετωσαν : κνρούντων Aesch. Choeph. 714, γελώντων καπιχαιρόντων Soph. Aj. 961, πεμττόντων Ο. C. 455, βοώντων Aristoph. Ach. 186, παραθέντων'ΝηΚ 456, άποδόντων, υμνύντων TliUC. V. 18, μετεχόντων Plato Protag. 322 d, ΰοκιμαζόντων Xen. Mem. i. 4, 1, εγγραφόντων, όψειλόντων, law in Demosth. 43, 71. ch. xiv. IMPERATIVE.—-THIRD PERSON PLURAL. 307 d) Doric by the side of -ντω : ΰώόντων, εχόντων, επιόντων, βέντων, established from inscriptions by Ahrens Dor. 296. It will be seen that there is not the slightest reason for calling these forms A ttic, as is often done. They are called so by the grammarians simply because the Atticists recommended them for practical purposes, as distinguished from the forms in -τωσαν, which were afterwards more common (Gregorius Corinth. § xcvii.). 3) Forms in -ντον occur only on the Lesbian inscription C, I. 2166, where Boeckh reads φερόντων, φνλασσόντων, καταγρέντων (cp. επιμελέσθων). The copies of the stone do not always agree, but in some instances all give -ντον (φυλάσσοντον), which is defended by Ahrens Aeol. p. 130. In Conze's Tit. 8, 2 1. 6, 8 (Reise auf Lesbos) also -τείχοντον and -ντον are to be regarded as traces of this formation, as Wald rightly maintains (Addi- tamenta ad dialectum Lesbiorum et Thessalorum cognoscendam. Berol. 1871). 4) Forms in -ντωσαν occur only in the isolated εόντωσαν (Anecd. Delph. ed. E. Curtius xiii t 15, xxix. 17, xxxix. 20). 5) Forms in -των. Of these I know only two instances : έστω ν and It ων. εστων in Homer only α 273 θεοί h } επ\ μάρτνροι εστων, for in A 338 τω δ' αντίο μάρτνροι εστων it may be 3 dual ; but it is completely established in Plato Legg. 759 ovtol ϋε εστων εζηγηται δια βίην, Rep. 502, Xenoph. Cyr. iv. 6, . 10, in no. 32, 9 of the Delphic inscriptions published by Wescher and Foucart, and in the inscription from Chios in Cauer's Delect. Inscript. No. 133 1. 20 ; so that the εστων recorded once or twice in Archimedes, though regarded with suspicion by Ahrens Dor. 321 f., is certainly not to be tampered with. — 'ίτων occurs only in Aesch. Eum. 32 ίτων πάλω λαχόντες, ώς νομίζεται. In both forms the ν alone has evidently the function of denoting the plural as distinguished from the singular. We 50 have an analogy in the Oscan form eituns, which occurs several times, if we take this, not, as was formerly the case, as 3 plur. indie, but with Sophus Bugge Ztschr. xxii. 390 as 3 plur. imperative. 6) Forms in -τωσαν. From Thucydides l onwards these forms are used in Attic by the side of those in -ντων, and by degrees they supplanted the latter : μαθέτωσαν Thuc. i. 34, φερέτωσαν Plato Legg. 759, παραλαμβανέτωσαν Xenoph. Cyr. vii. 2, 14, (in a law) μενετωσαν Demosth. 21, 94, ίτωσαν Eurip. Iph. Taur. 1480, εστωσαν Ion 1131 (Dind. εστία). Cp. περιμαζάτωσαν Menander fr. 109. — Besides these αίρίτωσαν, εστωσαν, παραμεινάτωσαν, ποιησάτωσαν, λαβέτωσαν, παρεχέτωσαν are established by Dorian inscriptions (Ahrens Dor. 296, Ber. d. k. sachs. Ges. d. Wissensch. 1864 p. 228). If we turn from this statement of the facts to their explanation, all the forms evidently fall into two groups. In the first case the 3 plur. 1 Cobet Nov. Lect., p. 327, corrects away forms in -τωσαν from pre -Macedonian prose- writers. But there are a large number of them. χ 2 308 MOODS OF THE PEESENT AND SIMPLE AOEIST STEM. ch. xiv. is characterised by the ντ common to this person in the indicative also : λνόντω, λΰοντωι•, λύοντον (?), λυόντωσαν. # The second group is formed from the 3 sing. : ίτων ϊ-ωσαι\ The comparison of Latin forms like eunto, legunto is of itself enough to leave no doubt that the former way is the older. As in the 3 sing, we traced the ending -ro>=Lat. -to back to -w-r=Lat. to-d, it is probable that ->τω also lost a final dental, so that we may give -ntat as the conjectural primary form of the termina- tion. Benfey ' On Plural Endings ' p. 33 thinks he can quote one ex- ample of this termination in Sanskrit: hajantat Naighantuka ii. 14. In this termination the plural is evidently denoted just as it is in the indicative, while the imperative is denoted as in the singular by the long a and the repeated t. 2 Hence there is no difference of principle between the two numbers in their mode of formation. As to the forms 51 with an added ν and σαι•, it is certainly the most natural thing to ex- p^iin both elements from the analogy of other plural forms, supposing that the custom of pronouncing a final ν in a 3 plur. like ελεγον, εβόων, ε -oiovr, and a σαν in such as εΰοσαν, εποίησαν produced a similar ending here also. This was doubtless most naturally suggested to those who spoke Attic ; for the consciousness that ντ belonged to the plural could hardly have been clearly retained among them. Even the genitives of participles like λεγόντων, ποιονντων, from which the grammarians derive the imperatives, may have contributed to produce this result. This explanation is strongly confirmed by the evidently later second group, for in this the mark of the plural lies exclusively in the appended ν or σαν. The «diddle forms are limited to four : *νσθω, σθω, σθων, σθωσαν. The most remarkable, in which we can still recognise the effects of an interior plural v, has come to light within the last few years, in a single in- stance, which is however four times repeated : it has been pointed out already in Stud. ii. p. 450 and above p. 69. The inscription on bronze, discovered at Tegea, and published by Eustratiades in the Αρχαιολογική έφημερίς, Τίερίοΰος Β, τεύχος ΙΓ (1869) ρ. 344 [cp. also Cauer, Delectus Inscriptionum p. 4] has been ascribed on good grounds by Kirchhoff (Monatsb. der Berl. Akad. 1870, p. 63) to the first half of the fifth cen- tury B.C., and assigned to the Laconian dialect. There we read on the second side — el μίν κα ζόη αυτός άνέΚίσθω i.e. if he, the depositor of the sum of money here in question (Xuthias), be alive, he is himself to recover the sum ; then follow the conditions in case of his death : al be κα μη ζόη τόϊ viol άνελόσθω το\ γνήσιοι and again three times άνελόσθω after the plural subjects rat θυγατέρες, τοι νόθοι, το\ ασσιστα πόθικες (?). Hence άνελόσθω is the plural to άνελέσθω. Now if we remember that the thematic vowel appears as ο only before nasals, and elsewhere as ε, we see that άνελόσθω is evidently for *άνελόνσθω, and hence it is to άι>ελε'σθω precisely as λεγόντω to λεγέτω or as λέγονται to λέγεται. It is the very termination of this form, which Ahrens Dor. 297 justly held that we ought to expect : ' in subjectivo ' 2 The repetition of the suffix comes out with especial clearness in the Urn- brian forms etuto, habctutu (habituto), stahituto, tusetutu (tursituto), which in form approach the Latin itote, habetote, statute, torretote, but are third persons plural (Aufr. and Kirchb. i. 143). ch. xiy. IMPERATIVE. — THIED PERSON PLURAL. 309 — so Ahrens calls the middle — * e ΰιΰόσθω et κρινέσθω plurales forinae 52 ΰιοόισϋω et κρινόνσθω fieri debebant.' A remarkable confii*mation by a later discovery of a form which had been merely inferred. After what has been said of the previously discussed forms in -ω, we shall have no hesitation in principle to assume here too the loss of a r, and hence to assume a *ελό-νσθωτ, which is evidently for *ελύ-ντ-τωτ. In this form the internal ν denotes the 3 plur. precisely as in λεγύντω. The middle element is expressed in the σθ, the imperative in the last two letters. On p. 64 we traced the ending ->.ται, e g. in λέγο-νται back to the three pronominal elements n-ta-ti ; the termination -νσθωτ points to four : n-t-ta-t(a). Hence the 3 plur. of the imperative possesses one such element more than the 3 plur. ind., just as the 3 sing, imper. λεγέ-σθωτ possesses one more than the 3 sing, indie, λέγε-ται for *\eyf-ro-rt. If however it seems to any one more probable that such a curious form should not be based upon a very ancient tradition from the freshest formative force of the Indo-Germanic language, but that it originated much later in the endeavour to mark the plural in the imperative middle also in a manner analogous to λέγονται as compared with λέγεται and έλεγοι το as compared with έλέγετο, I can make no objection to this view. In any case we must go back to an -ονσθω, for without the ν the analogy is a very weak one. On the other hand the final r in this view may have been foreign to this form. This unmistakeable Laconian form supplies us with a most welcome confirmation of a Heraclean form. On the first Heraclean table 1. 127 we read : ει τινές, κα μη πεψυτεύκωντι καττάν συνθηκαν, άγγραψάντω και έπελάσθω τα έπιζάμια τα γεγραμμέια. As the neuter plural is joined to a plural verb on these tables, έπελάσθω can only be plural. Now it might be supposed that έπελάσθω was contracted from έπελαέσθω, and belonged to the second class of plural imperatives, which, like κρινέσθω and others to be discussed immediately, do not differ from the singular imperatives, but Ahrens Dor. 195 rightly saw that this would contradict the Dorian laws of contraction. For as the imperative of οράω in Epi- charmus is 8ρη, and as the Heracleans contract επιβάη into επιβί} (Meister Stud. iv."394), we should certainly have expected *έπελ>Ίσθω. 53 On the other hand έπελάσθω is explained quite simply from *έπελαό-σθω, just like ας from άος, φιντία from φιντίαο. Meister is right in following this acute explanation, which Ahrens discovered without the help of the Laconian form. — Finally we have to take account of two Attic forms of the same kind, first established by KirchhofFs excellent Inscriptiones Atticae Euclidis anno vetustiores. I owe the notice of them to Paul Cauer's kind communication. In no. 32 A 16 we have και σνναγόντων και συικληόντων τάς θύρας του οπισθόδομου και συσσημαινόσθων [Kirchhoff appends a? to the ο] τιίις των της 'Αθηνάς ταμίαις,; and in 78, 5 [οί στ]ρατηγοι χρωσθω[ν. We find also έπιμελύσθ[ων in C. I. A. ii. 92 1. 5. Evidently these forms mutually support one another. The second form of the 3 plur. imper. middle is not distinguished from the 3 sing. It occurs only in four instances from a single not very ancient Corcyraean inscription C. I. Gr. no. 1845 : ΰιΰόσθω, έκλογιζέσθω, κρινέσθω (1. 125 κρινέσθω έκαστοι) ένίανειζέσθω (Ahrens Dor. 297). Pro- bably this formation rests upon a confusion with the 3 sing., from which it was no longer possible to distinguish the plural by a different vowel. The third form in -σθων is from Homer onwards by far the most •common : επέσθων I 170, πιθέσθων I 167, ϊηριαάσθων Φ 467. Kiihner i. 310 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. c&. xiv. 537 quotes instances from Herodotus, Sophocles, Aristophanes, Thucy- dides, Plato and Xenophon ; and Ahrens Dor. 297 others from Dorian inscriptions of different countries (Ther. πορευέσθων, Cret. ησθων, στανυέ- σθων, Rhod. άποΰόσθων). In Delphian inscriptions we find έφελέσθων, άνθελέσθων (Berichte etc. 1864, p. 228). The fourth form in -σθωσαν is related to that in -σθων precisely as the active -τωσαν is to -των. The form is un- Homeric, and unknown also to Herodotus (Bredow de dial. Herod. 337) and apparently to the tragedians, but in Attic prose it is used by Thucydides (ώφελείσθωσαν iii. 67), and it is not unknown to the Doric dialect : Cret (Dreros) Ιασσάσθωσαν, Corcyr. άπολογιζάσθωσαν, Ther. πορεύσθωσαν, and often in Archimedes. Cp. Ahrens. 297. The third and fourth forms have evidently come from the singular by the addition of ν and σα ν as plural signs, precisely in the same manner as was the case partly in the active. 64 E) Dual Forms. The second person of the dual of the imperative active and middle is tolerably common in Homer : έφομαρτεϊτον και σπευίετον θ 191, άποτίνε- τον θ 186, εμβητον και σφώϊ, τιταίνετον Ψ 403 — ερχεσθον Α 322, σίτου θ' απτεσθον και χαίρετον ΰ 60, μάχεσθον Η 279, φράζεσθον Υ 115 ; and quoted also from Attic writers : χαίρετον Soph. 0. C. 1437, έίπατον Aristoph. Av. 107, Plato Euthyd. 294, άκούετον Aristoph. Plut. 76. On the other hand Kontos in Λόγιος 'Ερμής i. 66 maintains that there is no other instance of an early date of the 3 dual imper. act. than κομεί- ,των θ 109 {τούτω μεν θεράποντε κομείτων), where this is established as the reading of Aristarchus, and has good M.S authority. In the 'Ομηρου επιμερισμοί (Cramer Anecd. Oxon. i. 397) we read ' σημειοϋνται ώς προσ- τακτικον υπάρχον το κομείτων, αλλ' ουδέποτε ΰυϊκω προστακτικά τρίτου προσ- ώπου έχρήσατο "Ομηρος.' For in A 338 : τω δ' αυτω μάρτυροι εστων need not be regarded as a dual. Besides this Kontos can only quote a 3 dual ΰιαφερετων from Maximus Tyrius 20, 1. Certainly this passage, as well as that from the Epimerismi, shows that the Greeks of a later date in- corporated the forms in -των in their paradigms. There is a noteworthy passage of Suidas quoted by the same scholar : εχετον άντι του έχέτωσαν, ■ΰυϊκώς 'Χέγετον τοϋτο ΎιμαΊος καϊ * Αντισθένης, εχετον ΰέ και Κλείταρχον αυτοϊς νοοϋντα είς μίαν και την αυτήν.' It seems to me by no means proved that here, as has been more than once conjectured, we should write έχέτων and λεγέτων j we may rather appeal in support of this isolated -τον in the 3 dual imper. to the uncertainty in the use of the dual (cp. pp. 52 and 307) which is adequately explained by the rarity of the usage ; indeed in the above-quoted verse of the Iliad some copyists actually wrote κομείτην by an error. Besides the doctrine of the grammarians is here too supported by the analogy of Sanskrit, which has for the 3 dual as distinguished from the second person "the well-established ter- mination -tarn, e.g. i-tdm -^^Ί-των, pa-tam, slda-tam. Cp. Delbriick Verb. p. 61. But whilst with the Indians the termination of the im- perative coincided with that of the preterite and the optative, a distinc- tion of vowel was produced in Greek between -την and -των. 65 The 3 dual middle cannot be recognised, as it is identical with one form of the third person plural : λυέσθων. For everywhere a plural form can replace the dual form. ch. xiv. CONJUNCTIVE. 311 Π. CONJUNCTIVE. As we always start from the simplest and most transparent forma- tions, it will be our first duty in the case of the conjunctive to examine those forms in which the principle of formation of this mood comes out quite clear and unconfused. These are those in which the pure root is lengthened by a short α -sound (ε or o), provided with the primary personal endings, 3 and employed conjunctivally. We have already repeatedly quoted 'ί-ο-μεν by the side of 1-μ^ν as an instance of this method of formation. In Vedic Sanskrit — for such forms are quite unknown to the post -Vedic language — there are according to Delbriick p. 193 only a few instances of the kind : e.g. from han 'kill.' conj. han-a-ti (= Zd. jan-a-i-ti) with ind. han-ti ( = Ζά. jaih-ti) : from as 'be' conj. as-a-ti (= Old Pers. ah-a-tiy, Zd. ahh-a-t) be he ind. as-ti he is ( = 0. Pers. ας -tiy, Zd. ας -ti). Still the principle of formation is completely established by perfect and aorist forms which will occupy us further on, and also by the analogy of the Persian languages. Hence the most primitive conjunctive is distinguished from the cor- responding indicative in no other way than the thematic indicative from the primitive indicative. We may state this thus : conj . han-a-ti : ind. han-ti * .* ind. bhar-a-ti : ind. bhar-ti, or, putting it otherwise, bhara-ti may be at the same time conjunctive and an indicative by-form of bhar-ti. On this fact is based the explana- 56 tion, which in ! Zui* Chronologie ' 2 (pp. 49 ff) I endeavoured to give of the origin of the conjunctive. It is highly probable that formations, which are completely alike in appearance, were also in substance and originally alike, or in other words, that the α (ε, ο) of the thematic in- dicative, and the a of the primitive conjunctive at first served the same end. On p. 9 (cp. p. 138) we took the thematic vowel to be a stem- forming element, by which the nominal character of the stem was more sharply denoted. Hence if bhara-ti meant originally ' bearer he,' from this on the one hand the meaning ' he is a bearer ' with the force ' he is engaged in bearing ' and so the durative force of the indicative might be developed, and on the other hand the meaning 'he is destined to bear/ ' he is to bear,' i.e. the conjunctive application. The Greek language has preserved a not wholly insignificant number of such primitive forms, though only in its oldest phase, the Homeric dia- lect. All these forms were entirely misunderstood by the older gram- marians, and were wrongly taken as conjunctives 'with a shortened mood-vowel.' Though this assumed shortening in the conjunctive of all moods which is elsewhere always inclined to lengthening, must have appeared to every thoughtful scholar an extremely dubious process, even 3 Delbriick has proved with certainty that even beyond the sphere of Greek the conjunctive had at first only the primary personal endings, and that hence we cannot talk of a ' conjunctivus imperfecta, ' as the Sanskrit grammarians have hitherto called the shorter forms. Cp. ' The Old-Indian Verb,' p. 192. 312 MOODS OF THE PEESENT AND SIMPLE AOFJST STEM. ch. xrv. comparative philology only arrived by degrees at the correct view, because it was only by degrees that the forms of the Yedic dialect which bore on the question came to light. In Bopp (Yergl. Gr. § 716) Greek forms of the kind mentioned are not yet put in their right place. So dependent are we all upon the discovery of facts. A more correct view was given in my * Tempora und Modi,' and afterwards by Schleicher Compend. § 289. Since then Westphal especially has done good service for the Greek conjunctive, and also Joh. Paech in his doctoral disserta- tion ' de vetere conjunctivi Graeci formatione ' (Breslau 1861), which is evidently due to Westphal's suggestions. 4 Besides Herm. Stier in my "' Studien ii. p. 125 ff. has thoroughly discussed various sides of the Homeric conjunctive formation. All Greek conjunctives may be best divided into three classes : 1) Those in which the mood-element remains as a short α -sound ; 2) Those in which the sign of mood consists in the lengthening of an already existing α -sound : 3) Apparent exceptions to the first two methods of formation. 1) Conjunctives with a short α -sound inserted. With regard to these forms comparative grammar finds itself in the most decided antagonism to the doctrine of the old grammarians, which has maintained itself with slight modifications up to our own time. The old grammarians, in their absolutely un-historic way of regarding the question, starting from the Attic dialect, were obliged to set down in the first place the first of the three recorded forms of the 1 plur. conj. aor. θώμεν, βέωμεν, θείομεν, and to regard the other two as affections of this primi- tive form. In this respect the fragment of Herodian (ii. 267 ed. Lentz) is instructive. We see from this, that no fewer than three phonetic affections and one hypothetical intermediate form (θεΐώμεν) were necessary for this grammarian in order to get from his starting-point, the Attic θώμεν to the Homeric θείομεν; viz. first ΰιαίρεσις : θώμεν θέωμεν, second πλεονασμός : θεωμεν *θείωμεν (the latter form being merely an assumed one), third συστολή : θείωμεν θείομεν. The current modern grammar substitutes for the very dubious expression πλεονασμός extension or lengthening, and can get no farther (cp. Buttmann i. 2 516, La Boche homer. Untersuchungen p. 152 ff.). For comparative grammar, on the contrary, the third form is not only the earliest recorded, but also the most original, setting aside a slight modification, and the others are to 68 be explained from it. It needs no argument to show that the latter view, which starts from the oldest form recorded in Greek literature, and from the primitive form resulting from a comparison of this with the forms preserved in Sanskrit and Persian, is the only historical, and consequently the only scientific view. The old view was only possible by reason of the often-mentioned error that the poets ' metri causa ' allowed themselves all conceivable Procrustean operations. 4 The view expressed by Paech and repeated by Westphal, that I gave the true explanation only for the one form ίομ^ν as compared with 1μ*ν, has no justifica- tion. For on p. 240 of that work of mine ιταρστ^τον, δώο /xev, θάομςν, δαμ«ί«τ€, and on p. 247 βίίομ^ν, στςίομ*ν are also mentioned. But I gladly admit that these scholars have essentially promoted our insight into the structure of the conjunc- tive, especially by a more correct explanation of the sigmatic aorist forms with a ehort vowel. ch. xiy. CONJUNCTIVE. 313 The Homeric poems offer the following eleven conjunctives with a short vowel from present and primitive aorist stems. Postponing for the present the difficult question as to the way of writing the vowel which precedes the mood- vowel, we will quote the forms in alphabetical order : 1) άΚ-ε-ται conjunctive to άλ-το he leaped (cp. pp. 90, 130). The breathing is given differently in the 'two passages (Λ 192, 207) in which the form occurs. Herodian wrote αλ-ε-ται, as he wrote αλτυ and άλμενος-, good M.SS. have αλεται (cp. La Roche), which La Roche and Bekker rightly follow. 2) επί-βη-ο-μεν ζ 262, κ 334, κατά- βη-ο- μεν Κ 97, with the variant βεί-ο-μεν to be discussed hereafter. 3) βλή-ε-ται ρ 472, conjunctive to βλήτο (cp. p. 132), with the variant βλησεται, which I mention only because the unfamiliar character of these old forms elsewhere too led the copyists into similar mistakes. 4) γνώ-ο-μεν π 304. 5) Ιω-ο-μεν Η 299, 351, π 184. 6) ερεί-υ-μεν Α 62 ; cp. ρ. 213. The corresponding indicative must have been *'έρη-μι. 7) θεί-ο-μεν A 143, Ψ 244,486, ν 364 κατα-θεί-ο-μεν 264, άπο-θεί-ο- μαι Σ 409, κατα-Οεί-ο-μαι Χ 111, τ 17. 8) ί-ο-μεν with a short ι (e.g. άλλ' "ιομεν Ζ 526) 21 times, with a long ί (e.g. Β 440, I 625 at the beginning of a verse) 8 times according to Stier Stud. ii. 129. 9) κιχεί-ο-μεν Φ 128. 10) στή-ο-μεν (v. 1. στεί-ο-μει) Ο 297. παρ-στη-ε-τον σ 183 (ν. 1. στησετυν). « 11) φθί-ε-ται Υ 173, φθι-ό-μεσθα 3 87. To these must be added also 1) Two passive aorist forms : ΰαμη-ε-τε (M.SS. ΰαμεί-ε-τε) II 72, 59 τραπεί-ο-μεν Γ 441, Χ 314, θ 292, and perhaps νεμεσσηΰεί-ο-μεν, which I. Bekker has adopted in Ω 53 (νεμεσσφείομεν ημείς for the traditional νεμεσσΐ]ΘΖμεν οι ημείς}. 2) Two perfect forms, the common έίΰ-ο-μεν by the side of the indi- cative ϊ?-μεν and πεποίθ-ο-μεν κ 335 by επέπιθμεν, to which we shall return when treating of the perfect. 3) The numerous conjunctives of sigmatic aorists, like ερνσσ-ο-μεν, άμεί-ψ-ε-ται, which will require thorough discussion hereafter. The whole group consists of forms in which the mood- vowel is not lengthened either by position or otherwise. For the 1 sing, we could not imagine a form *ΰω-ο-μι or anything of that kind, because the vowel in this personal form is always long, but for the second and third we might certainly expect *ΰω-εις *ΰω-ει coming from *£ω-ε-σί *?ω-ε-τι, and for the 3 plur. *Ζω-ου-σι proceeding from *£ω-ο-νη ; but nothing of the sort occurs: instead of these we have forms like γνώ-ω, γνώ -rjr, γρω -y, γνώ-ωσι. Perhaps the fault lies with the record. For as all such forms, if they were living at all, must have been already antiquated at the time of transcription into the new alphabet, we have no better authority for the correct reproduction of the sound of ΓΝΟΕΙΣ ΓΝΟΕΙ ΓΝΟΟΣΙ, ΘΕΕΙΣ ΛΑΜΕΕΙΣ than for that of ΗΕΟΣ. On the other hand the transition into the all-absorbing o-conjugation by the addition of a the- matic vowel to the root- vowel is established by numerous facts. I need only mention βάω, ελάω and other formations discussed on p. 148. Hence it is difficult to decide. 314 MOODS OF THE PKESENT AND SIMPLE AOK1ST STEM. ch. xiv. There are only two points still to be discussed : the quantity and the quality of the stem- vowel. The quantity offers no difficulty after what we have seen on p. 135. For if, as seemed probable to us there, the vowel of such aorists was originally long and only sometimes shortened, all forms are thus most simply explained. In ϊω-ο-μεν, στη-ε-τον, θη-ης we see the original length retained, so that we cannot talk of any lengthening at all, any more than in γνώ-ης, γνώ-ωσι, or in βλή-ε-ται. 60 Some slight difficulty is presented by 'ί-ο-μεν, for no reason for the length is discoverable : we should therefore have to find this in the analogy of γνώ-ο-μεν,Οεί-ο-μεν. But in this case, as we saw, the long vowel is the exception, the short one predominates. Besides it is an indisputable fact that, compared with the more fixed quantity of the hard vowels, t is everywhere subject to much greater variations in respect of length and shortness. The question as to the quality is much less easily settled. As the <>- sound always appears as ω, the i-sound as ι, only α (η) and ε remain to be discussed. But in these the tradition varies most perplexingly between η and the diphthong ει. Untold quantities of dust have been stirred up around the forms belonging here, and the unlucky theory of lengthening combined with an over-hasty attempt to arrive at a general rule has cast more darkness than light upon this subject. Westphal, Method. Gr. i. 2, 286, was the first to view the case more correctly, without, however, making it quite clear. We must in this question deal at the same time with the forms with the short mood-vowel, and with those which, according to the analogy of the thematic conjuga- tion, show a long vowel instead; and hence we must put βείομεν and θείω side by side with στί\ομεν and στήωσι. There is, however, by no means a large number of forms in question, but only 9 aorist and present stems and 5 passive stems altogether : viz. 1) The stems βλη, δαη, ή, έρη, θη, κιχη with an e in the stem. 2) The passive stems ΰαμη, μιγη, σαπη, τραπη, φανη, of which the same holds good. 3) The stems βη, στη, <ρθη, with an α in the root. Here are altogether 14 stems. The M.SS. give us no sure basis in this question, as even the best of them show the greatest fluctuations, but with a decided preference for the sound ει, and besides, as can be seen from the proofs in La Roche (Textkritik, 405 ff., Homer. Unters. 152 f.) fall into frequent confusions of conjunctive and optative forms (μιγηης, μιγείης, μιγείης). On the 61 other hand we have at least for a certain group of forms positive infor- mation as to the doctrine of Aristarchus, i.e. we know 1) That Aristarchus wrote ψανήη (X 73), σαπηη (Τ 27), θήης (Ζ 432), as Didymus testifies at all three places (όντως Ά. cia των (ϋύο η); 2) That in Ψ 244 he read θείο με ν according to Aristonicus * ή διπλή 6τι συνίστάλκεν το θείωμεν' (ορ. Ε. Μ. ρ. 727, 30) ; 3) That in Ρ 95 he read περιστηωσ according to Didymus * Άρίσταρ- \ος cia του η.' Of these three decisions of Aristarchus modem scholarship has, strange to say, accepted the first two, but rejected the third, on the strength of a phonetic theory, which, as we shall see, is entirely arbitrary. This theory is stated in the following words by I. Bekker, Horn. Bl. i. p. 227, * this same ε , if the verse requires that it should be long, passes before ο ch. χιγ. CONJUNCTIVE. 315 and ω into ει, before η into »/.' La Boche, Horn. Unters. p. 152, blindly follows, adding nmch that has no bearing on the point. As the funda- mental assumption of a lengthening of the stem-vowel is false, this is enough to make the whole theory collapse. But even without this in- sight into the genesis of the forms it is not hard to refute it. La Boche himself gives numerous exceptions to this asserted rule : e.g. Όΰυσήος, Άχιλήος, "Αρηος, ττόληος. How can we speak of a disinclination to the combinations ηο, ηον, ηω in a dialect which gives us forms like νηός, νηόν, νηών, ηοΰς, ήοΊ, ηώ, ΰηομεν, αίζηών, παιηονα in abundance? Even I. Bekker did not venture to meddle with forms like these, nor with participial forms like κεκαφηάτα, κεκμηώς, τετιηότι, and fell into self- contradiction, when in spite of this, supported by Zenodotus, he advocated the reading τεΰνειως instead of the τεθνηως of Aristarchus. Evidently Aristarchus knew nothing whatever of any such doctrine ; in the other grammarians no trace of it is to be found ; and scholars who lay especial stress upon the weight of tradition, ought least of all men to favour a priori such an invention. Hermann Stier, in the paper mentioned above, has justly expressed himself against this theory, but he might well have rejected it more decidedly. As in our view the length of the stem- vowel is original, we shall, to begin with, gladly accept a good tradition, where such is to be 62 found. Hence we write with Aristarchus in Ρ 95 πεοιστήωσ, which is formed precisely as γνώωσι, ϊώωσι, the latter also in Hesiod Theog. 222. In the same way we read also in Ο 297 στη-ο-μεν like γιώ-ο-μεν, ^ώ-ο-μεν. As no one disputes στη-ης Ρ 30, στηη Ε 598, and τνηρστη-ε-τον σ 183, we here get the simple series : yvco -ω (ΤΤη -rjs γνώ -fls στη -rj γνώ -rj στη -e-rop στη-ο-μςν γνώ-ο-μ€υ στη-ω-σι γνώ-ω-σι Hence it at once becomes probable that ε-βη-ν, which is quite parallel to ε-στη-ν, formed its conjunctive in the same way, i.e. βή-ω, not βεί-ω, though all M.SS have the latter in Ζ 113, the only passage in which the form is found, νπερβήη and εμβηη, as is commonly read inl501, Π 94, κατα-βή-ο-μεν in Κ 97 with the best of all M.SS. the Yenetus Α, έπι-βή- υ-μεν ζ 262 with the codex H(arleianus), supported' by the reading of the important codex M(arcianus) επιβήσομει>, as La Boche gives it at this place, or έπιβήομεν with superscribed σο, as he quotes it in Horn. Unters. p. 151 j and of course also φή-η λ 128, ψ 275, and <ρθη-η Π 861. The case is somewhat different with the ε -stems. Still I do not see why we should not accept — against the opinion of Stier — Aristarchus's άν-ήη Β 34, almost the only form which has authority, άφήη Ρ 631, for which the M.SS. give a specimen collection of senseless readings (άφήει, άφείη etc.), μεθήη ε 471 in spite of the ει of the M.SS., θηγς Ζ 432, Π 96, βίη κ 301, ο 51, for each of which there is some slight support, and in the same way ΰαμή-ης Γ 436, μιγίινς ε 378, σαπήη Τ 27, ψηνητι Χ 73. Copyists have everywhere a tendency to ει, but the more recent editors rightly follow Aristarchus. There is no dispute either about β\η-ε-ται p 472. Hesiod gives only the one form θείη Opp. 556, where ει is generally written. There remain still the 1 sing, and 1 plur. : for these the tradi- 316 MOODS OF THE PEESENT AND SIMPLE AOEIST STEM. ch. xiv. 63 tion only knows ει, hence ΰαείω Κ 425, εψείω A 567, μεθείω Γ 414, κιχείω Α 26, ερείομεν, θείο/ιεν, κιχείομεν, τραπείομεν. The diphthong could only be explained as the weakening of an η, for the older form doubtless had this vowel. But as we have seen how often this ει creeps in erroneously for * an >/" which has other evidence in its favour, it is not too bold to assume that the diphthong in these eight forms only originated in the incorrect notion of the copyist, that ετ/, ε ω etc. were the normal forms, and that ει was to be regarded as the usual Ionic lengthening for ε. We made a similar conjecture on p. 103 for ειηται, more correctly ήαται. As soon as we write all forms with η, we have the strictest analogy between the stems in o, a, and e. It follows from what we have said that there are very few cases of a real lengthening left. Two of these have been already mentioned, 'ί-α-μεν (p. 314) and φή -yg. We are probably not wrong in seeing in the η of the indicative φη-μί the source of the length in the conjunctive. The third form of this kind is the entirely isolated μετ-είω, 'όφρα ζωυίσι μετείω Ψ 47 (cp. Χ 388 ζωοϊσιν μετέω). Here too it is natural to employ the same principle of explanation as in φή -rj. Gottfr. Hermann thought he had discovered a 3 sing, ε'ίη or ήη to tnis 1 sing, έίω (Opusc. ii. 32). Hence in I 245 he wrote ταντ αϊνώς δειδοικα κατά φρένα, μη οι άπεΐλάς ςκτέΚέσωσι θεοί, ημίν be δη αίσιμον ei^. (cp. ρ 586). The M.SS. however have the optative ε'ίη, and this can be easily explained. I. Bekker wavered between the two readings. In 1843 he wrote at both places Λη, in 1858 είη, in 1861 (Horn. Blatter i. 228) he preferred ηη. The more recent editors are doubtless right in retaining the optative. — A conjunctive, in which lengthening after the fashion of the singular indicative is unmistakeable, is furnished by εΐω=ΐω in Sophron. (fr. 2 Ahrens), quoted on p. 300 in comparison with the imper. ει. An altogether abnormal mood-form am-i-y (άνα- σταίτι, M.SS. άνασταίη) has been adopted in Pind. Pyth. iv. 155 (cp. Ahrens Dor. 133). Here it has been supposed that α was lengthened to at, and the analogy of the similarly isolated Homeric παραφθαίγισι, ex- plained in another way on p. 40, has been pressed into the service. But the assumption is completely incredible. Could άνα-στά -r} (cp. Homer ffrrjy) be the correct form 1 64 2) Conjunctives in which a previously existing α -sound is lengthened. The rule that the short thematic vowel of the indicative is lengthened in the conjunctive, is so well established from Homer onwards, that we need hardly give any instances. Conjunctive forms like τύχωμι, άκ-ούι/ς, upi'iyy, ψεύγωμεν, παμαισχνιητε, ελκωσι, ψερητυν- — άγωμαι, νέηαι, φέρητηι, πειθώμεθα, επησθε, πί\ωνπ.α — ϊ£ω, ελθρς, X«/3ij, πό0ω/ιε>', τάμητε, ώάγωσι — λάβωμαι, λίπηται, γενώμεθα, πίθησθε, τράπωνται are found by hundreds even in the Homeric poems. These forms follow exactly the same prin- ciple of formation as the corresponding forms in the Vedic dialect, so tliat e.g. the following Homeiic conjunctive forms agree letter for letter with their Indian equivalents pointed out by Delbriick : ch. xiv. CONJUNCTIVE. 317 a-(fa-ti — αγη-σΐ ζ 37, ayrj Η 335 vida-si = 'l8tj-s (for ^ιδη-σι) vida-ti = Xdrj (for βώη-τι) bhara-ti = φέρν (for φςρη-τι) voJca-ti = €Ϊπτ] (for feiirq-rL (from υα-ναΚά -ti) from ^-^επη-τι). In Zend the same rule holds good, so that e.g. bard-t (with a secon- dary ending) is parallel to the quoted form bhard-ti=(f)Epri. A more complete agreement cannot be imagined, and hence the occurrence of this conjunctive formation for the earliest period in the life of language is put entirely out of doubt. I lay stress upon this fact, because we can see from it how extremely improbable it would be that, instead of the long vowel which characterises this mood, that which is especially opposed to the usage of the conjunctive, the short thematic vowel, should capriciously make its appearance. It was only the incompleteness of the material then accessible, especially in the case of Sanskrit, that could have led me to the view expressed in £ Tempora und Modi,' that the conjunctive was to a certain degree only an experimental mood, and that the long vowel might occasionally be shortened again. The Italian languages show the same formation of the conjunctive. The long a of the Latin /eras, Old Lat. ferdt, feramus, ferdtis, audidmus, docedmus, and similarly in the passive, are now, in opposition to earlier opinions to a different effect, generally recognised as identical with the same sound in Sanskrit and Zend. Of the same formation are Oscan 55 forms like deicans■=L•^ά>. dicant and Umbrian like facia=zfaciat. The interchange between and e, in a manner corresponding to the indicative, is characteristic of Greek: by means of .this with true Greek delicacy the originally existing parallelism between indicative and conjunctive is restored, while the contrast between ferimus and feramus does not allow us to recognise the original principle of formation so completely. If we now ask what was' the relation between this formation of the conjunctive and the primitive formation previously discussed, we can hardly content ourselves with the bare fact that the conjunctive was distinguished from the indicative in both instances by the addition of a short a. For it would be hard to conceive of the modal force of such an addition. For the primitive formation represented by 'ί-ο-μεν we thought (p. 311) we could discover a probable explanation in the assump- tion of a nominal stem. For thematic verbs such an explanation is impossible, for we can hardly suppose nominal stems with a long a. It rather seems that we have here merely a formation upon analogy, the linguistic instinct finding only a quantitative difference in the primitive forms after their origin had become obscure, and hence in the need of setting a conjunctive by the side of *bharati (=<ρερε-η) which had be- come fixed as an indicative, creating a *bhard-ti which differed only in quantity. For bhara-ti : bhara-ti * \ as-a-ti : as-ti. I will confess how- ever that I am myself not quite contented with this explanation, and that a more satisfactory justification of this long vowel would be veiy wel- come, if it could be discovered. 3) Exceptions and Variations. Yerbs without a thematic vowel show, with the exception of the previously quoted relics of a more primitive formation, the greatest ten- 318 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiv. dency to form their conjunctive after a thematic fashion. It is the same feature of the history of language as that of which we have learnt to recognise so many instances in the indicative. Just as on p. 169 we saw that from the suffix -va comes on the one hand -vo (Ιάμ-να-μεν but κάμ-νο- μεν), but on the other -va-o (*<$αμνά-ο-μεν), so here too this twofold possibility presents itself, a) treatment of the final vowel after the fashion of the thematic vowel : Ιΰνα-μαι COnj. ΰΰνωμαι ; and 66 b) addition of a thematic vowel, which in the conjunctive is naturally long : <ρά-μεν, couj. *φά-ω-μεν contr. φώ-μεν. The method of formation denoted a) is limited to disyllabic stems, the final vowel of which could evidently the most easily pass into the analogy of the thematic conjugation. Two forms of this land are found in Homer : δύ^αι Ζ 229 according to Herodian, while Tyrannion wrote ΰυνήαι, and κέρωνται Δ 260. A third instance επίστηται Π 243 is very doubtful. Aristarchus according to Aristonicus and Apollonius in the Lexicon took the form as an indicative with irregular lengthening. Zenodotus absurdly read έπιστέαται. The conjunctive can hardly be defended here, but certainly the η in the indicative is also extremely singular, ΰίωνται with the indie, ΰίενται would be taken in the same way, if the optative cu'oiro p -317 did not show that the verbal stem else- where too passed into the thematic conjugation. Hesiod Scut. 110 has the form μα'ρνώμεσθ 1 , the parallel to the optative μαρνοίμεθα λ 513. In Attic writers forms like επίστηται, Ζυνησθε, ΰύνωνται are universally recognised as conjunctives; so too πρίωμαι. Here the conjunctive force attaches itself everywhere to the letters η and ω introduced from the thematic conjugation. But there is one more rare method of formation of this mood, accord- ing to which other vowels also appear in the lengthened mood-syllable. Bergk de tit. Arcadico (Hal. 1861) p. xv deserves the credit of having first collected such formations. They are the following : 1) δέάτοι Arcad. conj., inscr. from Tegea (Fleckeisen's Jahrb. 1861, p. 587) line 19 : οσα ay ΰέατοί σφεις ζαμία. 2) Ιυνάμαι inscr. from Drerus (Gott. Nachr. 1855 p. 104) 1. 41 : σπενσίω δτι κα Συνάμα ι κακόν. 3) εράται Pind. Pyth. iv. 92 : οφρα τις . . εράται (v. 1. εράται). 4) ϊσάντι C. I. no. 3053, 1. 11 inscr. from Cnossus : δπα ων 'ίσαντι (ν. 1. ισώντι) i.e. όπως ουν εϊίωσι) cp. Stud. i. 1, 246. 5) καθίσταται C. I. no. 2671 from Calymnia 1. 42 : όπως μή ΰια ψάψυν των πραγμάτων κρινόμενων εις π\είω ταραχα,ν 6 ΰάμος καθίσταται. επισυνίστατοι inscr. from Tegea 1. 19. 67 παρίσταται inscr. from Andania 1. 72 (Sauppe παρίσταται) ; αν δε μη παρίσταται επί ΰοκιμασίαν. 6) πρυτίθηντι inscr. from Andania ed. Sauppe Gott. 1860 1. 89 ΰσα κα οι θύοντες ποτι r£ κράνα προτίθηντι (Sauppe προτιθήντι). Cp. 1. 93 άνατίθηται. 7) κατασκενάσθηντι ib. 1. 93 όπως κατασκενάσθηντι (Sauppe -θήντι) θησαυροί. 8) προγράφηντι ib. 1. 162 δ αν προγράφηντι (Sauppe -φηντι). The strangest of all the forms is ίινται=ώσι in the inscription from Andania 1. 85 όσοι κα ήνται εκ τας άμιτέρας πόλεος, for which Sauppe writes ηνται. But the sense appears to require the former, and the form is equally strange in either case. It looks as if to εντί, which is several ch. xiv. CONJUNCTIVE. 319 times quoted as Doric (Ahrens p. 321) a middle *ενται was formed. To this ενται ήνται as a conjunctive is related precisely as προτίθηντι is to προτί&εντι. In several of these forms it is not impossible to assume a contraction, especially for ΰννάμαι, 'ίσαντι, in which ά may be the Doric contraction from κείωνται, and Veitch quotes κεηται, Ιιακεησθε from the best Attic prose writers, and προσκέωνται from Hippocrates, while κήται, which has been so confidently introduced into the Homeric text, is nowhere established beyond a doubt, and as a form surpassing the Attic dialect in its disfigurement it appears altogether unsuitable to that early time. 6 We come now to a question much discussed, and answered in very different ways, that of the accentuation of a number of conjunctives of the conjugation in -μι. Ought we to write τίθηται or τιθήται ? From ' La Roche Ztschr. f. ost. Gymn. Sept. 1874, p. 412, defends κηται, which is found at Τ 32, Ω 554 in A and, he says, at τ 147 in N, referring to κίονται X 510. - Hartel 'Homerische Studien ' iii. 10, argues for Ktitrai with ct for the most part short. ch. xiv. CONJUNCTIVE. 321 the grammatical point of view the question shapes itself thus : is τίθηται formed according to a), i.e. upon the analogy of ΰΰνηται, or according to b) i.e. upon the analogy of ζυμβλήται 1 As with regard to accentuation we have no more trustworthy source than the old grammarians, we have to deal in the first place with their doctrine. Unfortunately they are by no means at one on the question. We have tolerably good informa- tion for the present-forms. We know from the scholion on Ζ 229 that Άρυτταρχοι• και ol άλλοι, in opposition to Tyrannion, who actually wrote όυνήαι, took such forms as proparoxytona, and from Herodiani. 462 that as distinguished from ΰύνωμαι, επίστωμαι, ΰιΰώμαι, ιστώμαι were regarded as regularly correct. Only those middle conjunctives, which had no active form, were accented as proparoxytona, the rest as properispomena. We cannot indeed discover any internal reason for this distinction, but per- haps the rule was nevertheless based upon the actual usage, and it gains a firm support from the fact that the same holds good for the optative. On the other hand the Anecdota Oxoniensia ii. 344, 28 and i. 469, 7 give us a canon for the aorists which Lentz similarly ascribes to his Herodian (i. 469, 7) : παν υποτακτικόν εις μαι λήγον επι δευτέρου μέσου 71 αορίστου εν τϊ} συνθέσει προπαροζύνεται οίον θώμαι άπόθωμαι και ΰιάθωμαι, (Τχώμαι άπόσχωμαι. Hence e.g. in Λ 799 we must accentuate άπόσχωνται. We can again discover no reasonable ground for the different treatment of the two tenses. Any one who regards the doctrine of the ancients as the unerring standard of our accentuation, must therefore adopt this contradictory fashion. But the editors have rarely done this; they have generally preferred the contracted forms, so that e.g. προσθτ} is commonly written in Herod, vi. 109, on the strength of all the M.SS., προήται in Demosth. 19, 118, προήσθε in Thuc. i. 71, 4 by Bekker and Classen (Poppo πρόησθε) etc. Even the latest thorough discussions of these questions by Bellermann in the Ztschr. f. Gymnasialwesen xxiv. p. 331, though containing welcome and very abundant information as to the evidence of the M.SS., and by v. Bamberg ib. xxviii 28 ff., have not led to any important results. There remains finally only the fact in the history of language that from an ancient date there has been a wavering between the fuller contracted and the (so to speak) slighter forms, for which it is no longer possible to determine definitely the extent and the canon. The same question returns in the optative. But in this mood we have, at least, a certain fixed point in the supplanting of the earlier ει by the diphthong οι. For it is clear that by the change of vowel the passage into the o-conjugation was completed, so that we have still less reason to wonder at πρόοιτο, επίθοιντο, than at πρόωνται, επίθωνται. For this reason πρόσχοιντο, with which we may also compare πρόσχοιμι, is the only accentuation for which, there is authority. Now between the optative and the conjunctive undoubtedly there is an analogy, and this is the reason why έπίσχωνται, πρόσχω and the like are generally written, and we understand the view of the grammarian who in his rule gave the preference to this analogy. Finally we must touch upon one more peculiarity of the thematic formation. We have repeatedly expressed ourselves in opposition to the assumption that the long thematic vowel, in which lies the distinctive mark of the conjunctive, can be occasionally shortened again. Such an assumption had a kind of probability only so long as the short vowel in forms like ϊ-ο-μεν, ε'ίΰ-ο-μεν was not understood to be original, and that 72 Ύ 322 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiv. in the sigmatic aorists, to which we shall return, could not be explained. After the disappearance of these apparent analogies, we shall hardly make up our minds to regard a short vowel in the place of a long one in the present forms of thematic verbs as possible. In fact, in face of the enormous number of regular thematic conjunctives, there are only 9 forms which are suspected of having been abnormally shortened. Herm. Stier Stud. ii. 138 points out 8, to which we must add κελενομεν in Ψ 659=802, a verse which Stier has probably intentionally passed over. Of these 9 forms, in the first place one, viz. ερείομεν A 62, has been quoted above (p. 313) as a regular formation on the analogy of verbs in -μι. The two conjunctives, which we find in immediate succession Β 232 f. ϊνα μίσγεαι εν φΐΧότητΐ ην τ αντος άπονόσφι κατίσχεαι may, as Stier saw, be made regular by writing an η, μίσγηαι, κατίσχηαι. We find a shortened η in /3ε/3λί7αι Λ 380. How easily might the copyists, . misled by the false notion that the conjunctive admitted either quantity, make an error in transcribing the Ε ! One apparent present-conjunc- tive may be taken as an aorist, viz. £ 672 ως αν επισμνγερως ναντίΧΧεται εΊνεκα πατρός, whether we write ναυτίΚεται with Paech, or with Stier assume an Aeolic form like όφέλλειεν Π 651. The latter course is the simpler. It is different with the form βονλεται A 67' al k4p πως άρνών κνίσης αίγων τε τελείων βούλεταί άντιάσας ήμΐν άπο Xoiyov άμνναι. Westphal and Paech regard this as analogous to αλ-ε-ται, presupposing an indicative *βονλ-ται corresponding to the Latin vol-t. But as the ου of this verb, as we saw on p. 172, came from compensatory lengthening, and presupposes a primitive form *βοΧ-νο-μαι (cp. also Gust. Meyer * Die mit Nasalen gebildeten Prasensstaamme ' p. 46), the assumed βονλ- μαι vanishes. A present formed without the fhematic vowel could at the utmost only appear as *βολ-μαι. But there is not a trace of this to be 73 found anywhere in Greek. Hence I regard it as probable, that the true reading is βοΰλητ άνπάσας, as Stier also suggests. The custom of εκ πλίφους γράψειν might easily give rise to the error. — Thus four forms are still left. Among them is S7 484 τω και κε τις εύχεται άνηρ γνωτον εν\ μεγάροισιν άρης άλκτήρα Χιπεσθαι, where Gottfr. Hermann Opusc. iv. 41 regarded the conjunctive with κε as intolerable, and by an easy emendation, confirmed by one M.S., wrote και re τις. Cp. La Koche ad loc. εύχεται is here decidedly in- dicative. There is still less difficulty about Ψ 659=802. άνδρε δύω περ\ τώνδε κεΧεύομεν, ωπερ αριστω πύς /ιάλ' άνασχομενω πεπΧηγε'μεν. Here the paraphrase published by Bekker takes the form as indicative (προστάσσομεν) and there is nothing to prevent us from following it. In Κ 361 ως δ' ore καρχαρόδοντε δνω κννε, είδατε θηρης ή κεμάδ" 1 ψ λαγωον ε'πείγετον εμμενες αΐε'ι ch. χιν. . CONJUNCTIVE. 323 the only difficulty about the indicative arises from the fact that δ δε τε προθέησι μεμηκώς follows. But Paech very properly reminds us of the reading recorded as that of Aristarchus. χωρον αν νληεντα^ δ be προθέησι μ. If by a slight emendation we write 6 τε προθέησι, we get a relative clause with the conjunctive, introduced into a simile which has the indica- tive, a construction for which Paech justly compares Ν 62, p 518; the alternative which he offers, to take έπείγετον as the conjunctive from a form *έπειγ-μι, «which is without any analogy, has nothing to recom- mend it. — Finally, we have still to discuss Μ 42 • ως δ' or' αν εν τε κννεσσι καϊ άνδράσι θηρητηρσιν κύπριος ψ λέων στρέφεται σθένεϊ βλεμεαίνων. Stier justly refuses to defend an indicative after οτ avhj κ 410-12, where we ought rather to read with Bekker σκαίρωσι. But here too Paech has made a happy suggestion, writing for ώς ci' οτ α ν by a slight change ώς V οποτ (cp. Λ 492). Thus all instances are easily set asitfe, and it is to be hoped that the ' shortened mood- vowel ' has played out its part. There remains however one irregularity to which the friends of 74 irregular shortenings might appeal. On Dorian inscriptions, especially on the Heraclean Tables and on inscriptions from Thera, but also on the Arcadian inscriptions from Tegea, a remarkable fluctuation appeal's in the 3 sing, conj., and we find sometimes the regular η, sometimes ει, sometimes η, e.g. κόπτη side by side with νέμει, φέρει and άμμισθωθή. The inscription froin Andania published by Sauppe has no other active conjunctives but those with ει : θελεί, έχει, πέσει etc. and gives the peculiar middle conjunctive form συντελείται 1. 39 as well as γίνηται 1. 2. It will be sufficient with regard to this to refer to the thorough discussion by Ahrens Dor. 294, and to Meister Stud. iv. 390. I agree with the latter in taking ει to be a phonetic weakening from η, for which he well compares the later Attic βούλει for the earlier βουλή. That there is no question here of a confusion between the indicative and the conjunctive is shown most plainly by forms like εΙ = >) and λάβει which have no cor- responding indicatives. For the form without 1 1 would assume, not as Westphal does (Method. Gr. i. 2, 62) a special form *φέρη-τ with an originally secondary personal ending, but a purely phonetic process, which, as Meister well explains, is confirmed by more than one example. Long before the irruption of itacism we notice in dialects which were not regulated by a local written language, as a kind of prelude to the later and more thorough changes of vowel, various dislocations among the long e-sounds, with which we may to a certain extent reckon ει. Apart from these 3 pers. sing., there is not the slightest trace of any un- certainty in the use of the conjunctive vowels among the Dorians, as is proved e.g. by Heraclean forms like πεψυτεύκωντι, άφομοιώσωντι, ΰέκωνται, and Messenian in the inscription from Andania like όμόσωντι, λάχο)ντι. Cretan expressions, which might be quoted against this, like δ ς £έ κα μή φυτεύσει, δτι δε κα μη πραζόντι I take with Kleemann (reliqu. dial. Creticae Hal. 1872 p. 9) as future, like the Homeric : ο'ί κέ με τιμήσονσι. τ 2 324 MOODS OF THE PKESENT AND SIMPLE AOEIST STEM, ch. xit. 75 HI. OPTATIVE. If we compare an optative form like λέγο-ιψεν with the correspond- ing indicative form λέγο-μεν, we are presented at once with the vowel t, as the sign of the optative, just as the short a- vowel previously appeared as the sign of the conjunctive. But a further examination shows that language did not content itself with this vowel alone. If we compare Xiyo -ι-μεν with λέγο-ιε-ν and ΰο-ίη-ν, evidently a triple form of the mood- sign results : ι tc ιη. In Sanskrit two of these three can be established with certainty. The verbs with a thematic vowel have, with the excep- tion of the 1 sing., i as the sign, e.g. bhare-ma i.e. bhara-i-7na — φέρο-ι-μςν, on the other hand in the other main conjugation jd appears ; e.g. Ved. jd-jd-m from the rt. jd go 3 sing, s-jd-t for as-jd-t = Gr. i -ίη for *<σ~(η. Schleicher thought he could recognize the intermediate ja in the 1 sing. of the thematic verbs, e.g. bhare-ja-m. But Delbriick regards this form as a purely phonetic and specifically Indian modification of *bhare-m, and hence assumes that from the i of the diphthong ai (Ind. e) before m an α -sound was naturally developed. I must reserve my assent to this assumption, until there is some other confirmation of such a remarkable phonetic process. Another trace of this ja seems to be offered by the 3 plural. The Indian form bhare-jus is generally referred to *bhare-jant, which exactly corresponds to the Gk. (pipo-uv^=Za. barajen. But it remains doubtful whether the a is a constituent of the personal termina- tion, as in ά8-αη(ί)=ήσαι>, or of the mood-element (cp. p. 49). Schleicher is inconsistent, on p. 699 dividing s-j-us for *sjant, but on p. 703 assuming *ia-jt-v as the earlier form of είεν. The remaining cognate languages give us, in their less known or less finely distinguished quan- tity, not more than two forms, or in part, only one, thus 0. Lat. s-ie-m for *es-ie-m = e -ίη-ν by the side of fere-mus for fera-i-nvw = φερο-ι-μεν and similarly ste-t, ste-mus i.e. sta-i-t, sta-i-mus 76 unlike the Gk. στα-ίη. Gothic has in the 1 sing. pres. e.g. b-aira-u, if this is rightly traced back to *baira-ju, and in the preterite e.g. ber-ja-u t ber-ei-8 etc. a trace of ja, while in the present elsewhere only i appears, e.g. baira-i-ma^-tyipo-L -μεν. In Slavonic and Lithuanian only i or even Ϊ is recognisable as a relic of the mood-syllable. With this mood-sign are united the secondary personal endings aa a rule : and by the employment of these an analogy is formed between the optative and the preterite, which is especially important for the usage of the Greek moods. But it is very remarkable that it is in Greek, which retains the primary endings for the conjunctive more consistently than any other language, and which has worked out most delicately the distinction of meaning between the conjunctive and the optative, that we find a surprising exception, the 1 sing, of thematic verbs in ο-ι-μι, e.g. φέρο-ι-μι. We discussed this case on p. 28, and there regarded the primary ending as an instance of high antiquity, as a relic of that earlier period in which the division of primary and secondary terminations had not yet ch. xiv. OPTATIVE. 325 been fully established. On p. 31 we discussed the traces of the regular ν in τρίψοιν, αμάρτοιν. The completely isolated Homeric 3 sing, with a primary ending παραφθαίησι Κ 346 we thought on p. 40 we must regard as an aberration of the linguistic sense on the part of a later imitative poet. As to the origin of the optative formation, comparative grammar has from the first been inclined to recognise a significant element in the syllable ja, which acts as forming the optative by the side of i and ja. The various attempts at a further explanation turn essentially upon two points in dispute, viz. first upon the origin of this syllable, and secondly upon the relation of the shorter form i to the longer. As to the origin, the relation of the optative to the future is of much impor- tance for this. The future in -sjd-ml, e.g. dd-sjd-mi (3 sing: dd-sj-a-ti)= Dor. Ζω-σΙω contains, as is generally admitted on'the ground of the close connexion of meaning between the optative and the future, the same ja as that here discussed, only in composition with the rt. as. In other words dd-s-jd-mi comes from the rt. da and *as-jd-mi, and this *as-jd-mi takes its stand with its full personal ending by the side of *as-jd-m the 77 primitive form of the optative of the rt. as be, which makes use of the secondary ending. This combination led me in my essay ' Zur Chrono- logie 2 ' p. 59 ff., following the previous investigations of others, to the hypothesis that the optative syllable is essentially identical with the present expansion ja. Hence as we believed we could explain the conjunctive from a particular kind of the present indicative, so we see in another expanded indicative formation the source of the optative. Hence, according to this view, the modal application of certain verbal forms arose throughout from the temporal, especially from the denotation of what we call ' kind of time/ Benfey (Gottingen 1871) has subjected the origin of the optative to a thorough investigation. I can agree with him in several negative contentions. Thus with Benfey I regard the hypothesis of Schleicher, that the ja of the optative is connected with the pronominal-stem ja, as unsatisfactory. For though Schleicher certainly had not in view the relative, and evidently later application of the stem ja, but probably only thought of a nominal stem-formation after the fashion of the present stems in a, na, nu, yet it is impossible to assent to such a view, so long as the meaning remains completely unexplained. On another point too I agree with Benfey : we cannot be satisfied, as Schleicher is, with the simple assumption of 'intensification' for the relation between ja and ja. For although sometimes else- where, e.g. for indicative forms like φη-μί by the side of φα-μίν, we cannot arrive at the ultimate reason for the difference of quantity, as we saw on p. 98, yet the case is different with this mood-syllable, inasmuch as it appears in its twofold quantity under much more mani- fold relations. On the other hand I cannot continue to follow Benfey's acute discussion, where he regards as the proper source of the optative formation the preterite of an intemive form of the rt. i, known only to Sanskrit, and defends this by laying special stress on the length of the i only recognisable in Sanskrit. We have seen repeatedly that the assumption of a preterite, originally not augmented, is quite arbi- trary. Besides Benfey himself cannot deny that there are optative 78 forms with full personal endings, and is at one with me in holding that the primary -mi especially attached itself to the 1 sing, from the earliest 326 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xit» times. But how can we talk of preterite forms in the optative, if we da not find that even the secondary endings, which besides the augment are theonly signs of the preterite, always cleave to the optative 1 The pre- terite ought therefore to be left wholly out of the question, and we ought to speak only of forms with full and with abbreviated personal endings, as we actually find duplicate forms of the kind side by -side among the Indians. Just as little can I believe in an intensive formation as a component of the compound. Intensive formations are not very commonly used in any of the cognate languages. It is therefore a very bold presumption that an intensive of the rt. i was so familiar in the period before the separation of languages that it could become incorpo- rated with a preceding verbal stem into the unity of a new form» Besides this formation would be manifested as such in many forms only by the length of the i (prim. i + i), & length of which no traces occur except in Sanskrit. The whole hypothesis is hence quite devoid of any firm basis. Again it is evidently in no way easier to arrive at the meaning of the optative from the intensive of i than from the root i or ja itself. An intensive of i must have meant ' to go eagerly after some- thing,' ' to desire eagerly.' But such a meaning only suits the 1 sing, of the opt., where of course e.g. a hypothetic da-jja-m=lo-'ir\-v (or da-jdmi) may be explained very well from the meaning ' I desire eagerly or strive to give.' On the other hand for the second and third persons we could not get from ' thou strivest to give,' * he strives to give ' either the optative meaning in the narrower sense ' mayest thou give,' ' may he give,' or the potential * you will doubtless give,' ' he will doubtless give.' Hence if we do not wish to assume that the meaning of the mood first stamped itself upon the first person, and thence extended itself, to a certain extent abusively, to the others, we must give up the notion of looking for intensive forms in the elements of the optative. It would be easier to get from the unstrengthened root of going to the 79 main functions of the optative through the intermediate conception of a tendency towards something. In this way the force of wishing would be iiiven originally for the first person exclusively : ' I am going to give ' in the sense of c I am inclined to give ; ' for the second and third the force of the potential, which borders on the future : ' thou art going to give ' (cp. French tu vas donner) i.e. * thou wilt doubtless give,' 'he is going to give/ i.e. * he will doubtless give.' We can also well understand how in a later period of the language, after the origin of the form, which had by this time become an actual modal form, had disappeared from consciousness, the distinction of the persons, which we have presupposed, also completely vanished, and the first person might be used with .a potential force as much as the second and third with an optative force. However, even the Greeks of the historic time had the dualism of the modal function hovering before them, when they, partly under our very eyes, created a special exponent by means of the particle av for the one main appli- cation, viz. the potential usage, and thus acquired the. evident advan- tage of being able to distinguish rigorously in independent sentences between the potential and the strictly optative functions. But there is another point to which due importance has not been given in earlier attempts to explain the optative. In giving in Chronol. 2 59 the comparison (a)8-ja-n(t) : a8-a-nt(t) = svid-ja-nti : bhara-nti ch» χιτ. OPTATIVE. 327 I did not altogether overlook the difference of quantity in the a found in many forms, but I did not lay proper stress upon it. The present formation -as-jd-mi (Lat. ero for *esio) which underlies the future ending sjd-mi (Gr. -σίω) agrees indeed in the 1 sing, with the optative (a)s-jdm (=Lat. siem for *es-ie-m), but this is to a certain extent, so far as the quantity is concerned, merely accidental. The 3 sing, is in the future ending *(a)s-ja-ti, but in the optative (a)s-jd-t. The same distinction appears in most of the remaining personal forms. I have already mentioned that Benfey justly brings this distinction into greater promi- nence than his immediate predecessors. Benfey, following a casual suggestion of Bopp's (§ 715), conjectures that the long a in (a)s-jd-t is related to the short a in . (a)s-ja-ti just as bhard-ti is to bhara-ti ; in 80 other words that it betrays a conjunctive formation. If this combina- tion is correct, there are properly in the optative two forms differing in principle, in the first place indicatives of the present formation in ja for the whole thematic conjugation, e.g. *bhara-ja-mi (Skt. bhare-ja -ηι), and • secondly conjunctives of the same formation for the conjugation without the thematic vowel, e.g. *da-jd-mi (Skt. de-jam, Gr. ΰο-ίη-)>). As it is indisputable that the meanings of the moods only gradually got marked off and not in the least in opposition to each other, we cannot be much surprised if formations of such a different origin came to coincide in force. For according to what seemed to us probable, under the head of the conjunctive, as to its origin, there was originally no greater difference between bhar-ti, he bears, and bhara-ti, (if) he bear, than between ' he bears' and 'he is a bearer.' Hence it seems to me not unreasonable to suppose that at the time when the optative began to form itself, the usage of language still wavered in its decision as to whether it should develope the new form from an indicative, or from the conjunctive which was not by any means sharply opposed to this, but which was already a distinct form. Thus we can explain the fact that relics of both methods of formation are preserved to us. Optative forms of the so-called second main conjugation in Sanskrit in fact cannot be distinguished phonetically from conjunctives of the so-called fourth class. E.g. ac-jd-m might just as well be a conjunctive with a secondary ending from an indicative ag-jd-mi, as an optative of the rt. ac without present expansion (which it really is) : though in the forms from α -stems or the so-called first main conjugation phonetic modification prevents confusion with indicatives of the fourth class. Even in the period of the united life, in those roots, which employed no thematic vowel, the heavy form with the originally conjunctive d established itself (-jd-m, jd-s,jd-t etc.) ; but in thematic formations, the stem of which was already weighty enough of itself, the lighter (-ja-m, *ja-s (is), *ja-t (i-t) etc.). If this view is correct — though no one can fail to see how hard it is for us in all * these questions to arrive at the ultimate facts —we get a glance into the gradual growth of the forms, and obtain this series : present expansion (kind of time) — conjunctive — optative, 81 the former developed from an earlier, the latter from another and later method of expansion, partly by the help of the earlier conjunctive. It is true that our view rests upon the assumption that for certain forms the mood-syllable was originally not i but ja, and that the third form of the mood element, the simple i, is weakened from ja. But the latt&r presumption has so many analogies in its favour that it will hardly 328 MOODS OF THE PEESENT AND SIMPLE AORIST STEM. ch. xiv. meet with any serious opposition, and if we can explain the syllable ja in optative forms, without having recourse to the assumption of a purely phonetic expansion of the a, which is certainly not free from objections, such an explanation will, I think, deserve the preference. We shall come back to the question of the relation of ja (u) to i in the consideration of some particular forms. Here I only wish to bring forward one more point, which seems to me not wholly unimportant. Delbruck in his « Altindisches Verbum/ p. 195, points out that optative forms can be • in part very rarely ' quoted from the Yeda. Evidently the conjunctive in this ancient record of Arian language is a much more common mood than the optative. Now if we were right in assuming that the latter only by degrees, so to say, separated itself from the indicative of the i- or ja-cl&ss, it is probable that this separation only brought about as its latest effect the formation of optatives with the mood-sign ja (i) from those stems, which had this syllable already as a present-expansion or a so-called class-sign. And hence it is noteworthy that forms of this kind, e.g. pac-je-s i.e. *pac-ja-i-s from a still earlier *pak-ja-ja-s (videas), according to Delbruck, though they do occur, are very rare. Perhaps we may explain this fact by saying that in the time of the Vedas the last step in the path of separation had not yet been very long effected, and that therefore, in other words, the mood as such still showed signs of its origin from the tense. For a mood can only be regarded as completely established for the linguistic sense, when it can be formed from every tense-stem. The Indians seem as a rule never to have attained with full clearness to the consciousness of the difference 82 in principle between tense-forms and mood-forms, for they possess in their terminology no expression for mood whatever. We now turn from these attempts to get an insight into the first establishment of the mood-forms, to the actually extant Greek optatives. The forms arrange themselves simply in three groups, the first contain- ing those of the so-called verbs in -μι, the second those of the thematic verbs, and the third the deviations and the variations between the first and the second. 1) Optatives of the Verbs without a Thematic Yowel. The mood-syllable has in the three persons of the singular the form -cj}=:Skt. ( /a, Zd. yd, Lat. ie, while the e in the 3 plur., with the excep- tion of the later formation in -σαν to be mentioned afterwards, is always short. Evidently βαϊεν as 3 plur. is related to βαίην as 1 sing, precisely as ετίθεν to ετίθην, Dor. εφάν to tyav ; and as we explained the short vowel in the preterite of the indicative by the influence of a double consonant once present, and forming position, we naturally do the same here. Zend offers the similar ending jl-n ; Sanskrit has altered the α (here too shortened) into u and nt to 8 (-jus). The agreement of Greek with Latin in the e-sound, as compared with the Indian a, deserves notice. The only form with a belonging here hitherto discovered in Greek is the 3 sing. εα=εϊη with the 3 pi. συνίαν (cp. αποτίνοιαν) on the old Elean fparpa C. I. G. 11 [Cauer Del. Inscr. p. 135] with the Boeotian irapi'tav (Keil Sylloge p. 1 4, 3, 4). But we must not fail to take into account also the so-called Aeolic forms of the sigmatic aorist, like τΐσεια, τίσειας, τίσειαρ, to which we shall return in a subsequent chapter. ch. xiv. OPTATIVE OF NON-THEMATIC VERBS. 329 Vedic forms, which entirely agree with Homeric forms, are de - ja - m = δο-ίη-ν ο 449 dh&ja-m = θε-ίη-ν Ε 215 $ne-ja-s = yvo -ίψς Γ 53 dhejurs = θε -h-v Δ 363. The e of the root-syllable for an original d rests upon a specially Indian phonetic affection. Compare further (a)s-ja-t = e -Χη 0. Lat. (e)s-ie-t (a)s-jiirs = eUv (Zd. qh -β-ή) 0. Lat. (e)s-ie-nt. For the contraction of the radical υ with the mood-element ι, shown °* in the Homeric αναΐύη ι 377, ΰύη ν 286 σ 348, the Yedic bhu-jd- ma equivalent in formation to a Greek *φν-ίη-μεν is instructive. The Homeric form Ιείη Τ 209, at first sight a surprising one, of which we have the plural περι,-ιεϊεν in C. I. G. 1688 1. 18, according to Boeckh's reading, has its counterpart in the Yedic 1 sing, jd-jd-m. Ιε-ίη-ν is regularly formed from the stem ίε, which is to Skt. jd as θ ε to dhd. h=jd is expanded from I. Cp. p. 121. On the other hand the unmistakeably synonymous form έίην I 496, Ω 139 is quite unparalleled. 7 Buttmann A. Gr. i. 2 541 conjectured that the diphthong here rested upon a con- fusion between ει and i. For *'τη-ν=1-ιη-ν (cp. Skt. 1 plur. i-jd-ma) would be really the complete analogue of Ιύην. The mistake might have been occasioned by the ει of the indicative. But it is also conceivable that έίην was only a variant for Ιείην, so that both would correspond to the above-mentioned Yedic jd-jd-m, the former with the loss, the latter with the vocalisation of the J. Other Homeric forms of similar formation, in some cases retained also in later times, are βα-ίη-ν Ώ 246, επι-βα-ϊε-ν θ 512, στα-ίη a 256, τλα -u-v Ρ 490, φα-ίη-ς Γ 220, φθα-ίη Ν 815, άφ-ε-ίη Γ 317, κιχε-ίη Β 188, αλο-ίη-ν or αλωην Χ 253 (cp. ο 300, where La Roche regards ή κεν θάνατον ψνγοι η κεν uXwrj as possible [cp. Merry on t, 183]). For the first two persons of the plural the Homeric language has only forms without the e, like ψα-ϊ-μεν Β 81, lo-l -μεν Ν 378, Ιιΰο-ϊ-τε λ 357, άπο-ΰο-Ί-τε χ 61, θε-ϊ-μεν μ 347, έπι-θε-Ι-τε Ω 264, εΛ-τε φ 195, ίκΰνμεν Π 99, and so in the passive aorists following the same rule κοσ- μηθεϊμεν Β 126, πεψνθεΊμεν π 305, ΰιακρηθεΊτε Γ 102, ω 532. Duals of such optatives do not occur in Homer. There is a remarkable coin- cidence in the fact that in Old Latin also the fuller formation of the only verb, which has at all retained it, the verb substantive, is limited to the singular and the third person plural, while in the first two persons of this number we find only % in the place of ie : thus c -ϊη-ν (for €σ-ιη-ν) = s-ie-m (for es-ie-m) €-'ίη-ί ( „ 4σ-ιψς) = s-ie-s ( „ es-ie-s) e -ιη ( ,, έσ-ιη-τ) = s-ie-t ( „ es-ie-t) 84 €-Ie-v ( „ eV-ie -ϊτ) = s-ie-nt ( „ es-ie-nt) but €-1-μςν = (e)s-i-mus e-?-rc =(e)s-i-tis. Cp. Neue Lat. Formenlehre ii. 2 592 f. • Haupt Opusc. iii. 341, defends the derivation from eivcu be. 330 MOODS OF THE PBESENT AND SIMPLE AOKIST STEM. ch. xiv. Differing in this from the usage of Homer and the Old Latin, the language of Herodotus and the Attic writers (cp. Kiihner Ausf. Gr. i. 2 543 f.) frequently gives fuller forms ; but the shorter ones continued to be largely used side by side with them, and are regarded by the gram- marians as better Attic. Thomas Magister (ed. Bitschl, p. 54) says oi γαρ 'Αττικοί εν πάσι πληθυνηκοϊς των ευκτικών — παραληγομένοις τφ η άποβάΧλουσιν αυτό, οίον τνφθέϊμεν, ποιοΐμεν, βαϊμεν. The addition π\ήν φυλάττεσθαι δει το κακόψωνον, ένθα αν τύχτι, ως επι τον βαϊτε και στάΐτε καϊ των όμοιων deserves no serious consideration. Compare ξνμβαίημ€ν Thuc. iv. 61 ξνμβαΐμ^ν Eurip. Phoen. 590 ίμβαίημω Xen. Anab. V. 6, 12 Βοίημϊν Xen*. Cyr. v. 3, 2 δοΐμ*ν Plat. Rep. x. 607 d Soirjre „ „ iv. 6, 47 Herod, vii. 135 άφ€ίητ€ Demosth. xix. 71 a^eire Thuc". i. 139 γνοίημΐν Plat. Legg. xi. p. 918 ξνγγνοΐμ^ν Soph. Antig. 926 θείημβν Plat. Protag. p. 343 e προςθέίμεν Plat. Men. 84 wpBtufn Dem. xviii. 324 /caratfen-e Dem. xiv. 27 τιθ€Ϊμ€ν Plat. Rep. x. 605 φαίημςν Eur. Ion 943 (Dind. ξνμφάιμίν) φαϊμςν Plat. Phil. 63 φθαίη™ Herod, vi. 108 (φαΐτε and the like are not found). It is only from ειμί that the forms without η are considered the rarer; still είμεν Eur. Hipp. 349, Plat. Theaet. 147, ε'ίτη ν the only dual form of this kind which I have been able to hunt up, Plat. Tim. 31 (είητον Plat. Euthyd. 273e), είτε Soph. Antig. 215 (M.SS. ήτε), are the commonly received readings. The fuller post-Homeric forms agree with the Indo-Persian method of formation, thus e.g. εΐήτην withVed. (a)s-jd- tam, είητε with (a)s-ja-ta, Zd. qh-jd-ta. However it seems to me very doubtful whether the fuller forms, recorded from a later time, rest upon 85 a primitive tradition : I incline rather to the conjecture that they owe their revival to the analogy with the singular forms. 8 The love of securing uniformity is always a main feature in later times. Our view is further confirmed by the relation of the two forms of the third plural. The Homeric language has only one instance of the forma- tion in -ιη-σαν, which most resembles the singular : στα-ίη-σαν Ρ 733 (παρασταϊεν θ 218, περισταϊεν ν 50), while there are ten instances of the shorter form. In this case there can be no doubt as to the more recent character of the form, for, as we saw on p. 50, the termination -σαν is everywhere of later origin. It is only from Herodotus onwards that forms like εϊησαν Herod, i. 2, ii. 6, ψαίησαν Thuc. viii. 53, γνοίησαν Demosth. xxxiii. 15 became by degrees somewhat more usual. Whether the particle εΤεν, so common in Attic, has really come from the 3 plur. opt. of ειμί , as is frequently assumed, I do not undertake to determine. As the Attic writers especially connect the neuter plural with a singular verb, the plural, for the subject of which we could only supply ταντα or αυτά, is surprising. "We might also suggest a shortened έίη, afterwards supplied with an appended riasal. But the question deserves further 8 This is supported by the result, at which La Roche Ztschr. f. Oest. Gymn. Sept. 1 874 p. 426 ft. arrives, that it was mainly the prose writers who brought the longer forme into more general use. ch. xiv. OPTATIVE OF THEMATIC VERBS. 331 investigation, in which it would be necessary to take into consideration also the particle ει a. Finally in the middle the Greek language knows no duplicate forms. Everywhere ι alone assumes the function of denoting the optative : ΰο-ί-μην, φα-Τ-ο, θε-ϊ-το, γνο-ϊ-το. In the Homeric ΰαινϋτο Ώ 665 and the corresponding 3 pi. laivvar σ 248 the ι (cp. ΰύη p. 329) has left the only trace of its existence in the length of the preceding v. And here Greek again coincides with Sanskrit and Zend, where I is the only sign of the optative in the middle, so that the Greek τιθε'ιτο has the advantage over the corresponding Yedic dadhlta at least in the retention of the radical vowel. After we have taken this rapid survey of the facts of the case, we are led on to the question of the historical relation of the shorter forms to the 86 longer ones. With regard to Greek, it is very rarely that any doubt is expressed upon the point. The old grammarians talk, as we saw, of an αποβολή of the η, for which Choeroboscus (Herodian ed. Lentz. ii. 824, 31) uses the expression συγκοπή. The accent also points to the priority of the longer form, for it is only thus that we can understand the dif- ference between 3t£otre, ΰιΰοΊτο and λάβοιτε, γένοιτο. We must not in- deed deny that ΰιΰοΊεν, which cannot in any case come from ΖιΙοίησαν, is also properispomenon. But here, as elsewhere, there is no difficulty in the assumption that analogy was an essential source of the accent. Among modern grammarians the view that the shorter forms are con- tracted has been the prevalent one. Ahrens 'Ueber die Conjugation auf MI' p. 15 was the first to set against it another, starting from the shorter forms, and regarding the η of the longer as ' strengthening.' But such an assumption is so entirely opposed to the conception which runs through the whole of the modern science of language, that it will hardly find any adherents. Benfey has disputed the contraction from another side. He lays great weight upon the fact that the sign of the optative in three languages, viz. in Sanskrit, Zend and Greek, consists for the middle in the simple ϊ, while for the active, at least over a wide range, it consists in ja> and that there is not a single trace of the full syllable jd in the middle in any one of these languages. The last-mentioned fact loses somewhat of its importance from the circumstance that in Zend we find at least one isolated form, quoted by Benfey himself, dai-d-ya-ta answering to a Greek *()ι-δο-ιε-7-ο, from which it is plain that at any rate the syllable jd, accord- ing to our view the earliest form of the mood-suffix, was not wholly foreign to the Arian middle. Now the middle terminations are through- out heavier than those of the active ; and thus there was more induce- ment to contraction. Hence we can hardly be surprised to find that this process went on independently in the three families of speech. For we might venture to assume the shorter form even for the period of their common life. From the standpoint of the classical languages it would be a strange notion to derive ΰοϊμεν, σταϊτε, ΰιακρινθεϊμεν and Lat. slmus, 87 sltis, vellmus from forms like ΰοίημεν, siemus etc., but to deny the like origin for ΰιΰοΊτο, θεϊσθε. The optative forms like αγαιτο, επίσταισβε, which are otherwise accented, will be discussed under 3). 332 MOODS OF THE PEESENT AND SIMPLE AOKIST STEM. ch. xiv. . 2) Optatives of the Verbs with a Thematic Vowel. The thematic vowel invariably combines with the modal ι to form the diphthong oi } which answers to the Sanskrit and Latin e (=ai) ή>4ρο-ι-μ*ν, Skt. bhare-ma, Lat, fere-mus, for it is an established fact that the Latin futures characterised by the long e are in origin optatives. Vedic forms, completely corresponding to Greek or Latin forms, are 2 sing. Skt. ρας -jes - Lat. spec-ie-8 3 sing, vide-t = Gr. p ίδοι 2 dual vdTietam = Gr. ρΰποιτον Zend supplies 2 sing, apa-baroi-it - άποφέροις Lat. au-feres 3 plur. bara-yen m φ4ρο-κν "We discussed on p. 328 the character of the vowel in the 3 plur. It is noteworthy that in Homer the regular form of the 3 plur. mid. is that in -οιατο : γενοίατο, σπερχοίατο, βιψατο Λ 467, πειθοίατο, τοζαζοίατο, with the one exception of A 344, where the harsh• hiatus οππως oi παρά νηυσ\ σόοι μαχέοιντο ' Αχαιοί points to a corruption. Thiersch proposed μαχεονται. We should pro- bably write with Struve and Ahrens μαχεοίατ. The new Ionic follows the Homeric language : Herodotus has βονλοίατο, άγοίατο, άπικοίατο, άνιωατο (iv. 130), μηχανωατο (vi. 46). It appears at first sight pro- bable that this α is the same as that which occurs, though with less con- sistency, and not after thematic vowels, in the indicative of the same dialects : Homer, βεβληαται, κέαται, Herod, επιστέαται, έκΰιΰόαται. In the active we find, from the group of forms here under considera- tion, only the three isolated dialectic forms quoted on p. 328 answering to the α of the middle, along with the widely extended sigmatic or so-called Aeolic aorist-forms in -εία, -ειας, -ειαν (τίσειαν A 42) also mentioned there, so that we can hardly refuse to recognise m as one phase of the optative syllable ja, though a rare one. Thus the e of the Zend hara-ye-n also, and that of the common Greek <ρέρο-ιεν, can hardly be different. But by the side of -oiav and -οιεν we find also a third termination : oi-p f in which the vowel has entirely disappeared. There are frequent instances of a third person plural, in which the optative sign consists merely in t, in the Delphic inscriptions : thus παρεχοιν (Anecd. Delphica ed. Ern. Curtius no. 12, 13, 31, Wescher-Foucart no. 32, 9 j 33, 9; 38, 11 etc.) ποιίοιν W.-F. 24, 7, θίλοιν 43, 20. These same inscriptions give also παρέχοιεν and παρεχοισαν. In παρίχοιν we have certainly not to recognise an originally distinct method of forma- tion, like ποιησαι by the side of ποιήσειε, but a contracted form, which throws light again upon εΤμεν as compared with έίημην and the like. Without noticing the Delphic forms I. Bekker Homer. Bl. 112 and 219 conjectured the termination -iv for the 3 plur. in some passages of Homer and of Attic poets, especially ν 382 rovs ζ(ίνονς iv νηϊ πο\νκ\ήιδι βα\όντ(ς ts 2iK(\oi>s πίμψωμςν, oBtv κ* rot άξιον $\φοιν ch. χιν. VARIATIONS OF THE ORDINARY OPTATIVE. 333 where the M.SS. have the singular. Most recent editors have followed Bekker, for the singular in this connexion gives no sense. Bergk Poetae lyrici 3 p. 487, who calls άλφοιν (incorrectly as we have seen) a ' forma inaudita quam finxit I. B.,' attempts to get rid of the difficulty by striking out the preceding line as an ' imperita rhapsodi alicuius inter- polatio.' The other passages are all not convincing, so that Bergk is light in *saying that this form has but weak support in literature. 3) Fluctuations and Variations. We have here to do in part with the same phenomena, which we learnt to recognise above p. 320 if. in the conjunctive. But in the optative some points present themselves in a greater diversity of shapes, inasmuch as in the active there is a greater variety of endings. The fluctuations in the case of the optative are of two kinds. We find on 89 the one hand in verbs in -μι by-forms following the thematic rule, on the other hand, what is more surprising, in thematic verbs by-forms with the peculiarities of the verbs without thematic vowels : thus we have εοιμι as well as ε'ίην, and on the other hand ποιοίην as well as ποιόίμι. The manner in which the thematic method of formation gradually overgrew the other shows itself in two ways in the optative, just as it did in the conjunctive. Either the thematic ο takes the place of the radical vowel, or it is suffixed to this vowel. Compare επί-Θοιντο and θεϊντο with ε-ο-ι and ε'ίη. The accentuation of the optatives of the first kind was discussed on p. 321, so that I need not return to it here. Among the optatives of this kind we find two of a peculiar character : ιοίην and σχοίην. If we remember the Homeric form Ιείην, mentioned on p. 329, which approximates to the infinitive \ivai, it is very natural to bring Ιοίην into a special relation with this, giving the equation Ιοίην '. Ιξίην = θοίμΐθα Ι θείμεθα and thus to regard the ο not as suffixed, but as originating from the ε. I am not shaken in this view by the fact that ΰιεζωίην and the like cannot be quoted from any author earlier than Xenophon and Isocrates (v. 98 ; vi. 42). The form might have been long in use, and it might be by pure chance that it does not occur earlier in the extant texts. For on what analogy could 'ί-ο-ι-μι (if we are to derive it from a stem expanded by an o) become Ιοίην 1 We shall see immediately that this fuller form only makes its way into the thematic conjugation in a strictly limited circle of forms, in the case of which it admits of complete explanation. We have further the entirely isolated Homeric επισχοίης & 241 τω kcv επισχοίης λιπαρούς πόδας €Ϊλαπινάζων, The Yen. Α. has the reading επισχοίες, the Palimps. Syr. επισχοίας. From the scholium we see that Herodian read επίσχοιες, an unexampled form, which he supposed he could explain either by the ' pleonasm ' of the ε or from a shortening of the η. επίσχοιας is quite a's unprecedented : it might perhaps be compared with μείνειας and other so-called Aeolic forms of the sigmatic aorist, which we touched upon on p. 332. επισχοίης, 90 the reading of Alexander of Cotyaea, finds support in the Attic σχοίην (Isocr. i. 45), σχοίη (Plat. Apol. 34 a), σχοίησαν (Hyperides pro Euxen. col. xlii. 19). Now if we remember that the aorist-stem σχε with its originally radical ε presupposes an early inflexion after the analogy of 0ε, 334 MOODS OF THE PRESENT AND SIMPLE AORIST STEM. ch. xiv. ε, and has actually retained σχές in the imperative, we may compare with the equation ίοίην : Ιίίην = βοίμεθα '. θείμεθα tie second σχοίην : * σχείην — Ιοίην I Ιύην. We may therefore say that ο in σχοίην comes from ε in *σχείην just as ε-σχο-μεν presupposes *έ'-σχε-/χεΐ', and as παρά-σχε with a thematic ε presupposes παρά-σχε-ς with a radical ε. Whilst in the ending -ιη-ν a relic of the old method of formation is still preserved, in νπέρ-σχοι £ 184, ενί-σποι Ε 107, επί-σποι β 250, in all middle forms developed from such stems, like άνασχοίμην λ 375, σχοίατο Β 98, and in Ζίοιτο ρ 317 by the side of the indicative Ζιέ-νται (cp. above p. 120) it has altogether dis- appeared. The corresponding forms from the roots ε and θε cannot be quoted from authors earlier than the Attic writers, 9 and even here there are considerable variations in the M.SS. between ει and οι, as also in accentuation, thus Plat. Gorg. 520 πραο2το, Dem. xviii. 254 προοϊντο (Westermann προέϊντο), άφίοιτε Plat. Apol. 20, κατάθοιτε vith the v. 1. καταθεϊτε Dem. xiv. 27, πρόσθοιτο Dem. vi. 12 according to Voemel, επιθοίμεθα with the v. 1. επιθείμεθα 4 Thuc. vi. 34. l Here belongs also κάθημαι, the vowel of which, after the remembrance of an originally existing ς had become obscured, was treated like a final vowel. Forms like καθήμην, καθήτο, καθήμεθα, which we ought perhaps to admit in Arist. Lys. 149, Ran. 919, Xen. Cyrop. v. 1, 8, according to the traces in the M.SS., change into καθοίμην (Plat. Theag. 130), καθοϊτο (also κάθοιτο), though Cobet Nov. Lect. 225 in his rigour will not allow these in Attic. 91 It is much less common for the primitive verbs to become thematic by the addition of a vowel. Here belong the Homeric forms ε-ο-ι-ς I 284, ε-ο-ι I 142, λ 838 and the quite isolated ενέοι Herod, vii. 6. The introduction of the vowel may have been facilitated by conjunctives like ε ω and the participle εών, the only form in use. The forms εοις and έ'οι find a support in the Latin -ses, -set of pos-sem, es-set (for ed-set). Besides these I can only compare θέοιτο which appears occasionally in Herodotus (προςθέοιτο i. 53, νποθέοιτο vii. 237; on the other hand προθείτο iii. 148). For 'ίοι (S 21), according to what has been said above, is rather to be taken as moulded on Ιοίην for ϊείην, and is related to Ιοίην as υπερσχοι (dis- cussed above) is to έπισχοίην, though certainly the participle Ιών sug- gests the parallel with έ'οι, ιών. These scanty traces of an added vowel make it extremely improbable that, as has often been assumed, in all the optatives of the verbs in μι with the sound οι the analogy of the con- tracted verbs was followed. The course of formation was not προςθεΊτο ιτροςθέοιτο προςθοΊτο, but προςθεΐτο (and πρόςθειτο), προςθοϊτο (and πρόςθοιτο). This view solves for us a riddle, otherwise hardly intelligible, in the fact (one which contravenes the whole course of the verbal structure), that the contracted- verbs, as well as the contracted futures which are analogous to them, though undoubtedly thematic in their formation, yet deviate into the analogy of the non-thematic verbs. The older grammar regarded the so-called Attic forms like ποωίην, τψψης as by no means • ύίΓοτίθοιτο Herod, iii. 41 has weak authority: Stein writes inrcrieero. 1 The Cretan form συν4θοντο C. I. 2664, 2, which Ahrens regards as an err, points to a still wider extension of the thematic o. ch. xiv. VARIATIONS OF THE ORDINARY OPTATIVE. 335 surprising by-forms of the shorter like ποιοΊμι, τιμωμι. But a little re- flexion upon the course of the developement of the Greek verbal structure shows us at once that such forms are really quite as astonishing as an occasional *λεγέΐ'αι by the side of λέγειν or άγωμ,ι by the side of άγω would be. But it has been fully proved above p. 246 ff., that the con- tracted verbs, for which the Aeolians retained very extensively the analogy of the verbs in μι, show even outside this dialect unmistakeable traces of such inflexion, and that various phonetic phenomena point with certainty to a general extension of the Aeolic inflexion in earlier times. It is therefore a priori probable that this phenomenon also may be φ ranged with the others. Before we discuss this any further, it will be necessary to point 92 out the actual facts as to the occurrence of these forms, for which col- lections have been made by Fischer ad Vellerum ii. 345, Matthiae i. § 198, 2, Kiihner i. p. 544, Wecklein Curae epigraphicae 31, La Roche, Ztsch. f. o. Gymn. Sept. 1874, p. 424 ff. In Homer there are only two instances : φιλοίη ΰ 692, and φοροίη ι 320, while optatives like φθονέοιμι λ 381, καλέοι, επιφθονέοις, ομοφρονέοις,νεικείοι, εωμι π 85, εω ν 12, ΰρώοιμι ο 317, ηβωοιμι Η 157 are far more common. On the various forms of the verbs in α ω in Homer Mangold Stud. vi. 208 ff. gives full informa- tion. The so-called Attic forms appear also here and there on Ionic inscriptions : thus C. I. 3044 άπειθοίη, άνωθεοίη (1. 51) by the side of ποιοϊ (1. 43) ; and in Herodotus, but here too still as rare exceptions ; thus ποιοίη vi. 35 (v. 1. ποιοΊ, ποιέΐ)), ενορωη i. 89 by the side of ποιέοιμι v. 106, καλέοι i. 11. From Hippocrates too a few instances are quoted. They occur, as isolated forms, among the Dorians, e.g. C. I. 2556, 47 άΰικοίη, Epicharm. fragm. 33 • tls hi κα \ώη γενέσθαι μη φθονονμ^νος φίλος ; where, according to the principle established by Hugo Weber in his paper on the particle κα, we must write καν. In the Argive treaty of alliance (Thuc. ν 79) we find ΰοκοίη. On the other hand we have in Alcman p. 89 Β. 3 νικώ, in the Locrian inscription of Chaleion or Oean- theia 1. 2 συλώ, 1. 4 άΰικοσυλω, 1. 6 μεταΐοικεοι, 1. 8 προζενεοι, and on the Delphic inscription C. I. 1688 έφιορκέοιμι (cp. εψιορκίοιμεν in the Cretan inscription published by Bergmann 1. 71). We see from this that the name 'Attic' for the fuller forms is only so far correct, that they be- come more common and almost the rule in Attic. For from the time of the tragedians the longer forms occur, and in prose they become in time much more common than the shorter ones. The following statement may make this plain : Aesch. Suppl. 1064 άποστβροίη Prom. 978 νοσο"ιμ % αν „ Ag. 1049 άπ€ΐθοίης Sogh. Ο. R. 1470 δοκοϊμι Soph. Antig. 70 δρωης ,, Phil. 895 δρώμι „ Track. 902 άντωη Eur. Hel. 1019 άδικοίημ^ν Aesch. Suppl. 1014 ίντνχοϊμεν 93 (v. 1. άδικοίην νιν) „ OyC 132 δρωημεν Soph. Phil. 1393 δρωμ € ν (Dind. ξυνδρωμεν) Arist. Vesp. 279 άντιβολοίη Soph. Trach. 1235 νόσοι „ Nub. 1255 ζωην Thuc. ii. 5 προχωροίη, δοκοίη Thuc. ii. 100 δοκοί „ viii. 66 σιγφη Plat. Rep. vii. 516 a καθορφ 336 MOODS OF THE PEESENT AND SIMPLE AORIST STEM. ch. xit. Plato Orat. 391 άγαπωην Plato Rep. viii. 557 άποροϊ „ Menex. 247 νικωητ€ „ Legg. ii. 664 π^δώ „ Gorg. 486 Ιλιγγιωης Theogn. 107 άμως Isocr. vii. 50 ίπιτιμωηρ Dem. iv. 51 νικωη Aesch. ii. 157 μαστιγοίηρ Arist. Pax 1Q76 νμεραιοΐ (v. 1. μαστιγοΐμι) „ „ 102 δοκοίησαν regularly -yep, -oitv? To understand the longer forms we must start from the primitive forms of the contracted verbs, which recommended themselves to us on p. 248 as the most probable, viz. *φιλήη-μι } *ΰονλώη-μι, *νικάη-μι. The original optatives were thus *<ριληε-ίην, *Ζου\ωε-ίην, *νικαε-Ίη-ν. These forms were variously modified. By contraction there came about the Aeolic forms ψιλείην, for which there is good authority, as well as for οίκείην (cp. Ahrens Aeol. 140), ΖουΧοίην, νικαίην, the last two not being established by quotations, but presumable on analogy. On the other hand here, as in so many other forms, the very common ο made its way into the place of the ε : *φιληο-ίην, *ΰονλωο-ίη-ν, *νικάο-ίη-ν ί shortened into φιλεοίην, ΰονλοοίην, νικάοίην, and contracted into φιλοίην, ΰονλοίην, νικωην. It is noteworthy how rare the examples of this formation without contraction are, like the above-mentioned Tean ανωθεοίη. This might suggest that φιλοίην arose in the same way from the Aeolic ψιλείην as σχοίην from *οχ*Ιην 9 and for the Homeric 94 ψοροίην this view would find special confirmation in the Aeolic infinitive ψορηναι. But the optatives of the verbs in -αω with their -ωην point without any qualification to -αοιην, and it is probable that the two other classes of contracted verbs went the same course. The contracted verbs in -ε ω were followed by the contracted futures, which indeed in every respect are guided by their analogy. Thus εροίην, ψανοίην and the like need no further explanation. The case is different with the optatives of the perfect like πεποιθοίη. We shall come back to these forms in dealing with the perfect. We may simply mention here that the vowel of the perfect-stem in είΰείην, ΰεΐιείην is treated exactly like a si em-vowel, so that here too no difficulty is presented by the method of explanation just set forth. There is thus only one group of forms left, which is surprising, the Lesbian Aeolic optatives in -οιην or -οην from thematic aorists (Ahrens Aeol. 132). For these we have really two witnesses, the scholium on & 241, mentioned above, in which there are quoted as analogies to the Homeric επισχοίης from the work of Alexander of Cotyaea περί παντο• ΰαπών, Ίθ(ην και άγαγοίην παρά Σαπφοϊ* and * πεπαγοί η ν παρ 1 Έ,νπόλιΙι,' and Etym. Magn. p. 558, 28 "Εστί λάχοιμι, λάχοις, λάχοι• τοντο γίνεται κατά πλεονασμον τον η Άττικώς λαχοίη' είτα προςθέσει του ς λάχοιης ' και τροπή τον ς /«ς* ν, γίνεται το. πρώτον' καϊ αποβολή Αωλικρ τον ι λαχόην (cp. Anecd. Oxon. ii. 204, Choeroboscus ii. 772). Hence the fragment of Sappho preserved by Apollon. de syntaxi 247 (fr. 9 B. 3 ) is undoubtedly read correctly thus : aW «γω, χρνσοστίφαν* Άφρόδιτα, rovdc τον πάλορ λαχόηρ. 2 Cobet Nov. Lect.p. 364 limits still further the use of the shorter forms, saying • Tragicis licet νοσοΊμι et δοκοΐμι et similia dicere, populo et comicis et oratoribus non licet, sed %οκοί•ην, ροσοίηρ et similia sola in usu sunt.' ch. xiv. MOODS OF THE PEESENT AND SIMPLE AORIST STEM. 337 Of the two other forms the one has been already set aside by Ahrens Dor. 330. For it is very improbable that the Attic comedian Eupolis should have formed an otherwise unheard-of aorist ireV«yov, and much more -probable that πεπάγοίην is to be taken as the optative of the per- fect πεπη-γα in a comic imitation of a Doric dialect. Meineke in the Addenda to the larger edition of the Comedians v. p. 10 to a certain extent admits this. Hence χενάγοίψ goes along with the above-men- tioned πεποιθοίηΐ' and the like. Is it possible that άγαγοίην, which is also quoted from Sappho, is merely a copyist's error for λαχοίηνΐ In 95 any case these aorists are quite isolated, and although in the case of the Aeolians, who formed optatives in -ιην from all derived verbs, it would not be quite impossible to imagine an extension of this analogy, we may still venture the -conjecture that λαχόην is perhaps by no means identical with λάχοιμι, but is formed from a derived *λαχόω, Aeol. *~λάχωμι, and is thus completely regular. For a derived *λαχόω by the side of λάχος would be quite as conceivable as θυόω, νεψόω by the side of θύος, ιέφος. The loss of l has its analogies, as Ahrens has shown in § 16, in a widely extended uncertainty of the ι between vowels among the Aeolians, which points to a semi- vocalic pronunciation, and of which we found a trace in the Elean εα=εϊη on p. 328. For barytone verbs Ahrens Aeol. 133 quotes regular forms like δανοις, χαίροισθα as at the same time Lesbian. 338 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM. CH. XV. CHAPTER XV. VERBAL NOUNS OF THE PRESENT AND SIMPLE AORIST STEM. Of the forms of the two - stems hitherto discussed there remain now only the infinitives and the participles, i.e. the formations which on p. 2 we described as verbal nouns. Though the more uniform extension of the participles through the various branches of the Indo- Germanic languages makes it hardly doubtful that the verbal adjectives or par- ticiples were fixed at an earlier date than the petrified case-forms of abstract substantives, which we call infinitives, we will still keep to the traditional order, and begin with the latter. I. INFECTIVES. In the formation of the infinitives we meet with a variety, which is quite surprising, when compared with the uniformity in the form of the moods. Not merely do the different branches of the cognate languages show important differences in the form of the infinitive, as we shall have to explain further on, but even the Greek dialects, which elsewhere show hardly any but phonetic variations in the formation of the verbal forms, differ considerably in the formation of the active infinitive from the stems here mentioned and also from the perfect, while for the middle and, to notice this at the same time, for the sigmatic aorist, they offer us the usual spectacle of essential identity. We shall return further on to the significance of this remarkable difference, which hitherto has re- ceived but little attention ; but our task for the present is to represent the variety as accurately as possible. The various infinitive forms of the Greek language C£m be best arranged in five groups : 1) those in -μεναι and -μεν : e.g. φάμεναι, εΧθεμεν, τεθναμεναι. 2) those in -ναι : e.g. γνώ-ναι, <ρά-ναι, γεγονέ -vat. 3) those in -εν, -ην, -ειν : e.g. Cret. φέρει', Aeol. έίπην, Homer. θείειν, φνγέειν. 4) those of the sigmatic aorist : ΰεϊζαι, άμνναι. 5) those in -σθαι throughout the whole middle voice. It is only the first three groups which need a more thorough dis- cussion. What we are struck with at once is a very extraordinary variation, in the case of the primitive verbs, i.e. those which have no thematic vowel, between the first and the second group, in the case of the thematic formations between the first and the third group. In the language of Homer this variation reaches its height. We may see in this one of the many proofs of the unmistakeable fact that this language established itself in the mouth of epic singers under the crossing influences of different dialects. From the root ες there are five forms : ϊμμενα^ εμεναι, εμμεν, εμεν, and είναι, the last already the most ■ch, xv. INFINITIVES. 339 common. In the case of the derived verbs it is not uncommon to find three forms originating in the crossing of the Aeolic formation with the Ionic after the analogy of the conjugation in -μι : φορη-μεναι, φυρή-ναι, φορε-ειν, so that here the first three groups are represented in one and the same verb. 1) First Group: Infinitives in -μεναι and -μεν. 97 Between the longer and the shorter form there is this difference, that the former is decidedly more archaic, i.e. more usual in the Homeric dialect, but cannot be elsewhere proved to have been in use, except by a few traces in the Lesbian Aeolic dialect, and in the elegiac poets in- fluenced by the Homeric language. As instances of the various present and aorist forms of the two chief conjugations, which belong here, we may quote εμμεναι A 117, ΰιεζίμεναι Ζ 393, παριστάμεναι η 341, ζενγνν- μεναι Γ 260, κιχημεναι Ο 274 — καταβημεναι Μ 65, ανστί\μεναι Κ 55, £ο- μεναι Α 98, θέμεναι Β 285, γνώμεναι Β 349, καταΰΰμεναι Γ 241, άλώ- μεναι Φ 495 — αειίεμεναι θ 73, άρχέμεναι Ύ 154, ληγέμεναι Ι 257, ποντο- πορενέμεναι ε 277 — ειττεμεναι Η 375, ελθεμεναι Α 151 — όνθεμεναι i.e. ava-di -μεναι and εμμεναι on the inscription from Cyme C. I. 3524 1. 53, 51, αττνίόμεναι on that from Mitylene 2166, 40; εμμεναι appears re- peatedly on the Lesbian inscription discovered by Conze, and in Sappho fr. 136. For forms like φνλασσέμεναι in the Elegiac poets cp. Benner Stud. i."2, 31. — The forms ε ί μεναι as Megarian in Arist. Ach. 775, and αλεζέμεναι in the Spartan treaty Thuc. v. 77 are quite isolated, and Ahrens doubts both. The infinitives in -μεν are formed by Homer less commonly from verbs in -μι than those in -μεναι : εμμεν Σ 364, μεθιέμεν Δ 351, 'ίμεν Α 170, όρννμεν Ι 353— Ζόμεν Δ 379, μεθεμεν Α 283, επιπροέμεν Δ 94, οντάμεν Ε 132, σγεμεν θ 254. Those from thematic verbs are more common : (σγεμεν A 323, άγορενέμεν Β 10, άγαπαζέμεν Ω 464, Ιιωκέμεν, καιέμεν, νιφέμεν, πολεμιζεμεν, σπευΐέμεν, φερέμεν — είπεμεν Η 373, ελθέμεν Δ 247, φαγέμεν κ 386. On the other hand the Dorians have made the form in -μεν the regular one for non-thematic verbs (Ahr. Dor. 315). This form may be established as Cretan (ΰιΰόμεν C. I 3048V Laconian (άποΰόμεν 1334), Heraclean (ήμεν tab. Heracl. i. 75 etc.), Delphian (είμεν Wescher-Foucart i. 2, 7), Locrian (έζεΊμεν Stud. ii. 453, 1. 3, 8), Corcyraean (άναθέμεν C. L 1841), and Syracusan (προΰιΰόμεν Epich. fr. 71, Ahrens), and is abundantly represented in Pindar by the side of the rare Ionic form ΰονναι (τιθέμεν Pyth. i. 40, στάμεν Pyth. iv. 2) : cp. G. A. Peter de dial. Pindari p. 63. The few instances of the kind in the Attic drama are discussed by Gerth Stud. i. 2, 257. The 98 Aeolians are much less consistent. "We have indeed Boeot. εΐμεν= είναι C. I. 1562, 63, καρτερημεν Athen. χ. p. 417 b, and Thessalian Ιόμεν (Ahrens Dor. p. 529, 1. 12), but the Lesbian Aeolians always said επαίνην (and α'ίνημι), άντλην, νίκαν, ομννν (Conze) : we also find in Conze xii. C. 1. 12, πάρψ—παρέϊναι (from παρίημι : αϊ ΰεϊ πάρην αϊ/τοις τας ΰίκας), which modifies the rule of Ahrens Aeol. 315, that mono- syllabic stems always had the ending -μεναι. There are further. the Arcadian forms ηναι, άπειθήναι, κατυφρονηναι (Michaelis in Fleckeisen's Jahrb. 1861, p. 594), which are of importance in forming our judgment on the Homeric dialect. Evidently the same variety prevailed among the Aeolians as with Homer. For there is an agreement also in the fact ζ 2 340 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM. ch. xv. that at any rate with the Boeotians and Thessalians there are infinitives in -μεν from thematic verbs. Boeot. φαγέμεν (Ahrens Dor. 523),. κριϊΰέμεν Strattis in a Boeotian passage (Ahrens Aeol. 210), φερέμεν (Kecueil d'inscriptions de Beotie par Decharme, no. xxv), Thessal. νπαρχέμεν, εγγραφεμεν (Ahrens Dor. p. 529, 1. 20, 27), all forms which ■ have their parallels only in Homer. There are still a few peculiarities to be discussed. For Homer Bekker (Homer. Bl. i. 69) has discovered the law that 'the language of Homer knows no infinitive in -μεν, in which the penultimate has a double consonant or a long vowel : ' hence while we find γιώμεναι, there is no *γνώμεν, no *στήμεν, *1νμεν, *φορήμεν, and none from passive stems like *?>αμήμεν, *ΰοβήμεν, though forms of this kind are common among the Dorians. To the Aeolians, however, such forms seem to be quite as unknown as to Homer. Length by position appears in ε μ μεν, which occurs five times in JJomer (Σ 364, ξ 334, ττ 419, r 289, χ 210) with the variants εμμεν and είναι, and also in 'ίΰμεν (Λ 719). — Three Homeric infinitives have a surprising long vowel before the endings -μεναι OV -μεν, τιθήμεναι Ψ 83, 247, 'ίμεναι Υ 365, and ζενγννμεν Π 145, which violates the rule just given. Of these τιθήμεναι has been left unassailed, for it is evidently formed upon the analogy of καλή μεναι Κ 125, κιχήμεναι Ο 274. For 'ίμεναι, however, G. Hermann Opusc. i. 242 has proposed 'ίμμεναι, supported by an unintelligible passage in the 99 Et. M. p. 467, 20, for ζενγννμεν ζενγνύμμεν which has no support what- ever. I. Bekker (Horn. Bl. i. 69) approves of both, but has only ad- mitted the former into his text. The model from which this suggestion is derived is άρόμμεχαι Hes. Op. 22, where the reading is by no means certain, so that we ought perhaps to write simply άρω μεναι, a form with an Aeolic colouring after the pattern of the Homeric φυρή μεναι (cp. αλώμεναι). For the two Homeric verbs we must conjecturally assume a transition to the analogy of fhe thematic verbs ; i.e. ϊμεναι is contracted from *ιέμεναι (cp. Ιων, ίοιμι), ζενγννμεν from *ζενγννεμεν (cp. ζενγννον Τ 393). If anyone finds this assumption over-bold, he may remember that ζενγννμι and ορννμι are the only two verbs in -μι which form an infinitive in Homer. The view that syllables naturally short can become long, by what is supposed to be the easier road of the doubling of the consonant, rests upon a false application of the irregularities occurring with the augment, discussed on p. 78. — I. Bekker (Horn. Bl. i. 147) shows that the forms in -ή μεναι as well as those in -εμεν were favoured by the preference for a dactylic ending at the close of the fourth foot. The most curious of all infinitives are those in -μειν, quoted from inscriptions among some of the Dorians, especially among the Rhodians and their Sicilian colonists; these all come from roots without the thematic vowel : Rhod. προτιθέμειν C. I. 2525 b, 1. 99, ε'Ίμειν 2905 c. 7, Gelan αναθεμειν 5475, 1. 25, Agrig. ε'ιμειν, άναθέμειν, αποΐόμειν 5491, 1. 18, 23, 24. In Epicharmus, too, Ahrens Dor. 315 thinks he finds traces of an ending -μειν, and in the appendix he adds άνοικοίομηθήμειν from an inscription from Telos near Rhodes published by Ross. This termination is evidently confined within very narrow limits. It deserves notice that these same inscriptions give us in part the strange form παρείσχηται, παρεισχήσθαι (Ahr. 341), that is, another instance of ει in the place of a wide-spread ε. ch. xv. INFINITIVES. 341 2) Second Group : Infinitives in -ναι. The forms like ψάναι, hdovai, τιθέναι, dtiKvvvai, so common in Attic, are as entirely unknown to Homer as to the Doric and Aeolic dialects. Even of the perfect infinitive in -έναι, like ειΰέναι (Homer, 'ίΰμεΐ'αι), γεγονέναι, as Meister Stud. iv. 422 has already noticed, there is not the «lightest trace to be found in Homer, any more than with the Aeolians 100 and Dorians (cp. Ahrens Dor. 331). A deep rift here runs between the Greek dialects. For the Homeric we may, in accordance with what has been noticed above (p. 340), lay down the following rule, for the distri- bution of the three terminations -μεναι, -μεν, -ναι : -μεναι is used, evidently as an archaism, with the most different stems, -μεν only with stems with a short vowel, -ναι only with those with a long vowel. Hence γνώ μεναι and γνωναι φορημεναι and φορήναι Oepevai θίμίΡ and θάναι occur but not *γνωμ*ν but γνωναι not ουτάναι but ovrapevai or ovrapev not eldevai but tdpevai or 'Lbpev not τεθνάναι but rtOvapevai or τεθνάμςν. There is only one exception to this rule, viz. Ιέναι, but this occurs very frequently, though often with the various readings 'ίμεναι and "ιμεν. If we remember that here the form alone in use later on differs from the genuine Homeric form only by a single consonant, it becomes not impro- bable, as Leo Meyer Yergl. Gr. ii. 279 has already conjectured, that this exception was unknown-to the genuine language of Homer. The col- lections given already on p. 339 f. show that the Dorians and Aeolians do not share this disinclination to infinitives in -μεν with a long penultimate. The disinclination is specifically Homeric, and quite leaves the impres- sion of a custom fashioned in the circle of the epic singers. Infinitives in -ναι after a preceding long vowel are tolerably numerous in Homer, quite in accordance with the later language, as άήναι γ 183, άλώναι Μ 172, βήναι Μ 459, βιώναι Κ 174, γνωναι β 159, Ιιαγνώναι Η 424, Ιονναι Λ 319, Ινναι Β 413, είναι Ν 369, θε'ιναι Ζ 92, στήναι Φ 266, περιφνναι ω 236, and more common than all είι αι. The present form ύιϊοϋναι occurs as a quite isolated case in Ω 425, and must be placed among the numerous linguistic peculiarities of this book. All these forms are not established as Lesbian Aeolic ; how slight the traces of them in Doric are is shown by Ahrens p. 316. Even in Pindar, according to Peter p. 63 hiKvvvai 101 {fr. 171 Boeckh) is quite isolated. On the other hand we find in the Arcadian dialect -ήναι=ειναι (Teg. 10), άπειθήναι, κατνφρονήιαι (ib. 49) answering to the Homeric tjvai and φορήναι (cp. above p. 339). 3) Third Group : Infintives in -εν (-ειν, -ην). In the formation of the infinitive from verbs with a thematic vowel the Greek dialects agree very rnuch more than in that of verbs in -μι. The old forms in μεναι and -μεν grow antiquated much sooner after a thematic vowel. From the very earliest times we meet with infinitives in -ειν (Aeol. -ην, Dor. -εν) everywhere on Greek soil. Even in Homer this ending is far more common in tbe present than the m-formation. If I may trust my collections, there are 151 infinitives in -ει ν (including 342 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM. CH. XV,. the contracted verbs) against 84 m-formations. In this statement all compounds are reckoned along with the simple verb as one. It is espe- cially noteworthy that the contracted verbs can only form their infi- nitive in two ways, either in the Aeolic fashion : χαλημεναι, φορήναι, or with the ending -εν, whether uncontracted, περάαν, όχέειν or con- tracted νεικεΊν, πεφαν. We might indeed conceive of forms in -εεμεν, -αεμεν or -ειμεν, -άμεν, but nothing of the kind is heard of. Evidently the formation of the infinitives in -μεναι and -μεν established itself at a time in which there were not yet any contracted verbs in the later sense of the word. When these came up and formed their infini- tives in a manner corresponding to the other flexion, the forms of the third group, which by this time had been quite usual, were selected ; by the side of these it was only Aeolic forms like ψορήμεναι which held their ground here and there in the usage of the minstrels. The present infinitives in -ειν are Ionic in the wider sense, hence they are also Attic ; but besides they are also current in the so-called milder Doric (Ahrens Dor. 170), and hence they have been recently quoted e.g. from the Locrian dialect (Θύειν, αγειν, φάρειν, Allen Stud. iii. 237). The Lesbian Aeolians used -ην instead (Ahrens Aeol. 89). On inscriptions we find σνμφέρην (C. I. no. 2166, 29), άρκέην, εχην, έπιγρά- φην (no. 3524, 17, 29, 37), φεύγην (Sauppe de duabus inscr. Lesbiacis Gott. 1870, p. 24), in M.SS. άγην (Sappho 1, 19), κρέκην (ib. 90), φρον- 102 τίσΐην (41, 2), on the strength of which -ην is universally written in the infinitive in Sappho and Alcaeus. The same termination is frequently denoted as Doric by grammarians ; but, as Ahrens (p. 158) shows, this can at most be the case for some branches of Doric, especially for La- conian. In Theociitus -ην as an infinitive termination is not certain. Instead of this we find the shorter ending -εν (Ahrens, p. 176) abun- dantly established from different sections of Doric. There is the evidence of inscriptions for the Cretan forms φέρεν, τίκτεν, άναγινώσκεν, σπείρεν (Helbig p. 33), the Heraclean άγγράφεν, εχεν, υπάρχεν (Meister Stud. iv. 421), the Theraean έγγράφεν, the Delphic ay εν, φέρεν, θύεν. Finally the Arcadian Ιμφαίνεν, έπηρειάζεν, υπάρχεν (Gelbke Stud. ii. 26) are to be noticed. Even the conjugation vowel ε of the derived verbs cannot produce a long syllable. In the Theraean dialect there are δωικέν, λει- τουργέ ν, in the Delphic ένοικέν, in Cretan νοέν, ποιέν, in the Pytha- goreans κρατέν, θεωρέν, as against contracted forms of the «-conjugation,, like the Delphian έπιτιμήν, συλήν. The reduction of a double ε to a single one before final consonants is one of the recognised peculiarities of Doric. We come now to the aorist infinitives. For these we must refer in the first place to the thorough investigation by Renner in Stud. i. 2, 32. In Homer we find the ending -εειν according to Renner's calculations 102 times, but only in 14 places before a vowel, e.g. μ 446, είσιΰέειν ου yap κεν ύπεκφύγοι αίπύν ολεθρον, Δ 263 εστηχ ώσπερ έμοί, πιέειν, ore θυμός άιώγη. There are further three other forms, the contracted, e.g. ελεϊν, αΰεϊν, πιέϊν, which in some verbs is the only form of the kind admitted by the metre, e.g. άλαλκεΊν, αμαρτείν, ειπεΊν, πεπιθεΊν, while- other stems on the contrary exclude the open form, e.g. δραθέειν, έρυκα- κέειν, πραθέειν ; secondly, -έμεν : άλαλκέμεν, φαγέμεν, σχέμεν, πιέμεν ; most rarely -έμεναι : άλαλκέμεναι, έλθέμεναι. According to my calcula- tion there are 19 different aorist infinitives, for the most part in very common use, in -εειν, 21 in .«if, 10 in -έμεν, 4 in -έμεναι. Hence her0 ch. xv. INFINITIVES. 343 too, as in so many other cases, the form which in later times is the only- one in use, is even in Homer the most usual. Hesiod, according to Forsteinann de dial. Herod, p. 33, in the Opera, and the Theogony, has only the contracted form (αμψιβάλεϊν, ενισπείν, έλθε'ιν, ih'iv, θαιεΊν), in the Scutum, which for the most part comes nearer to the Homeric language, 5 aorist infinitives in -εειν occur in 6 passages : 103 ελεεϊν 337, λιττέειν 332, μαπ εειν 231, 304, *iiur 252, πραθέειν 240 ; twice before consonants (λιπέειν και, πραθέειι> μεμαώτες), four times before vowels in the main caesura of the third foot. The traces of the ter- mination -εειν in elegiac and iambic poets are extremely faint, while the Alexandrine poets in their imitation of Homer by no means reject them. There is absolutely no instance of an infinitive in -εειν on an inscription. They have disappeared from the text of Herodotus in consequence of the investigations of Bredow (p. 324 ff.) and Stein, the M.SS. evidence for them, which in Hippocrates also is weak, not recommending them. The Attic ΙΰεΊν, ΧαβεΊν, μαθεϊν etc. point by their accent to contraction. In the same way in the Laconian passage in Aristoph. Lys. 1004 σιγήν (= θιγεί »/) is read according to the traces of the Kavenna M.S. ; and the Dorian aorist infinitives in -εν, like the present infinitives of the con- tracted verbs are accented as oxytones : τεκέν, εζεΧέν, εμβαλέν, συναγά- γεις while the Asiatic Aeolic law of accentuation requires us to write ϊείπην, άποθάνην, νπόσχην (Sauppe de inscr. Lesb. p. 24). Reserving for the present the proof of this view, the fact may be here provisionally mentioned, that all these endings may be easily explained from a primi- tive form in -εεν. One extraordinary formation still remains, viz. the few instances of an infinitive in -v instead of -ναι from primitive aorists. There are pro- bably not more than two which can be established, πάρην=παρεϊναι, mentioned above p. 339, and ΰϋν=ΰϋναι (Conze, Reiseauf Lesbos). Since, as we saw, the Arcadians had forms like άνειθήναι, against which there are Lesbian forms in -ην, it is natural here to conjecture a loss of the syllable at, which would also serve to explain the Aeolic passive infini- tives like μεθύσθην=μεθυσθϊ)ναι. φΰν is quite isolated, and extremely doubtful in Parmenides v. 65 ed. Karsten, where the M.SS. give the unmetrical verse. vorepov ή πρόσ&ν τον μηδενός άρξάμενον φϋναι. την is just as doubtful in Anthol. xi. 140, where the M.S. has ois ου σκωμμα Χίγςιν, ου πέιν φιλον while Planudes gives πίν. We should probably read πιείν with synizesis. Finally Ιοΰν twice occurs before vowels, viz. Theogn. 104 τοϋ μεταΐοΰν 104 εθέλοι and Phoenix Coloph. v. 20 νόμος κορώνα χείρα £οΰν επαιτοΰση. Perhaps it is best to indicate by an apostrophe in both cases the elision of the at. As the fourth group of the infinitives, the aorist infinitives in -σαι, show no variations whatever within the range of Greek, and the fifth group containing the middle infinitives in -σθαι very slight ones, there is no reason to occupy ourselves more in detail with pointing out these forms, and we may at once turn to the question of the origin of the different forms, a question which in some points is one of difficulty. 344 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM. ch. xv. The nature of the infinitive, both formal and syntactical, has been thoroughly discussed of late, especially in the writings of Wilhelm de infinitivi forma et usu, Eisenach 1872, and Jolly Geschichte des Infini- tive im Indogermanischen, Munich 1873. By way of elucidation of the Greek infinitives the following important facts are brought out by this discussion : 1) In every infinitive form we have to expect, besides the verbal stem, which in many cases is still further characterised as a tense stem, a nominal suffix and a case suffix. 2) The same multiplicity of nominal suffixes, which has already met us in Greek, prevails in the infinitive in several of the cognate languages, above all in the Vedic dialect ; and this is sufficient to warn us beforehand to use the greatest caution in the attempt to prove the identity of forma- tions which only distantly resemble each other, without adequate reasons. 3) In Greek we can discover with certainty only such case suffixes as denote the locative or possibly the dative. The locative is adapted for the infinitive, inasmuch as it denotes the sphere within which an action takes place, the dative, inasmuch as it denotes the object with reference to which an action is done. Now if we consider in the first place the first group of active in- finitives, the polysyllabic forms in -μεναι, with which we were concerned on p. 339 find their complete correlate in the Yedic forms in -mane, of 105 which Delbriick Ind. Verbum p. 226 however quotes only five instances. But of these, five, two correspond in stem also to Greek forms : da-mane =ΰό-μεναι, vid-mane^'d -μεναι. We have further the Zend infinitives in -maine (Wilhelm p. 14) like ctao-maine praise from rt. ctu=Skt. stu. There can be no doubt about the origin of these forms in Sanskrit -man is here one of the most common suffixes for the formation of abstract substantives, and -man-e the dative of this suffix. From the stem vid-man there is an instrumental, from the stem da-man a genitive as well as the dative used as an infinitive. The case is the same with the Zend -maine, the dative of the nominal suffix -man. Hence if ΰή-μεναι is identical with da-mane, which does not differ at all phonetically, it follows that co -μειαι also is a dative. And such is the opinion of Bopp Yergl. Gr. iii. 2 324, Leo Meyer Vergl. Gr. ii. 281, Delbriick Ztschr. xviii. 82, Wilhelm p. 14. I have hitherto in agreement with Schleicher (Compend. 3 p. 401) especially in the < Elucidations ' 2 p. 197 (E. T. p. 221, but cp. pp. 198-202 of the third German edition) held a different view, influenced chiefly by three considerations. In the first place it seemed to me im- probable that the infinitive ending -με rat contained a different case from -εναι. But the latter e.g. in ειΐ-εναί seemed from its similarity to the Skt. ved-ane the locative from a stem vedana, to be established as a locative. This argument is not valid, because, as we have seen already, -f ναι is post-Homeric, and, as will be shown, is very different from this -arie. Secondly, while Latin has a suffix -men, corresponding to the Skt. neuter -man, Greek gives only neuters in -μα (from μαν). Hence we should rather have expected -μαν-αι. On the other hand it seemed to be easy to connect -μεναι as a locative of a feminine stem -μένα with the Greek participles j and there was little to surprise one in the middle force of the latter, inasmuch as it was evidently only by degrees that the distinction of the genera verbi became established, while in abstract ch. xv. THE ORIGIN OF THE INFINITIVE TERMINATIONS. 345 substantives like π\ησ-μονη, which are unmistakeably akin, there is nothing of a middle character (cp. Jolly Infinitiv p. 85). There is no great weight in this consideration, because the heavier vowel α in these much-used forms, after their connexion with those in μα(ν) had become obscured — cp. ποιμήν, πο ι μένος — might easily have been 'thinned' into ε. Thirdly -at nowhere occurs as a suffix of the dative. But as -at is 106 the regular representative of a Skt. e, e.g. in the personal endings -rat =te, -vrai=nte, the assumption that the old -ai in this place retained its full strength, has at least some analogies in its favour. The impro- bability that the Skt. da-marie and U μεν α ι agree purely by accident, and the impossibility of explaining the Sanskrit form otherwise than has been proposed, i.e. as a dative, remain the principal facts. Hence I agree now in regarding the datival origin of the Greek forms in -μεναι as established. Then comes the question how the shorter -μεν is related to this -μεναι. The almost universal opinion, to which I have hitherto un- reservedly given my assent, is to the effect that -μεν simply arose from the fuller -μεναι by the loss of the -at. A confirmation of this view could apparently be derived from the accent. For the accentuation οντάμεν, είπεμεν, άγέμεν, which violates the ordinary rule for the accent of verbs, points decidedly to some loss at the end. But it is still difficult to assume that the long diphthongal ending has been dropped. The loss of the -cu could almost be supported by the elision of the -at in verbal endings before vowels, and by the consideration that the consciousness of the meaning of this element must have become very faint even in the earliest times. A further support, which this view seemed to find in the assumption that the ending -εν was in the same way a shortened form of -ε ναι, quite fails us. For, as we shall see, -εν never occurs side by side with -ει at, and is a completely distinct formation. And it is only in the case of the Lesbian Aeolic infinitives in -ην by the side of -ηναι (μεθύσβην) mentioned on p. 343 that there is any more pro- bability in the loss of the diphthong, although even here another view is possible. Under these circumstances it seems to me very doubtful whether we can refer -μεν to -μεναι, and it appears more probable that, as others also have conjectured, -μεν has lost a simple -t. Thus ΰύμεν-αι would be dative, Ζόμεν for Ιόμεν-ι locative of the nominal stem ΰομεν. The loss of the t has its complete analogue in the preposition εν by the side of ενί, and a more distant one in the 2 sing, in -εις for *εισι, *εσι and in the plural datives and locatives in -oig, -αις by the side of the 107 earlier -οισι, -αισι. In this way we might explain also the Bhodian forms in -μειν : βέμειν would be to θέμεν from *βέμεν-ι as fir to iv from ενί, that is to say, we should have here an example of epenthesis. How- ever I should not like to attach too much weight to this latter conjec- ture. Naturally we cannot suppose that there was any more difference of meaning between -μεναι and -μεν, than in Sanskrit between the infini- tive forms originating in the dative and those from the locative. The category of the infinitive, which Jolly very justly places along with the adverbs, rests rather upon the petrifaction of the case-suffix. We come next to the second group, the infinitives in -ναι. For the correct explanation of these forms the statistical data given on p. 340 ff., which have not as yet received due attention, are of the very greatest importance. Previously, starting from the Attic ειίεναι, I regarded the ε as an essen- 346 VEKBAL NOUNS OF PKESENT AND SIMPLE AOEIST STEM. ch. xv. tial element of these infinitives ; and hence I compared e.g. hilivai with the Skt. locative vedane, which is used in later Sanskrit as an in- finitive. But it is an error in philology to compare an evidently late Greek form with a Sanskrit form which is likewise somewhat recent. And in fact the traces of Vedic infinitives in -arie, as Delbruck p. 225 shows, are extremely scanty. The two which rest on certain evidence, dhurv-ane ad p'ugnandum and turv-dne ad vincendum, are regarded by Wilhelm p. 15 as datives from stems in -an. If we consider how in the Vedas the most different stem-formations are in isolated instances used in particular cases in a manner resembling the infinitive, we shall admit that this agreement may very easily be purely accidental. On the other hand two facts of the Greek usage can hardly rest upon chance, viz. firstly that the Homeric language, with the exception of Uvea, mentioned on p. 341, and probably incorrectly recorded, has only infinitives in -rai with a long penultimate ; and secondly that the penultimate, contrary to the general course of verbal accentuation, in all Greek dialects has always the accent : γνωναι, υντάναι, τίθεμαι. The two facts combined make it probable that this termination has suffered the loss of a 108 syllable before the v. The vowel before the v f which in Homer is always long, and which in all Greek without exception is accented,, always belongs to the stem, and not merely in hdo -ναι, φά-ναι, τεθνά-ναι^ φορή-ναι, where this is self-evident, but also in ειΐέ-ναι, which is con- nected with εΙΙέ-ω, εΐΰε-ίη-ν. Hence we cannot talk of an ending -εναι τ but only of -ναι. Now some might wish to make use of these facts to show that the -ναι, to which we are thus brought, is a shortening of -μεναι through the intermediate stage -μναι. We should thus have . series like γνω-μςναι * γνω-μναι γνωναι φά-μ€ναι * φα-μναι φάναι. The syncope of the ε would find support phonetically in forms like μέΰι-μνο-ς, μέρι-μνα, τέρα-μνο-ς and in Latin words like alu-mnu-s, colu-mna. But it would be difficult to adduce a single Greek analogue for the reduction of μν to a mere ν : μν is an extremely favourite com- bination in Greek. The form lovvai leads us on another track. Benfey Orient u. Occident i. 606 (cp. Wilhelm p. 17) compared this with the Vedic ddvdrw, which is indeed according to Delbruck p. 225 quite isolated, but which occurs frequently, and can only be explained as the dative of a nominal stem da-van, derived from the rt. da. This com- parison remained till lately a mere conjecture, but it has recently found a striking confirmation in the Cyprian hfivai, which occurs twice (cp. Deecke u. Siegismund Stud. vii. 248). Now are we to suppose that lovvai was formed in a quite different manner from ΙιΙοΰναι, or that the isolated Homeric lilovvai is an absolutely distinct formation from ΙιΙόναι τ and both again from γνωναι, θ είναι, φάναι 1 In this way we arrive at a certain degree of probability that all infinitives in -vat have originated in Η ναι. As όονναι came from lap ε ναι, so θεϊναι would have come from θεΐεναι, γνω-ναι from yvia-hvai. The post-Homeric forms ΙιΙόναι, φάναι, είδε- ναι would be later formations, arising at a time in which the medial / was completely extinct. But at least in the accentuation of the penul- timate there still remained a reminiscence that it was once an ante- penultimate. Cp. 'ί-κο-μεν for ε-κ^-μεν from the rt. κοϊ (Princ. i. 186), ch. xv. . THE ORIGIN OF THE INFINITIVE TERMINATIONS. 347 κλόνι-ς for κλοβ-νι-ς (ib. i. 184). I do not venture to give this explanation as a certain one. But it will be admitted that it agrees with the 109 recorded facts and with phonetic laws. It would not be absolutely impossible that ψά-ναι, τώέ -vcu and the like should be locatives formed by means of the suffix -να j but then the accent would remain entirely unexplained. All the earlier discussions of the forms of the third group start from the arbitrary assumption that the ν of the ending -εν was a mutilated -ναι, hence that λέγειν originated in *λεγειναι. Now under the head of the μ -formation it seemed to us that there were serious objections to the admission of such a loss of the end of the word : and hence all proba- bility for such an assumption in the case of the third group is wanting. For in no single instance within the limits of the present and aorist formation is there even the slightest trace of a longer form by the side of the shorter known to us. Forms like *\εγειναι or *λεγεναι or *ψ>ν- γειναι, *φνγεεναι or anything of the kind that has been imagined, no- where exist. For the immense mass of verbs with a thematic vowel the infinitive in -ειν (Dor. -εν, Aeol. -ην) is established beyond doubt from the earliest times. To derive very ancient forms common to all Greek, like λνειν (λνεν, λύην) from forms like γεγονέναι, ειΐέναι, in which we take the ε to be an element of the tense-stem, and which make their appearance only at a comparatively late date, is moreover forbidden by the accentuation. Besides we could not thus explain either the doubled vowel in ψνγέειν (or φνγέεν), or the length of the final syllable in λνειν, λνην. For the assertion that the vowel of the preceding syllable was lengthened in compensation for the dropped ai, happily belongs to a point of view which has long been passed by philologists. Leo Meyer Yergl. Gr. ii. 281 traces back forms like λνειν, φνγεϊν to λνέ-μεν, φνγέ-μεν. Certainly in this way the diphthong of the ordinary form and the doubled ε of the Homeric would both be explained. But it is quite im- possible to adduce any analogy for the loss of a consonant so much liked and so common as μ from the middle of a word between two vowels. No one will appeal to the loss of m in the Skt. e=*me of the 1 sing, middle, for. we have here to do with a fact which falls within the life of the Greek language; and still less to the explanation of the Greek 1 sing, mid. in -μαι from ma-mi, defended on p. 55, for here too, even in the 110 earliest times of the Greek language as we know it, no pi remained be- fore the i. The loss was occasioned by the aversion to too much repe- tition of the same sound in formative syllables. Forms like θέμα, 'όνομα, εύρημα are among the commonest in Greek; and as the infinitives in -μεναι and -μεν are evidently akin to the substantives in -μα, we cannot see what can have induced the language to get rid of this μ only in the infinitives, while everywhere else it had no objection to it. For this very reason this theory was very early opposed by another, which re- garded the suffix -εν as completely independent of -μεν. As early as my essay ' De nominum Graecorum formatione,' p. 56, I expressed myself in favour of this, though I there expressed several erroneous opinions on the subject in question. Schleicher agreed with me in principle, tracing φέρειν back to a primitive form *φερεναι, totally distinct from φερέ-μεναι (Comp. 3 p. 411). But in the first place *ψερεναι is, as we have seen, a pure fig- ment of the imagination, destitude of any analogy ; and in the second place even with the help of this hypothetical form Schleicher can only 348 VEKBAL NOUNS OF PRESENT AND SIMPLE AOPJST STEM. ch. xv. explain the diphthong of φέρειν by assuming *φέρεη as an intermediate step between *φερεναι and φέρειν. But in what other instance have we a final ο ι shortened to ι 1 An advance is marked by Scherer, Zur Gesch. d. d. Spr., p. 474, who, though he also refers φέρειν to *ψερενι, does not take this hypothetical form as the shortening of an equally hypothetical *φερεναι, but as the locative of a neuter stem φερ-εν. But this view also is untenable j for, to say nothing of the fact that such stems in -εν (cp. Lat. ungu-en) are unknown in Greek, it would be absolutely im- possible to explain from a primitive form *φερενι the Boric φέρεν and Aeolic φέρη> } or from an analogous *φυγενι the Homeric φυγέειν and Attic φυγε'ιν. The forms of the Greek dialects, like Aeol. φερην Dor. φερεν Ion. φέρειν point with certainty to a primitive form *φέρεεν, where we must take φέρε as the stem, εν as the ending, as also in the Homeric aorist, the termination of which will require a more thorough discussion imme- diately, we find both e's side by side. This is the place to return to those remarkable forms, which have not hitherto received due consideration in the discussion of the infinitive 111 formation, as we promised to do on p. 343. We saw that the ει of the present infinitive in -ειν points to contraction. But φέρειν cannot have come from *φερε-ειν, as we might at first conjecture, for then the form would have to' be perispomenon • and even supposing that the Tonic diphthong « had arisen otherwise than by contraction from ε ε, the Doric and Aeolic forms φέρεν ψέρην would be inexplicable. But if φέρειν goes back to φερε-εν, for φυγείν we have evidently to expect not φνγέειν but *φνγέε'ν as the earlier form. For between the present and the thematic aorist stem there is not the slightest difference of declension in respect of personal endings, moods and verbal nouns, with the one exception that the aorists in the infinitive and participle are inclined to accentuate the element which does not belong to the stem. This tendency to which we shall return under the head of the participle, gives us the inestimable advantage of being able to recognise φυγείν at once, by the accent, as a contracted form. That these active aorist infinitives are peris- pomena, and the corresponding middle forms are paroxytones is the well- established doctrine of the old grammarians. We may here simply refer to Arcad. 173, 20, Herodian ii. 185, 25. Certainly this doctrine was based upon the observation of the living language. There were only a few Homeric, i.e. no longer living forms, on which there were doubts and controversies (cp. Gottling Accentlehre, p. 56, which we had occasion to mention on p. 276. In the old copies of the Homeric poems the infinitives under dis- cussion must have been written ΦΥΓΕΕΝ ΙΔΕΕΝ etc. For the μεταχα- ρακτηρίζοντες of the Attic time, who themselves said φυγείν, Ιΰεϊν, and who, as may be concluded from numerous facts, were of opinion that the poets not uncommonly allowed themselves the * pleonasm ' of a vowel before another of the same sound, it was extremely natural to reproduce these forms by φυγέειν, ίϊέειν, and so not merely to satisfy the require- ments of the metre, but also to approximate to the Attic practice. The forms in -ttv, required by grammatical analysis, adapt themselves to the verse, as we have intimated above (p. 342) in the great majority of 112 instances without difficulty. Where a consonant immediately follows ' Λ THE OKIGIN OF THE INFINITIVE TEKMINATIONSf / . 34&/Γ such forms, this is self-evident. Of the 14 passages in Homer, which contain such infinitives before vowels, there are 7 in which the length of the termination -εν may be completely explained by the main caesura of ' A ► the third foot, verses like y^J Σ 511 ψ διαπραθεειν η ανδιχα πάντα δάσασθαι y + besides Τ 15, Ψ 467, Ω 608, α 59, ι 137, λ 232. There is nothing more surprising in the quantity of the syllable -εν here than in Β 228 πρωτίστω δίδομεν, eur' αν πτολίεθρον ελωμεν. Again in the 5 passages in which such an infinitive stands in the caesura of the fourth foot, there is not much to astonish us in the lengthening *of -εν, e.g. ε 349 αψ άποΚυσάμενος βαΧεειν els οΐνοπα πόντον Cp. A 263, Κ 368, X 426. Compare Η 418 (cp. 420). άμφότερον, νεκυάς τ άγέμεν, έτεροι δε μεθ' νΚην. In θ 508 ερύσαντες with its Ρ would be quite enough to lengthen the final syllable of *βαλέεν (cp. Stud. vi. 266). Two verses remain in which the polysyllabic character of the word makes such a phenomenon explicable even in the second foot : μ 446 εϊσιδέειν, ov yap κεν νπεκφνγοι αίπνν οΚεθρον τ 477 πεφραδεειν εθελονσα φιΚον πόσιν ένδον εόντα. The lengthening of a final syllable short in itself would here come under the same head as the cases, quoted by Hartel Horn. Studien i. 2 116, e.g. εΐΰόμενος Άκάμαντι Ε 462. The examples from Hesiod are all in the caesura of the third foot. It will be seen how easily the forms in -εεν can be introduced into the text of Homer and Hesiod. But we have not quite finished even yet with these forms. The stem-forming suffix and the case-ending have yet to be determined. Scholars used to be disposed to put this infinitive ending side by .side either with noun-formations in -εν, -ov, or with those in -ενα, -ova. But -εν occurs in Greek only in a few obscure substantives like ανχ-εν, -ov somewhat more coirimonly, e.g. in εϊκ-ον, but without anywhere sharing anything in common with infinitives. - Feminine substantives like ηΰ-ονή come nearer in meaning. But who can believe that a suffix with one initial vowel, having for its primitive form -an or and, was from the first added IIS immediately to verbal stems ending in a vowel. It is by no means usual in the primitive formation of words to occasion accumulations of vowels and contractions. It looks much more as if the concurrence of vowels in φενγε-εν, ψνγε-εν, as in countless other cases, e.g. in γένει, 'ίκεο, ενχεαι, φάεα, νείκεε, had originated in the loss of a consonant. The question then arises, what consonant 1 According to Greek phonetic laws, as we have seen, we cannot suppose the loss of a μ, which has often been con- jectured : but we must take our choice among the three spirants j, v, and s. I do not see anything in favour of the first two ; for we should have to imagine a locative from the suffix -van, -vani, corresponding to the dative -vane (da-vane), but there is no hint of anything of the kind. On the other hand the third would find some support in the Vedic ending -sani. Delbruck p. 227 quotes 8 forms of the kind from different stems, e.g. pra-bhu-shdni from the rt. bhil=zGr. φν. There is, it must 350 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM, ch. xv. be admitted, no such infinitive from a stem in a thematic vowel, as we might expect as an immediate parallel for the Greek forms, no *bhara- sani as an analogue to φερε-(σ-)εν. If. this comparison is correct, we must assume a loss of the final locative i } i.e. the same process which we regarded as probable for the ending -μεν on the analogy of εν as compared with the Homeric IvL In this way we might even find some support for the diphthong ει by the side of the thematic c, in opposition to our pre- viously expressed view : ψυγέ-ει ν might have come, by means of epenthesis, from bhuga-sani, like είν from ενί and possibly the Rhodian -μειν from mani. However, there are weighty reasons against such a combination. For as the Doric ψνγέν, 6ιγήι> cannot possibly be traced back to ψνγέειν, θιγέειν, we should have to regard the ending -ενι, -ειν with a transposed ι as a peculiarity of the Ionian main dialect. But even within this dialect the ει of the ending ειν is evidently not genuine. The contracted verbs show no trace of the ι in forms like τιμάν, μισθοϋν, and the accentuation of φέρειν makes a contraction from *φερεειν inconceivable. We must therefore reject the hypotheses of an epenthesis of the ι. 114 According to the earlier hypotheses as to the infinitive suffixes the most usual Greek infinitive formation seemed to find its most complete analogies in the Teutonic languages, the German ending -an being com- pared with the Greek -εν, though this created fresh difficulties. If our parallel between -εν and -sani is correct — and I do not see anything which could prohibit such a comparison — Greek here approximates to the oldest Indian. It is not without significance from this point of view, that this infinitive ending is in the Vedas one of those ' which,' as Jolly, p. 132, puts it, 'deserve the name of infinitive in every respect,' i.e occur also as imperative. This discloses also a kinship with other Indian and Latin forms. It is natural to conjecture that the ending sa-n-i is connected in its first element with the suffix -se, which is added rarely, e.g. in (fi-slie from the rt. tfi, conquer, directly to roots, and more often to stems in a, and then (cp. Delbriick, p. 223) is taken as -ase, e.g. tflva-se, live (pres. (jlva-mi). This formation has long been recognised as the analogue of the Latin infinitive, e.g. da-re for da-se, vwe-re for vlve-se. Thus *λεγε-σεν, conjectured as the earliest Greek form, and lege-se the earliest Latin form come tolerably near to each other. The Latin, as has long been recognised, agrees more exactly with the ending of the Greek infinitive of the sigmatic aorist, which we set down as the fourth group of Greek infinitive forms. We shall enter upon this more fully under the head of the sigmatic aorist. Finally we have still to discuss the fifth group, the middle infinitives in -σθαι, which are common to all Greek from the oldest times. The few deviations like the Locr. ελέσται=ε\έσθαι, Cret. άναιλίθαι, and the doubtful ΐτθαι• καθίσαι Hesych. have been already mentioned on p. 69 f. The Vedic dialect frequently makes use of the termination -dhjdi, which is undoubtedly to be compared with -σθαι, though no middle force is perceptible in it. The latter is something peculiar to Greek; and Jolly's investigations show that there are few languages in which the infinitive undergoes that more delicate development, which renders it capable of expressing ΰιάθεσις. The Latins have formed for themselves 115 for the passive a form in -ier, subsequently l, which in spite of all the pains spent upon it has not yet been completely deciphered. The Greeks selected one out of the numerous adverbialised infinitive-like formations ch. xv. PAETICIPLES. 351 and by degrees confined its usage to an exclusively middle sense. The similarity of these forms, after they had been modified by phonetic changes, to Greek middle forms like φέρεσθον, φέρεσθε, ψερέσθω, ψερέσθων undoubtedly favoured this limitation of usage. We have Indian parallels to Greek infinitives in (Delbriick p. 226) : hhdradhjai = φίρίσθαι sahaclhjdi = ί'χβαθαι sattaclhjai = €π€σθαι 9 It is worth noticing that such forms are also developed from expanded and derived stems, e.g. piba-dhjdi like γίγνε-σθαι, prna-dhjai like τέμ- νε-σθαι, mandaja-dhjdi like τιμάσθαι. The same termination appears in Zend as -djdi or -dhjdi, agreeing in usage with the Indian form ; verez- jeidjdi, as Jolly Inf. p. 87 writes, or verezidjdi as Spiegel and Justi read, the infinitive of the rt. varez (verez) which corresponds to the Greek hpy or ϊρεγ do, agrees letter for letter with the Greek ΐρέζεσθαι. The same language supplies the isolated form buzdjdi, or as Justi writes it bu-zlidyai —φνεσΰαι. In the sibilant which is found here before the termination Jolly I.e. 1 recognises a correlate of the σ, which regularly appears in this place in Greek. He conjectures that the sibilant is a relic of the rt. as, to be, ' be.' It seems to me more natural to think of the suftix -as or s(a), which, we have already seen, was contained in the Indian and the Latin, and probably also in the Greek forms. On the analogy of simple formations like ψέρεσθαι, such a suffix might also have been connected with more complicated tense-stems, e.g. in λέξεσθαι, Χέζασθαι. However, on p. 71 we found a means of explaining the Greek -σθαι, according to which the σ might be of purely phonetic origin. Whether the entirely isolated Zend form necessitates a different analysis, 116 I will not now discuss, any more than the further question what relation it has with the 0=Skt. dh, Zd. d or dh. It may be regarded as estab- lished that this consonant here essentially forms a stem, and therefore has the same force as in the substantives μισ-θό-ς, στή-θ-ος, πλή-θ-ος, αχ-θ-ος, and also that -dhjdi is the dative of a stem in -dhi. But whether there is any connexion between this dh and the rt. dha place, do, as has often been conjectured, is a question which cannot be settled without more profound investigations of the Indo-Germanic stem formation. Cp. moreover Wilhelm de infinitivo, p. 23. II. PARTICIPLES. As compared with the variety and obscurity of the infinitives, the participles of the first two tense-stems present a very simple appearance. No one can fail to see that a definite and somewhat strictly limited por- tion of adjectival formations, even in the time of the Indo-Germanic unity, approximated to the verb in respect of government and varia- bility, and continued to exist in the separate languages with forms clearly marked and but slightly altered, while in the case of the infini- tive very different attempts were made, of which only a small portion found more extended acceptation. 1 In the like manner Justi Handbuch des Zend, p. 372, says ' the inserted zh belongs originally to the termination.' 352 VERBAL NOUNS OF PRESENT AND SIMPLE AORIST STEM. ch. xv. A) The Active Participle of the Present and Thematic Aorist. The stem-forming suffix was originally after consonants -ant, after vowels -nt. As the latter suffix is unpronounceable, we shall have to start with -ant, and to explain the shorter suffix by supposing that the a disappears after a preceding vowel in the stem. The phonetic relations are precisely the same as in the formation of the 3 pi. in -anti, -ant and -nti, -nt (p. 46). The following may serve as instances of the corresponding usage of this suffix for the same purpose. Skt. atfant, Gr. άγοντ, Lat. agent „ arhant, Gr. άρχοντ „ bharanf, Zd. barant, Gr. fapovr, ~L&t.ferent } Ch.-Sl. berg (nom. sing.) „ jant, Gr. lovr, Lat. eunt „ sant (for asant), Zd. Kent, Gr. iovr, Lat. -sent (prae-sent). 117 The different colour given to the vowels in Greek and Latin deserves notice. Greek prefers the dull ο with the consistency which distin- guishes this language. Latin is less regular ; and here and there, e.g. in euntis, voluntas (for volunt-tas), voluntarius and in sont, which ac- cording to Clemm's acute explanation (Stud. iii. 328 ff. [cp. iv. 205, viii. 344]), is identical with sent, the duller vowel appears by the side of the prevalent e. The initial vowel of the participle was after- wards completely lost in the Greek verbum substantivum. The Attic ώ ν is to the Homeric and Herodotean εών, just as Ion. δρτή is to Att. εορτή or as Att. ος to Horn. εός. It is an instance of the phenomenon of hyphaeresis, thoroughly discussed by Fritsch Stud. vol. vi., esp. pp. 111-113. Westphal's attempt (Griech. Gr. ii. 106) to trace back 6ντ to sant breaks down from the fact that an initial sigma does not thus simply disappear. The Doric form εντ (dat. pi. εντασσιν tab. Heracl. i. 104) is formed upon the analogy of θεντ, τιθεντ, with the loss of the final ς of the root. The Homeric forms ε με vat, ε μεν as infinitives are similar. The accentuation of the suffixed syllable, which has become the rule in the aorists : ιΰών, λαβών, may be placed among the accentual ten- dencies mentioned on p. 348. But forms like Ιών, εών, εκών and Indian present participles belonging to the so-called sixth class, like rdhdnt (rdh, ardh grow), dvish-ant (dvish hate) rutfdnt (rug break), show that the occasion for this was not an internal one, connected with the essen- tial character of the aorist, but an external one, connected with the slight phonetic substance of the stem- syllable. It is worth noticing how far back this twofold accentuation goes in point of time. Another variation, which appears in the Asiatic members of our group of lan- guages, on the other hand, has left no traces in Greek, so far as participles proper are concerned, the variation of the nasal in the participial suffix. In Sanskrit only a small portion of the cases retain the n, the majority reject it : nom. plur. bhdrant-as = φέροντες, but ace. plur. bharat-as= other poets Cp. εσσοημένον ' τεθορυβημενον, ωρμημένον Hesych. There is absolutely no support for a lost consonant in the case of the χ 31 roots λα/3 ?> λαχ, λεγ, as has been shown, in Princ. ii. 144, 111, i. 454. εμμορε and ε'ίμαρται have been compared with the Skt. rt. smar, think, and an indication of their connexion has been found in the rough breathing of είμαρται, which would thus be from *σε-σμαρ-ται. But the difficulty in the way of this combination has been already pointed out in Princ. i. 413. This verb is connected rather with a rt. μερ divide (μέρος, merere) the meaning of which is far removed from that of thinking. The dialectic forms εμβραται * ε'ίμαρται Hesych., έμβραμένα ' ειμαρμένη Hesych. and Ε. Μ. p. 334, 40, 2 ώβρατο' ε'ίμαρτο Hesych., are also to be taken into account. Besides, we cannot deny the appearance in other instances also of a rough breathing of later origin. For σενω (Princ. i. 475) the loss of a spirant after the σ cannot be proved with certainty, in spite of εσβενα. The key to all these perfects is hence doubtless to be .sought in metathesis, a notion which has been established by Brugman • 2 εμβραμενα. Παρά Ί,ώφρονι η ειμαρμένη. Και Λάκωνες ούτω Κε^ουσιν. 362 PEKFECT STEM AND FOKMS CONSTKUCTED FROM IT. ch. xyi. Stud. iv. 102, 124, and Siegismund v. 211, following Pott Et. Forsch. •ii. 2 389. Bailly is of the same opinion for a part of the verbs. From *λέ-ληφα, *λέ-λ»;χα, *λέ-λεγμαι, *μέ-μορε, *σέ-σν-μαι came *e\i-\i](f>a etc. by the prefixing of an irrational initial ε, then by the suppression of the second vowel *ελληφα, εμμρρε, εσσυμαι, and finally in the case of some verbs at a later date, when the first λ gave up a part of its ' voiced sound ' to the preceding vowel, έίληψα, ε'ίληχα, εϊλοχα, εϊμαρται. Hence εμμορε isto εϊμαρται much as the Lesb. aor. ενεμμα is to the Attic ενειμα. Following this view, one might be inclined to hold even that ελλαβε (explained otherwise on p. 78) was a metathesised *λέλαβε (cp. λελαβέσθαι) and that εμμαθε came from *μέμαθε. In the following perfects ε appears as the syllable of reduplication before vowels.• It has long been recognised that this anomaly, like the syllabic augment before vowels, is to be explained from the dropping of a spirant, to which in the case of reduplication was added later on the loss of the same letter at the beginning. As in the case of most of the stems which belong here, the original initial letter was pointed out on p. 79 f., we may content ourselves here with a brief statement. "We have to do with the following forms : 132 1) έαγε Hes. Opp. 534 έπϊ νώτα εαγε; then in Attic poets and Plato: κατεαγότες C. I. A. ii. 61, 42, κατεαγυτα Herod, vii. 224. The initial digamma may still be detected in Sappho 2, 9 γλώσσα εαγε, where Bergk on the analogy of Lesbian instances to be mentioned immediately conjectures γ\ώσσ έναγε. 2) εάδα I 173 τοισι ΰε πασιν εαΰότα μνθον εειπεν ) then in Alexan- drine writers. The stem expanded by ε has retained its } among the Locrians^fa3r/9ora inscript. of Naupactus (Stud. ii. 445) 1. 38. 3) Ιάλωκ-α. ενάλωκα was quoted on p. 79 as a certain proof of the digamma ; εάλω /ca is Attic from Aeschylus onwards, while Pind. Pyth. 3, 57 uses αλωκα, and ήλωκα occasionally makes its appearance in Herodotus and Attic writers, generally with the variant εάλωκα. 4) εελμαι. Ω 662 κατά άστυ έέλμεθα, Ν 524 Διός βουΧ^σιν εεΧμένος. There is the plupf. ήέΧμεθα' συνήμμεθα Hesych. and εόΧει Pind. Pyth. iv. 233 (Boeckh), and with a stem expanded by ε εόλητο Apoll. Rhod. iii. 471, to which belongs εόληται' τετάρακται, and ενΧητο' επεφυρτο, ετέτακτο Hesych. 5) εεργμένοι Ε 89 γέφνραι εεργμέναι Ισχανόωσιν. This verb was passed over in treating of the augment, because even in forms without the augment or the reduplication there are traces, of a double ε : Apoll. Rhod. iv. 309 εέργεται, so that the ε -may possibly be of the same kind as in εε£να. But the f is well established. Cp. Princ. i. 222. 6) εερμένος. σ 296 ηΚέκτροίσιν εερμενον ) and also εερτο' έκρήμνα Hesych. Cp. Ι/ειρε p. 81. 7) εεστο (Μ 464) may find a place in our list here, though it is possibly not for h-ha -το, but for ε-βεσ-το, being thus a past tense from the unreduplicated Ησ-μαι, εσ-μαι (3 sing, επίεσται, oracle in Herod, i. 47). 8) εοίκα, common from Homer onwards : he has also είκτον I 27, είκτην A 104 and the middle forms Huero ν 31, εικτο Ψ 107. The / is probable, but not cei-tainly established. Cp. Princ. i. 309. 9) εολπα in Homer, Hesiod and Alexandrine poets: X 216 νϋν &) νώί γ* εολπα, Υ 186 χαλεπώς 3ε' σ' εολπα το ρεζειν, plupf. εώλπει <(> 96. Here as in no. 5 the double vowel extends itself also to forms like εΆπεται ch. xvi. BEDUPLICATIOF WITH AN INITIAL CONSONANT. 363 Ν 813, εελπυίπην θ 196, in which the first ε can only be a prothetic 133 vowel. For the origin of the verb see Princ. i. 328. 10) εοργα. Γ 27 οσσα εοργας. The plupf. εωργει £ 289 was dis- cussed on p. 81. Cp. p. 86. Herodotus also has the form εόργεε i. 127. 3 11) εονρηκα (Hippocr. ονρηκα) has been already discussed p. 80, and quoted from Aristophanes. 12) εώνημαι in Aristoph., Plato, Lysias, Demosthenes with the plupf. εώνητο, Aristoph. Pax 1182 τω Μ σιτί' ουκ εωνητ. Cp. p. 80. An active εωνηκώς is quoted from Lysias in Bekker's Anecd. p. 95, 25. 13) εωσμαι (Herod, ώσμαι). Thuc. ii. 39 άπεώσθαι and similar forms elsewhere in Attic prose. Plutarch has also εωκα. Two perfects have distinctive peculiarities of the same kind as those mentioned on p. 81, viz. 14) άν-ε'ωγα, quoted from Hippocrates and post- Attic writers, άνέωχα from Demosthenes and Menander, άνέωγμαι, άνεωγμην common from Euripides (Hippol. 56 ου γαρ oW άνεωγμένας πύλας) onwards, with the un-Attic bye-forms ήνοικται, ήνοικτο. 15) εώρακα with pluperfect forms (εωράκεσαν Thuc. ii. 21) common in Attic prose, with εώραμαι since Isocrates. With a short second syllable in Aristoph. Thesm. 33 ονχ εόρακας ττωποτε. The root is fop, Princ. i. 432. In both verbs the interior strengthening is the result of a trans- ference of quantity. It is different with 16) ε'ίωθα, from Homer onwards (E 766 ή με κάκιστ είωθε κακής όΰννιρσι πεΧάζειν) though he has also the form εωθα, which is usual in Herodotus : θ 408 αιεί yap μοι εωθεν ενικλάν ' οττι κεν ε'ίπω. After Homer's time there is also the plupf. ειώθειν, in Herodotus εώθεα. The first letter of the root was discussed on p. 85 : ε'ίωθα is from an original ε-σΐοθα. The diphthong is due to compensatory lengthening for the σ 134 which has been lost before /, so that ε-σϊοΰα became in the first instance έί-ΐοθα, while the ω is due to the influence of the A Cp. Brugman Stud. iv. 170. With regard to the iEolic ενέθωκα i.e. ε-ΗΒω~κα, where the loss of the σ has left no trace behind, we may refer to p. 85. Finally in the following perfects the reduplication syllable can only be recognised from the contraction. The case is quite like that discussed on p. 83, where the syllabic augment lay hidden in the syllable εί, only with the difference, that the perfects have lost a spirant not merely after, but also before the ε. ειλίσσετο goes back to an earlier ε-ΗΧίσσετο, εΐλικται to βε-ϊέλικται. 1) ε'ίακα, έίαμαι, quoted from Demosthenes (8, 37; 45, 22). In opposition to what was conjectured on p. 84 as to the origin of this verb, attempts have been made from two different sides, and in a somewhat inconsistent manner to prove that ε'άω, in which verb the loss of a / after ε is quite certain, comes from *<τε/αω. S. Bugge in Fleckeisen's Jahrb. vol. 105 p. 95 compares f' /αω with the Lat. de-sivare desinere, recorded 8 %opra(C) ' €δο£ e, κρςμ,νάται (better κρημναται) Hesych. looks like a perfect. Lobeck Ehem. 124 suggests &pape and the like, but from the root ap we cannot get to syl- labic reduplication. Can we suppose that in this form there is concealed the rt. va7' choose, to which αίρίω belongs (Princ. ii. 180) ? The Homeric Μ ?jpa ' άλος ηολιής ειρυμεναι, ch. xvi. EEDUPLICATION WITH AN INITIAL VOWEL. 365 θ 151 νηνς re κατείρνσται, with the forms of the plupf. Σ 69 ε'ίρυντο, 13 30 ειρνατο. 9) είστίακα, ικίτίαμαι good Attic, cp. p. 86. Β) Reduplication where the Stem begins with a Yowel. "With an initial vowel there are two possibilities for reduplication. Either the initial vowel alone was repeated, which then however always remained the same as the vowel of the stem-syllable, and did not, as with an initial consonant, sink to ε. There is evidence of the possibility that both vowels existed independently for a time side by side in Gothic perfects like ai-alth, ai-aik, ai-auk, which may be compared in respect of their form with l -άλλω, Ί-ανθ-ο-ν (Fritzsche Stud. vi. 325). Hence for the root ad eat we may probably presume a form a-ad-a. In Sanskrit we find only forms like ada (cp. Lat. edi) ; and as in Greek the vowel of such perfects is only distinguished from the initial vowel of the stem by its length : Dor. 'οιλλαχα, Horn. (εφ-)^πται } we shall have to assume, as in the case of the temporal augment (p.. 87) that the rule for this re- duplication was framed before the splitting of the vowel. For such forms cannot be explained from the contraction of a syllable ε'-. The Indian perfect knows only this kind of reduplication : rt. ar raise, has in the perfect ara. There are no Italian perfects either of any other formation from the roots here in question. On the other hand Greek possesses a second fuller formation of the reduplication-syllable, which is known by the name of Attic reduplication. It consists in the repetition of the initial vowel along with the consonant which follows it. "We actually find as against the Vedic ara (i.e. as it were *ώρα) a Greek ορ-ωρα. This reduplication is called Attic doubtless only, as Buttmann Ausf. * Gr. i. 2 327• saw, because it here and there (e.g. τ/λεσ^αι for the old άλήλεσμαι, ήρεισμαι for ερήρεισμαι) had disappeared from the usage of the later Greeks, and hence was to be recommended as a good and genuine Attic formation. Noun- formations, which are based upon the same principle, are collected by Fritzsche 1. c. p. 287. Many, like άγαγύρτης, άλάλαγξ, όΰοΰνσταί, are connected with kindred verbal forms, but others 137 like επ-οψ^ηρ-ιιρα, όλ-ολ-νζω by the side of ululare have arisen inde- pendently. Viewed precisely the primitive form of the first method of reduplication, e.g. a-ara or a-dra, is related to the second e.g. ar-ara exactly as the less complete in the Lat. me-mor to the more complete in μέρ-μερα (φροντίΰος ίίζια Hesych.). If we count in the initial smooth breathing, the reduplication-syllable in each of the two cases of the second series contains three elements, in each case of the first it contains two. We have met with the same process already in the reduplicating aorists, where forms like άγ-αγεϊν, ήκαχε, ήραρον, ωρορε occur. In this instance it is not unknown to Sanskrit. Delbriick p. 111. quotes from the Vedas dm-ama-t from am damage, and with a weakening of the second a to i an-ina-t from rt. an breathe, ard-ida-t from rt. ard (causative =shatter), which are only recorded by the grammarians (cp. above p. 290). From Zend Bopp Yergl. Gr. ii. 2 529 quotes the 3 pi. perf. ir-irith-are from the stem irith dissolve, as an isolated kindred formation. The Greek perfect formation is separated from that of other redupli- cating tenses by delicate distinctions. Here the distinction lies in the rhythm. The aorists leave the stem-syllable short : ωρορε ; the perfects 366 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. cfr, xvi. bring it out vigorously by its length : ορ-ωρε. The rule for the perfect has an unmistakeable analogy with the manifold lengthenings, by which in compound words the beginning of the second constituent is brought into prominence: ποΰηνεμος, υπώρεια, άνώννμος. Even Bopp, Vergl. Gr. ii. 2 529, refused to regard the long vowel after the earlier fashion as 'aug- ment ' ; and it cannot be taken as a temporal reduplication in addition to the syllabic, especially as it reappears, as Bopp also pointed out, just as much in nominal forms like αγωγή, άκωκή, έΰωΰή. The long vowel agrees besides with the extensive tendency of the Greek perfect to prefer a long stem-syllable, which we shall have to treat at greater length further on. In the feminine of the participle the short vowel is preferred, so that apapvla is not distinguished in rhythm from αραρών. The carrying out of this rhythm of the perfect is, it appears, specifically Greek, while 138 evidently the kind of reduplication here under discussion belongs in its origin to an earlier period of language. There is not, however, perfect consistency with respect to quantity, as is shown by άκάχημαι and άκηχεΰα.™, άΧάλημαι, άλαλυκτημαι, ερέριπτο. The need of bringing into definite prominence the essential differentia of the perfect gave the impulse to this specifically Greek expansion of old germs. We saw on p. 356 how in another direction also the Greeks extended this differentia further than the limits which can be demonstrated from other languages. As we now turn to the task of giving a list of the perfects with Attic reduplication, we are met by two facts in the chronology of the language, which have hitherto received but little notice. The first con- cerns the relation of these perfects to those in Homer which are characterised merely by a long vowel, the second the nature of the stems • in which this kind of reduplication is effected, and the closely connected question as to the antiquity of this method of formation, regarded from the point of view of the more general history of the Indo-Germanic languages. If I may trust my collections, there are in the Homeric poems only 25 perfects from stems beginning with a vowel. Of these 19 show the Attic, 6 at most the ordinary reduplication. The former will be given later on along with the later formations of the same kind : the 6 with the ordinary reduplication are : κατ^κισται π 290, τ 9, εψ-ήπται Β* 15 and frequently elsewhere, with εψήπτο Ζ 241 and άν-ήφθω μ 51, 162, ήσκηται Κ 438, with επησκηται ρ 266, ήσγυμμένος Σ 180, άφ-~ιχθα<. ζ 297, perhaps 'ίσασι Ι 36, for the long ι can hardly be understood otherwise, επφχατο Μ 340 with παρφχηκεν or παρώχωκεν Κ 252 with the variant παροίχωκεν, which has good support from the Alexandrine time, and probably deserves the preference. The cases in which reduplication is entirely omitted with an initial vowel, like οντασται, άΰηκότες will have to be discussed later on, along with the same phenomenon in the case of stems beginning with a consonant. From the point of view of the Greek language we might thus be led to the opinion that the method of for- mation which was afterwards widely employed, was in Homer's time 1 39 something novel. But a glance at the universal rule in Sanskrit is enough to show us that in this case the language of Homer does not in the remotest degree reproduce the original. The perfect is in epic poetry generally not a very common tense. This is the main reason why the instances are so few. It is only in Attic, as will be seen repeatedly, that the perfects were freely formed, and that too especially in later times. Besides, most of the forms with an initial vowel have come from derived ch. xvi. REDUPLICATION WITH AN INITIAL VOWEL. 367 stems, in part unknown to the Homeric poems, like ήγγελται, ήΰίκηκα, ώμολόγηκα, ήρώτηκα, ήζίωκα, εζήτασται, νβρικα etc. In view of the in- creased needs of later times, long after Homer a rich abundance of such forms was produced in imitation of a not very numerous stock of primitive creations. But it is worth noticing that what is the rule later on, in Homer appears as the exception. If we ask now in the second place in what sort of stems the Attic reduplication appears, there is no lack of very ancient ones among their number, and it is quite what we should expect to find these in Homer. Among such perfects proceeding from roots are ακαχμενος, αρηρα, ε$ΐ}3α, odojSaj ολωλα, οττωττα, ορωρα. But the great majority of the perfects with Attic reduplication came from disyllabic stems, and betray thereby iiheir relatively late origin. The disyllabic stem has originated either by prothesis as in άγερ (άγηγέρατ Υ 13) άλιφ (άλήλιόα), όρεγ (όρωρέχα- ται Π 834), and probably also in ενεγ/,•, εριπ, or by evident expansion of the stem, as in έλ-υ-Θ (ελήλνθα, Horn. είλήλονθα), άρ-ε (αρέσκω^ άρήρεκα), αίρε (άραψηκώς and even αναιραιρεμενος) or by unmistakeable secondary formation as in άΧάλημαι, αλαλύσθαι, άλαλνκτημαι, συν-αρηρακται {αράσσω), άρηρομένη (αρόω), ελήλιγμαι [ελίσσω). ■ In the Case of two verbs we can prove that there was once a digamma : έμήμεκα (cp. Skt. vam, Lat. vomo), and 'ελήλιγμαι (rt. /ελ) already mentioned. Some stems e.g. that of ομνυμι, όμώμοκα, that of όρύσσω, όρωρυχα, of ελέγχω ελήλεγκται and of ερείΰω ερηρέΰαται remain obscure, but these too do not at all look as if they were particularly old. The curious and perhaps justly suspected forms given by the Et. M. p. 372, 42, ερηρστηκα (from ερωτάω) and ετητόμακα (from ετοιμάζω) give us the extreme in- 140 stances. In short, the result at which we arrive is that the great ma- jority of these forms, and among them not a few which are found in Homer, have been created in imitation of a few old forms by the luxuriant productive force of the Greek language. What Windisch Ztschr. xxi. 410 has already remarked as to this method of formation is thus completely confirmed. The Attic reduplication has therefore a special interest as bearing on the history of language. It shows us what the Greeks were capable of at a comparatively recent time, and may serve as a warning n©t to go too far in explaining Greek forms from pre- Greek models. The Homeric language, in which many at every step conjecture reminiscences from the grayest antiquity, evidently is in the middle of this period of recent imitation. We may here as in other cases cast our eyes back from Homer to a still earlier poetry, which was .actually creative in moulding characteristic forms of language. The established instances of Attic reduplication are as follows, Homeric examples being marked with an asterisk : 1) *άγηγέρατ (plupf.) Δ 211, Υ 13; other forms, like αγηγερμενος, άγηγερκας are not quoted from any author earlier than Appian. Op. on no. 17 εγείρω. 2) άγήοχα, first in Aristotle (συναγηόχαμεν Oecon. Β p. 1346, a, 28) and in the spurious documents in the speech De Corona (39, 73), άγηόχει in Polybius, συνηγηόχει ' εκόμισε Hesych., συναγάγοχη, inscription from Thera, C. I. 2448, ii. 10, <τυ»/αγαγοχε7α, i. 28. 3) *άκάχημαι θ 314, άκηχέΰατ Ρ 637, άκαχείατο Or ακαχηατο Μ 179, άκηχεμενη Ε 364. The reduplication in this case extends very far, for άκάχοντο, άκαχίζω, άκηχεΰόνες ' λυπαι (Hesych.) are quoted. 11) ά; 12) a, 368 PERFECT STEM AND FOKMS CONSTRUCTED FROM IT. ch. xvr. 4) *άκαχμένος common in Homer, e.g. Ο 482. 5) άκηκοα common from Aeschylus (Pr. 740) and Herodotus onwards» 6) *άλάλημαι Ψ 74, άλάλησο γ 313, άλάληντο also Eur. Andr. 30β> (cbor.). 7) *άλα\νι:τημαι Κ 94. 141 8) άλάλυκτο from the Horn, άλνάσω Quint. Smyrn. 13, 499; 14, 24» 9) άλχιλύσθαι * ψοβεΊσθαι, άλνειν Hesych. 10) άλήλεσμαι Herodot. vii. 23, Thuc, άληλεκέναι Nicarchus AnthoL xi. 251. άλήλιφα Demosth. άληλιμμένος Thuc. iv. 68. άραιρηκώς, άραιρήκεε, αραίρηται, άραίρητο Herod., αναιρερη μένος by itacism for άν-αιρ-αιρη-μένος inscription from Thasos, discussed by Berg- mann Hermes iii. 238. On the other hand there is %ρηκα, τίρημαι from Aesch. and Thuc. onwards. 13) *άράρα Pind., conj. άρήρτι ε 361, άρηρός Λ 31, άραρυΤα Ο 737 and often. In Aesch. (Prom. 60) and Eurip. there are isolated forms of the kind, άρηρεμένος Apoll. Rhod. iii. 833 (for which Quint. Smym. has άρηραμένος). προσαρήρεται (conj.) Hes. Opp. 431, ήρήρειντο Apoll. Rhod. iii. 1398. 14) συν-αρήρακται ' σνγκέκοπται Hesych. 15) άρήρεκεν quoted from Sext. Emp. ed. Be^kker p. 652, 29. 16) *άρηρομένη Σ 548, Herod., άρήροτο Apoll. Rh. iii. 1343. 17) *έγρηγόρθασι Κ 419, 2 pi. imp. εγρήγορθε [Η 371 etc.], inf. middle, έγρήγορθα'ι Κ 67, from Aesch. onwards έγρήγορα (έγρηγορώς Eumen. 685), έγρηγόρειν common in Attic prose. [Cp. Lobeck on Phryn. p. 119.] The way in which we regard the reduplication in this verb depends upon the question how we explain the ε in εγείρω, whether as the remains of the doubling preserved more completely in the Skt. (ja-yar-ti he wakes (Princ. i. sub voc.) or as a prothetic vowel (Fritzsche Stud. vi. 322). In the former case k-ypiyyop-a would be twice redupli- cated, like δ£-δί-δαχ-α, only that the perfect reduplication made itself visible in the second syllable, in the latter ε- in the perfect would also be prothetic; in both cases we should have to explain γρή-γορ-α as by metathesis from γερ-γορ-α (Siegismund Stud. v. 169). The adoption of the p into the reduplication-syllable is in any case*very singular, for we should have expected *ε-γέ-γορα or *ε-γή-γορ-α. Hence the forms quoted have, strictly speaking, only the appearance of Attic reduplication : a remark which also holds good for no. 1. We cannot however doubt that to the instinct of the language all these words seemed formed alike. 142 The long vowel of the second syllable proves this. I have discussed in Stud. vii. 393 the trace of a Laconian regularly reduplicated ΙηγορεΊν • έγρηγορέναι. — The form εγ-ήγερ-μαι recorded from Thuc. (vii. 51) on- wards (with έγ-ήγερ-κα in the post- Attic period) is more regular than the active form. 18) *ε^δα, κατά τανρον ε%ηοΧ>ς Ρ 542, έ^νίαι Hymn, in Merc. 560* ίδήΰοται χ 56, εΰήΰοκα Attic prose, with the mid. έΰήΰεσμαι. 19) *ε\{ιλαΓαι Π 518, ήληλατο Ε 400, επελήλατο Ν 804, εληλάΖατ η 86, with the active έλήλακα and the corresponding plupf. from Herodotus onwards. 20) εληλεγμαι occurring from Antiphon onwards, with its plupf. 21) εληλιγ μένος Pausan, x. 17, 6, but everywhere else from Hesiod onwards εΐλιγμαι. ch. xvi. KEDUPLICATION WITH AN INITIAL VOWEL. 369 22) *ελήλνθα. άπεληλνθα Ω 766, επελήλνθα h 268, then in Herodotus and Attic writers. The expanded Homeric form εϊληλονθα Ε 204, A 202 etc. (with εληλονθώς Ο 81) is altogether abnormal. I do not know that any attempt has ever been made to explain the diphthong ε<. 23) εμημεκα, εμημεσται from Lucian and Aelian, the plupf. εμημέκεε quoted from Hippocrates. 24) ενήνοχα common in Attic prose, with the mid. ενηνεγμαι, Herod. viii. 37, εζενηνειγμένα. — Hesychius gives the regular form κατήνοκα • κατενηνοχα. 25) *έρηρέΰαται Ψ 284, έρηρέ(ίατ(ο) η 95, ήρήρειστο Γ 358, έρηρεισμένος Herod, iv. 152. The active ερήρεικα is post-Attic : — ηρεικα, ήρεκϊμαι occur from Polybius onwards. 26) έρηριγμένος Aristot. 27) *έρέριπτο Ε 15 ; post-Attic έρήριπται, ερήριπτο. 28) *ερήρισται from ερίζω Hes. fr. 219 τψ <Γ οντις ερηρισται κράτος άλλος. 29) *6()o)hi ε 60, ι 210, and afterwards similar forms in post- Attic writers, ωΰωΰε » σέσηπεν (probably ώΰωΰει ' σεσηπει), ώζεσεν. 30) *όΰώΰυσται ε 423, ωΐνσται \ ήχθέσθη Hesych. 31) *σνν-οκωχότε ' επισνμπεπτωκότες Hesych. In our M.SS.atB 218 there is συν-οχωκότε which Buttmann i. 2 331 derives from the former by a transference of the breathing from assimilation to εχω. The chief support for this view is to be found in the substantives όκωχη, σννοκωχή, while we may also very easily get from όχέω to a by-form όχόω, and 143 thence to a regular *ώχωκα and without reduplication οχωκα. Cp. εποχημέιος Nonn. viii. 229. 32) *όλωλα Π 521 and often elsewhere in Homer and afterwards common, plupf. όλώλει Κ 187 etc., όλώλεκα from Herodotus onwards. 33) όμώμοκα, ομώμοται or όμωμοσται from Eurip. onwards. 34) *οπωπα Ζ 124 etc. όπώπει (f> 123, both thenceforward in poets and Herodotus. 35) *ορωρε Η 388, όρώρει θ 59, plupf. also in the tragedians, όρώρηται conj. Κ 271. 36) * όρώρει only in Ψ 112, plupf. from a presumable ορωρα from the rt. /op, pres. Όρομαι : cp. Lobeck on Buttmann ii. 2 260, Princ. i. 430. 37) *υρωρέχαται Π 834, όρωρεχατο A 26. 38) όρώρυχα Pherecr. (Com. ii. p. 327 v. 19), όρώρυκται Herod, iii. 60, with ωρνκται ii. 158 (?), όρώρνκτο from Herodotus onwards. 39) νφηφασται, quoted by Herodian ed. Lentz ii. 950 : ον?εις 'Αττικός -παρακείμενος άπο τον ν άρχεται, αλλά μόνος 6 νφήφασται (cp. Suidas S. v. νφηφανται Bekk. Anecd. 20). On the other hand έννφασμένος Herod, iii. 47, and the like in Xenophon and Antiphanes, Com. iii. p. 52. The Et. M. gives νφϋφασται, which certainly diminishes the authority of the former form, already attacked by Lobeck ad Phryn. p. 33. I have attempted in Princ. i. 369 an explanation, by referring νφ-ηφ to the rt. vabh O. H. G. weban ; νφ would then be an early weakening of f αφ, in the second syllable the fuller form would be preserved, but with the loss of the F. We met with something similar above p. 292 in the redupli- cated aorist. In Sanskrit the syllable va is reduplicated by u, e.g. in u-vatta. Under any circumstances the form is singular. Β Β 370 PERFECT STEM ASTD FORMS CONSTRUCTED FROM IT. ch. xvi. C) Loss of Reduplication. It is hardly probable a priori that such an essential characteristic as reduplication could be simply lost. A prominent feature of Greek formations throughout is the retention of all significant elements. Even in recent formations we have just seen that the tendency to reduplication continues to be living. But as in the case of the augment so in redupli- cation there is an essential distinction between forms beginning with a 144 consonant and those beginning with a vowel. It is only in the latter, which indeed for the most part exclude any prominence of the sign of the perfect, as well as of the preterite, as a distinct syllable, that the loss of the reduplication is at all common. In cases of an initial consonant there are only very few instances of this phenomenon, and these belong either to the rusty antiquities of the language or to the isolated ventures of particular writers. The retention of the reduplication is in Sanskrit also the almost invariable rule. Delbriick Altind. Verb. p. 121 men- tions and discusses the altogether isolated exceptions, and opposes the earlier and looser assumption, adopted by Corssen in order to make out reduplication to be something quite unimportant for the Latin perfect. It is especially noteworthy that reduplication is among the Indians occa- sionally suppressed, only in the more polysyllabic dual and plural forms, which also furnish the much-discussed by-forms with e (tenus from tatdna). The preference of Latin for compact forms, cooperating with the altered laws of accentuation, produces in Latin a further extension of this phonetic decay. The same thing happened in Teutonic, and as Windisch informs me, also in Keltic. The Greeks and Indians, with their delicate feeling for articulation, have, as is so often the case, retained the most faithfully the earliest forms. a) Loss of Eeduplication with an initial consonant. 1) γεΰμεθα only in Theocr. xiv. 51 in the proverb μϋς γεύμεθα πίσσας (cp. Diogenian. Cent. ii. 64 άρτι μυς πίσσης γένεται). It is most natural with Meineke to take γενμεθα as a perfect. Still we can hardly reject as impossible the explanation of the word as a present for γενόμεθα (cp. λονται, λοΰνται), especially since Ζεΰμενονζ=Ιευόμενον, ΰεόμενον has been proved to be probably right in the last verse of Theocr, xxx. ■ 2) ϊέχαται, ποτιίεγ μένος Buttmann ii. 2 149 takes as perfect forms with the reduplication lost, because the meaning ' await ' which is con- nected with these forms (M 147 σΰεσσιν εόικότε, τώτ εν ορεσσιν ανδρών ήϊε κυνών δέχαται κολοσυρτον ιόντα, Ι 191 ΰέγμενος Αιακίΰην οπότε \ή£ειεν άείΰων, and often ποτ ιΐεγ μένος) is elsewhere limited to the perfect forms with reduplication : Κ 62 ΰεΰεγμένος εις ο κεν ελθτις, Ε 228 ήε συ τόνΰε δε'δε£η. Op. above pp. 104, 131. Kiihner points to the post-Homeric 145 use of the present δέχομαι, e.g. Eur. Or. 1217 δόμων πάρος μένουσα παρ- θένου δέχου πόδα. And it cannot be denied that in the non-reduplicated forms προσδοκά ν and προσδέχεσθαι the meaning * await' has been attained without the help of reduplication. Now as we have also forms from the it. δεχ without a thematic vowel in δεκτό, δέΕο, δέχθαι (pp. 104, 131), the difference in meaning can hardly induce us to follow Buttmann in separating δέχθαι, receive, from δέχθαι (for δεδέχθαι) await. 3) θύμμενος, burnt, damaged, only in Et. M. 458, 40 : θΰμμενον : ch. xvi. LOSS OF EEDUPLICATION. 37 ί ΰηΧοΊ το νπο πνρος βεβΧημένον η κεκακωμενον. καϊ πάρα γεωργοΊς οι νπο- κεκανμένοι υπό πάχης άμπεΧώνες εντεθνμμενοι καλούνται. The present is τύφώ. To assume an aorist *εθΰμμην like έΰέγμηι; which would require to have a passive meaning like βλήσθαι, is not a desirable course : it would be better to suppose that the word, belonging as it does to the language of peasants, lost its reduplication, and subsequently also the accent of a perfect. 4) ελειπτο Apoll. Rhod. i. 45, 824 according to Buttmann i. 2 318 and Lobeck on Buttmann ii. 2 17 a plupf. for εΧέΧειπτο. We discussed this form in another connexion on p. 131. . 5) επαλιλλόγητο Herod, i. 118, of which Buttmann says 'perhaps the somewhat clumsy compound furnished the reason for the simple augment.' It would not be very venturesome to alter a form so isolated into πεκαΧιΧλόγητο. There are also three forms only recorded by Hesychius : άπότμητατ άποκέκοπται, ΰ ι ακόρι στα ι' ΰιαπαρθενενεται, επιτευκται' εν επιτυχία εστί (M.S. έστω), all three guaranteed by the alphabetical order, but without any information as to their source. φΧασμένος' τετνφωμένος is declared by Meineke on Theocr. xiv. 51 to be a copyist's error for πεφλα- σμένος, though it is not more surprising than the other three examples. We will discuss οίδα under b). b) Loss of Eeduplication with an initial vowel. There are precisely the same difficulties in the case of reduplication with an initial vowel as those discussed on p. 91 ff. in the case of the augment. It is therefore not necessary either to enumerate all cases separately, or even to discuss in detail the combinations of letters before which this license is found. The Herodotean ε'ίκασται by the side of γκασται in the tragedians is not really different from είκαζον by the side 146 of γκαζόν, though in small matters like these there are here and there small peculiarities. In the preterite there is ηυρον as well as ενρον, but in the perfect there is probably only ευρηκα, εύρημα ι. We may divide the forms that belong here into three groups : (1) those with an initial diphthong or vowel long by nature, (2) those with a short vowel lengthen ed by position, (3) those with a vowel which remains short. The first and second groups are very widely extended over Greek of all times and dialects; the third alone is somewhat surprising, and is coiTespondingly rare. To the first group belong the numerous com- pounds with ευ like ενΰοκίμηκα, ενεργέτηκα, εντΰχηκα, ενώχημαι, in which at the most an internal reduplication is sometimes attempted, but also, what we should not have expected, a number of perfects, for which an initial f is established. Among them o/oa takes a place of its own, inasmuch as the corresponding Skt. veda and Goth, vait also show no trace of the perfect reduplication. It is probable that here, not unin- fluenced by the completely present meaning, the sign of the perfect was lost even in the time of the common language. Had we not the parallels of the cognate languages, from the point of view of Greek it would be very natural to explain the loss of the reduplicated ε in oloa by Herodotus's oIku (e.g. iv. 82) by the side of the ordinary Greek εοικα, and to place the loss of the ε in a time when the / had already dis- appeared. Strictly speaking, oUa belongs therefore to the cases of β β 2 372 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvl dropped reduplication discussed under a). But we. place it here because jof its resemblance to some other forms. Hesychius has preserved for us also the corresponding middle form ϊδμαι with the explanation γινώσκω, οϊδα. Since ΐδμαι is to «Ιία as ή'ίγμαι is to eouca, there is no reason with Mor. Schmidt to regard this form as corrupt. A form corresponding to the Herodotean oltca occurs also in Alcman, where, however, the reading οίκας ώραΐω λίνω (fr. 80 Be. 3 ) is not quite certain, because of the variant εικας (cp. εϊκώς). On the probability of a f in this verb see Princ. ii. 309. The analogy of these two present perfects will meet us again in a different connexion. As phonetic changes show themselves very rarely in 147 the case of an initial ει, απειλημένος (Herod, ii. 141), εΤργμαι (Aristoph. Xen. Aesch.) are less surprising, though here too there was originally Ρ (Princ. ii.TtO, i. 222). In the familiar Homeric οντασται (Λ 661) we could not expect the diphthong ων, which is unknown to this dialect. There is more to surprise us in the Herodotean οίκοίομηται (i. 181), oiKurrat, οίκητο, οίνωμένος, by the side of the Attic ωκοΐόμηται (Thuc. vii. 29) ωκισται, ωκηκα (Soph.), but the diphthong is quite in harmony with the Ionic usage elsewhere, οικοΖομηται is, however, also found on. the Heraclean Tables (i. 137), and Meister Stud. iv. 423 quotes οικημαι from Archimedes. Of perfects with an initial long vowel it is worth while noticing 'άρημένος (Σ 435), the origin of which is obscure [cp. Merry on C 2]. The long a of the reduplication syllable reminds us of that in the augmented syllable of άλτο. There is also άΰηκότες in the formula κα- μάτω άΰηκότες ήΰέ και νπνω (Κ 98, 399) with the variants άΰηκότες, άΰοη- κότες. La Boche, Textkrit. 179, recommends the reading of Herodian with the rough breathing (rt. ά^ cp. ΐιΰ-ρύ-ς, άδ-ινό-ς from a=sa, in Lat. satur) and one δ. It is doubtful, however, whether the form with Zl 9 which has quite as much authority, is not more correct, as in the cognate #SV Ε 203; cp. Princ. ii. 290. In the second group we may mention as Homeric forms ερχαται and ε ρ χα τ ο κ 283, ξ 73, the latter by the side of έέρχατυ κ 24:1, which agree well with άποεργει θ 325, εργαθον and the aor. ερζαν, and quite correspond to Herodotean forms like κατέρΐ,αι,απεργμένος, and εσσαι from the rt. hg (εννυμι) ω 250, from which an oracle in Herod, i. 47 gives the 3 sing, έπιέσται with the plupf. εσσο Γ 57, it 199, εστο ρ 203 and frequently. The course of the language must have been this — that first there was real reduplication Ρε-Ρέρχαται, Η-ίεσ-σαι, then with the disappearance of the internal } (cp. εάΐα) Ρε-έρχαται, Η-εσ-σαι, then with aphaeresis of the ε Ηρχαται, ίέσσαι, and finally when the initial f also was sacrificed to the dislike felt by the Greeks for soft spirants, ερχαται, εσσαι. Forms of the kind mentioned last but one have left traces enough in Homer. While then in these cases a vowel did not originally stand at the beginning, Herodotus furnishes a number of perfects of the kind without reduplication from stems, for which we cannot imagine a conso- nant to have been lost : αμμένης i. 86, απαλλαγμένος ii. 167, άρΎμίνος 148 i• 174, αρμοσμένος ii. 12i, καταρρωΖηκας Hi. 145, αρτεαται i. 125, εργασται iii. 155 [but cp. Princ. i. 221] ορμεατυ i. 83, εσσωται vii. 10,4 [εοσωμένος viii. 130J, άπεψημένος i. 188. Here the disinclination to long vowels before more than one consonant has evidently hindered the application of re- duplication. A doubtful instance of the kind is όψρυωμένος in Timon Phlias. p. 28 Wachsmuth (Diog. Laert. ii. 126). For ώψρνωμένος which ch. xvi. POSITION OF THE KEDUPLICATION. 373 is not without M.S. authority, might also be brought into the verse by synizesis on the analogy of Ίΐλεκτρνωνος (Hesiod. Scut. 3). The most surprising group is the third. For we can see no reason for omitting the reduplication where the initial vowel is short. Hence the cases of this kind are not numerous, and are almost exclusively Ho- meric archaisms. There is an exception in the very surprising Hero- dotean αλισμένος (iv. 118 and elsewhere) : cp. Bredow de dial. Herod, p. 292 ; like οιΰα, and εσσαι with an old digamma. Homer gives άλιτη- μενος δ 807, with little trace of the perfect in accent or meaning, and justified also by the impossibility of bringing ήλιτη μένος into the hexa- meter, and άνω-γα (e.g. S 105), which was retained in the language of the Attic poets and Herodotus (iii. 81). The origin of the word is obscure, but it is not improbably connected with ανάγκη. Hence per- haps α»'-ωγα is based upon Attic reduplication, like άκ-ωκή, the nasal in the stem-syllable disappearing after it had produced a softening of the consonant. The Oscan angit (tab. Bant. 2) with angetuzet (ib. 20) which has been identified on insufficient grounds with the Lat. agere, having the meaning enjoin, order, might be cognate. άμφ•ιαχν~ια Β 316. The word could not be otherwise brought into the verse : besides the whole verbal stem is already reduplicated, for ί-αχ for Ρι-Ραχ evidently goes back to the stem preserved in ήχος, ήχέω (cp. Fick Worterb. 3 204). — Ιέρευτο only in Ώ 125, justified by the metre. — From later poets we have further απάμειπτο Anthol. Pal. xiv. 4, άμειπτο Nonnus Dionys. xliv. 241, but there is no trace of a pluperfect meaning in either place, and it is doubtless better to regard the word as an aorist of late formation like ελειπτο (p. 131) : — On the other hand εποχημένος αρματι κύκνων Nonnus Dion. viii. 229 is a certain instance of the mutilated perfect. D) Position of the Reduplication. 149 The numerous irregularities which are to be found in compound words in respect of the position of the reduplication do not come within the scope of our investigation, any more than the similar phenomena in the case of the aorists, which were briefly mentioned on p. 94. It will be sufficient to notice some of the rarer instances. Here, as in the augment, the careful endeavour not to omit the expansion which is significant of the tense is prominent. This endeavour is here even greater than in the case of the augment, so that, with the exception of the few instances already noticed, the reduplication is never wanting even in compounds, though sometimes, as a result of this strong tendency to mark the form distinctly, it is doubled. The fundamental law, according to which every verb not compounded with a preposition is treated as a whole and altered at the beginning, produces such clumsy forms as ΙεΙνστυχηκα (Plato, Lysias), πεττολιορκημένος (Thuc.) ήναντίωμαι (Time. Dem.) But the preposition here too is marked off as an unes- sential member of the verb, not only when the verb is used also without a preposition, as in ajr -όλωλπ, άνα-ΰεΰρομε, έπι-τέτραπται (Homer), but also when it either occurs only with a different meaning, e.g. in h -ωκηκα (Plat.), κατα-πεφρόνηκα (Orators), or does not occur at all, e.g. in άπο- Χελόγηται (Orators), παρα-νενόμηκα (ib.), εγ-γεγυημαι (Plat. Dem.), εμ- πιποΰισμενος (Aesch. Prom. 650), έπι-ώρκηκε (Xen.), εν-ΰεΰιωκότα (tab. Heracl. i. 120), εν-τεθύμημαι (Thuc). The theoretical distinction of 374 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch.'xvi. decomposita and composita had evidently never very ihuch life in it in the genius of the Greek language. It is comparatively rare to find other first elements treated like the prepositions in compound verbs, and this is apparently the case only where the whole word begins with a vowel. To this class belong αντευπεπυίηκεν Dem. xx. 64, ίπποτετρόφηκα Lycurg. 139, οΰοιπεπορηκαμεν Philippides Com. iv. 471 (Herod, viii. 129 has on the other hand Βωΰοιπορήκεσαν). Here again we may recognise a luxu- riant creative impulse. Because the reduplication could not get its full rights with an initial vowel, it was carried out syllabi cally within the word. This view is confirmed by the fact that sometimes under similar conditions we find double reduplication, either in such a way that there 150 is a vowel at the beginning of the first element, and the internal redu- plication is syllabic : ώΰοπεποιημένη Xen. Anab. v. 3, 1 (Kriiger ώδο- ποιημένη) or conversely; ΰεΰιήτημαι Thuc. vii. 77, πεπαρφνηκα (Aeschin. ii. 154) or thirdly, with a repeated vocalic reduplication : ήνώχληκε (Dem. xxi. 4), έπηνώρθωμαι (Dem. xviii. 311). In the last two examples the treatment of the prepositions εν and ανά shows that they were no longer felt to be such. This exception to the fundamental law is illus- trated by numerous instances elsewhere, and there is really very little to surprise us in it, for the coalescence of elements originally independent is one of the most frequent phenomena in the history of language. Cases of the kind are furnished by πεπρωγγύηκα tab. Heracl. (Meister Stud. iv. 424), which we may call the counterpart to the previously mentioned έγγεγνημαι, 4 μεμετιμένος (Herod, vi. 1), where the stem- vowel too is very remarkable, ήμψίεσμαι (Hipponax 3, Aristoph., Plat.), ήνοισται * προςηνέχθη, προςενήνεκται Hesych. Ε) Significance of the Reduplication in the Perfect. The significance of the reduplication in the perfect may be very clearly recognised from the Greek use of this tense. But it has been frequently overlooked by our grammarians, because they could not shake themselves free from the notion that the perfect must have been origin- ally a past tense, though in Greek above all languages the employment of the primary personal terminations and the absence of the augment in the perfect indicative, and also the numerous perfects whose meaning is unmistakeably present, might have guided them aright. Yet even Buttmann was misled by his notion that reduplication was the source of the augment (i. 2 313) into confusing the perfect with the past tenses, and thus had much trouble with cases ' where the perfect has the force of a present.' He admits (ii. 2 89) that in perfects like μέμηλε 'the present proper and this derived present force approximate so nearly that the usage of the language confused them.' For several mimetic perfects, 151 like κεκραγα, he gives up the attempt to derive them from an earlier perfect force. Bopp was the first to discover the true point of view, which deserves the more cordial recognition in that the usage of the Indian perfect did not give at the time when Bopp wrote his Compara- tive Grammar, the slightest trace of a present force, with the exception perhaps of the anomalous veda=J : o't^a f Goth. vait. At that time the Old Indian perfect seemed to be a purely past tense, and hence its usage * Perhaps ήγγραμμίναν, as Naber Mnemosyne i. p. 105 is inclined to write for the recorded ϊτγραμμ4ναν in the second Cretan inscription published by him, line 4, in the sense of ίγΥ^η/ρομμίναν, belongs here. ch. xvi. SIGNIFICANCE OF THE REDUPLICATION. 375 tended rather to increase than to diminish the old confusion. Hence it was not by means of Sanskrit, which in so many other instances served to clear up matters at once, but by a general estimation of the Indo- Germanic formations, in which the Greek usage, the isolated Latin perfects odi and memini, and the numerous Teutonic so- called ' preterite presents ' carry great weight, that the founder of comparative philology was led to the correct view, which he expresses in § 515 in the following words, ' The reduplication-syllable merely serves to intensify the concep- tion, and to lend to the root an emphasis, which is regarded by the genius of language as a type of what has already become and is com- plete, opposed to that which is viewed only in the course of becoming, and has not yet reached the goal. Both in sound and meaning the perfect is akin to the Skt. intensivum, which is also reduplicated, and in which the vowel of the reduplication is intensified for the sake of emphasis.' It was only later on that Bopp found some support for his view in the Vedic use of the perfect, quoted in ii. 2 p. 466 note ; i.e. cases in which ' the perfect denotes the completion of an action ' ; 5 and discovered (p. 531) some ana- logies for this in the use of the corresponding Zend forms. Spiegel too (Grammatik der altbaktrischen Sprache p. 318) takes the reduplicated perfect of Zend as the tense of completed action, and quotes instances in which this form has quite the force of a present. My statement in Tempora und Modi p. 172 f. was in accordance with Bopp's first edition. Kiihner Ausf. Gr. ii. 2 126 ff. has made no use of these explanations. More 152 recent investigations of the Greek perfect, especially the dissertation of Warschauer 'De perfecti apud Homerum usu' Breslau 1866 and the paper by Richard Fritzsche * iiber griechische Perfecta mit Prasensbe- deutung ! (Sprachw. Abhand. aus G. C.'s, gramm. GeseUsch. p. 43 ff.) have carried the question further and treated it in detail. Hence i content myself with putting together in a narrow compass the most important points, and am often only able to repeat what I then stated. A very considerable number of perfects have quite the force of a present, and among them very many Homeric ones. In some instances it is easy to derive the present meanirig from the temporal force of the perfect which afterwards became the rule, that of a present of completed action : e.g. κέκτημαι (cp. Goth, aig) ' I possess ' from the notion i I have acquired,' έ-γνωκα=ηουί Ί have recognised.' But in the case of a very large number of verbs, especially verbs, as Fritzsche p. 48 well remarks, which denote an action as distinguished from a state, we cannot get out of the difficulty thus. For a completed action ceases, while a con- dition brought to a state of completeness may endure. Viewed tem- porally πέπληγα can only mean ' I have struck,' which carries with it the meaning of ' I am striking no more,' and any one who wishes to derive the whole usage of this form from such a fundamental meaning must take X 497 χ(ρσϊν π^πληγως και οναδείοισιν ένίσσων as denoting that the action of the first participle is completed, while that of the second endures, an explanation which a little reflexion upon the connexion of the passage will at once show to be quite impossible. One cannot see bow κέκραγα ' I cry ' is to be explained as ' I have cried.' 5 The investigations of the use of the tenses in the Veda, in which Delbruck is engaged, will not fail to show how far such isolated observations are confirmed. 376 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. The imperative κεκραχθι if from a perfect of completed action could only mean ' have cried ' i.e. ' have done with your crying '; but it means just the opposite; e.g. Aristoph. Yesp. 198 tvbov κβκραχθί της θύρας κςκΚξίσμίνης. When it is said of Thersites Β 222 o£ea κεκληγως λεγ' όνείδεα, 153 the meaning is certainly not that Thersites first cries out and then utters abuse. In Ρ 264 the perfect βεβρυχ$ μέγα κνμα after ώς οτε is certainly not connected with the present βοόωσι to convey the meaning that the roaring of the waves is past when the beach resounds, but that both continue side by side, just like the battle-cry of the Trojans, which is more vividly presented by this comparison. In short the explanation of the present perfects from perfects of the usual kind is in many cases impossible, in others only to be carried out by unnatural devices wholly at variance with the simplicity of the earliest language. Scholars have been led to these attempts at explanation only by the erroneous notion that there were no other means of accounting for them, and that the perfect form must necessarily have from the first the assumed temporal meaning. But this is not at all the case. On the contrary we have seen repeatedly that the original force of the reduplication was intensive and that the perfect was a present to start with. By means of redupli- cation intensive presents were formed at early periods in the language. Some of these presents followed the rule of the present formation. To these belong the reduplicated presents of the primitive formation like Ι&ωμι and the not very numerous forms with a thematic vowel and a similar expansion, among them especially verbs of the I-class, like γαργαίρω, τιταίιω, and also the Sanskrit intensives, which are charac- terised by a heavier reduplicatior, e.g. dar-dhar-ti he holds zealously (it. dhar), α-ρϊ -pet he swelled (rt. pi), nd-nad-ati they sound loudly, and some similar Greek forms like νψνίω by the side of νεω, Ιει-Ιίσσομαι , κωκύω, μαι-μάω (Fritzsche Stud. vi. 300 f.). Others, which did not establish themselves as present forms, presented themselves above on p. 288 ff. as reduplicated aorists. Now for some of such forms a distinc- tive method of treatment came in, that is, there were certain peculiarities of the reduplication vowel, the personal endings and the thematic vowel, through which gradually a special category arose. This new category of individualised presents we call perfect. In it the derived and transferred meaning, that of completed action, comes to be predominant. 154 But it cannot surprise us that at a time when all this was in a state of flux, a series of reduplicated forms took the same outward form, while the£ remained true in their meaning to the original destination of this device of language. I expressed this in the Tempora und Modi thus (p. 176) 'the word-forming reduplication creates in κεκραγα and μέμνκα the same terminations as the reduplication of inflexion does in λέλοιπα, βεβρωκα.* All these views have been abundantly confirmed and ren- dered much clearer by the more exact investigation of the Vedic forms. In the Vedic dialect the limit between intensive present and perfect forms, as Delbriick shows fully on p. 135, wavers in many ways. In short it comes out from all this as clearly as possible, that tJie Indo- Ger manic perfect only by degrees parted off from a reduplicated present, ch. xvi. INTENSIVE PERFECT-PRESENTS. 377 and from this it follows that where we find in the perfect a present meaning, which agrees with the force of reduplication elsewhere, 6 there is not the slightest reason to regard this as anything secondary, but rather on the contrary to recognise in it something extremely ancient and primitive. The very considerable stock of these noteworthy present perfects may be classed as follows. 1) Mimetic verbs. Perfects of this kind follow the analogy of presents like μορμύρω, κικλήσκω. Fritzsche p. 48 quotes eight mimetic perfects in Homer with the most decidedly present meaning : βέβρνχα (Ρ 264), γέγωνε (ε 400), ιαχυϊα mentioned above p. 373, κέκληγα (Ρ 88) with the later by-form κίκλαγγα (Arist. Yesp. 929), λελ?/ι,-ώς X 141 (λελά*α in the tragedians), μεμηκώς Κ 362, αμψφέμνκεν κ 227, τετριγνϊα Ψ 101. There are further the later κέκραγα, which is so common in Attic prose that the redu- 155 plicated verbal stem is employed not merely in the future κεκράζομαι, but also in nominal formations like κεκραγμός, κεκράκτης and the comic compound κε^αίι^ά^ας (Aristoph. Yesp. 596), κεκριγότες Arist. Av. 1521, and κεχληΰέναι ' ψυψεϊν Hesych. (Fritzsche p. 51). 2) Verbs of sight and smell, wherein we include those which denote the excitement of these sensa- tions, and those which denote the receptive activity, as under the first head. Language indeed often denotes both by the same stems (Princ. i. 140). Reduplicated presents of this kind are furnished by μαρμαίρω, παιφάσσω, οε^αλλω, παπταίνω. Here belongs ΰέδορκα, which means only look, glance, beam : ττνρ όφθαλμοϊσι ΰεΰορκώς τ 446, το λτλεος τηλόθεν ΰ^ορκε Pind. 01. i. 94, πρόσωπον μήτε δεΰορκος μήτε σνννουν Aristot. Physiogn. p. 808, a, 4 (a face neither expressive nor intelligent). There are also λέλαμπε which in Eur. Androm. 1026 can hardly be taken otherwise than as a present, and must certainly be taken so also in Troad. 1295, and Όδωΰα, the plupf. of which is Homeric : ι 210 όΰμή δ' ηϋε'ια από κρητχφος όΐωίει. οπωπα on the other hand appears from Homer (e.g. Ζ 124) onwards so often as the present of the completed action that even in Soph. Antig. 1129 σε δ' νπερ ΰιλόφον πέτρας στέροψ οπωπε λιγι-νς we shall not explain οπωπε by ορά as the scholiast does, but re- gard it as a genuine perfect 'has seen thee.' We may however place here εοίκα, which has always a present force, only that the fundamental meaning of the root was perhaps a very different one (Princ. ii. 309). 3) Verbs of bodily actions. We have reduplicated presents of this kind in βαμβαίνειν chatter, ποιπνΰειν pant. The following perfects belong here : ΰεΙρα-γμένος grasp- 6 We may quote here the words of Lobeck ad Sophoclis Aiacem v. 380 on the nature and force of reduplication : ' In παιπάλλω clare apparet vis reduplicationis intensiva. Nam ut προπρό, πάμπαν, aurauros, ijmpjtus, quisquis, undeunde nuda soni eiusdem iteratione plus significant quam simplicia, ita verba quae motum crebrum et quasi coruscantem demonstrant, reduplicationem tanquam propriam notam continuatae actionis recipiunt.' Reduplicated presents have been dis- cussed on pp. 105, 179, 209, 212, 215, 217, 221, 226. 378 PEEFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. ing (κονίης £. αιματοεσσης Ν 393), ΰειΰεχαται η 72 with the past ΰει- ΰεχατο (Δ 4 τοι δε χρνσέοις Ζεπάεσσιν ΰειΐέχατ' αλλήλους) in the same meaning of greeting, which attaches to the reduplicated inchoative form ΰεώίσκομαι or δεΰίσκομαι (cp. above p. 197), κακώς κεκαφηότα θνμόν Ε 698, κεχηνότα Π 409, and in the other forms also in Attic writers of the wide-opened mouth, λελειχμότες licking Hes. Theog. 826, which 156 Fritzsche well compares with the equivalent Skt. intensive le-lih, πεπο- τήαται of the fluttering of birds Β 90 αϊ μεν τ' ένθα αλις πεποτήαται, αϊ Ιε τε ένθα, πεψρικυϊαι chilled Η 62, cp. Pind. Isthm. 6, 40, but also in an im- material sense Λ 383 οϊ τέ σε πεγρίκασι, πεψρικώς shuddering Demosth xviii. 323, σεσηρεναι grin Hes. Scut. 268, Aristoph. Pax 620. — We may also without violence place here the almost synonymous Homeric pluperfect, ελέλικτο (from ελίσσω) A 39, and όρωρέχατο A 26, for ' to stretch one's self' is a bodily action ; but in Π 834 όρωρέχαται πόλεμίζειν is to be regarded as a feeling of the mind. Perhaps also τέθητα is of the same kind, if it denoted originally the staring gaze of astonishment. 4) Verbs of mental states. These have certainly to a large degree been developed out of the preceding group, for the action of the body was the sign of a certain excitement of the spirit. But the transference of meaning cannot always be pointed out so clearly as in the case of πέφρικα and όρωρέχαται. Here belong άκάχημαι Τ 335, άλαλνκτημαι, Κ 94, γέγηθα θ 559, ΰείΰια and ΰείΰυικα, to be compared with ΰειέίσσομαι, εολπα Χ 216, ερριγα Ρ 175, κέκηΰα Tyrt. 12, 28 Be. 3 κεκοτηότι θνμψ Φ 456, κεχαρηότα Η 312, λελίη- μαι Μ 106, λελιμμένος (rt. λιψ) Aesch. Sept. 380, μεμαώς and μέμονα (cp. μαιμάω), μεμηλώς Ε 708, from the middle form of which the reduplicated present μέμβλεται Τ 343 has been formed, /ιέ/χν^^αι in ordinary Greek, μέμηνα Aesch. Prom. 977, τέτηκα Γ 176, τετίψ'ηες Γ 30, τετιημένος θ 437, τέτλαθι Ε 382, πεψνζότες Φ 6. Some of these perfect stems also pass into noun-formation, as is shown by άκηχεΰόνες ' λύπαι. — In Latin memini and odi belong here, the former of which by its imperative memento, a form unique in all Italian languages, is proved to be a genuine present. Several present perfects in Teutonic languages also denote emotions of the mind, as Goth, man —μεμονα, Lat. memini, 6g I dreaded=a*:axi7pn. 5) Intensive perfect-presents from other verbs. The remaining forms may be arranged in two subdivisions : a) Intransitive, 1 57 which naturally denote a condition which is one degree stronger and fuller than that expressed by the corresponding present form (cp. Butt- mann ii. 2 89). Here belongs άλαλ^αι by the side of αλάσθαι (γ 313), differing from it much as ' roam about ' does from ' wander.' The mean- ing of βέβριθα comes out clearly, if we compare r 112 βρίθγσι It ΙίιΊρεα καρπφ with Π 384 ως δ' νπυ λαΐλαπι πάσα κελαινι) βέβριθε χθων, though the intensive force of the reduplication is less evident elsewhere, e.g. ο 334 τράπεζαι σίτον και κρειών ήΰ' οίνου βεβρίθασιν } SO the difference ch. xyi. INTENSIVE PERFECT-PRESENTS. 379 between ή άγορα πέπληθε (Pherecrates Com. ii. 265 πριν άγοραν πεπλη- θέναι) and πλήθει (πληβούσης άγορας) is at most very slight, and sb with τέθηλε and βάλλει. We see more of the intensive form in Η 345 άγορή γένετ Ίλ/ον εν πόλει άκρτι δεινή τετρη-χνϊα, and in the familiar use of πέποιθα by the side of πείθομαι, κεχρημέιος ρ 347, and κεχρητο π 398 are stronger than χρώμενος and εχρητο. Warsehauer in the essay above- mentioned p. 19 wishes to take the Homeric βέβηκα often as an intensive βαίνω, and to explain thus the remarkable use of the pluperfect 'de deorum gradibus aeterni roboris plenorum,' e.g. A 221 ή h 1 Ούλνμπόιδε βεβήκει. Certainly the same form occurs often enough of men in the same sense, where an especially vigorous movement does not suit the context, e.g. Ζ 495. But it is certain that εβεβήκει, when thus applied, approximates very closely to the imperfect, and that an entirely false sense is introduced into the passages by those who try to find in it a real pluperfect of previously completed action. In the same way we have a perfect with the force of a present in Δ 11 τω δ' άντε ψιλομμείδης 'Αφροδίτη αιεϊ παρμέμβλωκε. b) Transitive. In A 113 προβέβονλα has decidedly an intensive tinge, which has entirely disappeared in the German perfect-present ' ich will,' and so in βεβρώθοις Δ 35, which approximates as nearly to the verbs of bodily action as βέβουλα does to those of mental condition : there are also κεκοπώς σ 335, πεπληγνΐα Ε 763. A very clear instance, out of which a temporal perfect cannot be extracted by any kind of dodges or devices, is given in κλήρω νυν πεπάλασβε Η 171. Reduplication has here much the same effect as hi παιπάλλειν. The intensive force has disappeared in arwya and γεγριφώς ' 6 ταΐς γερσϊν αλιεύων (Hesych.). It is worth noticing that verbs denoting a state* appear far more often in these forms 158 than others do. An action often seems all the more energetic, the more it is complete within itself. This is the sense in w T hich Rumpel ' Die Casuslehre ' p. 118 f. speaks of the ' intensive pregnancy ' of the intran- sitive. The circumstance that so many perfects of the earlier stamp are used intransitively, is closely connected with this. To this list belongs probably the Indo-Germanic vaida, perfect from the root vid see, with the heightened meaning of inner sight or knowledge (Skt. veda, Zd. vaedd, 2 sing, voigtd, Gr. Ρυϊδα, Goth, vait), which lost its reduplication, it is true, in the very earliest times, but was certainly characterised originally by this device of language, just as much as the Homeric δεδαώς, in order to distinguish the clearer inner sight from the external. . For the current explanation f. I have seen " and hence ' I know ' has much too sensualistic and empirical a flavour to suit very ancient times. In this examination we have omitted all forms in which it is pro- bable or easily possible that the present meaning originated from a fully developed temporal perfect. We might in this sense distinguish between primary and secondary perfect-presents. Those are secondary in which the meaning attaching to the present, and indeed also to the aorist forms, of the gradual or the sudden origination or of the coming into being of the action is excluded in the perfect : this is the case certainly in ε-στη-κα which presumes the ' petrifaction ' of the movement which is presented in ΐστασθαι, στή να ι, and in βέ-βηκα in the meaning I have 380 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. stept out, I stand firm, whence the adjective βεβα-ιο-ς is derived from the perfect stem βεβα. To this class belongs also the familiar Homeric άμψιβέβηκα A 37 ος Χρνσην άμφιβέβηκας, which I formerly explained erroneously as ' wander round.' Passages like Ρ 359 and the substan- tive άμφίβασις Ε 623 prove that the present perfect denotes a firm and, properly speaking, striding position to protect an object. This was the explanation given by Aristarchus, as Aristonicus testifies : κατά μεταφορά ν εκ τετραπόδων νπερμαχεϊν. This view is now with justice generally adopted by most commentators. Hence this is a secondary perfect- present. We must come to the same conclusion about the Lat. co-epi (Lucret. [once : and four, times in Plautus : cp. Munro on Lucr. iv. 619]), 159 contracted coepi, like incepi. Of the Greek perfects with a present force I am inclined to interpret in the same way έγρήγορα, τεθάρσηκα, τέθνηκα, νενόμικα, πεπόνημαι, πεφόβημαι, έσπούΰακα, πέψυκα. It is some- times difficult to decide, as so often in questions of syntax, because we have no criteria for the time in which a definite usage established itself. It is not uncommonly just as possible that very ancient tenden- cies continued to show themselves in a usage, as that later ones were at work. So much is certain. Even the language of Homer is acquainted with the employment of the perfect form to denote completed action. In clearly recognisable distinction from the extremely numerous in- stances of a purely present usage, just discussed, we meet with perfects like Τ 122 ήδη άνηρ yiyov έσβλός — Ένρνσβενς, Ρ 542 ως τις τε λέων κατά ταϋρον εΰηίως, Ε 204 ες"ΐλιον είλήλονθα, Γ 57 κακών ενεχ δσσα εοργαζ) Α 125 άλλα τά μεν πολίων έζεπράθομεν, τά ΰέΰασται, Γ 134 πόλεμος δε πέπανται, μ 453 είρημένα μνθολογεύειν. Those are note- worthy instances in which one and the same perfect is used in two ways, e.g. ρ 190 3/} γαρ μέμβλωκε μάλιστα ήμαρ by the side of the purely present •καρμέμβλωκε quoted above. Probably this latter usage is much older. Perhaps we should represent to ourselves the way in which the present of completed action was developed out of the intensive present somewhat as follows. By the side of the intensive present there was in many cases from the first a non-intensive present formed from the same stem. The dis- tinction between the two necessarily became more precise, as time went on. The non-intensive or ordinary present, e.g. ολλνμαι, έρχομαι, γιγιώσκω, inasmuch as it expressed a less vigorous action to start with, by degrees, as compared with the perfect present, shifted into the form for the (so to speak) incipient or growing, attempted, intended, preparatory action ; . while on the other hand the perfect present e.g. ολωλα, ελτ/λυβα, εγιωκα denoted the opposite to all this, the completed, full, finished action. Owing to the fact that a past tense, the pluperfect, was very often formed from the perfect stem, and not uncommonly a future also, this contrasting usage got more and more established ; and as the instinct of language is directed, as time goes on, much less to the ancient, delicate and, so to speak, qualitative and more physical distinctions between syno- 160 nymous forms, than to easily conceived distinctions useful for the con- nexion of speech, the perfect was used more and more to express a stage of time distinct from the present. The last step in this development, viz. the change of the present of completed action into a past tense of action generally, is prepared for perhaps in later Greek prose — a question which still calls for thorough investigation — but is never earned out in ch. xvi. PEKSONAL TERMINATIONS OF THE PERFECT- ACTIVE. 381 Greek : [cp. Winer's Grammar p. 340, with Moulton's note (Ε. T. ed. 2)]. The existence of the aorist, which in Greek is so fully developed, pre- served it from this. On the other hand the so-called perfectum his- toricum of the Romans, the prevalent corresponding usage of the perfect in Sanskrit, and the narrative use of the Teutonic perfect, to which has been given indeed the name of preterite, or even of imperfect, represent the final point in the long series, in which the different stages can be clearly distinguished. How the present of completed action can change into a past is shown most strikingly by the exclusively popular use, especially in South Germany, of the periphrastic perfect ' he has done, said ' etc. in narrative. Doubtless the comparison of languages which are not cognate would also be able here to explain and confirm much. Thus W. v. Humboldt in his work 'TJeber die Yerschiedenheit des menschl. Sprachbaues' p. 267 mentions that in the Huasteca language the same syllable serves to denote the violence of an action, and to express the past. We have here the same starting-point and the same final point as in the Indo-Germanic languages, and we may probably conjecture that there were also similar intermediate stages between the two. II. THE PERFECT ACTIVE. In no tense is there such an essential difference between the form of the active and that of the middle as in the perfect. Almost without ex- ception the perfect middle follows the primitive method of formation of the verbs in -μι, while the active, though it has also preserved some relics of this method of formation, in the vastly predominating number of verbs goes back to a disyllabic stem ending in a vowel. The only 161 distinctive mark common to all perfect forms is reduplication. Now that we have treated of this generally, we must divide our subject-matter, and discuss the perfect active by itself. In doing so we start with the terminations, and then turn to the formation of the stem. It is only later on that we come to the perfect middle with its essentially different characteristics ; and this, like the active, we shall deal with first in the indicative. The forms proceeding from the different perfect stems, viz. the pluperfect, the moods and verbal nouns of the perfect, and the future of the perfect stem form the close of the whole widely ramifying dis- cussion. A) Personal Terminations of the Indicative. The indicative of the perfect has long since taken its place in the system of the Greek verbal forms among the main tenses, i.e. by the side of the present and the future. The primary personal endings, which have not suffered the slightest alteration in the middle, and in the active only differ very slightly from those of the tenses mentioned, clearly point to this common character. In Greek it is only the infinitive and the par- ticiple of the perfect which have terminations entirely differing from those of the present. In the cognate languages, on the contrary, the case is quite otherwise. The Sanskrit and Zend perfect points, it is true, by the diphthong e in the middle, which is characteristic of primary forms, to the fact that the perfect is one of the present formations, and the ending 382 PEEFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. of the 2 sing. mid. -se agrees completely with the Greek -σαι. But in the active we can notice no remarkable likeness of the personal terminations to those of the present. The 1 sing, and 3 sing. perf. are without any terminations ; the 1 plur. has not -mas, but the secondary -ma, the 3 pi. -us, which occurs also in various past tenses. The Latin perfect too does not correspond at all completely in its terminations to the pre- sent ; though this is less surprising here, seeing that in Latin the dis- tinction between the two kinds of personal terminations is almost entirely obliterated everywhere. It is however noteworthy that the only ter- minations which are quite peculiar to the perfect distinguish this tense quite as definitely from the present, as from all other tenses. In -the 2 sing. Skt. -tha, Zend -tha, Gr. -σθα, Lat. -sti 162 correspond. We discussed this termination on p. 34 ff. ; it may be placed indifferently under the head of the primary or the secondary. Moreover, olaQa=vettha is the only instance showing the two termina- tions with the same root. But for the Latin perfect, forms like vidi-sti, legi-sti as contrasted with vides, legis are of much importance in deter- mining the connexion of this tense. For the Greek, the question is whether the greater agreement of the Greek perfect with the present in its terminations is a very ancient feature which has been here preserved, or a later formation, which only arose on Greek soil. I have no doubt about the answer to this question. We have had reason to think that the perfect was originally nothing but an intensive present. How then could the almost complete identity of the terminations in Greek, and the extensive resemblance in the case of the middle in Sanskrit rest upon chance ? It is therefore extremely probable that the deviating forms of the perfect active in the two Asiatic languages are based upon later mutilations. We found a complete loss of the ending in the 1 sing, present also. Now evidently reduplicated forms might more easily than others suffer such a loss, because of their length and heaviness. But in the case of Sanskrit this abbreviation in the active is a very ancient one. Delbriick (p. 46) also gives instances of the 2 plur. entirely with- out an ending from authorities as early as the Vedas, e.g. dadd like a . possible *ΰέΰοτε. Who would venture to regard such forms as com- plete 1 They have evidently lost their termination. Now why should not the mi of the first person, and the ti of the third person in the sin- gular have disappeared just as well % Other attempts to explain forms like yeynva, yf'yoi'i=Skt. ηγε-7α. Now that we have thus demonstrated in the Greek perfect various traces of a stem-forming e, sometimes long, sometimes short, we may return to the Latin perfect, and find in the relics of an e in the Greek perfect a very important additional reason for connecting this tense with that which bore the same name in Latin. For there are not wanting 178 cases, in which a Latin vowel fluctuating between i e and ei answers to a Greek e which has come from an original a, as e.g. in the negative particle ne (mi, ni) by the side of the Gr. rq in νήποινος, filius (Umbr. ace. pl.feliuf) by the side of θηλαμώι; semi and sin-ciput by the side of Gr. ήμι. Thus the parallel between Aeol. Ρ<άΙη-μι, ΐοίΰη-σθα and Lat. vldl, vldlstl is completely justified. In the 3 pi. indeed the ordinary form viderunt has the long e. 3 Had the Lesbian Aeolians formed a 3 pi. from the disyllabic stem ϊυιΖη on the analogy of ϊ-σαντι, it must have appeared as ϊοιδή-σα ισι. In other formations from the Latin perfect stem, as in the 1 plur., the short vowel comes out distinctly. There is really hardly any more reason for holding that vidimus is shortened from vidimus than for holding that in γεγόνα-μεν or Skt. tfagani-ma or Goth, vitum the short vowel took the place of an earlier long one. Con- junctives like vldS-rim, futures like vide-ro, pluperfects like vidg-ram show the short e, and thus resemble Greek forms like πεπονθέ-μεν, γε- yovi -τω. Finally the fluctuation of the quantity in the 3 pi. now first becomes properly intelligible. While in the other forms the usage became firmly fixed, here we may recognize both analogies side by side, the formation from a stem in e, which became the prevailing one : vide- runt, steterunt, and that from one in £ : steterunt. Stetirunt is to stete- runt as πεπονθέ-μει> is to πεψυτενκή-μεν. In this as in other cases I • Joh. Schmidt's conjecture (Vocal, ii. p. 345) that the length of the vowel is due to the influence of the following r must remain improbable, until some rea- son is discovered why in numberless other instances, e.g. in the infinitives in ere, in the conj. impf. in trem, in widely extended formations like generis, viscera, and in the whole past tense tram the r had no lengthening force. ch. xvi. PEBFECTS WITH A PRESENT INFLEXION. 393 incline more and more to the view that the quantity of the formative syllable was not always from the first quite fixed. Quantitative inde- finiteness or the existence side by side of a long vowel and a short one, seems to me, especially for early times, not at all an inconceivable thing. By these considerations however we are brought back to another analogy for the e in the perfect. We learnt previously the existence of Old Indian intensive forms from stems in I, but declined to regard the agreement in this vowel as 179 the sign of any particular identity of formation. Now, after we have seen how many different traces of the long stem -vowel can be pointed out by the side of the short one, the question is whether this Old Indian long ϊ was originally the same as the Greek η, i.e. whether they do not both go back to a long a. The question is connected with the nature of the % in Sanskrit generally, which takes the place of a even e.g. in the present formation of the so-called 9th class, jund-mi 1 pi. junl-mas. Delbriick p. 151 says ' originally the a extended to all the persons : the separation of nd and ni is of later date.' In these forms too, as in the perfect, there are by-forms with a short a. It is therefore very natural to conjecture that we have to do here with the same phonetic processes, and that the % of the reduplicated Sanskrit- intensives as the representa- tive of an earlier d may really be compared both with the Doric η and with the Latin i (ei, e). Finally we have still to discuss a treatment of the perfect stem, which has hitherto only been mentioned in passing, viz. the complete transition into the analogy of the thematic present. The present-like nature of the perfect made this danger evidently a very near one. The active perfect had, so to speak, to steer between Scylla and Charybdis. The one rock, on which it might split, was the analogy of the past tenses, the other danger the transition into the forms of the present of the incomplete action. Both are as a rule happily avoided. Still the attractive force of the present seems to have been not small, especially in the earliest times. Hence the forms of the kind in the epos. Later on the language of literature almost entirely resisted this tendency, but in Doric and Aeolic dialects the false tendency was developed. The evidence of the latter fact preserved to us is comparatively extensive, so that its range was probably tolerably wide. The Homeric perfects with a present inflexion are the following. άνώγει Ζ 439 and elsewhere, also in Herodotus (vii. 104), with the past ήνωγον I 578. The imper. άνωγέτω β 195 may just as well belong to the true perfect ; the infin. hruyiatv (N 56), to be explained in Homer as 180 an instance of the present inflexion, comes to coincide with the Doric πεπονθέμεν discussed above p. 391. We saw on p. 373 that άτωγα is in origin a perfect. How else would its perfect inflexion be possible 1 There are many perfects that have become presents ; but presents can hardly have become perfects. — hthw, e.g. Ε 44, is shown to be a perfect by its reduplication, which in this case is strengthened after the fashion of the intensives (cp. ΰειΰίσσομαι). Evidently ΰείΐω is for *ΰει?ίω, which would be to the extant fafiia (N 49), ΰείΰιας (σ 80) as άνώγω to ονωγα. — κεκλήγοντες, the reading of the best M.SS. in Μ 125 and elsewhere, and proved by Didymus on Π 430 to have been one of the two readings of Aristarchus, while κεκληγώτες (cp. τετριγώτες etc.) was the other, is rejected by I. Bekker Horn. Bl. i. 94, but there is strictly 394 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. speaking nothing surprising in it, except that it stands alone among Homeric participles. For in view of the variety of the Homeric forms who could find anything surprising in the difference from the singular κεκληγώς 1 κεκλήγοντες ocelli's with the same variant in Hes. Scut. 379, 412. — We have also in Hes. Scut. 228 the uncontested ερρίγοντι εοικώς. It is worth noticing that all these four epic perfects are always used as presents, and hence it was all the easier for them to take the form of the present. The perfects in ω are most commonly denoted as Sicilian and speci- fically Syracusan ; for which cp. Ahrens Dor. 329. Herodian (ii. 830) ειώθασιν οι παρακείμενοι τρεπειν το ά εις ω καχ ποιεϊν ενεστώτα, πολύ ?έ τοιούτον εθος πάρα Σνρακοσίοις' ολωλα όλώλω, ΰίΰοικα ΰεΰοίκω, κα\ το κέκλνκε (imperat.) $ε παρ* Έπιχάρμω άπο θέματος τον κεκλνκω. — δεοΌικω occurs in Theocr. χν. 58. These too are purely present in meaning. The second and third persons in -etc and -ει, belonging to them with the variants -ης and -η, have been already mentioned on p. 390. Among these there are some perfects of present meaning e.g. γεγάθει=γέγηΟε Epich. 75, but also unmistakeable perfects of completed action e.g. άλιψβερώκει Sophron. fr. 63, πεψνκει Theocr. v. 33. To this group of forms belong also the Doric infinitives in -ειν, also mentioned above, of which γεγόνειν 181 (by the side of εφηκότα) and [άμφιι-]βατήκειν are recorded on a Rhodian inscription (C. I. 2905 B, 1. 6 and 7). γεγάκειν occurs in Pindar 01. vi. 49, κεχλάδειν fr. 57 Be.', πεψνκειν in Epicharmus fr. 97, ΰεΰύκειν in Theocr. i. 102. Hesychius gives also ίηγορέϊν, probably an error for ήγόρειν, with the explanation εγρηγορέναι Λάκωνες, a word discussed in Stud. vii. 393, and κατατεθηπειν θαυμάζειν. Participles of the like kind are quoted by Ahrens from Archimedes : μεμενάκονσα, άνεστάκονσα. We may add from the Delphic inscriptions published by Wescher and Foucart (no. 190, 15) τετελεντακονσας. Pindar follows the same fashion in the two present-perfects πεφρίκοντας Pyth. iv. 183, and κεχλάΐοντας ib. 179 with κεχλαΰώς 01. ix. 2. Outside Sicily therefore these forms are only in use here and there among the Dorians. On the other hand the Lesbian Aeolic usage in the participle seems to have been more regular (cp. Ahrens Aeol. 148). The ScholionV. on Π 430 describes κεκλήγοντες as Aeolic, with which cp. Herodian ii. 306, 35. πεφνγγων=πε<ρενγώς is quoted there and in several other places from Alcaeus, and also νενοή- κων, ειρήκων. πεπληρώκοιτα is found on the Lesbian inscription C. I. no. 2189 1. 9, έπιτετελέκοντα in Conze (Reise auf Lesbos) No. XVII. 1, ευεργετήκοισαν X. 1, κατεληλύθοντος VIII. 2, 9} hence Κ aibel justly supplements πεποήκ[ων] in the Lesbian inscription discussed by him (Ephem. epigr. II. xx.). In all Greek dialects the perfect-present ήκω is common ; it has never quite lost its meaning of completed action, and hence it was rightly explained, even by the ancients, as a perfect. Thus in Anecd. Oxon. i. 212 we read εκ τον ΐημι ησω ηκα και ήκω ' έζ ου και δασύνεται' από $έ τον παρακειμένον τον %κα γίγνεται κατά Σνρακηνσίονς ρήμα ήκω, ώς πεποίηκα πεποιήκω. This view is confirmed, as Ahrens Dor. 345 has noticed, by the gloss of Hesychius ικαντι* ηκονσι. For we have more than one testimony to this t as Doric, even in an inscription C. I. 2140 b, according to Ahrens's probable explanation of the conjunctive, ov γαρ μή σννείκη, by συνήκη in the sense of προσηκγ ; and ε1κω=ήκω occurs in Epi- charmus fr. 19, 13, ε\κε=η*κε in fr. 24, 2. Hence *^κα — for we must 182 assume this to have been the earliest form of the word — was probably ch. xvi. CHANGES IN THE VOWEL OF THE STEM-SYLLABLE. 395 the perfect of the rt. k (ϊκέσθαι). We may doubtless place here also the feminines of participles of the shorter perfect formation, like the poetic βεβώσα, first found in υ 14, then in Soph. O. C. 312, γεγώσα Eur. Med. 405, and the common Attic εστώσα (in Herodotus, e.g. v. 99 εστεώσα) and τεθνεώσα (e.g. Lys. 31, 22). It would be certainly a mistake to explain this ending -ωσα by going back to -νσια, the earlier form of the termination -via. We have here to do with later formations. They are all evidently contractions from -αονσα. As εητηνία and the like sounded too archaic by the side of εστώς, while a contraction of ηνι was hardly possible, εστώσα was formed on the analogy of πμώσα and similar forms universally familiar. Posidippus went even further in giving after the Syracusan fashion εστήκω for the 1 sing, εστηκα (Athen. x. 412). c) Changes in the Vowel of the stem-syllable. It is entirely in harmony with the origin of the perfect from an intensive present, that language is in many cases not content with re- duplication in this tense, but that as a rule the stem-syllable, if not already long by nature or position, undergoes a strengthening of the vowel. This strengthening of the vowel of the stem is the counterpart to the raising of the vowel of the reduplication syllable, which is found espe cially in intensives. It is a very ancient phenomenon. In the preference for a fuller vocalism of the stem-syllable Greek agrees with Sanskrit, Gothic and old Erse. The following forms may provisionally serve as examples of this : Gr. rt. φαν perf. πβφηνα \ 7 ed. rt. tan S s. perf. tat ana „ „ λΐ7Γ „ \ί\ονπα „ „ riK (free) 1 s. perf. HreKa „ „ φυγ „ πέφευγα „ „ rutf (break) 1 s. perf. rurotfa. A special agreement of Greek and Sanskrit is found in the fact that the intensification of the vowel in both is attached to the polysyllabic stem- formation. Delbruck p. 119 lays down the rule for the Vedic dialect, * If the stem-syllable is short, the termination is immediately added : if the stem-syllable is long, an intermediate i comes in between this syllable and the termination.' Without agreeing with the assumption of an 183 inserted letter, we may regard the facts observed in the following way. In the Vedic dialect intensification of the vowel occurs only when the syllable of the root receives a kind of protection in the vowel which is added, and which, according to the view we have previously developed, forms the stem. And precisely the same holds good of Greek : veda= folia, vid-md =ΡίΙ-μεν. Differences appear indeed in individual instances, for the a is treated in particular Sanskrit forms differently from the i. On the other hand there is one exception, which is found in both languages in common : οισθα has the fuller sound, just like the Vedic vettha, in spite of the immediate addition of the terminations. The instances of the agreement of Greek with Gothic and Erse are seen mainly in the dif- ferentiation of the a, to which we shall return. Latin, on the other hand, in direct opposition to Greek, shows a preference for a short vowel in the root-syllable, as is seen from pipigi as compared with πέπηγα. As we have preserved to us from Old Latin the form tu-tudi (cp. Skt. tu-toda), which was afterwards replaced by tu-tudi, we may conjecture that the Graeco-Indian rule was originally not unknown to Latin, but 396 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch.xyi. that afterwards, not without the influence of the accentuation, it changed to the direct opposite in the case of the not extremely numerous perfects, which retained the reduplication. Yowel -intensification in the perfect is evidently an archaic procedure, and hence, as Uhle (Sprachw. Abhandl. p. 63) well puts it, ' the capacity for intensification is so to speak innate in the root.' It cannot be subsequently transferred to later forms, and is confined within veiy definite limits. The Greek perfects may be divided, in respect of vocalism, into three groups : those with complete intensification (λέληΗα, εοικα), those with half intensification (γεγοια, ερρωγα), and those without intensification (λέΧαμπα, γέγμαφ(ΐ). The last group is the most varied, because the strengthening of the vowel is omitted from very different reasons. The first group embraces not only the forms in which the short vowel of the root passes into the corresponding long vowel, which happens only in the case of a primitive a, but also those in which a diphthong appears, 184 which is only the case where there is t or ν in the roo L . From the Greek point of view these two processes, which Sanskrit grammar sharply distinguishes, coincide here, just as in the present (cp. p. 150 ff.). The ά of the Doric λέλάθα, the at of πέποιθα and the εν of κέκενθα are completely parallel. The distinction between the so-called guna and vrddhi is significant only for Sanskrit, not for comparative grammar, and the impossibility of separating these two stages of intensification, which is clearly perceived in every enquiry which extends beyond the limits of Sanskrit, is a main obstacle to all attempts to explain phonetic intensifi- cation from accentuation. From the Gothic we may compare with these groups for the most part such words as ala preter. 61 (sprang up), bi-leiba (cp. λείπω) preter. bi-laib, biuga (cp. φεύγω) preter. baug (cp. πέφενγα). The ά of the root becomes ά or in Ionic η in the following perfects : ϊέϊηα Ρ 253 for ^δε-ΰηϊ-α, cp. COnj. aor. οάηται for *ΰαί-ηται. ΰεΰηχα by the side of ύάκνω, Babr. 77. εάγε i.e. fc -Μγε, Herodot. εηγε cp. above p. 362, with aor. pass. Ιάγη (Ν 162). εαδα cp. p. 362, beside εϊ/αΰυν, ah (p. 79). ε'ίληψα, Dor. ειλάφα (p. 361) by the side of λαβείν. είληχα (ib.) by the side of λάχεΊν. λελάχασι' τετενχασι Hesych. Cp. Merzdoif, Commentationes philologae semin phil. Lips. p. 54. επτηγα in the Attic orators, beside πτήσσω επτάκον. κέκηΰε Tyrt. 12, 28 Be. 3 beside κεκάΖον. κέκηφε' τέθνηκε Hesych. beside the Homer, κεκαψηοτι θυμω. κεκληγύς Β 222 (cp. κέκλαγγα Xenoph.) beside εκλάγον (cp. above, p. 285). κέχάια Dor. (κεχάναντι Sophron. 51 Ahr.). κεχηνότα Π 409, other forms in Attic writers, beside ϊ-χ&νόν (above p. 288). κεκραγα beside εκράγον• For the quantity of the α in the present cp. Uhle Abhandl. p. 68. €7π-λ£λά0α Doric (Pind. 01. 11, 3), λεληθα in Herodot., beside λάθεϊν. μεμηκως Κ 362 beside μεμάκυ'ια and μαχών. 185 μέμηνα in Attic poets beside εμάνψ. πεπάγαισι Ale. fr. 34, 2, πέπηγε from Homer (Γ 135) onwards, beside ΐπάγηΐ'. ch. xvi. CHANGES IN THE VOWEL OF THE STEM-SYLLABLE. 397 πεπηλότι a doubtful reading in Nonnus Dion. xiv. 152 beside eVaXro, πάλος. πεπληγώς Χ 497, πεπληγυϊα Ε 763, other forms very isolated in Attic writers beside έκπλάγήναι, έκπλαΎησομαι. εκπεφάναντι Sophron. 75 Ahr., πέφηνα in the tragedians, more rare in Herodotus and in Attic prose, beside φάνηναι. σέσηπε from Homer (B 135) onwards, beside σάπηιαι. σεσηρώς Aristoph. Pax 620 ; other forms also later in isolated usage ; Theocr. vii. 19 σεσάρως, fern, σεσάρυία Hes. Sc. 268. τέβηπα ζ 168, Herod, ii. 156, beside τάφων. τετηκα Γ 176, τετακότας Eur. Suppl.1141 chor. beside τάκήνα:. τετρηχνΊα II 346 beside ταραχή. The ε of the root becomes η only in μέμηλε Β 25 with μψηλώς Ε 708 and plupf. μεμήλει Β 614, and also in Hesiod Opp. 238. The reading μεμάλότας in Pindar 01. 1, 89 is very doubtful (cp. Tycho Mommsen ad loc). But μέμηλε is intelligible only under the hypothesis that the older form of the root μαλ continued to survive by the side of μελ, like /3αλεΤν by the side of βέλος, καλεϊν and κέλεσθαι, σταλήναι and στέλλειν etc., and that from this came *μέμάλε, Ionized into μέμηλε. Choeroboscus in a passage cited by Lentz in his Herodian (ii. 795) mentions a τέθηλα as a perfect of θέλω, but nothing is heard of this elsewhere, any more than of his τέτνπα and other forms, which seem to bear the mark of a grammarian's manufacture. Of an ι in the root, raised to οι, we have the following instances : hi-doitca A 555 and elsewhere in Homer, common in the weaker form ΰέΰοικα from Theognis (v. 39) onwards, must be regarded as the perfect of the stem hue, which is expanded from £/, and underlies the intensive present όειΰίσσομαι (δειΰίζασθαι Σ 164) and ϊεΰείκελος' άει φοβούμενος Hesych. (cp. Principles ii. 308). For if we take κα as the formative syllable of the weak perfect, the diphthong is hardly intelligible. We 186 shall come back further on to this form as one of great importance for the explanation of the κ•. eoira common from Homer onwards, in Herodotus olua (iv. 82). Forms like είκνΐα, εικός, είξασι will be discussed later on. λέλοιπα in universal use from Homer onwards. ol()a the same, in complete agreement with Skt. veda and Goth, vait.» The weaker diphthong ει is in the moods etc. still more firmly established in this case than in εοικα. πέττοιθα from Homer onwards (Δ 325). πεφλοιΗναι' φλνκταινονσθαι Hesych., ΰιαπέφλοιΰεν ' ΰιακέχυται beside ε-φλίΰε-ν ΰιέρρεεν, φλώάνει.' ΰιαρρεΊ. Cp. Princ. i. 375. κέχλοιΰεν ΰιέλκετο Hesych. Cp. ib. ii. p. 302. Cp. χλιΰη. haix- γλοιΰως' διαρρέων υπό τρυφης. In Plutarch (Ale. 1) we find ΐιακεχλιΰώς in the same sense. Α υ in the root becomes εν in ε£ευχα, the earliest instance of which is in Philostr. Vita Apoll. 2, xiv. 64 (επεζευχότας). κέκενθα X 118 and also in Attic poets. τέτενχα, in Homer only τετενχώς μ 423, and so in Demosth. xxi. 150, in Herodot. iii. 14 ετετευχεε. 398 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch.xvi. πέφενγα from Homer (a 12) onwards in poetry and prose. We have entirely isolated instances in μέμϋκα, Homeric (Σ 580), with the aor. μνκε Υ 260, pres. μνκόομαι, and hence with an intensification of υ to ν (cp. p. 157). είλήλουθα by the side of έλήλνθα, the former only in Homer and once in Theocritus xxv. 35 : cp. above p. 368. Evidently εϊληλονθα is to the future ελενσομαι as πέπυιθα is to πείαω or to the present πείθω. It is the solitary attempt in the perfect to introduce an intensification of υ different from that of the present, just as σπονΰη by the side of σπεΰΖω is an isolated example of the like phonetic change in the formation of nouns. For σπουΐη is to σπεΰΐω as λοιβή to λείβω, while the case is different e.g. with• ακόλουθος by the side of κέλενθος. 187 The total of these perfects is 39. 4 It is only the monophthongal intensification of the a which is at all a common phenomenon : of this there are 25 instances. We come now to the second group, characterised, as we saw, by half intensification, that is, by the change of the e which prevails in the present stem into o. Within this group we can again make three divisions. By far the most numerous has ο in the perfect, by the side of ε in the present : in a few instances ω appears instead of ο by a phonetic affection : in a single instance η passes into ω. The whole phonetic change, which characterises this group, evidently belongs to a later period than the complete intensification. It presupposes the splitting of the a. and therefore has, so to speak, a European character. The diversity of the vowels, which proceeded from an old a, is in fact used, not in Greek alone, to serve, as it were, in the second line for the distinguish- ing of tenses. We learnt in the case of the thematic aorist to recognise the preference of this tense for a, and in the case of the present, the preference for ε . In the same way the perfect is fond of the o. But we cannot talk of a law. The usage may rather be formulated somewhat as follows. The language is not governed by this one tendency alone : it makes use willingly of this very manageable pigment, if so we may call it : but it can get on without anything of the kind. Scholars have long since compared with these Greek processes many phenomena of the Teutonic Ablaut ; and this is not the first case in which the teutonic vocalism has offered us some noticeable points of comparison. Only the latter is far more richly developed, and in the case of the perfect is all the more varied, in that a distinction is made between the singular and the plural. But there is agreement thus far, that, as was shown on page 145 the present stem has a preference for the e, which may become i, while the perfect stem on the other hand prefers in many cases a heavier vowel, which appears on Teutonic soil as a, and in Greek has assumed, though perhaps only at a later date, the duller colouring of the o. Possibly we have a relic of a relatively older time in the form τέτραφα from τρέπω which crops up in the Attic orators by the side of τέτροφα, and the α which appears in the perfect middle (τέτραμμαι, τετράφαταή and 188 which will hereafter be discussed, is perhaps not wholly unconnected with this α of the active. Hence we may so far compare the Gothic giba gab with τρέπω τέτραφα or τέτροφα, stila stal with κλέπτω κέκλοφα* According to Windisch there are many instances of a similar relation, * hvitpya, the ω of which depends upon the influence of the digamma, waaj discussed above, p. 363. ch. xvi. CHANGES IN THE VOWEL OF THE STEM-SYLLABLE. 399 in Erse also : e.g. the present con-dercar beside the perfect ad-chon-darc (vidi) agrees exactly with ϋέρκυμαι ΰέδορκα. The stem-vowel of the perfect was unmistakeably felt by the three languages to be a heavier one, so that this phonetic change is in some degree connected with that discussed under the first group, indeed has perhaps been formed in analogy with it, when the splitting of the a at a considerably later stage in language created new possibilities. Phonetic variation is as a rule more sought after in an intermediate stage of language than in the earliest. Latin is but little influenced by this tendency, and shows in its perfect hardly anywhere anything that can be compared with the facts mentioned. If any one should be inclined to think of isolated instances like pello pepuli, which might possibly be for an earlier* pe-poli, like te-tuli for te-tol-i, a glance at pulsus and the 0. Lat. at-tula-t would upset such a comparison. The Latin vocalism shows the archaic stiffness united with great weakenings. We miss in it almost the whole of the varied life of what I believe to have been the intermediate time. We now turn to the various perfects marked by an o, adding at the end the few instances in which o- is found over against an a in the present. 5 Dialectic variations present themselves here more frequently than in other perfect formations. γέγονα common from Homer onwards. ΙέΙορκα in posts from Homer (X 95) onwards. ^έΰρομα, αναοέΰρομε ε 412, επώέΐρομεν ν 357, rare later. εγρήγορα see above p. 368. * ε'ίλοχα (λελεγε, λε'λογα Hesych.) p. 361. εκτονα from Aesch. (κατεκτονας Eumen. 587) onwards in Attic 189 writers; άπεκτόνεε Herod, v. 67. εμμορι p. 361. ένηνοχε with ήνοκα Hesych. p. 369. εολπα p. 362. εοργα ib. ΰι-έφθορας Ο 128, and again in late prose. άν-ήνοθεν ρ 270, επ-εν-ήνοθεν θ 365, both also as preterites. The present * ανέβω (cp. avdog), which must be assumed, does not occur. κέκλοφας Aristoph. Plut. 372, also in prose. κέχοΰα, έγκέχοϊα Aristoph. Ran. 479. μεμονα in Homer (Π 435), other poets (ψμέμονεν Soph. Trach. 982) and Herodotus (μεμονέναι vi. 84). The present stem with ε does not occur with the same meaning, but we may compare μένος, Μέντωρ, μεμονα '. μεμάασι. * \ γέγονα : γεγάάσι. There is no need to assume a distinct μεμονα, which follows the meaning of μένειν remain, on the strength of Eur. Iph. A. 1495 Ίνα τε Ιόρατα μέμονε laia. κατα-νένογε ' συνονσίακεν Hesych. is of obscure origin. νένοφε " νενέψωται. Hesych. according to M. Schmidt, ξυννενοφυϊαν ' σκυθρωπή ν ib., perfect to ρέψω, which is adopted at any rate in the compound ξννηψει at the suggestion of Cobet Aristoph. Aves 1502 : cp. Dindorf on Aristoph. fragm. 142, 349. πέπομφα quoted from Herodotus and Attic prose. 6 The forms β4βροχα, %στο\α, (στόφα quoted by Choeroboscus (Herodian ed. Lentz ii. 356, 794, 837) and in part noted as Attic, I do not venture to place in this list, any more than ττ4ν\οχα quoted by Veitch p. 482 from one passage in Hippocrates, while everywhere else this writer has ir«r\e*a. 400 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. πέπονθα, from Homer (r 6) onwards, with the isolated participle πεπάβυϊα (ρ 555). For the present form with ε we have πένθος. πέπορΐα in Aristophanes. πέφορβα, only έπεφόρβει Hymn. Merc. 105. τετοκα, τετοκυίης Hes. Opp. 591, then in Herodotus and Attic writers. τέτροφα from τρέπω Aristoph. Nub. 858, ανατέτροφας Soph. Trach. 1009. In Attic orators τέτραφα crops up, for the most part as a variant for τέτροφα. The latter is preferred by I. Bekker and Sauppe. Weidner reads άνατετραφότας in Aeschin. i. 190. τέτροφα. from τρέφω, in Homer (intrans. φ 237), Sophocles (O. C. 186) and later writers. Here too, though first in Polybius, the variant τέτραφα is found. 190 The change of ε into ο is neglected probably only in the following isolated forms. βέβλεφα Antipater εκ του περί γυιαικος συμβιώσεως in Stobaeus Floril. 70, 13 : άποβεβλεφότεν. αν-ηνεχνίαν ' άναφέρουσαΐ' Hesych. Cp. ε νηνοχα and κατ-η ι οκα' κατε- νηνοχα ib. — υπείνεγεν ' υπέβαλε, υπεϊχε ib. is also probably of the same formation. κεκλεβώς=κεκ\οφώς inscription of Andania. λέλεγα* ε'ίρηκα Hesych. beside λέλογυς ' ε'ίρηκας and on the other hand σννειλεχώς ' συι-αγαγών Hesych., εζείλεγα Aristid. 49, 381. όρωρεχότες ' όρέγοντες Suid. πέπλεχα by the side of πέπλο\α Hippocr. πεπνρεχότες Aristot. Probl. xi. 22, πεπυρεχέι -cu Galen. Here are altogether only 8 instances, all of a later stamp and some of them quoted as rarities. We may say therefore that the change of ε into ο became the rule. Besides the 24 instances with ο by the side of ε, there are 3 more forms in which the ο answers to an « in the root : viz. κέκονα quoted in the Et. Magn. from Sophocles (fr. 896 Dind.), rare elsewhere, with the present καίιω. λέλογχα in Homer (λ 304) and other poets, and in Herod, vii. 53. πέποσχα Syracusan, Epicharm. fr. 7 Ahr. Here the present expan- sion has made its way into the perfect, as in the Skt. pa-praUUha (stem praek, perhaps for prak-sk) and Erse ne-naisc nexuit. Cp. p. 198. Finally there is one form in which under the influence of a digamma once present the ο appears as long, viz. ε'ίωθα, from Homer onwards : cp. Brugman Stud. iv. 176 and above p. 363. On the other hand the long ο answering to the long e is only to be found in one quite isolated form : ερρωγα, occurring from Aeschylus onwards in Attic poets and prose- writers, while the Heraclean tables give the form έρρη'γεΊα mentioned above p. 391. Cp. Hesych. κατερρηγότας' Ζιερρηγμένονς. There remains the third group, i.e. perfects which from very different 191 reasons do not distinguish this tense-stem by any change of the vowel. We may here again make two subdivisions, viz. perfects whose stem- syllable is already long, and perfects whose stem-syllable remains short. There is no small number of perfects based upon stems which are always long, and which therefore remain unaltered. ch. xvi. VOWEL OF THE STEM UNCHANGED. 401 1) With a stem- syllable naturally long. προ-βέβουλα A 113. If the explanation of the diphthong given on p. 172 was correct, the present formation has here found its way into the perfect, just as in πέποσχα mentioned above and in Lat. iunxi. βέβρϊθιι in Homer (Π 384) and other poets, beside the present βρίθω. βέβρϋχα in Homer (P 264), Sophocles and later prose writers, by the side of βρϋχάομαι. γέγηθα from Homer onwards (Θ 559), Dor. γέγάθα (Pind. Epich.), beside γηθέω. γέγωνε also in poets. ίεΰίωχα only in Hyperides Lye. p. 29, 6 ed. Schneidewin, beside διώκω. ΰεΰουπότος Ψ 679 beside Ιουπέω. epplya beside ρϊγέω : cp. above p. 360. εσκηψα first quoted from Hermippus in Diog. Laert. i. 117 (επεσκη- φέναι), beside σκηπτω. κεκρ'^ότες Ar. Αν. 1521 beside κρίζω. For κρίκε Π 470 cannot be at once assumed to belong to the same verb. κεκνφα, ανακεκνφαμεν Eur. Cycl. 212, connected with κνφός, κύψων. κεχλάΰώς Pind. 01. 9, 2 and elsewhere, a completely isolated form. — A second perfect of the kind κεχληδέναι ■ ψοψεϊν Hesych. is discussed by Fritzsche Abhandl. p. 51. πέπληθα Pherecr. Com. ii. 265, and later writers, beside πλήθω. πέπράγα from Pindar onwards. πέπϋθα, only recorded in κατα-πεπνΒα ' κατερρνηκα in Hesych. ; hence there is no evidence of the quantity of the v. But it is certainly long as in πύθω. πέψρϊκα cp. above p. 378. Editors write ψρΊσσον in Hes. Scut. 171, and the ι is long also in φρίκη, φρικώδης etc. ϊ\ρχα. ήρχότι in a psephisma in the Yitae decern oratorum, Plut. 192 Moralia ed. Wyttenbach iv. p. 414 F. 2) Stems with α long by position. κέκλαγγά quoted from Aristoph. (Yesp. 929) and Xenophon, evidently formed in imitation of the present κλαγγάνω, and of κλάγζω, έ'κλ crybaby the side of εκλάγον. The Homeric form κεκληγώς was quoted above on p. 396. κέχανΰα Ψ 268, plupf. κεχάνΐει Ω, 192, bears exactly the same relation to χανίάνω. The short α of the root appears only in the aor. εχαΐον. λέλαμπε twice in choric songs of Euripides, Andr. 1025, Troad. 1295 ; cp. above p. 377. μεμαρπως Hes. Opp. 207, in Quint. Smyrn. άμψιμεμαρφε. πεφΰγγων quoted from Alcaeus (Ahr. Aeol. 148); cp. φνγγάνω. The stem-syllable of the perfect remains short with any degree of regularity only in the following cases : 1) In the forms quoted above without a thematic vowel : ΰείΰιμεν and ΰειΰίμεν, ΰείΰιθί, ε'ικτην, τεθναμεν, τεθνάμεναι, τέθναθι, ϊΰμεν, 'ίστε, 'ιστοί', μέμαμεν, μεματον, εσταμεν, εστάθι, τέτΧαθι; τετΧάμεναι. The only words which show intensification without the support of an added vowel are εοιγμεν and ειληλουθμεν. From this consideration we derive a confirma- tion of our view that the ε of εοικέναι, πεποιθέιαι is stem-forming, and not an element of the termination. It cannot be a matter of chance, for the perfect middle, which always adds the terminations directly, is as a rule disinclined to the long vowel: πεφνγμένος, ήίκτο. A small number of 402 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. intermediate forms with the diphthong ει from stems in ι deserve notice, viz. είΰέω, εΐΰείην, ctderm, ειΰώς — εικεναι (Eurip. Aristoph.), εΐκώς (from Homer onwards, Φ 254), ε'ίξασι (Attic), — πέπεισθι Aesch. Eumen. 599, if we are not there to read with Gilbert πέπισθι. This ει takes a similar position as a rarity by the side of οι to that which α holds from roots with f. by the side of o, the more so that both sounds, ει as well as a, are common in the middle : λέλειμμαι, εστραπται. Deviations from a rule are often signs of antiquity. Is it possible that ει originally prevailed 193 also in the perfect as an intensification of *, and only later became weakened into οι, not without the cooperation of the desire to distinguish the perfect from the present 1 2) The stem-vowel further remains short often before the termina- tion -vta in the feminine of the participle : άρηρώς — apapvla (Ο 737), but αρηρυία Hes. Theog. 608, ΧεΧηκώς — ΧεΧακυΊα (μ 85), μεμηκώς — μεμακυίαι (Δ 435), μεμάώς and μεμάότες (Β 818 μεμαύτε Ν 197) — μεμανΊα, τεθηλώς — τεθαΧνΊα Ι 208, with which πεπαθνία, πεφυνΊα also agree, but there is no consistency, as is shown by ποτι-πεπτηυΐαι ν 98, τετρηχυία Η 346, τετρι- . γνϊα Ψ 101, πεπΧηγυϊα Ε 763. The feminine to (Ιΰώς is now written in our editions sometimes εΐΰυϊα (ου πριν άΖυϊα τόκυιο Ρ 5), sometimes Ιΰυϊα (έιΊ φρεσι πάντα ΙΙυΙα ν 417); the feminine of εϊκως, besides the solitary ειοικυίαι 2 418 is only είκνϊα. Perhaps the pronunciation was Ρεϊώνϊα and Ρώυϊα, fefiKvia and ΐικυϊα. This variation in quantity is limited to the Epic dialect (cp. Uhle Abhandl. p. 68). 3) The rhythm of the perfects with the Attic reduplication, which, when the second syllable is long, prefer the third to be short, was men- tioned above on p. 365. Hence we have άΧηΧιφα, εΧήΧνθα, έρίφιπα, όρωρεχα and others of the kind. 4) The stem-syllable of many perfects with an aspirate remains short, even without this kind of reduplication, as in ήΧλαχα, βέβλαψα, βέβΧεψα, γ έγραφα, ΐεΰίύαχα, λέλαφα, πεφΰΧαχα, έσκαψα. In many instances we know nothing as to the quantity, e.g. in ΰιαπέπλιχε' διαβέβηκε, ειαπε- 7τλίχώς* ΰιεστώς, κεχηνώς Hesych. It has often been assumed without any sufficient reason that in such cases the vowel is short. "We can hardly say more than that the short vowel before the aspirate is the more common. The reason for this peculiarity is undoubtedly to be sought in the nature of the aspirates. Uhle op. c. p. 70 asks very justly : * Is it mere chance that the double consonant preserves from aspiration, as it does from intensification ? Or may we suppose that the aspii-ate was felt to be a kind of double sound 1 ' I think we can unhesitatingly answer this question in the affirmative. In fact according to the view established in Principles ii. 9 ff. the aspirates actually consisted in early times of an explosive sound immediately followed by a breathing, and hence were heavier than all other consonants. It is very noteworthy besides that the 194 dental aspirate Θ, the one which is never denoted graphically by TH, in regard to this perfect formation does not hold quite the same place as χ and : and for our view of the whole phenomenon it is of much importance to remember that it is quite unknown to the older language. The old perfects εϊληψα, ε'ίληχα, τετρηχα, τετενχα, as well as κέκενθα and Χέληθα have the intensification of the vowel before the aspirate and prove that we cannot speak of a definite law, but only of a phonetic tendency prevalent especially in Attic. 5) Finally there are still two isolated forms with short vowels to l>e , in κλέπτω π. As there is no trace of any word *κλέβω, it seems to me more probable that the softening of π to β was limited to the perfect, παρτετύμβει shows by its reduplication that it is a perfect, and by the apocope of the preposition and the present ending that it is a Doric form. Taking into considera- tion τνμβογέρων' έσχατόγηρως καϊ παρηλλαγμέιος τρ ΰιανοία. [Anecd. Bekk. 65, Phot.], which can hardly have anything to do with τύμβος grave, I conjecture a perfect τέτυμβα with the intransitive meaning so common with the older method of formation, direct from the rt. τυψ, which has elsewhere only in τνψόω, τετύψωμαι, derived from τύφος, the meaning ' to be surrounded with mist, darkened.' Two epic forms resembling each other, but deviating widely from the formation of these perfects, are πεφνζότες and μεμυζότε. πεφυζότες we find thrice in Book Φ of the Iliad (6, 528, 532), once in Book X (1=Φ 6). In Princ. ii. 96 I think I have disproved the view that the £ here arose from a combination of the final consonant of the root with the / of the participial suffix For. The ζ is there explained by the analogy for φϋζα. Granted that there was a masculine φϋζο, a perfect stem πεφνζα might have been formed, just like the present and aorist stems θερμό-, χραισμο-. We might also conceive a present *φυζω= Lat. fugio : the expansion would then have made its way into the perfect, just as in the Syracusan πέποσχα, quoted above p. 400, or like 203 the nasal of λέλυγχα, πεψύγγων, κέχανΐ-α. μεμυζότε we know only from a quotation of Antimachus in Eustathius (on Od. ν 401) going back to Herodian (i. 444). The old grammarians regarded these forms, like Hesiod's λελειχμότες or λελιχμότες (Theog. 826), as syncopated from πεφυζηκότες etc., but no one will now believe this. For the third of these forms hardly any other explanation than that from a noun-stem is possible. e) The Perfect with κ. This form too, which is very widely extended in Attic Greek, we can see becoming more common by degrees in the period of the language ch. xvi. THE PERFECT WITH κ. 409 for which we have historic testimony. As will be shown in detail further on, there are only about 20 forms of the kind in Homer, while the total number of the perfects in κ quoted by Yeitch reaches, as I count them, to 268. If we remember that the meritorious work of this industrious collector excludes all quite regular forms, and therefore very many derived verbs, we may venture to assert that the total number of these forms may be fairly put at least at 300 ; and it is very remarkable that hardly a fifteenth part of these is Homeric. But the different law of formation is of still more importance than the numerical relation. The language of Homer has perfects with *,• only from stems ending in a vowel. Hence forms like the Attic πέπεικα, ίσταλκα, ήρκα are quite unknown to it. Even the existing perfects from vowel-stems like ΰείΰοικα, βέβηκα, πεφϋκα have by-forms without the /:, like ΰείΰιμεν, βεβάάσι, έμπεφνυΊα There is an especially large number of participles of the kind, like κεκμηωτας by κέκμηκα, τετληυΊα by τέτληκα. The formation with κ does therefore exist, in the time of Homer, but rather as experimental than as normal. And even the Attic writers have retained in the forms quoted above p. 387, like εσταμεν, τέθνατε, ΰεΰιως traces of the twofold formation. Thus the case is different with this whole class of perfects from what it was in the class previously discussed. "We see them gradually making their appearance side by side with the primitive perfects, and supplementing them in a particular direction. For in the case of vowel-stems we can as a rule only get a singular of the perfect by means of κ ; we never find a *βίβαα or *βέβηα, as was noticed on p. 386. This also serves to explain how it has been possible 204 for scholars to conceive the notion that this if, which comes in so con- veniently to distinguish the perfect forms, was developed purely phoneti- cally between the long vowels supplied by the intention of the language and the a. I myself in the Tempora und Modi p. 199 gave my assent to this view, propounded first by Thiersch and afterwards brought orward with greater weight by Ahrens 'Ueber die Conjugation auf MI im Homerischen Dialekt ' § 7, 2, according to which the κ arose ' for the avoidance of the hiatus ' or ' from the gaping of the mouth.' The more exact observation of phonetic processes, which has come up only since the appearance of that early work of mine, has since taught us that the 1 explosive sound of the vocal cleft ' commonly called spiritus lenis, which is heard between two vowels, when one immediately follows the other, cannot possibly grow into a guttural explosive. We shall hardly find any one nowadays adducing the forms ον-κ-ετι, μη-κ-έη in support of such a view. The κ here certainly rests on a pronominal particle added to the negation, which occurs more fully in ονκί and with aspiration in οΰχ/, ναιχί (cp. Poscher, Studien in. 144). Out of thousands of instances where vowels come into contact, this would be the only one in which the contact was avoided by such an insertion. If the science of language in its present position scorns the ' connecting vowel,' it is still less tender to what Westphal calls the ' separating consonant.' I have therefore long ago retracted my earlier view in different places, espe- . cially in the ' Elucidations ' p. 128. Another view, viz. that the r originated in Ϊ or even from σ, has been rejected in Princ. i. 79 as not made out. The labial spirant is just as far removed as the dental from the hard guttural explosive. Phonetic violences of this kind are now probably regarded as possible only by one or two here and there. 410 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. The only correct way of explaining this is to take the κ, which we never find coming from any other sound, by itself, or in other words to establish analogies for it as a formative element added in this place, and extending by degrees as time went on. In the Principles 1. c. I have 205 connected the κ with the root-determinative k which we have often found added, especially to roots ending with a vowel. Holding firmly to this further connexion, I now maintain with positiveness that the κ of the perfect is a stem-forming element, and as we certainly cannot separate the vowel following the κ from the consonant, the syllable ra is stem- forming. It is true that the stem needs such a suffix only as a noun- stem. But after learning to recognise, under the head of the formation of the present, a large number of suffixes, which though originally in- tended to denote radical noun-stems, still made their way into the verb, and after finding even in the thematic aorist traces of such suffixes, quite apart from the vowel, we cannot be surprised to meet again in the per- fect such a stem-forming consonant. 9 No objection can be brought against this view on the ground that the perfect-stem is already marked by reduplication. Reduplication and noun-suffixes are quite compatible, as is shown e.g. by άγ-αγΰρ-της (Hesych.), βέ-βα-ιο-ς, κε-κράκ-τη-ς, κε- κρύφ-αλο-ς. I therefore now regard the κα of U -ΰω-κα as directly analo- gous to the present expansions va, w, vo, αιο, το, σκο and to the το and ko which are demonstrated in some few examples, as we saw above p. 281, even for the thematic aorist. Supposing there was a noun-stem βΰ-κα Ion. βη-κα,. by the side of this a reduplicated βε-βά-κα, βε-βη-κα might have been formed, and such reduplicated stems, at a time when the verbal forms were still fluid, might have made their way into the per- fect, side by side with the shorter forms, just as the noun-stems ΰεικιν, λαμβαΐΌ appear by the side of shorter present formations, and forms like άλιτο, άμαρτο by the side of shorter aorist formations. Such forms characterised by definite distinctive sounds agree with the taste of cer- tain comparatively recent periods of language better than the most .archaic kernel-forms, which in any case must receive a more individual 206 shape, much as a richly developed national life demands definite customs and dress, and as similarity and indeed uniformity as a rule extend more and more widely in the course of the life of a language and a nation alike. Hence we cannot be surprised that this κ, which at first only appears here and there, by degrees gets the upper hand, and in time conquers for itself a wide region over which it prevails, and that the peculiarly marked perfects of the older stamp more and more become varieties by the side of it. From such points of view there is still some reason for calling, with Jacob Grimm, the earlier forms which give evidence of youthful creative force, strong, those which have been re- duced to uniformity weak. For the syllable κα in the perfect we may, with our present way of regarding it, adduce three sets of analogies ; viz. 1) Verbal, from Greek itself, i.e. other verbal forms in which we may recognise a suffix with κ. To these belong a) In the first place the three isolated aorists, which are always • I find myself here in agreement with Brngman * Zur Geschichte der prasens- bildenden Suffixe' in the * Sprachw. Abhandlungen ' p. 155, without however being able to accept his further conclusions as to the extension of such stem- formations. ch. xvi. THE PERFECT WITH κ. *Τ> <£\ X η /v > <£ classed with it, 'έδωκα, εθηκα, ηκα. We must -come back afterwards tovV these forms, which evidently follow the sigmatic aoiists as regards their ' ^j inflexion, because they can only be understood in connexion with other unsigmatic formations, in which o, so to speak, plays the part of the thematic vowel. r β) A small number of present forms, where the κ is evidently of the /•> nature of a suffix. This is most plainly seen in όλέκω. The inflexion of this verb is based upon the two-fold verbal stem όλ and όλε and the two-fold present stem όλλν (from όλνν) and ολεκ%• The stem-form with κ is in Homer limited to the present stem : όλέκουσιν Σ 172, πεζοί μεν πεζούς ολεκον Λ 150, όλέκονται Π 17, όλέκοντο Α 10; it also occurs a few times in the tragedians (όλέκεις Antig. 1285, ώλεκόμαν Trach. 1012) always in melic passages. In Herodotus first (i. -45) we come across the perfect άπολώλεκα, which thenceforward takes over the transi- tive function as compared with άπολωλα. Evidently the longer perfect is related to the corresponding present much as πέπληθα is to πληθω. On the Cyprian inscription of Idalium there is (1. 16) the form to-ko-i-je, which Deecke and Siegismund{Stud.vii. 243) reproduce by οωκοί η (Ahrens, probably more correctly, by δώκοιε Philol. xxxv. 68), and refer to a present 207 *ύωκω, which is to ΰέοωκα as όλέκω is to όλώλεκα. Even if the Cyprians had by the side of this optative only the indicative of a past tense *εΰωκον or even the common έδωκα, the form would• be of importance for the formation here under discussion. For mood-forms hardly ever came from aorists with κ. 1 By the side of 'ίληθι and Ίλάσκομαι we find 365 the conjunctive Ιληκησι : ει κεν 'Απόλλων ημΊν ιλήκφσι και αθάνατοι θεοί άλλοι. The meaning is as little that of a perfect, as that of the optative in the Homeric hymn, to Apoll. Del. 165 tXfaot μεν Απόλλων Άρτέμιδι ζ,νν. We are therefore justified in assuming a present Ιληκω, the κ of which does not extend beyond the stem of this tense. The case is different with ερνκω, which was thoroughly discussed on p. 292. Here the aorist έρύκακον shows the same guttural, and this appears also in ερνζω, ερυζα. The κ has therefore more of the character of an expansion of the root. For no one will doubt that ερνκω is an ex- tension of the stem ερν protect. — The same judgment is to be passed upon βρΰκω in its relation to the rt. βορ (βιβρώσκω cp. Princ. i. 78, ii. 80) and upon διώκω as related to ΰίο-μαι, δίε-μαι (ib. ii. 309). — Another group is formed by the presents ϋει-Ιίσσομαι (cp. above p. 221) and πτήσσω, which are evidently based on the stems δικ and 7rr«i.•. Both occur also in forms belonging to other tenses : ϊειδίζεσθαι, επτάκον. To these stems -Jw y the sign of the I-class, is added as a present expansion. These presents are therefore quite parallel to Latin presents like fa-c-io, and ία -c-io. Those who agree with me in referring fa-c-io to the rt./a=Gr. θε, can regard this form as simply the present to ε-θηκα, which we might transfer into hypothetical Greek in the form *θησσω. On the other hand we can hardly say now whether ΰεί-δοικα is a strong perfect from the rt. δικ or a 208 weak one of the rt. dt. We can here see clearly how all these cases hang 1 The relation of έδωκα, δέδωκα to the Skt. daq worship, offer (addqat, dadaqa) need not be decided here. — I may mention also that Bergk Jen. Litteraturz. 1875 No. 26 explains the sign, which Deecke and Siegismund take iorje, quite differently, viz. as pti, and hence instead of δώκοπ reads the words as δοκο7 ψί (~σ<ρίν). With such difficult texts it is impossible to arrive at complete certainty except by repeated examinations. 412 PEEFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. closely together. Possibly the Homeric εγρησσω also comes from a stem έγρηκ or εγρηκο, which is not far removed from the perfect εγηγερκα, though certainly εγρηγόρθασι might suggest another explanation; and the perfect with κ is not quoted from any author earlier than Josephus. — In Apollon. Rhod. we find υποίρησσων (cp. Homer, υποΰρηστήρ ο 330). Ηρη-κ -ju) would be related to ΰέΰράκα much as fac-i-o to the Osc. conj. fofaci-d. y) As thematic aorists with the suffix κο, we recognised p. 282 ημπΧακον, on which I did not venture to say anything positive, because of its etymological difficulties, and ε-πτα-κυ-ν. The latter is related to the primitive κατα-πτη-την (θ 136) and πε-πτη-ως (ξ 354) precisely as κέκμηκα to κεκμηώς or as βέβληκα to ί,νμβλητην (above p. 132). They of Ιι-έ- τ μάγε- ν appeared to us also of the same nature as this k, so that we have again the parallel : τέ-τμη-κε : τμη-σι-ς '.*. ε-πτακο-ν : πτη-την. l•) There are also some isolated forms, in which the κ is evidently a later accretion, but not preserved in its purity, viz. άλΰΐω by the side of άλνσκω and άλύω (cp. p. 194). A veiy conceivable aorist *ήλνκ<>ν or a present *άλνκω is alike unknown : it is only in sigmatic forms that the guttural comes out. It is difficult to determine whether the κ of the iteratives and of the inchoatives undoubtedly akin to them (cp. p. 187) which is always preceded by σ, is connected with the κ here discussed. 2) Nominal formations. In all kindred languages nouns, which are derived immediately from roots or verbal stems by means of the suffix ka, are rare. Greek forms of the kind are found in σω-κό-ς, γλαυ-κό -c, θή-κη (cp. Skt. dha-ka-s re- ceptacle) : Latin ones like prae-fi-ca (mourning woman, from rt. fa,fari), medi-cu-s are discussed by Corssen Aussprache ii. 2 306 ff. : a few others are quoted by Schleicher Comp. 3 461, with the Ch.-Sl. plu-hu nation from the same root as ])o-pul-u-s and πολ-ν-ς, πλή-θ-ος. 3) Cognate verbal forms in other languages. Verbal forms of the nature of this perfect evidently belong to the 209 youth of the various languages. It is possible indeed, as has been attempted here, to point the patterns of these forms, inherited from a still earlier period, which are as it were their starting-points, but we can hardly trace any adequate reason for the fact that out of many stem- formations this particular one became the more common in this place and for this especial purpose : we can at most detect certain points of connexion in the usage of a suffix. In the case of the thematic aorists we discovered the tendency to a stem-formation with r, which developed in the Keltic languages, and perhaps also in Oscan into a definitely marked preterite. We cannot be surprised then if within the circle of the more familiar and more thoroughly investigated Indo-Germanic languages we find no perfect with a h. There is however a trace of this in Oscan, though unfortunately only in the one word \ιοκακεπ=ΐΜί. (col)locavit, on the inscription from Anzi (no. xxxv. in Enderis). Cp. Corssen Ztschr. xviii. 210. Corssen, on the ground of the complete isolation of such perfects on Italian soil, formerly would not admit that the κ of this form had anything whatever in common with that of the Greek perfect. But after discovering, as he thought, a considerable number of Etruscan perfects (3 sing.) in -ce e.g. tur-ce said to be=r£- τόρινκε, lupu-ce, which according to him would correspond to a γέγλνφε (in form pretty nearly *γεγλύ<ΐ>ευ^) } te-ce answering in the same way to ch. xvi. THE PERFECT WITH κ. 41 3 τέ-θει-κε, he altered his view (Die Sprache der Etrusker i. 754), separating the Greek perfect from the Italian — without sufficient reason, as we saw on p. 392 — because of the vocalism, but explaining the k as something common to both. ' As the Italian character of Etruscan does not seem to me to be made out, in spite of Corssen's admirable carefulness in working at the remains of the language, and as the interpretation of in- dividual points is still far from certain, 2 I mention these Etruscan forms only with reserve. An enumeration of all the perfects in ra would have hardly any mean- ing in face of the later wide extension of this form. I content myself ■*" therefore with quoting the Homeric forms and making a few brief remarks upon the occurrence of this formation elsewhere. In the Iliad and £he Odyssey we find the following 20 perfects with ν : άΰηκότες Κ 98, άΰηκότα,ς μ 281. βεβηκας Ο 90, άμψιβέβηκεν Ζ 355, plupf. βεβηκει Α 221 etc. βεβίηκε Κ 145, Π 22. βεβλήκοι θ 270, plupf. βεβλήκει Δ 108 etc. βεβρωκώς Χ 94. ΰεΐάηκε 6 134, ΰεΰαηκότες β 61. ΰεΰειπνηκει ρ 359. ΰεϊνκε Ε 811 etc. ΰείΰοικα Α 555 etc. νπ-εμνημνκε Χ 491. εστηκας Ε 485, εστηκε Γ 231 etc., εστήκει Δ 329 etc. κέκμηκας Ζ 262. μέμβλωκε ρ 190, παρμέμβΧωκε Δ 11. μέμνκεν (from μΰω) Ω 420. πεψνκττ) Δ 483, πεψύκάσι cp. above p. 385 πεψύκει Δ 109 etc. τεθαρσηκασι Ι 420, 687. τέβνηκε Σ 12 etc. τέτληκας Α 228, τέτληκε γ 347. τετυχηκώς Ρ 748. παρ-ωχωκε (Aristarch., M.SS. παρώ\ηκε cp. La Roche) Κ 252. The dubious σνν-οχωκότε Β 218 was discussed above p. 369. Yowel stems of every kind have produced perfects in κ in such abundance in the post- Homeric time that it is completely superfluous to quote particular instances. Yery many might be adduced even from Herodotus, the earlier tragedians and Thucydides. We may say indeed that the Attic writers could form a perfect in ira from any vowel stem. From consonantal stems on the other hand the most numerous perfects in the Attic writers belong to present forms in -ζω, like γεγύμνακα, ήρπακα, ήτίμακα, εγκεκωμίακα, εζήτακα, τεθανμακα, κεκωμακα, ώνόμακα, εσποΰΐακα, πεφρακα (from φράζω Isocr.), — ε'ίθικα, τεθέσπικα, εψήφικϋ, ήρέθικα, κεκόμικα, 211 νενόμικα, ώνείΰικα, πεπόρικα, ττεφρόντικα, νβρικα, — πέπαικα, all good Attic. γεγνμνάκασιν, προντεθεσπίκει occur as early as Aeschylus (Prom. 586, 211), κεκομικώς in Herodotus (ix. 115). Most of the examples are furnished by Plato, X'enophon, and especially the Orators. Rott in the Ztschr. f. Alterth. 1853 p. 183 has justly observed that the suppression of the consonant of the stem in the future and sigmatic aorist favoured the 8 Cuno in his Etruscan Studies (Fleckeisen's Jahrb. 1874 p. 3.13) translates turce by dedicat. Eh'a Lattes (Memorie dell' Institute» Lombardo Vol. xi.) takes turce as a proper name Turciiis. 414 PEEFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. formation of such perfects. For if in earlier times forms like ΰεδείτνηχύ by ΰειπνησω and εοείπνησα, Χελυκα by λύσω, έλυσα were in use, it was natural to form a ιιρπακα for άρπάσω, ηρπασα. It was just as easy for πέπεικα (Orators) to be added to πείσω, έπεισα. The instances formed from other stems in the Attic period are not very numerous, κέ-κρϊ-κα and τέ-τά-κα are hardly to be regarded as consonantal, seeing that the stem without ν was in use also in κρΐ-τή-ς, κρί-σι-ς, τα-τό-ς, τά-σι-ς. πέφαγχα is quoted several times from Dinarchus. It is only from Aristotle onwards and in later comic writers that these forms with a care- fully preserved nasal like αιτέκταγκα became more common. From stems in ρ ήρκα, €φλ.•α (βιειμκότες Xen.), σεσυρκα, εώθαρκα are the only ones which are good Attic, of those in λ : ήγγεΧκα, t στάΧκα. We can plainly per- ceive here the gradual extension of the formation. • Our statement of the case hitherto has made the perfect with κ appear as a pre-eminently Attic form. But it would be an incorrect notion not to consider this formation as existing in the other dialects. Intimations to this effect have already been given as regards Ionic. We may further quote as Doric ευράκοιμεν with a noteworthy α in the Cretan inscription edited by Bergman Berl. I860 1. 71, κεκρικαν from the inscription of the Amphictyones 1. 58, and the Cretan απέσταΧκαν (C. I. no. 3058 1. 4), both already mentioned for their termination on p. 385, and also the Heraclean infinitives (cp. p. 390) πεψντενκήμεν, πρωγγνενκήμεν. Common forms like βέβακα, ΰέΰωκα etc. need no quotations. Wescher publishes in the Bulletin de l'ecole d'Athenes no. 6 an inscription from Mantinea with the strange form άνάκεικε—ά,νατέθεικε (1. 26). άκονκα is recorded Anecd. Oxon. iv. 188, 14 as Dorian, and also ά^εωκα perf. of άφίημι (Suid. s. v., Herodian ii. 236, 2). As here the stem ε, not unlike the Attic 212 πέπτωκα, πτώσις, πτώμα and the forms εόλωΐ', εάΧωκα, has passed into the derived o-conjugation, so has the stem εθ in εθωκατι" εϊώθασι Hesych. where the termination shows that the word is Doric. The Lesbian form for this is ενέθωκεν εΊωθεν, along with ενάΧωκε ν—εάΧωκεν (Herodian ii. 640, 10) with the Lesbian ν for / and the disappearance of this spirant at the beginning. An odd formation presents itself in the Pindaric infinitive γεγάκεινζ=γεγοι>ένάι 01. vi. 49, which is related to γεγάμεν 01. ix. 110 much as ΰεδοικέναι to the Homeric ΰεώίμεν. The strange κέκοκεν εγνωκεν I have attempted to explain in Stud. vii. 392 as the perfect to the aor. εκομεν ε'ΊΙομεν This too does not look quite Attic. Among the Attic forms two, viz. /oca and τέθεικα, show a surprising diphthong. Vowel stems have elsewhere regularly the long vowel before κ \ some instead of this a short vowel like λέλϋκα ; some fluctuate, like ΰέ-ΰε-κα and ΰέ-οη-κα. We must treat these relations of the vowels collectively in a subsequent chapter, for the same phenomenon reappears in various other tense-stems. The diphthong is in any case unusual. As regards εικα, which occurs frequently from Sophocles onwards along with the middle είμαι, the ει is here justified by reduplication. We saw above p. 364 that είκα probably came from ]ε-$ε-κα. It is therefore formed like Ζε-Ζε-κα ; and even if we should be wrong in the etymology of the word, in any case a spirant has dropped out between the two ε'β, and these afterwards coalesced into ει. The diphthong is therefore here completely intelligible. It is otherwise with τέθεικα* It is noteworthy • According to Herodian ii. 837 the form arose τροπρ Βοιωτικί) &airep ^pwts ffyitfCS. • ch. xvi. THE PERFECT WITH κ. 415 that we have preserved to us at least one trade of the more regular τέθε-κα : ανατεθέκαντι Ross. Inscr. I. no. 81, 1. 10; but τέθεικα and the middle τεθε'ισθαι are current in the Attic writers from Euripides onwards. It cannot be denied that the earlier η has in many cases given place to ει in later Attic : βασιλείς for βασιλής, rjhiv for gfiij, λύει as 2 sing, mid. for Xvy. But it required some other attraction, I think, to cause this one perfect to separate itself from the many hundreds of perfects in -ηκα. For in fact analogies which break through the ordinary 213 rule are like heavenly bodies which by their proximity divert the course of others. Are we to suppose that έικα was here the disturbing force 1 If we remember that the extremely common verbs ϊημι and τίβημι con- stitute a pair of twins, which resemble each other on almost every point — just consider forms like ίει ετίθει, οίμην θοίμην, ηκα εθηκα, ησω θησω^ εθήναι τεθήναι — this explanation does not seem to me improbable. And I cannot suggest any other. The fact that η and ει never differed very much in the pronunciation of the Attic dialect has of course also to be taken into account. Finally some forms have still to be discussed in which peculiarities of the strong perfect pass over into the formation of the form with r. We noticed above p. 397 that the conditions of the vowel in the older perfect e.g. in ΰέΰοικα are not absolutely unknown to the perfect with κ. In view of the preference of this tense for the o, the above-mentioned ω of πέπτωκα, and similar perfects by the side of πεπτηώς might even find* its reason in the analogy of the strong forms. It is therefore natural to explain thus the ο of the surprising form έΰήΰο-κα. This is quoted from Aristophanes, Xenophon and Aeschines. It has a precursor in Homer. In χ 56 we have in our M.SS οσσα τοι εκπεποται και έΰήΰοτηι εν μεγά- ροισιν, which has generally been adopted. But from Et. M. p. 316, 35 it appears that Aristarchus read εΰηϊαναι, Serodian εδηύεται. The plural would be strange by the side of εκπεποται. On the other hand εδήόεται is a quite regular formation from the expanded stem έ£ε, which is related to the Attic kh'jh -σ-ται as όμώμοται is to υμωμοσται, and it is highly ' probable that itylerui stood originally in the text, because the change from ε to ο is unknown to the perfect middle. The active to εΰηίεται might have been *εΖηΖεκα (cf. εμημεκα). But here the very same ο made its way in, which we meet also in the form without κ discussed on p. 403, iWdfa (C. I. Gr. no. 15). The *εΰή?οα which we must assume is related to ίϊηνοκα as άκηκοα is to the Doric άκουκα (Ahrens Dor. 337). — We have further the strangest of all perfect forms άγήοχα, a tolerably late formation, which occurs first in the spurious letter of Philip Demosth. 214 xviii. 39, then in Aristot. Oec. 2, 2, and more frequently in Polybius :and later prose-writers, in the place of the regular ήχα (cp. p. 406) for which there is plenty of evidence from the best Attic period. Important assistance is given for the explanation of it by the Dorian συν-αγάγοχα in the inscription from Thera C. I. Gr. no. 2448, i. 28, ii. 10, iii. 13 [Cauer Delect, p. 77 ff.] along with the άγήγοχε of an inscription from Sigeum (Buttmann Lexil. i. 297). The Doric άγ-άγ-ο-χα shows un- mistakeably three elements, the Attic reduplication, the vowel o, the termination χα. The reduplication presents no difficulty; the vowel ο is evidently of the same nature as in εΰήοΌκα, and hence the χ cannot be based upon the aspiration of the rt. ay, for in that case it would be, so to speak, retriplicated instead of reduplicated (άγ-άγ-οχ for άγ-άγ-αχ). 416 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. Evidently the χ is only a phonetic transformation of the κ. Or in other words : άγ-άγο-χα belongs to the perfect in m ; it is for *άγ-αγο-κα, which would be formed exactly like έΰ-ήΰο-κα. But the κ has become χ by a phonetic affection. This view is confirmed by the form ίεριτευχε in the inscription from Mantinea published by Wescher Bulletin de l'ecole d'Athenes no. 6, 1. 6, to which we may also add βεβληχνϊα (C. I, Gr. 2360 1. 7 ; cp. Stud. vii. 393) though this rests on somewhat doubtful authority. As the two latter forms admit of no other explanation what- ever, this remains probable also for άγήγοχα. In the later άγηοχα (Boeot. άγείοχα according to Et. M. 9, 34) there is one more irregularity, the suppression of the γ, of which this is probably the only instance in the Attic dialect. The Boeotians offer the parallel Ιών or ϊώί>=εγώ, the Tarentines ό\ίος=όλίγος (cp. Princ. ii. 247). I have given further reasons for believing that reduplicated forms hold a place of their own in regard to phonetic laws in my essay * Ueber die Tragweite der Laut- gesetze' Berichte der Sachs. Ges. d. Wiss. 1870 p. 16 ff. Here two gutturals were felt to be enough. III. THE MIDDLE PERFECT. Compared with the active perfect, the perfect of the middle is a simple verbal form. A single very archaic method of formation has 215 been carried out here quite independently, and as we saw on p. 354, in contrast to the active. We could only quote a few instances in the active where the perfect stem is immediately united to the personal endings ; in the middle this formation has become an inviolable rule. Forms like ΰέΰο-ται, λελν-νται, πεπυσ-ται .βέβακ-ται τέτνζαι ιένητται are the most primitive which could be expected from such stems. They are quite parallel to Yedic forms like άα-άΙιί-ά/ιν6=τέ-6ε-σβε, ri-rik-she= λέ-λειπ-σαι, bu-bhu(j-mahe (from bhug enjoy, cp. Homer, πεψυγ μένος). Perhaps the reason why the Greek language retained to all time in the middle perfect the primitive method of formation, which in the present and aorist occurs only to a very limited extent, and often only as an archaic rarity (e.g. ησται, ΰέκτο, α ν σο), is to be sought in the fact that the stem strengthened by the syllable of reduplication gave as it were more hold, and had volume enough to remain recognisable in spite of many mutilations of the final letter, unavoidable where consonants were brought into contact. For this very reason the middle perfect is quite specially the place for consonantal changes which are based upon assimilation. The faithful preservation of the personal endings gave to the stem of the middle perfect a mobility which is elsewhere unknown in the Greek verb, without any loss of clearness. While the active perfect, as we have seen, developed in different directions only during the course of the history of the Greek language, the middle is already in full use in the time of Homer, and remains essentially unchanged for all times, with the exception of the third person plural, which by degrees dies out in the case of consonantal stems. The traces of a thematic treatment of the perfect stem are extremely scanty in the middle, α as forming a stem occurs, with the exception of άρηράμενος, to be mentioned immediately, probably only in the quite isolated forms εκγεγάασΟε Epig. Horn. 16, 3, εκγεγάαντο (οι παρ θεού ch. χνι. • THE MIDDLE PEKFECT. 417 εκγεγάαντο) Anth. P. xv. 40, 20, 4 and in the strange σεσυανται' ώρμή κασι in Hesychius, which evidently belongs to σεΰω, εσσνμαι. We must here 216 include the α in the stem, and treat ντο, νται as the personal endings. For -avrai as a personal termination by the side of -area is quite unknown. — One might detect a trace of ε in the forms άκ-ηχε-μένη Ε 364, άκ-ηχέ-μεναι Σ 29 ; but άκηχέδαται, to be discussed hereafter, and the cognate forms άκαχήμενος, άκαχήσθαι show that the vowel is in a different position. With more justice άρηρε-μένος may be quoted here, but it occurs only in Apollon. Rhod. iii. 833 and Quint. Smyrn. ii. 265, iii. 632, has in its place άρηραμένος, both evidently isolated ventures of learned poets. Another completely isolated form is ΰεΰάασθαι which occurs once π 316 as the middle infinitive to the active ΰέΰαα. Nor can we deny a vocalic expansion of the stem in εΰήΰοται discussed on p. 415, while the Hero- dotean αν-έω-νται (ii. 165 =ζάνεϊνται) with the Heraolean άνεώσθαι, and the άψεωμαι quoted by Herodian (Meister Stud. iv. 433) maybe grouped with the vocalic expansions mentioned there, which are perhaps limited to the perfect only by chance [cp. p. 273 note, and for the Ν. T. άφέωνται Moulton's Winer 2 p. 96 f.]. — Some few forms show traces of the thematic vowel : thus μέμβλε-ται Τ 363, with its past μέμβλε-το Φ 516. We shall probably best take these forms as originating by metathesis in με-μελ-ται, με-μελ-το. Buttmann Ausf. Gr. ii. 2 243 regards the ε here as the short vowel of μεμέλητο quoted from Theocritus, and certainly the ε was originally attached to the root, but Lobeck adduces from Oppian Halieut. iv. 77 the form μέμβλονται, so that here the thematic nature of the vowel is unmiatakeable. — The ε of the Homeric όρωρεται, only occurring twice (r 377, 524), is undeniably thematic : it is indirectly confirmed by the conjunctive όρώρηται Ν 271. (Cp. Buttmann Ausf. Gr. ii. 2 56.) — We may further recognise an indubitable transition into the thematic conjugation in άναγέγράφονται in Archimedes (Ahr. Dor. 333), supposing the form, by the side of which the isolated άναγε- γράφαται is also recorded, deserves our confidence. — The same process may be seen, carried out extensively, in οίχομαι, if this verb with its perfect signification is to be justly regarded as a petrified perfect of είκω (rt. fac), Princ. ii. 115. With regard to the personal terminations we must notice the careful preservation of the suffix for the 1 and 3 pers. sing, as compared with 217 the mutilated e of the Sanskrit, which is used for both. The endings -σαι and -σο too of the 2 sing, are carefully preserved in the perfect stem, as everywhere in forms of primitive stamp, a fact established already on p. 59 f. ; hence άΧάλησαι ο 10, ΰεΰάκρυσαι Π 7, imper. άλάλησο γ 313, plupf. πέπννσο \p 210, κεχόλωσο Π 585. There are exceptions in μέμνηαι Φ 442, by the side of μεμνησαι Ψ 648, βέβληαι. Ε 284, A 380, Ν 251 and εσσνο Π 585, ι 447, the latter form (cp. p. 130) perhaps to be taken as an aorist. As the only traces of a contraction show themselves in Homer in μεμν -n Ο 18, Υ 188, Φ 396, w 115 and in Herodotus in the imper. μεμνεο (δέσποτα, μέμνίο των 'Αθηναίων V. 105), Lobeck's view (on Butt- mann ii. 2 244) is very probable, that here we must assume with Herodian a present form μεμνομαι, which is parallel with μέμβλεται and connected 4 The thematic perfect eKyeyaovrai used as a future (Hymn, in Ven. 197) is too strange not to be doubted. There seems to me much probability in Baumei- ster's conjecture koI παΐδ€5 παίζεσαι 9tafnrcpis iKyeyaovTes. Ε Ε 418 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi, with the proper names Μέμνων, 'Αγαμέμνων. — The loss of the σ in the press of consonants in the case of forms like πεφύλαχθε, λέλειψθε, άνήφθω hardly needs further mention. •The endings -αται and -aro for the 3 plur. have already been thoroughly discussed on p. 64 f., so we need not return to them here. There are only two kinds of formations which deserve to be mentioned, those with δ before the q, and those which are aspirated. The δ presents no dif- ficulty in ερηρέ carat (Ψ 284, 329) and is probably the consonant of the root also in ερράΐατ Μ 431 (Princ. ii. 293); on the other hand in the Herodotean άγωνίϊαται ix. 26, έσκενάΰαται iv. 58, εσκενάΐατο vii. 62, παρεσκευάΰατο V. 65 etc. κεχωρίΐαται i. 140, 151, ii. 91 it must un- doubtedly be the same dental element, which is present e.g. in κομιδή and occurs in the present in -ζω as a component part of the double con- sonant ζ. On this I based the combination discussed on p. 242, accord- ing to which the δ in the Homeric άκηχέΰατ (Ρ 637), εληλέΰατ (η 86) is to be taken in the same way, and therefore bears witness to a formation which would end in -έζω in the present, while άκηχε μένος goes back to the same derived stem, though in this case it does not sKow this c\ We should have to notice also απεσπάΰαντο' άπέστησαν Hesych. if the gloss had to be read απεσπάΐατο. But the alphabetical order does not agree with this, so that the tradition must be regarded as quite uncertain.— 218 Attention was called on p. 405, where we were treating of the active aspirated perfect, to the aspiration which appears in a number of such perfect forms before the a.- The forms of the perfect and pluperfect which belong here are as follows : ΰειΰέχαται η 72, Ζειΰέχατ(ο) Δ 4 (cp. I 671, Χ 435) ; plur. to ΰεί- ΰεκτ(ο) I 224 with the present ΰεώίσκομαι (rt. Blk, δεικ). ειλίχατο Herod, vii. 90 from ελίσσω. ερχαται κ 283, έέρχατο κ 241, ερχατο £ 73 from έίργω. έσεσάχατο Herod, vii. 62 from σάττω. Ιψαμεμίχαται Herod, i. 146. όρωρέχαται Π 834, όρωρέχατο Λ 26. πτάχαται Thuc. iii. 13,'Xen. Anab. iv. 8, 5 (άντιτετάχαται), ετετάχατο Herod, vi. 113, Thuc. v. 6, vii. 4. The evidence of the inscription from Methone was mentioned above p. 66. τετράφαται (έπι) Β 25, Theogn. 42, Plat. Rep. vii. 533, τετράώαθ' (plupf.) Κ 189. τετρίψαται Herod, ii. 93. Η we put these nine examples along with the other instances of the employment of the endings -arot, -aro, it results that only the following sounds appear before them 1) a vowel: βεβΚήαται, 2) a 2) a liquid : άγηγέραθ' (plupf.), έστάλατο Hes. Sc. 288, έφθάραται. 3) ί : έσκευάΐατο etc. 4) the aspirates χ and 0, which in γεγράψαται, κεκρνφαται Hes. Opp. 386, τετεύχαται Ν 22, έπώχατο (only Μ 340), which following Buttmann Ausf. Gr. ii. 2 189 we derive from επέχω, belong to the root, in all other cases have arisen from the corresponding hard or soft non-aspirated explosive. The only exception probably is άπίκατο Herod, viii. 6, for which we should have expected *απίχατο. We may mention as a remarkable in- stance of a form in -vrai faithfully preserved (cp. πέπαννται) the Doric &ποκέκ\φνται, as Ahrens (Dor. 346) justly restores the words in Epi- charmus, from «cXp'wssAtt. κλρ'ω, κλείω. ch. χνι. • THE MIDDLE PERFECT. 415 As to the vocalism of the stem-syllable, the middle perfect here 219 deviates not inconsiderably from the active. The intensification of α to η is here by no means firmly established. By the side of the Attic forms ΰεΰηγμένος, πέπληγμαι, είλημμαι, λελήσθαι (Soph, El. 342) we find the Herodotean ΰιαλελαμμένος (in. 117), the Homeric λέλασται (Ε 834). Over against the active πέφηνα are πέφασμαι, πέφανται* How little the middle perfect is disinclined to the α of the root is shown by βέβαμμαι, βέβλαπται, γέγραμμιχι, κεκασσαι (τ 82), μέμακται, πέπαλται, εστακται etc. — The intensification of ι to οι, though characteristic of the active perfect, is unknown in the middle. Instead of this we have sometimes the ει which is common in the present stem : ΐέΐειγμαι (from Sophocles onwards), for which Herodotus has ΰέΰεγμαι (άπεΰέΐεκτο iii. 88), ήρήρειστο (Δ 136), έρηρεισμένος (Herod, iv. 152) beside ορισμένος (Hesych.), λέλειπται. from Homer onwards (N 256), πέπεισμαι from Aeschylus onwards, while to the active tettxt coiTesponds the middle προςίμίαι. (Eimp. Ale. 1063), ηικτο δ 796 and εαζτο Ψ 107, to the active olca the middle, 'ίΰμαι• γινώσκω, oUa Hesych. In ητενχαται (Ν 22) ν becomes ευ, but it remains un- raised in τέτνϊ,αι (Π 622), τετνκται, τέτυκτο, τετνχθαι, τετνγ μένος. From ζενγννμι in spite of ζυγόν, εζνγην we find only έζενγμένος, and form like εζενκται, εζενχθαι, εζενκτό in Herodotus and Attic writers; on the other hand in spite of πευθομαι and πενσομαι there are only τζέ-κναμαι from Homer onwards (λ 505), πέπνστο Ν 521, and in the same way in spite of φενγω, πέφενγα there is πεψν/μενος (Ζ 488). — We find no trace of the change of ε into o, of which there are such numerous instances in the active, with the exception of Ihrjcorat mentioned on p. 415, but pro- bably wrongly recorded; 5 there is one certain instance of ω answering to an ε in άωρτο (Γ 272). The regular ήερμένος is quoted first from Apollon. Bhod. ii. 171. But here undoubtedly the ο is connected with the/ which was once found before the ε (Princ. i. 442, cp. Fick 3 216). We mentioned the entirely unique reading in Μ 340 ττάσαι (πΰλαι) γαρ επώχατο on ρ. 418. Buttmann to support his derivation compares νκωχα. . On the other hand we find in the middle perfect a preference for the vowel α answering to ε in the present, to which we detected only slight tendencies in the active. This preference is however limited to 220 the vowel preceding a liquid or a nasal. While in forms like βέβλεπται, κέκλεπται, πεπεμμένος, πέπλεκται, εσκεμμένος the ε of the present stem appears also in the perfect, it becomes α in ΰεΰαρμένος Herod, vii. 70, Aristoph., ΰεΰάρθαι (Solon fr. 33, 7 Be. 3 ), είμαρται from Homer onwards (ε'ίμαρτο Φ 281), εσπαρται in Herodotus and Attic writers, έστάλατο (Hesiod Sc. 288), εσταλμαι from Herodotus onwards, εστραμμαι first in Hymn, in Merc. 411, then in Herodotus and Attic writers, εφθαρμαι from Aeschylus onwards, έφθάρατο quoted above from Herodotus, πε- παρμένος first in Λ 633, έτέταλτο Β 643, έπιτέταλται and the like from Sophocles onwards, τέταται with loss of the ν preserved in πέψανται, from Homer onwards (λ 19), τέθραμμαι from Euripides onwards, τέ- τραμμαι from Homer onwards (τετραμμένος Ρ 227). The α in many of these verbs extends to the verbal adjectives and passive aorists, in some also to the active aorist, the preference of which for α was mentioned on p. 278, in several to the perfect with κ : εψθαρκα, τέτακα. 6 Cobet Mnemos. Nova series ii. 368 also holds that cS^tcu is the right reading. κ ε 2 420 PERFECT STEM AND FOEMS CONSTRUCTED FROM IT, ch. xvi. Finally we have still to discuss the changes which are brought about by the contact of the consonants with each other. The changes, based upon general phonetic laws, before the terminations beginning with τ and θ (σθ), such as occur in λέλεκται, πέπυσται, γέγραπται, τετράφθω (Μ 273), άψΐχθαι need no discussion. 6 With regard to the 2 sing, we may men- tion that the oldest form arising from assimilation is represented in Homer by κέκασσαι r 82 and πέπυσσαιλ 494. For the latter Plato Protag.. 310 has πέπυσαι as a result of the frequent reduction of a double σ to a single one. Somewhat more difficulty is presented by the terminatipns begiiming with μ. Complete consistency prevails only so far, that every labial is assimilated to the μ, thus even in Homer in κεκαλυμμένος Π 360 etc., λελειμμέιος ι 448, τετοαμμενος Ρ 227 etc. On the other hand a dental explosive remains in κεκορυβ μένος Γ 18 etc. and in Pindar's κεκαΰμένος (01. i. 27) by the side of the Homeric κεκασμένος Δ 339, 221 κεκάσμεθα ω 509, κεκασμένος δ 725. The κεκορυθμένος is contrasted especially sharply with the equally Homeric λελασμέθα Λ 313 and πέ- πνσμαι λ 505, to which are subsequently added πέπεισμαι, έωσμαι. We cannot talk of a phonetic law requiring the change into σ in the face of Homeric forms like όΖμή, οΐΰμα, έφετμή, στάθμη, σταθμός, αριθμός, γναθμός. The process is rather to be explained thus : the sibilant, which is more convenient before the μ, and which made its way also in the Att. οσμή and in the New Ion. ρνσμός=ρνθμός, by degrees became the rule with dental stems in the common verbal forms, which are especially im- patient of any harshness, while on the other hand the unique κεκορυθμένος maintained its ground only as an archaism under the protection of Epic formulae. — The treatment of the gutturals is to be regarded in the same way. There are many instances from noun-formation of the older method of treatment according to which the hard explosive κ or χ is re- tained before μ, as ακμή, αιχμή, ΰοχμή, πΚοχμός, αυχμός, νεοχμός. Among verbal forms we find only άκαχμένος from the rt. άκ Μ 444, Κ 135, which as a ' petrified ' participle takes a position of its own. Elsewhere without exception every guttural is represented by y, whether the stem has y, as in έζευγμένος Σ 276, or κ as in δεΰηγμένος, ΰέΰειγμαι, κεκήρυγμαι } πεφυλαγ- μένος (Ψ 343), or χ as in ήργμαι, άφΊγμαι, ΰεΰεγμένος (Δ 107), τετάραγμαι, Πειραγμένος Ν 393, τετυγμένος (Π 225). Hence μεμορυχμένα can hardly be the correct reading in ν 435 ; we ought rather to write μεμορυγμένα with Apollonius in his Lexicon. The ' sonant ' sound of the y, as our physiologists call it, was more akin to the nasal, and hence became the general representative of the gutturals, as σ did of the dentals. The ten- dency of language is here always towards a simple rule for verbal forma- tion. The only point left open to variation was the treatment of a ν before μ and before , αν ro αντοϊ μεμισθωσώνται, .%■ άρτνσωντι ΐ] αποΐώνται ταν επικαρπίαν. 4) The inflexion of "ισάμι, carried out throughout among the Dorians,, which I have treated more fully Stud. i. 1, 240 if. The recorded forms are ϊσάμι Epich. fr. 98 Ahr. Pind. Pyth. iv. 247, Theocr. v. 119. 'ίσαις or 'ίσας Theocr. xiv. 34. . Wi C. I. no. 5773, Theocr. xv. 146. 'ίσάμεν Pind. Nem. vii. 14. 'ίσάτε Periander ap. Diog. Laert. i. 99. Ίσαντι Epich. fr. 26, Theocr. xv. 64. There is also 3 pi. conj. Ισάντι or ϊσαντι C. I. Gr. 5013 (οπα ων ίσάηι=ίΐ'α ειΰώσι, Chishull ισώντι). — Inf. ισάμεναι according to Hesyclu γισάμεναι' ειΰεναι, partic. 'ίσας Apollon. de adv. 587, 8, dat. 'ϊσαντι Pind. Pyth. iii. 29. The σ, united generally with the. vowel a, reminds us at once of two* kinds of formations, on the one hand of the 3 pers. pi. in -σαν, like ε-Ιο-σαν, ε-ή>α-σαν (cp. pp. 48, 50). This analogy comes out still more plainly in the Homeric plupf. 'ίσαν, the past of t σασί. But on the other hand the syllable -σα suggests the sigmatic aorist. After the views expressed on pp. 11, 12 as to compounded tense-formation, I have no doubt that we have an instance of this here.. These forms have the especial interest, that they reveal the connexion of the Latin with the Greek perfect. 232: The Latin perfect makes use of composition in three ways, viz. first, in all perfects to form the 3 plur. ; secondly, to form the conjunctive ; thirdly to form the whole perfect-stem, 'ίσασι i.e. ΐώ-σαντι and ειί,ασι are the analogies for the first application, fil -σαντι differs from vide-runt i.e. vide-sunt only inasmuch as the Greek stem preserves its primitive form, while the Latin has become disyllabic. The relation is just the same between viderint and the Heraclean μεμισθω-σώνται, while the Latin perfect in -si, e.g.' lii-si and the corresponding Old Irish, e.g. gabsi cepit,. ijhi-sa-m fecimus resemble the completely sigmatic 'ίσάμί. We shall return to 'ίσΰμι under the sigmatic aorist; in this word, as the force is completely that of a present, it may be • doubted whether it was ever 428 PEKFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvt. characterised by initial reduplication. In any case we have in dealing with these words to take into account the forms which Delbriiek de- scribes (Altind. Verb. p. 181) as * double stems.' VII. THE PLUPERFECT. The pluperfect, as the past tense of the perfect, seemed, until recently, to be a tense limited to Greek and Latin, in the possession of which these two languages had the advantage even over Sanskrit with all its wealth of forms. W. von Humboldt had indeed, as I pointed out in my * Tempora und Modi ' p. xiii and 230, recognised that Sanskrit was not entirely destitute of forms of a similar stamp, which lay concealed among the reduplicated aorists. But it was in Grassmann's Lexicon to the Rigveda and in Delbriick's Old-Indian Verb p. 122 that an Indian plu- perfect was for the first time definitely recognised. Still it is only among the Greeks and Romans that this tense has a wider extension, and thereby receives its definite place in the system of the verb. The past tense from the perfect stem has three kinds of general distinctive marks ; the augment as a sign of the past tense, the reduplication as a sign of the perfect stem, and finally the secondary personal endings, which are the only ones possible for any augmented tense. Of these •233 three distinctive marks the augment is in the pluperfect the most moveable. We cannot be surprised to find that so light an initial syl- lable did not cling very firmly to forms which had already so much to carry, and in which the past-force remained for the most part re- cognisable even without this syllable. Hence from Homer onwards we find in the most different kinds of literature pluperfect forms without any augment, like αποτέβνασαν, as La Roche writes μ 393 with most M.SS., άμφιβεβήκει θ 68, τέτυξο θ 163, πεπτωκεσαν Thuc. i. 89, πεπόν- θειμεν Plato Phaedr. 89, άποπεφενγη Plato Apol. 36 a, ΰιαβεβήκει Xen. Anab. vii. 3, 20, πεπόνθεσαν Dem. xviii. 213, όπωπεσαν Herod, vii. 125. It will suffice here to refer to the most recent investigation of this ques- tion by La Roche in the Ztschr. f. d. osterr. Gymn. 1874, p. 408 ff. We may just mention ειστηκειν (cp. above p. 86) which occurs first in Hesiod (Scut. 269), and then often in Attic poets (Ar. A v. 513) and prose writers (Thuc. i. 89), without the short forms εστατον^ εστάτην, εστασαν ever having the augment, and everywhere constantly interchanged with the unaugmented forms. Three methods of formation are in use : viz, first, the primitive, in which we have only the distinctive marks men- tioned above ; secondly, the thematic, in which the perfect stem is ex- panded by an added vowel ; thirdly the compound, in which we may clearly recognize the appended stem of the verb substantive. In the middle the first- formation alone prevails : in the active the first and second are represented only within a very narrow compass, while the third has become the rule. A) Active Pluperfect. 1) Primitive Formation. All the forms belonging here, like επεπιΰμεν,είκτην, 1<τμεν, '«'σαι', have already been quoted on p. 385 ff. along with those of the primitive perfect, so that we have no need to say anything more about them. ch. χτι. THE PLUPEKFECT, 429 2) Thematic Formation. Here belong, as the clearest representatives of the kind, the two Epic past tenses (3 plur.) εμέμηκον ι 439 with the participle μεμηκώς and έπέφυκον Hes. Theog. 152, Oppi 149, Scut. 76. The case is the same with ανωγον t 331, άνωγε ε 276, inasmuch as we have repeatedly learnt 234 to recognise άνωγα as a perfect. But here the perfect too is sometimes inflected like a present (άνώγει). The analogy of the imperfect is in these pluperfects just as unmistakeable as that of the present in many perfect forms quoted above. Such pluperfects correspond exactly to Sanskrit formations like e.g. a-JcaKaksha-m with the perfect JcaJcaksha (rt. Uahsh see). As in Sanskrit the border line between these formations and reduplicated aorists is doubtful, so with the Greek πέπληγον we cannot determine whether it is the pluperfect to πέπληγα or an aorist. The form was discussed more thoroughly on p. 290. But there are also some past tenses of the like kind which occur only in the third person singular. As with these there is no difference between the ordinary perfect and the past perfect, we cannot talk here of a transition into the analogy of the present. The distinction between perfect and pluperfect consists here exclusively in the augment. But as the latter may always be omitted in Epic poetry, to which all these forms belong, the two coincide in form, and it is only the connexion which decides. Here belong άν-ήνοθς-ν x\ 266 οφρα oi αίμ en θερμον αν. if• ώτειλής έπ-εν-ήνοθε Β 219 ψεδνή δ' επενήνοθε λάχνη δείδιε Σ 34 δείδιε γαρ μη Χαιμον άπαμησειε σώήρο) εγέγωνε S 469 γεγωνε Ω 703 κώκνσεν τ αρ* έπειτα, γεγωνε re πάν κατά άστυ with the present-like οσσον τε γέγωνε βοήσας e.g. ζ 294. Cp. Buttmann Ausf. Gr. ii. 2 35 f. It is plain that we have to deal here only with perfects which have the force of a present. One feels, with formations of this kind, how slight are the means which language employs for the most important distinctions, and can realize the dangers to which the system of verbal forms must have been exposed in languages which do not possess either the augment or the distinction between primary and secondary terminations. Perhaps the lack of perspicuity in the forms just dis- cussed contributed to favour the extension of the third pluperfect formation. 3) Compound Formation. 235 The Ionic dialect from Homer onwards has preserved for us a plu- perfect which is characterised by the addition of two vowels, pronounced separately, to the perfect stem. Fanatical friends of the theory of the 1 connecting vowel ' necessarily found this inconvenient, inasmuch as it looked as if language here, in its excessive luxuriance, employed two ' connecting vowels,' though one of these could not help being so completely in the way of the other, that it really turned into a ' hindering vowel.' We have here to do with forms like the following : 1 sing. 7/£εα & 71, τ/δε' ivt ψρεσί θ 366, ή'£εα Theogn. 853, Herod, ii. 150, Callimach. fr. 297 ed. 0. Schneider. 430 PERFECT STEM AND FOKMS CONSTRUCTED FROM IT. ch. xvi. ήνώγεα ι 44, κ 263, ρ 55. πεττοίθεα Ι !.434, θ 181. ετεθήπεα ζ 166. εώθεα Herod, iv. 127 (the only example in Herodotus). 2 sing, ετεθηπεας, an old variant of θηήσαο ω 90, while Hesychius quotes ϊ/^εες; rjfcig, with which compare Et. M. s. v. επεποιηκειν, επεποι- ηκεα (p. 386, 15 sqq.). The double ε savours strongly of grammatical theory. 3 sing. #£εε Β 409, 832, Λ 330, Ρ 402, in all which passages I. Bekker to suit his metrical theory writes gf&y. From Herodotus Bredow p. 320, 26 quotes similar forms, e.g. άραψήκεε iii. 39, έγεγό^εε i. 11 and often elsewhere, όπωπεε i. 68. 2 plur. συρρέατε Herod, ix. 58, the only example of the kind for this person. It is a priori probable that the two vowels did not originally stand side by side ; and it was very natural to conjecture the loss of a σ, as I did in Tempora und Modi p. 333 following Pott, after Bopp Vergl. Gr. § 645 had proposed an explanation similar in principle, but somewhat more complicated in the carrying out. And as in the same Ionic dialect we meet in the imperfect of• the rt. ες the forms ήα or εα,. έ'ας (Herod), ήε, εαΓε (Herod.), it was not possible to avoid identifying the terminations of these pluperfects with the imperfect forms of the verb substantive, especially as Lat. videra-m and *e-fe&ea, rjdea 23$ videra-s and *c-fei8eas videra-t and *£-pi8ee, fidee videra-tis and *e-}retSeare, rjSeare agree exactly, except in quantity, and in Latin the connexion of the termination -eram with eram was still more evident. In spite of - this there is still one difficulty remaining. It is a question whether the ε is a part of the appended auxiliary verb, or of the perfect stem ending in a vowel. Schleicher Comp. 3 812 prefers the former view. It see'ms to me that the second deserves the preference, if only on the ground of the 3 plur. For here, on the analogy of 'ί-σαν, ε-ψα-σαν, ε-Ιο-σαν, 'ί-σαν they knew = fid- σα ν, we must certainly divide yh -σαν so. Again we came across a perfect stem in ε in the infinitive εΐΰέ-ναι, in the conjunctive and optative ε*£ε'ω, είΰε-ίη-ν. Indeed we learned above on p. 390 ff. to recognise the perfect formation with an e sometimes short, sometimes long, as a Graeco-Latin form. The vocalism of πεποίθεα too suits only that of πέποιθα or πεποιθεναι, not that of πεπισμεν. Com- position of a stem with forms of the rt. ες met us above p. 427 in the case of 'ίσαμι ; and we shall have occasion under the head of the sigmatic aorist to discuss this principle of formation in its various connexions. Evidently ήΰέ-ατε i.e. *ε-ϊεΐ()ε-σα-τε is to ΐσατε i.e. *Λ£-σα-Γε precisely as a form *είΰε-μεν (which is very conceivable, on the analogy of the Doric πεπονβε-μεν) would be to the Horn. inf. 'ίΐ-μεν. The same element is in the one case added to an e-stem, in the other to the unexpanded root. The position of the 3 plur. is evidently quite special. The -σαν of $ΰε-σαν, ε-πεπόνβε-σαν never loses its a. Obviously on the ground of ήσαν, εσαν -σαν established itself in time as a termination of the third plural, and constantly extended itself more widely. Thus this CH. xyi. THE PLUPEEFECT. 431 personal form took an isolated position, and whilst in early times, when for the 3 pi. only ecov, εφάν and the like were in nse, ε-Ρεί^ε-σα, ε-ϊείΰε-σα-ς, ε-ΐείΐε-σε etc. down to the 3 plur. ε-Ρείΰε-σα-ν were quite on the same footing, afterwards the a in all the other forms fell before the universal tendency to transform it into a breathing, but in the one instance was protected by the analogy of the numerous forms belonging to other tense-forms, which had meanwhile become commonly current.. We have repeatedly seen above, in the case of the 2 sing, of the middle, that 237 the σ between two vowels was not always treated in the same way. It is worth noticing that the language of Homer, though it has quite a number of primary forms of the 3 plur. like βέβασαν, ΰείΰισαν, μέμασαν, ϊσαν, has only one single form from an e-stem εοίκε-σαν Ν 102. Now that we have learnt to know the methods of forming the pluperfect in their clearly marked types, the question is to which of • these categories that form belongs, which occurs even in Homer in considerable numbers, and afterwards came to be the only one in use. The forms here under consideration, which Cobet Novae Lectiones p. 212 ff. has submitted to an acute and comprehensive criticism are the following : a) Forms in η. 1 sing. — in the Old Attic dialect the 1 sing, according to the evidence of the grammarians and the best M.SS. ended in η. One main authority for this is that of Choeroboscus in Lentz's Herodian ii. 326 : ol ΰε 'Αθηναίοι* το διαλυθεί' υπό των Ιώνων σνναιρονσι και ποιοϋσιν εις η επεπ οιήκη λέγοντες όθεν και το ' εκεχήνη εγώ 1 παρ* Άριστοφάνει εν ^Α,χαρνενσιν (ν. 10) ore δί) 'κ^χηνη προσδοκων τον Αισχνλόν. The unhistoric conception of the old grammarians here meets us in all its strength, or rather in all its weakness. According to them -fa is an Ionic resolution of the usual -ειν, but then -εα itself is again contracted, into -η. But we perceive a correct insight in the fact that this 1 sing, of the pluperfect is compared with the 1 sing ή— ή v. We have the follow- ing instances in which first persons in η rest on good authority or are restored with certainty : Soph. O. R. 433, 0. C. 944, Antig. 18, El. 1018, Eurip. Hippol. 405, Heracl. 987, Aristoph. Av. 511 ί)ΰη, Aristoph. Eccl. 32 εγρηγόρη, 650 έπεπόνθη, Vesp. 800 ήκηκόη, in Plato, whose usage is treated by Stallbaum on Sympos. p. 198 c, frequently ΐβη, Apol." p. 31, άπωλώλη and ώφελήκη, p. 36 άποπεφενγη Theaet. 208 αψεστηκη, Sympos. 1. C. έπεπόνθη, p. 217 εγκεχειρήκη, Republ. 472 εΰεΰοίκη. Afterwards ειν alone is used. 2 sing, ήείΰης X 280 (Yen. A according to La Eoche ήείΰεις), ηΰης a 337 [Bekker : M.SS. oUa Q cp. Merry's note in loc], εζήΰης Soph. 238 Trach. 988 (1) [Cobet εξ^σθ'], ρϊησβα τ 93, Cobet Soph. Antig. 447, Dind. Aristoph. Nub. 330 ; έλελήθης is given by Dindorf after Brunck at Aristoph. Equ. 822, 1044. 3 sing, ήείΰη ι 206, ^η A 70, Β 38 etc•. La Roche, in spite of the authority of Aristarchus in favour of the η (Schol. on Ε 64) is inclined to write ει everywhere, and so too Cobet and Kontos in the Λόγιος 'Ερμής • Cp. Moeris p. 197 ed. Bekker ήδη Άττίκώί, fteiv Έλλτ^κά)*. 432 PEEFECT STEM AND FORMS CONSTRUCTED FROM IT. ch. xvi. p. 61. Certainly the variation between ει and η, even in the same verse Ν 355 πρότερος γεγόνει και πλείονα ijhi, is surprising. The only form which is established with absolute certainty is the Doric άπολώλη tab. Heracl. i. 39. In Theocritus there is the greatest fluctuation in the M.SS. between ει and η ; cp. Ziegler on x. 38, xiii. 40. In Attic writers- there is nowhere any support for η in the 3 sing. (cp. Gerth Stud. i. 2, 222). 1 plur. ήίΰημεν (probably to be written ήείΰημεν)' ήπιστάμεθα, φΰειμεν Hesych. εκεκρατηρίχημες Sophron fr. 71. b) Forms with e. In three passages of the scenic poets editors now follow Elmsley (on Eurip. Bacch. 1345) in writing first or second persons of the pluperfect with a short vowel : viz. Aristoph. Lysistr. 1098 δεινά κα 'πεπόνθεμες Soph. Ο. R. 1232 Χείπει μεν oil• 1 α πρόσθεν η δε μεν το μη ου Eur. Bacch. 1345 οψ έμάθεθ' ημάς, οτε h' έχρήν ουκ φΰετε. In the first passage it is a Lacedaemonian who is speaking ; and this is the only reading which makes sense (M.SS. πεπόνθαμες). In the second there is no reason for altering the γΰειμεν of the MJ*>S. : in the third though the recorded έίΰετε is also intelligible, the meaning is better with γΰετε. — For the 3 plur. the forms in ε-σαν are, as has been said above,, the only ones which are well supported. From Herodotus onwards forms like εζεληΧάκεσαν Herod. V. 20, εώθεσαν (Herod.), εωράκεσαν (Thuc), άπηρκεσαν (Demosth.) are universally in use. But from this ending -εσαν, as Ellendt. Lex. Soph. (ed. 2 da ) p. 515 justly notices, scholars have too quickly arrived at an ε in other persons, though the state of the case is quite different there. For Soph. 0. R. 1232 cp. Wecklein Are Soph, emend. 24. 239 c) Forms with ec. 1 sing. The termination -ειν, which was universally used later on, is now banished from the text of the tragedians. In later comic writers and in prose from Xenophon onwards there is abundant evidence for -ειν. A list of such passages is given by Kontos in the Αόγιος Έρμης p. 44 ff. 2 sing. Here so far as the tradition goes we are entirely dependent on the copyists, to whom -ης and -εις sounded just alike, -εις is common in Attic orators from Antiphon onwards : Antiphon 5 § 15, Isaeus 3 § 41 etc.-^ -ήρηρειοθα Archil, fr. 94 Be. 3 is quite unique. 3 sing. The form in -ει is recorded more than 30 times in Homer, e.g. ήρηρει Μ 56, εβεβηκει Λ 296, δεΰήει, Β 93. It continues to be the prevailing form. But we often have evidence that the reading of ' Aristarchus was that in -ειν e.g. Ζ 170 δεΤζαι δ' ήνώγειν φ πενθερψ, though here the presence of the f makes it impossible to speak of the ν as necessary : ήνώγειν, όντως σνν τψ ν Άρίσταργρς' ήνώγεεν yap (Did.). Zenodotus and Aristophanes also, according to the same authority, read in S 412 στήθος βεβληκειν υπέρ uvrvyoc. Hence there is not the slightest reason to attack this form, as La Roche Textkrit. 195 does, where we have authority for it, though we may well doubt whether Bekker was ch. χτι. THE PLUPERFECT. 433 right in introducing the ν at the end of the verse. The form in ειν is expressly asserted to be Attic in Bekker's Anecd. p. 422 άπερρώγει και σνν τφ ν άπερρώγειν το τρίτον πρόσωπον, and Cobet 1. c. and Kontos in the Αόγιος Ερμής justly express themselves in favour of the adoption of this form in the Attic writers, in order to avoid the hiatus : thus, as earlier editors gave the line, in Ar. Nub. 1347 ως ovtos, el μη τω 'froroi&w, ονκ αν ην όντως ακόλαστος ι^ειν=εωκα Αν. 1298, yfciv Eurip. Ion 1187. The plural and dual forms do not occur in great abundance, but present no controverted points : φΰειμεν occurs in Aeschin. iii. 82, ηΰειτε Demosth. lv. 9. In the 3 plur. -εισαν was regarded as un- Attic. The form in -ειν for the 3 plur., three times used by Apollonius Rhod. ii. 65, iv. 1700 (yfoiv), 947 -ηρηρειν is very peculiar. The learned Alexandrine gets into fine trouble for it with the Dutch critic (Novae Lectiones p. 467) : ' Apollonius ipse turpiter impegit. Quid eo homine facias qui γΰειν dixit pro sciebant' ! Certainly Apollonius sometimes indulged in strange freaks, and perhaps this 3 plur. in ειν was only an erroneous 240 imitation of Homeric forms like βάν, εφαν, μίγεν, which even at the present day many regard as abbreviated from those in -σαν. We come now to the question how the forms in η, ε and ει are related to those quoted above in εα and εε. That the η of the 1 sing, in the Attic writers is contracted from the Ionic εα, and that the ει of the 3 sing, is contracted from εε, was the prevalent opinion among the ancients, and is generally retained in modern times. This view meets with no difficulties in 1 sing. Ion. -ea Att. -η 2 sing. Ion. -eas (?) „ -ης, -ησθα 3 sing. Ion. -ee(v), -ei(v) „ -ei(v), Dor. -η 1 plur. Ion. -ea/iei/ (?) „ -η μεν. On the other hand this theory does not at once fit the 2 sing, in -εις, though on p. 430 we learnt to recognise the trace of an Ion. εες, and the 1 sing, in -ειν, or the plural and dual forms in -ειμεν, -είτε, -ειτον, -ειτην. But as the -ει in several instances cannot be shown to exist till late Attic, it is a very probable hypothesis that this diphthong made its way only by degrees from the 3 sing., where it had a just claim to its place, and that thus we have an instance here of an analogy extending. The ν of the 1 sing, may be explained from the analogy of the imperfects and aorists. In the case of the ει this view gains in probability from the fact that in one form, viz. in the 3 plur., the extension of this diphthong, though a much later phenomenon, goes on as it were under our eyes; for here it is only in the post- Attic period that ει takes the place of ε. This theory is also favoured by the history of Greek vocalism. The less the distinction between the sound of η and ει demonstrably became in the course of time, the more easily could the latter intrude itself into the place of the former. In the declension of the stems in -ευ we have the same process. The Old Attic Ιππής is to the later ιππείς precisely as γίης to ηίεις ; and the change of the termination -y in the 2 sing. mid. into -ει is also similar. Of verbal forms we may also compare the imperfect of είμι, γειν by the side of ήια, rja, discussed on p. 121, especially in the clearest instance, the form of F F 434 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. ch.xvi. 241 the 3 sing, rjeiv by the side of get. This ephelkystic ν forms an important link in the chain of the facts here coming into consideration. For as its place is only after short vowels, in forms of the 3 sing, like ηΰει(ν) there is still a reminiscence of the earlier form yhe(v). There are however two kinds of forms which do not agree with the theory suggested. There is first the 3 sing, in -η elsewhere than in the Doric dialect. For a 3 sing, ήείΰη could only be explained in Homer as a con- traction from 7)είΰεα. But no 3 sing, in -εα is either recorded or credible. For everywhere a 3 sing, in ε answers to the 1 sing, in -α : γέγονα — γέγονε, ήα — ήε, έγραφα — έγραφε. But we saw how fluctuating are the authori- ties for this η, which Cobet altogether rejects. We have therefore an instance here in which the textual criticism which is directed only to- wards securing an uniform polish, and is averse from questions of origin, agrees very well with the efforts which we are pursuing. — The other formation is that with the short ε for the first two persons plural, men- tioned on p. 432. πεπόνθεμες cannot have come from *πεπονθέαμες nor γίετε from ήΰεατε. But the support for these too was not very strong. If there had been more certain evidence for these two forms, we could hardly have helped regarding them as not compounded. As a matter of fact from the e-stem of the perfect we might arrive on the one side at a pluperfect with a long vowel η, on the other at one with the short vowel ε. πεπόνθεμες would then connect itself with the Doric infinitive πεπονθέμεν and the ordinary πεπονθέ -vat, ήείΰη and ηΰη-σθα with the Aeolic ΐοίΰη-μι, ΐοίΐη-σθα and the Heraclean πεφντενκήμεν. If we re- member that the 3 sing, plupf. άπολώλη actually occurs in the same Heraclean dialect, which presumably used *όλωλήμεν for όλωλέναι, and that εκεκρατηρίχημες, quoted more than once above, belongs to the equally Doric dialect of Epicharmus, it would not be exceeding the bounds of probability to assume a half-obsolete simple pluperfect with an e, the stem of which differed as little from the perfect stem, as the primitive and that supplied with the vowel of the present. The 3 sing, in η, which Aristarchus decidedly upheld in Homer, would then be a relic of 242 this formation beyond the sphere of Dorism. Still, considering the rarity of the forms mentioned, and the uncertainty of the authority for several of them, I do not attach any weight to this hypothesis, but pre- fer myself to hold to the view that all active pluperfects, with the excep- tion of those mentioned under 1) and 2) are compounded, and have originated from the Ionic forms by contraction ; and that hence, not taking into account these exceptions, the whole active pluperfect of the Greeks, which was actually in use, is formed in the same way as the pluperfect indicative in Latin. B) Middle Pluperfect. As the whole perfect stem in the middle is treated quite differently from the active, so is it also in the pluperfect. The method of formation is throughout the primitive; and therefore, as all that has to be noticed in the terminations has been stated before, there is no need for us to occupy ourselvas more in detail with this tense. We may just mention that even in Homer the middle pluperfect is very common, and is formed from the most various stems. The following will serve as examples : τετυγμην l 234, κεχόλωσο Π 585, έψήπτο Ζ 241, ΰέδετο Ε 387, ηρηρειστο Γ 358, -ch. χτι. THE FUTUEE PEKFECT. 435 ελέλειπτο Β 700, επέπυστο Ν 674, βεβλήατο Ε 28, εσφήκωννο Ρ 52, Γ£- τάσθην Δ 536. VIII. THE FUTURE FROM THE PERFECT STEM. Whilst in respect of the formation of a pluperfect Sanskrit may be compared with Greek at least in virtue of certain tendencies towards such a formation, the notion of forming a future from the perfect stem — as even the most resolute enemies of all ' Graeco-Italic ' must allow — is limited to Greek and Latin. Still even this formation is not carried out completely. The two languages here mutually supplement each other, Greek making but few attempts at an active future of completed action, but in the middle forming this tense tolerably extensively with a prevalently passive signification, while Latin on the other hand only ■carried out the active to completion, so that in the former the active, in the latter the passive can only be expressed by periphrasis. Χελνκως 'έσομαι = solvero XeXvaerai — solutum erit t 243 The future of the perfect stem presupposes absolutely the earlier formation of a future from the verbal stem. Α πεπαύσομαι or κεκόψομαι is inconceivable without παύσομαι, κόφομαι. The terminations -σομαι, -ση, -σεται had long stamped themselves on the consciousness of the Greeks as belonging to the future, when the creative impulse of the language employed them also with the perfect stem. The futurum exactum is a strictly new formation, but the agreement of Greek and Latin makes it probable that the tendency to such a new formation belongs to the time in which Greeks and Italians were still united. But by the side of the agreement of the two languages in this formation there is also a significant difference. The Greek future perfect adds the future termination to the perfect stem in its primitive form, the Latin to the e-stem : lique-ro for Hique-so, but Χελείπ-σο-μαι, tetige-ro, but τετάκ-σο-μαι. The difference cannot surprise us, for under the perfect stem we saw different stem-forms interchange elsewhere. The Latin future formations are more nearly represented by forms like the Homer, κεκαΰη- συμαι (θ 353) which agrees in its method of formation with cecide-ro, πεψιΰή- σεται (Ο 215), which would correspond to a Latin fe-Jide-rit, the conjectural basis of fiderit. But the Greek forms quoted do not attach themselves to perfects actually in use, and hence cannot be regarded as proper future perfects, but only as futures from stems reduplicated like the aorist. — The circumstance, at first sight surprising, that this future in Greek is only united with middle terminations, is connected with the fact to which our investigation has more than once conducted us, that the middle perfect forms, especially in the Homeric period, were far more widely current than the active. But a still more forcible reason is to be found in the fact that the future of the rt. as be, from which this compounded tense came, in Greek has middle, in Latin active personal endings. When the active perfect later on spread and multiplied, the language was doubtless lacking in the fresh creative impulse needed to develope futures from these new formations. Thus the perfect with κ 244 remained confined to some few future forms, which we shall notice immediately. We begin with these relics of a Greek active futurum exactum. The t f 2 436 PERFECT STEM AND FORMS CONSTRUCTED FROM IT. en. xvr. most remarkable instance of the kind is the Homeric κεχαρήσω, of which we find the infinitive at Ο 98 : ονδβ τι ψημί πασιν όμως θνμον κεχαρησίμςν thus in a definitively active meaning as compared with the passive ψ 266 ου μίν τοι θυμός Κ€χαρησ€ται. This is the only instance of a future perfect of a sharply differentiated active form side by side with the passive. The fact that the perfect stem here has an intensive present force cannot hinder us from taking the two forms so, any more than με μ νήσομαι by the side of meminero. — We have h- ΰωσω recorded on very dubious authority, viz. by Macrobius in the Excerpta Parisina (Grammatici Latini ed. Keil v. p. 610), where Homer's verse ν 358 arap και ΰώρα ΰιΰώσομεν is quoted with the reading ΰεΰώσομεν. The words are : inveniuntur huius modi tempora figurata et ex verbis in ω exeuntibus, ut est ΰεΰοικήσω, quod proprium Syracusanorum est, et δεδαίσω, ut apud Dracontem . . . Buttmann Ausf. Gr. i. 2 432 explains this as being the reading of the grammarian Draco of Stratonicea. In the passage quoted a future perfect is quite unsuitable. But perhaps we may still conjecture in this notice a trace of the real existence of such a form, ΰεΰωσω would agree with dedero except as to quantity. On the other hand εστηζω and τεθνήΕω are actually in use by the side of the later and in Cobet's judgment (Nov. Lect. 264) quite un- Attic middle forms εστήίομαι, τεθνηξομαι. Thus editors write in Eur. I. A. 675 e«r« σν, χερνίβων γαρ εστηξεις πίΚας in place of the εστήϊτι of the M.SS., as in Ar. Lys. 634 we have ώδε F εστήξω, Thuc. iii. 37 καθεστηζει, Ar. Acharn. 325 ως τεθνή'ξων 'ίσθί ννν'ι, Plat. Gorg. 469 τεθνήΟ,ει οντος, Aesch. Ag. 1279 τεθνήζομεν. As a variant to τεθνήζει we find twice in Aristoph. τεθνησει in the Ravenna cod., a form which we cannot regard as correctly recorded because of its strange- 245 ness : though when Cobet finds in it an 'immanis barbarismus,' it must be remarked that from the shorter stems in use in εσταμεν, τεθναμεν such a form might very well be formed upon analogy. In the usual middle forms all is regular with the one exception of slight variations in quantity. By the side of ΙεΙεμαι, λίλϋμαι we should expect ΰεΰέσομαι, which has but slight authority, and λελύσομαι, but ΰεΰήσομαι and λελύσημαι are in use, evidently under the influence of futures from the verbal stem, like Ιηβω, λνσομαι. There are numerous forms with an initial consonant, as ΰεΰέζομαι Ε 238, κεκλήστι Γ 138, μεμνησομαι Χ 390, πεφήσεαι χ 217, τετενζεται Μ 345, κεχο\ω<Τεται Α 139, λελείψεται ίΐ 742, μεμίζεται Hes. Ορρ. 179, γεγράψεται Soph. Aristoph., πεπλήζομαι Herod., εσκεφομαι Plat., εστερησομαι Eurip. Ι. Α. 1203. With regard to stems with initial vowels (cp. Homer, εφήσεται Ψ 795) Cobet Nov. Lect. 241, following I. Bekker's precedent, has restored many forms of the kind for corrupt readings in Attic writers, as άπ?;λ- λάζεσθε Aristoph. Ach. 757, ήτιμώσεται Dem., ήρήσεται Plato Prot. 338, άνεωίεται Xen. Hellen. v. 1, 14. This future is as a rule but rarely formed from derived stems. Still we have κεκινίυνεΰσεται Antiph. v. 75, ϊιποκεκινίννεΰσεται Thuc. iii. 39, ΰιαπεπολεμησόμενον Thuc. vii. 25. The moods and verbal nouns of the future are very rare ; in Homer there is only μεμνήσεσθαι : r 581, 79. ch. χτπ. THE SIGMATIC AORIST. 437 CHAPTER XVII. THE SIGMATIC AORIST The verbal forms hitherto discussed are quite sufficient, in the case of a verb proceeding directly from the root and characterising the present stem by expansions of the various classes, to express all elements of meaning developed at all in the Greek verb, with the one exception of the future : 246 and even in the case of this tense, as we shall see in the following chapter, there is no lack of attempts to express it with the aid of the means already mentioned. It might be objected that passivity in an action viewed as aoristic is only to^ be denoted by means of the more recent formations, with which 'we shall have to deal in Chap. XIX. But even here the older language is by no means destitute of a simple means. Homeric aorists like βλήμενος, "Αρηι κτάμενος show that it was only by degrees that the middle aorist lost the capacity of expressing passive action. Hence it follows that all the verbal forms which we have still to discuss are strictly speaking superfluous, that is, are not called for by any one definite need for expression. A form like έτρεψα beside ετραπον is, so to speak, an alter ego, and in the same way τρέψαι is needless by the side of τραπε~ιν. Such luxury of form-creation has hardly come before us hitherto. Ϋον the variety of the forms of the present stem, though resembling it, is not of the same kind. In the case of the present, we thought we could detect slight differences of meaning once existing for the plurality of forms. In the region with which we are now concerned there can hardly be any thought of that. At most we may compare the i-class of the present-stems with the sig- matic aorists, if we were right in explaining this from composition. In the introduction (p. 11) we put all the forms in question here together as the auxiliary stratum. There too the guiding thought was expressed that all the forms still to be discussed are related to those already dis- cussed as a sentence with the copula is to one without it. In order to establish our view more firmly it is necessary in the first place to take a survey of the forms distinguished by the letter σ. In Greek alone these forms, here limited as a rule to the aorist and future, form a very considerable group of verbal forms, which evidently had a mutual influence upon each other, and thereby in the course of time developed constantly more definite rules for their formation, σ thus became one of the most important formative sounds for the Greek verb, of all consonants it is heard the most frequently, and there are a number 247 of verbs, which besides the present stem have only developed the sig- matic stem. These forms appear in a somewhat different light, when we compare the cognate languages, and especially Sanskrit. The sibilant is here also richly represented, and we cannot doubt for a moment that it had its place even at the time of the Indo-Germanic unity in definite 438 THE SIGMATIC AORIST. ch. xnu positions. But the employment of it is less simple and regular than in Greek. The other languages fall quite into the background, but present some striking analogies. The dental sibilant shows itself in very different applications, viz. 1) In isolated personal endings, as in the Greek -σαν (ε-ΰο-σαν by the side of ε-ΰο-ν) and in the perfect ε'ίζασι i.e. εικ-σα-σι, for which it will be enough to refer to p. 48 and p. 427. Forms of this kind are probably confined to Old Persian, Greek, and Latin. 2) To the formation of Tense-stems, i.e. a) To the formation of a past tense, with corresponding mood-forms, participles, and infinitives from the pure verbal stem. These are the proper aorist forms, to be analysed more precisely hereafter. In these we may notice again the following differences, with regard to which, so far as the Vedic dialect is concerned, I follow Delbriick Altind. Verb, p. 177 ff. o) A simple s is added, which is directly united with the personal endings, e.g. Skt. a-bhai-sh-ma we feared (rt. bhl), hd-s-mahi we sepa- rated ourselves (rt. hd=Gr. χα, χάσκω). The greatest resemblance exists between the conjunctives. The mood- vowel here is a, so that e.g. hd-s-a-te is formed like Ιυ-σ-ε-ται as a conjunctive. If Latin forms like au-s-i-m,faxit are aorists, they belong here, i being added as an optative element to the stem expanded by s just as in ed-i-m. β) ish is added, in which the origin of the i is not definitely estab- lished, e.g. a~(fan-ish-ta he was born. Perhaps this i is of the same nature as the f , which appears in ε-στόρ-ε-σα and similar forms, which are dis- tinguished however by the use of the vowel a. 248 y) sish is added, which ought probably to be regarded as a redupli- cated s (si-s). m ΰ) sa is added, e.g. a-dhuk-shat he milked (rt. duh), Zend ven-ha-t he struck (rt. van). This formation corresponds to Greek aorists like ε-ΰεικ-σε, ε-φϋ-σα etc. In the 1 sing, and 3 plur. this appears also in Sanskrit in the place of a, e.g. a-jan-sa-m I held, 3 plur. a-jan-san (rt. jam). In precisely the same way Church-Slavonic forms aorists like 1 sing, ja-su I ate (=*ad-sa-m) just as if in Greek we had * id-era (cp. £σ-σα=έδ-σ•α I seated), bi-chu I struck (=*bi-sa-m). b) For the formation of forms from the present stem. Delbriick p. 181 with Grassmann calls such formations 'double stems.' They seem to occur in Sanskrit only with middle endings, e.g. arJca-se I praise (rt.. arK). We might compare the Greek ε-τάνν-σσε by the side of ra-vv-rau But there is a still more exact correspondence, as I have shown in Stud, viii. p. 460 ff., with the Latin imperfect conjunctive e.g. lege-re-m for legese-m. The e, which goes back to a + i and answers to the at of the Greek aorist optative, is related to the i of the above-mentioned fac-s-i-m precisely as vehe-t is to ed-i-t (as conjunctive), ster-ne-re-m is formed in just the same manner as the Skt. gr-nl-shd I praise (rt. gir). 1 c) For the formation of a perfect stem. This occurs, with any fre- quency, as was shown above p. 427, only in Latin and Keltic. In Greek we can only compare*t<7a/u. 1 These «-formations from the present stem are related to those from the verbal stem much as the original forms in -a-jd-mi, postulated above p. 327 for the optative of the thematic verbs (e.g. *bddha-jd-mi for hodhejam), to those of the primitive verbs (e.g. *as-jειν of εΐΰείην. On the same fundamental notion Ahrens bases his doctrine of the Greek inflexions, in which with especial clearness and consistency for every oblique mood — if we may so call the conjunctive and optative — and for every past tense a primarium is laid down, though this is wanting in the case of the aorists : e.g. primarium [λάβω], conj, λάβω opt, λάβοιμι etc. past Ζ-λαβο-ν. In Chap. XIII. we saw that the thematic aorist is nothing but a system of forms from shorter stems, the present indicative, or as we may also say the primarium, of which has fallen out of use. Hence it follows that we are justified in assuming for the formal completion of such a system, the actual existence of a primarium, even though in any par- ticular case such an indicative present may never have actually existed, but we must look rather to extensive imitation of increasing analogy after the type once discovered. It is not at all an objection to this view that the strictly aoristic force developed itself only at a time when the primarium had been lost. Consistency evidently leads us to presuppose also sigmatic primaria, or presents indicative, thus e.g. for the conjectural 255 Indo-Germ. aorist *a-dik-sa-m a *dik-sa-mi. Following out this idea we may give as a stem for the whole system of forms which go together dik-sa (as well as the shorter dik-s), without however on that account ch. xvii. THE ORIGIN OF THE σ. 443 maintaining, that such a stem actually existed or that the union of the it. dih with s-a, which came from as-a, was brought about at a time preceding the construction of verbal inflexion. 2 There is little probability in the latter notion. But if we assume that at a time when the nominal stems were not yet provided with case-endings, a twofold present dik-s-mi and dik-sa-mi arose by the incorporation of a nominal dih ' point' (cp. Lat. jiJ-dic ' right pointing') both with as-mi and also with as-a-mi, which was probably in use side by side with it (cp. the opt. ε(σ)-ο-ιμι), we can thus explain, I believe, all the phenomena here under consideration. From the same time, in which the formation by the addition of forms of the verb substantive was still in full swing, the other sigmatic tense, mood, and personal forms may also date ; they are as it were isolated offshoots, of which some remained isolated, others sprang up abundantly, though in part at a much later time. This view presents the following advantages over the older view which starts with the past tense. In the first place, an easier ex- planation of the past tense. To get from dik-dsat he was pointing, to a-dik-sa-t, it was necessary to assume a transposition of the aug- ment from the middle to the beginning of the word, which is not wholly without difficulty, especially as we may conjecture that the augment-syllable a cannot have long preserved itself separate from the initial a of the rt. as. On the other hand a-dik-sa-t is very simply explained as a past tense subsequently developed from *dik-sa-ti. The augment was in that case prefixed to the verbal form precisely as in all other past tenses. A second advantage of this view affects the relations of the meaning. ' At first sight,' I. say in Chron. 2 p. 58, ' it is surprising to find a root of this seemingly durative meaning undertaking such a formation (that of signifying momentary action). For being is certainly, 256 it would seem, very strictly a remaining, an enduring in something. We should therefore expect the rt. as rather in present forms, like Lat. pos-swni (=potis-sum) r than in aorists.' Now Clemm's theory does give us such presents. The employment of the sigmatic forms to act as aorists would therefore be explained somewhat as follows. By the side of the many other ways of forming the present stem, there were, during that linguistic period in which the compound method of formation came up, probably the same period as that to which we owe the present stems in -ja, presents in -s-mi and sa-mi with their moods, and belonging to these past tenses in -sa-m. Perhaps even at a time when the consciousness of the origin of these forms was not wholly extinct a contrast arose between the forms in -ja-mi and those in -sa-mi, the former denoting rather the strictly durative force of going about a thing, striving after a thing, the latter rather being as opposed to becoming, that is, so to speak, the point of congelation of the action. At a yet earlier time the distinction of continuous from momentary action had presented itself to language, and had been denoted by the distinction of the expanded forms from the lighter ones. We can understand how these sigmatic forms attached themselves to the latter, the aorists, and how thus, perhaps first in the past tense, a form which properly meant ' I was pointing ' came to be ϊψφ, 2 C. PaulilKKuhn's Zeitschr. xx. p. 321 ff.) in his reply to Merguet on Latin verbal formation, in which there is much upon this question well worth notice, defends the view of such a composition of the stem. But auxiliary verbs can hardly have developed at all without inflexion. 444 THE SIGMATIC AOKIST. ch. xvii. used in the narration of mere facts, while afterwards the oblique moods also acquired a momentary force. The more the need for such a distinc- tion of the kind of time grew, the more necessary was it to supplement the old partly primitive partly thematic forms, which on account of various phonetic difficulties had not maintained themselves in the case of all stems, e.g. but rarely with those ending in a vowel. The sigmatic forms filled up this gap. We can now understand further how in con- sequence of this the sigmatic primaria died off, as being superfluous for the present just as the primaria */3i//«, *λάβω, *τύπω have died off, and how finally in this way the main requisite for aoristic use, that they were accompanied by no corresponding present, was attained for these sig- matic formations. 257 The numerous derived verbs of the Greeks acquired their aorists only in this way. I now believe that attaching equal importance to the form and to the meaning I can so explain the origin of the sigmatic aorist. I do not at all fail to perceive that such speculations as to the genesis of the forms of language do not guide us to the certainty which is attainable in other questions. But to give them up, and to content ourselves with the mere summary phrases of stem-formation or analogous formation would be ultimately to give up the proper understanding of the structure of the verb, and to sink down from the idealistic flight which distinguishes comparative philology, to the purely empiric record of facts. Now that it has been made probable that an indicative in -sa-mi with the character of a present once existed, the question arises whether we cannot point out some traces of it. Clemm 1. c. is probably right in suggesting the Sanskrit desideratives, in which the syllable -sa is added as a rule to the stem strengthened by reduplication, e.g. pipd-sa-ti he wishes to drink (rt. pa), di-drk-sha-nte (rt. darg) they wish to see, like a conceivable Greek *ΰι-Βερκ-σο-νται. Sometimes the reduplication is wanting : ap-sa-nta they wished to attain (rt. dp). The desiderative force is doubtless here the result of the reduplication. From the mean- ing ' he is a strong drinker' — the 'strong' lies in the intensive force of the reduplication — we very easily arrive at the desiderative force. After such an application had once established itself in forms of this type, they might originate even without reduplication, just as there are some perfects without reduplication, and many past tenses without augment. We have an exactly similar instance in the Lat. vlsere : visit corresponds to the Skt. vivid-i-sha-ti which differs but slightly in its manner of formation. From Greek Clemm quotes Ησα-μι as similarly formed. As far as form goes there is no objection to be made to this : and no one can deny that the meaning is purely that of a present, while the common Homeric 3 pi. Ρί-σαν suits very well as a past to the present-like Η-σαντι. But what makes me still hesitate is the fact that the 3 pi. Λσάσι, which cannot be separated from the Doric 'ίσάμι, belongs entirely to the system 258 of the indubitably perfect .fot£a=Skt. veda. For this reason 'ΐσάμι was quoted as one of the traces of a sigmatic perfect, for which naturally we must assume the loss of the reduplication. With greater justice we may find relics of the primaria of which we are in quest in certain formations which have allowed the α to pass into the ordinary form of the thematic vowel. — For instance, αλεΐασΰαι Ν 475 θηγει αΧίϊ,ασβαι μεμαώς κύνας ή$έ και άνδρας is incontestably an aorist. But how is this form related to άλεζέμεναι A 590 άλεξέμεναι μεμαώτα and ch. xvii. PEIMITIVE FOEMATIONS WITH A SINGLE but according to Delbruck p. 198 this explanation of the forms is not quite certain. Instead of this -σε -σον has become the prevalent termina- tion from Homer onwards (άκουσον Ζ 334, εασον Π 451, υμοσσον Α 76 etc.). The question is how the ν is to be explained here. Pott Et. Forsch. ii. 2 1, 660 sees in it an appended particle vv. In that case this Greek form would have originated in the same way as the Lithuanian im- 266 perative in -h e.g. bu-k be, for which see Schleicher Comp. 3 827. But we have no certain instance of such affixes in the structure of the Greek verb. Still less can we agree with Bopp, who finds in the ν the remains of the termination -θι, which, he thinks, first passed into -ς (*7•ύψα-0<, *τνψας) and finally into -v. For all these intermediate forms are purely hypothetical and the assumed phonetic changes undemonstrable. And as we have repeatedly seen that nasals are developed after short final vowels, and become more or less firmly established — especially in the 3 plur. of the imperative (cp. p. 308) — the simplest explanation is that suggested by Giese (Aeol. Dial. p. 110), according to which the ν οϊ άκσνσον, Μείζον etc. is a nasal after-sound, which became firmly attached, and which then had as its result the dulling of α into o. The Cyprian dialect has recently supplied us with an instance of such nasal after- sounds" in the genitives in ων e.g. Όνασίλων=Όοτ. Όνασ/λω, for which an etymological justification has been sought in• vain. The Syracusan imperatives θίγον, λάβον adduced by Giese as analogous were mentioned on p. 303. It is far more difficult to explain the 2 sing. mid. in -σαι. Here too there is no trace of any dialectic variation. In Homer we find forms like ϊέζαι (Ε 227), κόμισαι (Ε 359), λίσαι (Α 391), λόεσσαι (τ 320), πείρησαι (Α 302), ψράσαι (π 260), υπόκριναι (γ 535). The attempt of Bopp (Vergl. Gr. § 727) to argue from a conceivable active *τνψαθι to the purely hypothetical middle intermediate form *τύψασθι, and to get from this by the rejection of the σθ to τύφαι cannot at all satisfy us. Benfey (Kurze Sanskrit-Gr. p. 90) intimates that he regards -σαι as a conjunctive ending, for which he quotes the analogy of a purely imaginary Sanskrit -*svai. But τνφαι is entirely without any trace of a mood-sign. Following Benfey Misteli Ztsch. xv. 332 conjectures that -σαι is to be taken as a ' fuller ' imperative ending and that τνφαι is for τνφα-σαι. But we should certainly expect τνφα-σο, for the imperative is by no means fond of fuller endings for the second person. Finally Schleicher declares the riddle to be unsolved. There might possibly be some truth in the notion that we are to take not the simple ι added to the tense-stem, for which hardly anyone will claim the force of a middle 267 personal ending, but -σαι as such here. Only the form would then b^ proved to be an indicative, δέξαι, τύψαι might be indicatives used as ψε (Σ 240), ερζα (rt. hpy —χ 312), ύ 9 1« and άρζα (Attic), έ'0ελ£ε (Ο 322). There are only two kinds of stems which form an exception, inasmuch as in these more extensive changes and irregularities occur, viz. the stems of the verbs which in the present end in -ζω, and those in λ μ ν p. We will deal first with the former, and in doing so will not sepa- rate the aorist from the future, which in this respect quite corre- sponds to it. The simple rule, which might be supposed to be the only one adapted to the structure of the language, that the dental stem in the sigmatic tenses gives σσ or a simple σ, the guttural ζ, is not observed in the case of the verbs which in the present end in -ζω. In root-verbs, it is true, we shall as a rule, and especially for the Ionic dialect in the widest sense, regard a σσ or σ in these tenses as the criterion of a final dental in the stem, a ξ as the criterion of a guttural, as has been done in the survey of them on p. 222 ff. : σχίζω . (έσχισα), στίζω (στίζω), and this criterion is almost always confirmed by other evidence from the formation of nouns and verbs (σχίΰη — στιγμή). But in the case of derived verbs in -αζω and -ιζω nothing warrants us in regarding those in which ζ appears in the aorist and future as in their formation essentially different from those which show σ. It is the same Ιικάζω from which the Doric Ζικαζάμενος comes and the Attic Ιικασά- μενος ; and no one will desire for the language of Homer, in which, in the case of two verbs, by the side of the forms with ζ {αρπάζων Χ 310, ηρπαζε Μ 305, έψαρπάζαι Ν 189, άρπάζας Γ 444 — κτερεΐζω β 2£2, κτερείζαι α 291) we find rarer forms with σ (άνηρπασε I 564, ηρπασε Ν 528, κτεριώ Σ 334, κτερωνσι Λ 455, κτερίσειε γ 285) to assume two stems for the same word. We must say rather that from the time of Homer onwards a twofold method of formation has established itself. The relation of the dialects is broadly this : Boric has as a rule ζ, Ionic as a rule σ j but the Homeric dialect fluctuates. As to the Aeolic nothing positive can be said owing to the deficiency of information. The question now is, can this tolerably extensive difference in the for- 271 mation of much-used tenses be explained, or must we be satisfied with accept- ing it as a fact ? Two attempts at explanation have been made as yet. The one, adopted by me in Princ. ii. 247 f. after H. L. Ahrens, endeavours to explain the ς from the incorporation of the j originally belonging to the present stem, δικάζω is for 3ικα£/ ω, and this, as I think I have shown, is for Ιικάβω. The type εΐίκαζα according to this view arose at a time in which the palatal spirant still remained pure, the type εΰίκασσα (later εΐίκασα) at a time when the £ had already made its way in. The former formation would thus be the earlier, and hence there are still numerous traces of it in Homer, the second the later. The other explanation is that from analogy. This is developed and supported most clearly by Cauer in the Sprachw. Abhandl. p. 127 ff. According to this view the historical process was somewhat as follows. There were from very early times radical presents in ζω with the future in ζω and others with the future in σω, and also certain derived verbs in -ζω (e.g. αρπάζω), whose stem originally ended in a guttural (άρπαγ-η). In determining the form of the derived verbs, which as a rule were later, language became confused, and did not keep the two types rigidly distinct. Hence on the analogy of στάζω έσταζα, αρπάζω ηρπαζα was formed also άβροτάζομεν (Κ 65), on the analogy of στίζω στίζω πολεμίζομεν (Ω 667). In the Principles [ii. 454 THE SIGMATIC AOBIST. ch. xvii. 248] 5 I uttered a warning against the assumption of formation on analogy, and demanded from those scholars who inclined to it the proof of definite verbal forms from which the analogy was drawn. This latter demand is now partially satisfied by Cauer's investigation : whether quite satisfactorily or not remains to be seen. First we must take a somewhat more exact survey of the facts given. In Homer there are more than 50 presents in -ζω with a dental future and aorist formation, and 28 with I. The various needs and con- veniences of the verse are satisfied in the case of the former group by the interchange between the archaic σσ and the later σ : e.g. ίεινίσσομεν η 190, ζείνισ Ζ 217, πελάσσαι Ψ 719, πελασαν Α 434, κόμισσα Λ 738, κόμισεν Ε 456. The choice between £ and σσ has therefore nothing 272 whatever to do with the convenience of forming the verse, in which many scholars so readily find the occasion for irregularities. Among the 28 instances of the second group there are 8 radical verbs, for which we must lay down a guttural stem, viz. the three mimetic verbs κλάζω, λίζω (λίγζε Δ 125), μύζω, and also ελελίζω shake (cp. p. 226) πλάζω^ ρέζω, στάζω and σφάζω. There are further 7 derived verbs, for which Cauer's derivation from a stem-noun in γ for an earlier *,• has some probability : αρπάζω (αρπαγή : cp. Lat. rapax), μαστίζω (cp. μάστιζ), ελελίζεσθαι to turn (cp. ε\ι£, ελίσσω), σαλπίζω (cp. σάλπιγξ), στηρίζω (cp. στήριγζ), perhaps also άλα7τά£ω, if we derive it with Fick i. 8 24 from the stem of the Sanskrit alpaka-s weak, ολολύζω (cp. όλολνγίι) which is possibly connected with μαρμαρύσσω μαρμαρυγή and the Sanskrit forma- tions in uha mentioned on p. 257. In the case of οϊμώζειν and \ΰζειν we might certainly explain the forms with £ from the analogy of the simple mimetic verbs. There still remain then 1 1 instances unex- plained, εζεναρίζω, πόλε μι ζω come from the stems εναρο, πόλεμο in the same way as οπλίζω, Ιεινίζω from όπλο, ϊ,εινο, and yet the former make εϊ,εναρίίει Ύ 339, πολεμίζομεν Ω 667, Β 328, the latter έψοπλίσσωσι ω 360, Εείνισ' Ζ 217. The remaining words are άβροτάζω, άποβρίζω (άποβρί- Ιαντες ι 151, μ 7), ΰαίζω, ΰνοπαλίζω, εγγυαλίζω, κτεράζω, μερμηρίζω^ πελεμίζω, στυφελίζω. αποβρίζειν, which Cauer quotes without proof as a primitive stem-verb, is possibly connected with βριθύ-ς and βαρν-ς (Princ. ii. 77), but in any case the guttural stem is not proved etymo- logically. These 1 1 instances must according to Cauer have followed the analogy of the former 17. Certainly the resemblance of the 11 to- the 17 is not particularly striking. It would be conceivable that even among the 11 there were some verbs in which the guttural had its source in a word-forming suffix just as in μαστίζω and σαλπίζω and in ψορμίζω, which in Homer only occurs in the present stem. What appears in the language of Homer as an exception, though a tolerably wide-spread one, in Doric has become the rule. The Heraclean tables give us the following instances (Meister Stud. iv. 427 f.) : άγγωρί- ζαντες, εγδικαζαμένοι, ΰοκιμαζ,όντι, εργάζονται, έμερίζαν, ετερμάζαν, κατεσω- Ιαμες (as well as κατεσψσαμες i. 47, 51), έρίζαιτες, ωρϊζ,αν. It is un- necessary to reproduce all the other examples collected by Ahrens Dor. 89 L 273 It is sufficient to point out that the most different Doric districts, viz. Laconia, Crete, Southern Italy, Corcyra, Rhodes and Thera agree in this. The ξ is least established from the region of the northern Doric. The * [The discussion of this question in the present work is now (Grundzuge*: 187D) substituted for that in earlier editions of the Principles.] ch. xvii. AOKISTS IN fa FROM NON-GUTTURAL ROOTS. 455 Delphic inscriptions published by Wescher and Foucart have in no. 154 νυσφίζαιντο, but in no 3, 8 ηγωνίσαντο. We find άγωνίσσατο on the by no means pure Doric inscription I in Jahn-Michaelis Bilderchroniken p. 84. The Locrian inscription of Naupactus gives us reason by the form ψάφιξις to conjecture that εφαφίϊ,ατο also was said in that district. On the Messenian inscription of Andania we read 1. 2 Όρκιζάτω and 1. 36 χωραΐάτω from the otherwise unauthenticated χωράζειρ, which Sauppe is probably right in translating ' set up,' but 1. 4 έξορκίσειν. The Cretan forms of the kind are collected in Helbig de dial. Cret. p. 20. We may quote εμφανιζάρτωΓ,έινβρίζτις, πειράζομεν, συναγωνιζαμένων, χαριί,ίομεθα. — The Aeolic dialects differ widely from each other. The Boeotians instead of the σσ which arises out of the combination of the dental with the second a said ττ. That is to say, where the σ came in contact with the explosive sound hardened to r, they allowed the latter to prevail. So Ahrens Aeol. 177 explained the κατασκενάττη, i.e. κατασκενάσαι of an inscription, and the reading of the Ravenna M.S. in the Boeotian - passage in Aristoph. Ach. 884 κήπιχήριτται τω ζένω, where the at in the place of the true Boeotian η is due to the Attic writers. Since then we have recently acquired the form καταΰονλίττασθαι 6 i.e. καταϊουλίσασθαι on an inscription from Orchomenus published by Decharme in the Archive des missions scientifiques for 1868, and κομιττάμειοι Athenaion in. 482. — In Asiatic Aeolic we find instead of this ττ the σσ familiar from Homer in the inscription from Erythrae published by Kenner in the Sitzungsber. der Wiener Akad. 1872 p. 335, 1. 55 χαρίσσονται, on one of those published by Conze (Reise auf Lesbos) ΰικάσσω, but also ΰιαρπάζαις. The decree of Tenedos (Philol. xxv. 191) has the forms εφαφίσατο and εΰίκασε. — On the other hand we read on the inscription from Tegea 1. 5 and 28 παρετάζωνσι from παρετάζω (Gelbke Stud. ii. 38), and the Arcadian dialect here agrees with the Cyprian, as it often does. For 274 the inscription from Idalium twice gives (Stud. vii. 252) the conjunctive tlopvltf 7 i.e. εξορίξρ or in Attic ε'ίορίσ^, where the ν is perhaps to be ex- plained from the primitive form οργίζω (in inscriptions opfog and υρβος). This confirms my conjecture (Stud. vi. 372) that the gloss in Hesychius νξον βοηθησον is Cyprian and identical with the Laconian σοϊΐον (cp. Heracl. σωίζαι), the rough breathing taking the place of σ. 8 Now that we have thus ascertained the facts of the case, I must confess that I cannot find any decisive argument against the explanation of this surprising ζ which I previously gave after Ahrens. The J, which was originally pronounced before the thematic vowel of the present, might as a palatal spirant easily harden into an explosive sound before σ, and this with the σ produced ξ, precisely as in modern Greek the verbs in -εύω, i.e. evo form their future in -ενσω, i.e. epso. For the fact that j even between vowels was not far removed from the γ proofs have been brought forward in the Principles 1. c. ; among these the most remark- able are the Heraclean forms ποτικλαίγωσα=πρυσκ\είονσα tab. ii. 69, ποτικλαϊγοΐ'=προσκλε~ων ii. 107, brought into this connexion by Meister. 6 The editor did not understand the form. He accentuates it καταδουλιττα- σθαι. 7 Ahrens however now gives a quite different explanation of this word. 8 We may place here also the curious form 4ζώ\ί-γξ€ • ώλισθςρ Hesych., which may possibly be taken as an aorist to a present *6λίζω as a byform of ολισθάνω. Cp. 7Γ€λίγξ<Μ • έπιδραμέΐν. σωλίΎ^αι • (?) δραμέΐν. 456 THE SIGMATIC AORIST. ch. xvii. Although I have no objection on principle to admit the operation of analogy in such cases, still the facts are, I believe, more easily explained by assuming from the earliest times two ways of treating the j. This naturally does not exclude the notion that such Doric forms, in part only making their appearance at a late date, as are unmistakeably based upon a dental stem, e.g. καθίζω (rt. ed, fut. καθιξεϊ in Bion ii. 16, κάθιζα ς Theocr. i. 12 9 ), παίζω (rt. παώ, fut. παιξοϋνται Laconian in Xenophon Conv. 9, 2 and often in later writers, cp. Traiyvtov), really owe their ξ first to the analogy of the other forms which were established earlier. 275 This has already been shown in Cauer p. 143 f. The converse change is also explained there. It cannot be denied that the analogy of the count- less verbs with ζ in the present and σ in the sigmatic tenses according to the usage of Attic Ionic, suppressed certain fully justified formations with £, and put in their place the more common forms with a. A be- ginning had been made here by the Homeric ήμπασε side by side with άρπάζαι, already mentioned. Homer gives from μνζω επέμυζαν (Δ 20, θ 457) for which we must assume a root μνγ. The same may be said of (Τκαζω, σαΧπίζω, στηρίζω, συρίζω. The liquids p and λ do not agree well with a following σ, the nasals still worse. As a rule the sibilant has disappeared here in consequence of progressive assimilation. Still there are some instances of the full form, all from stems ending in p or λ. No stem which preserves a final nasal before σ is known to me. For κενσαι Ψ 337, like έπερσε a 2 &ηά"άμερσε θ 64, has lost a dental explosive before the σ, and therefore remains un- changed. We may conjecture that the sound of the σ here was sharper. From *εσπενΰ-σα came first, we may be sure, *εσπεν-σα, but even in Homer we have the still more softened εσπεισα (*> 55). A special place is taken by τέρσειν dry, imper. aor. τέρσαι Nicand. Ther. 693 because the σ of the root coalesces with the σ which forms the stem. The case is different with the following forms : άέρση Panyasis in Athen. ii. 35, v. 14. μη σ' νβρις eVi φρςσι θυμον depart αρσαι from rt. ap fit : επήρσεν & 167, 339, άρσε 45, άρσον β 289, 353, άρσαντες Α 136, αρσάμενος Hesiod. Scut. 320. εΧσαν 1 Λ 413, inf. εΧσαι Α 409, Φ 225, έέλσαι Φ 295, ελσας η 250, Callinus fr. 1, 11, ελ<τα(ς (partic.) Pind. 01. χ. 43. άπό-ερσε Ζ 348, άπο-έρση Φ 283, άπο-έρσειε Φ 329 ci -έρσης from έίρειν string, fasten properly, is used in Hippocr. θερσόμενος τ 507. 276 εκέλσαμεν μ 5, επ-έκελσεν ν 114, κελσάσησι ι 149. Here the σ re- mained even in the tragedians (Soph. Fr. 804), and in the future κέλσω too, while the nearly akin όκέλλω has ώκειλα. εκερσεν Ν 546, κέρσε Κ 456, Hes. Scut. 419, κέρσειε Aesch. Suppl. 665, κερσάμενος Pers. 952, opt. fut. κέρσοι Mosch. ii. 32 (?). κνρσας Γ 23, Ψ 428, also in Hesiod. Theog. 198 (προσέκυρσε), He- rodotus (ενέκνρσε iii. 77) and the tragedians (κνρσω Soph. O. C. 235). ύρσομεν Δ 16, ορσωμεν Η 38, ορσητε Ψ 210, also in Hesiod, Pindar, and the tragedians (ορσω Soph. Antig. 1060). • Also κάθιξον κάθισον Hesych. 1 We may mention here also <λ<τατο, said to be for ήλάσατο in Ibycus fr. 54, and ^Κσάμην in 8imon. Amorg. 17 in a verse not yet fully explained. ch. χγιι. AORISTS FROM LIQUID BOOTS. 457 εστελσεν' εστειλεν Hesych. εψερσεν εκύησεν, quite analogous to the Skt. a-bhdrsha-m from the root bhar, which in the most different languages serves to denote bring- ing forth and pregnancy. Cf. Princ. i. 373. ΰια-φθέρσει Ν 625, φθέρσαντες Lycophron 1003. φνρσω σ 21, ψνρσάμενος Nicand. Ther. 507. There are altogether only 14 forms, 11 with p, 3 with λ before the σ . From these we come next to those with the final letter of the root doubled. For these the form ετέρρατο' εζηράιθη is instructive, the σ here, in spite of its power of resistance noticed above, having fallen a victim to the process of assimilation. Among the instances of the retention of the doubled consonant the solitary Homeric όψέλλειεν Π 651, β 334 is of especial importance. Asiatic Aeolic gives us more of the same kind (Ahrens Aeol. 50). We have the evidence of inscrip- tions for απέστελλαν, now found also in the inscription from Erythrae, (Kenner u. s. p. 335), επαγγελλαμένων, and εκριννε in the Lesbian inscriptions edited by Conze. The remaining forms rest upon the state- ments of the grammarians. Cp. Herodian ed. Lentz ii. 306, 33. So εγεννατο, σπελλάμειαι' στειλάμεναι Hesych., ενεμματο, κτένναι, imp. aor. ερρον, 3 sing, όρράτω, according to the probable reading of Ahrens (Herodian ed. Lentz ii. 503, 13). How single consonants came from these double ones, with the compensatory lengthening differently regu- lated in different dialects, needs no further explanation. Doric forms like the Cretan παραγγηλωντι (C. I. G. 2556, 43), αποστηλάντων (Naber Mnemos. i. 114 ff. 1. 1), πεί-ιαλαι* μεταπέμφασθαι Hesych. have not been very commonly preserved. It is worth while noticing the form (ρΰέραι=φθείραι on the inscription from Tegea (Gelbke Stud. ii. 38), 277 where the double ρ has passed into the single ρ without any compensatory lengthening. To the great regularity, with which the laws of compensatory lengthening are observed in the Attic-Ionic dialect from Homer onwards, as is shown by numerous instances like ήγειρα, ήειρα, εΰειρα, εκάθηρα, ηλατο, ϊηλα, έστειλε, φϊλαι, είειμα, εγημα, ενειμα, απεκτεινα, εγείνατο, άρτΰιας, ϊηνα, εκλίνατο, the occasional variations in the treatment of α before ν and p form a remarkable exception. In Homer perhaps the only word belonging here is ay^pavy Φ 347, which savours of Atticism, as distinguished from μιψτ} Δ 141. In Attic writers we are less surprised at the preservation of the ά after ι and ρ (περάνας, έπιανα), since this takes its place in a more extensive series of analogies, than at the occurrence of forms in which the ά appears after other consonants. We have good evidence for ισχι/ά^ασΆββοη. Eum. 267 — cp.Aristoph. Ran. 941, εκέρίανα Soph. fr. 499 Dind., Andoc. i. 134, Dem. ix. 29, εκοίλαναν Thuc. iv. 100, όργάνειας Soph. O. R. 335, πεπάναι Aristoph. Vesp. 646 chor. Lobeck ad Phryn. 25, Paralipp. 21 sqq. has already expressed the view that it is but useless trouble to attempt to find any definite reasons for this inconsistency. Only one thing is certain, that the later writers, evidently incline more to the ά in consequence of the popular language, which here and there Dorised. Perhaps also σί//χαι/αι in Xenophon — though this form is disputed — is to be set down to the score of the many different deviations of this πολύτροπος from genuine Attic. The reten- tion of the α must have been favoured by the analogy of the numerous other forms in which there was always a. 458 THE SIGMATIC AORIST. ch. xvii. We come now to the treatment of the σ in stems ending in a vowel. The regular forms with the σ retained, are here really the most difficult. In view of the tendency of the Greeks to allow a σ between two vowels to pass into an aspirate, the retention of the sibilant in the same position in the case of the aorist is surprising. For the cases cannot be numerous in which a σ which is not weakened from r (ψησί) or σα (ορεσι) remains unchanged in such a position. I have called attention to this fact, which has hitherto been too little noticed, in my essay * Ueber die Tragweite 278 der Lautgesetze' (Ber. d. k. Sachs. Ges. d. Wissensch. 1870) p. 24 f. and have there quoted the parallel between the σ of the aorists and that of the 3 pi. -σαν in ε-Ζο-σαν, ε-φα-σαν. The occasional preservation of the σ in second persons like Ιΰνασαι, ήπίστασο is not quite comparable, for the original ending here contained a } as well as σ. So too in the 3 plur. of those primitive past tenses, as in that of the pluperfects discussed on p. 430 f., no reason presents itself for deducing an original double consonant. I think I can find the reason for this anomaly in the history of sounds in the endeavour after perspicuity. The tendency to more convenient intona- tion was not operative here, because its satisfaction would have led to harsh and obscure forms like *εΰοαν, *ϊαν, *εδούλωα, *εστηα. But I think that I have now discovered another reason in the case of the sigmatic aorists, viz. the action of the other sigmatic aorist forms and the sigmatic future. Aorists like έπραξα, kypa\f/a, ήλπισα from con- sonantal stems were common enough to stamp the sound of σ for the instinct of the language as one distinctive of this tense-formation. Besides, the similarity between the sigmatic aorists and the futures must have become evident at an early date, so that the two tense-systems, one of them revolving about the stem γραψα, the other round γραψ, must have been felt to be parallel. But in the future the σ, arising from aj, was of a more fixed character, which secured its preservation even between vowels. I think we may assert that the parallelism between the future and the aorist here exerted a preservative influence. The change of the σ into an aspirate is even in the Laconian dialect,, to which it was quite especially peculiar, a phenomenon of later origin in aorist forms. In the Et. Magn. 391, 20, forms like ποιήαί are mentioned. We find now νεικάαρ i.e. νικι)σας and νεικάαντερ = νικήσαντες on inscriptions (Hermes iii. 449 f.). The chorus of the Laconians in Aristoph. Lys. 1247 begins with ορμαον i.e ορμησον. Also from the- Cyprian glosses of Hesychius Mor. Schmidt Ztschr. ix. 367 establishes the same change. We may regard as certain: εναυυν ενθες Κύπριοι , cp. είανσαι ' εζεΧεϊν, i -μίτρα-υν' υπόζωσον ΤΙάψιοι, ιμπάταον' εμβ\εψον τ ινκαταπάταον' εγκατάβλεφυν, ϊμαΰν ' πάταζο ν, σία ι' πτύσαι ΙΙάφιοι. In 279 the newly decyphered Cyprian texts no trace of this process has as yet appeared. It is more difficult to investigate the aorists with a characteristic a y which have no σ. These extend over the most various Greek dialects. Five of them form a distinct group, avor/ having preceded the σ in all. I quote first the forms coming into consideration. άλεύατο Γ 360 etc., ήλεΰατο Ν 184 etc., άλεΰεται conj. £ 400, άλεηται δ 396, χ'ιλέαιτο Υ 147, άλεναι Χ 285, ρΐ. άλέασθε δ 774, άλενασθαι μ 159, 269, Hes. Ορρ. 798, άλέασθαι Ν 436 etc., άλευάμενυς Ε 444 etc. From the active we find also the regular forms &\ενσυν Aesch. Sept. 141,. Suppl. 528, άλενσατε Sept. 86. ch. xvii. AORISTS WHICH SHOW NO σ. 459 οατέασθαι only Hes. Opp. 767 (?) έκηα A 40, κατέκηε Ζ 418, conj. κηομεν Η 377, opt. κηαιεν Ω 38, imp. κήυ>' (f) 176, inf. κακκήαι λ 74, κηαντες ι 231, κηαντο Ι 88, κηάμενοι Ι 234» There are also κεας Aesch. Ag. 849, Soph. El. 757, εκκέας Aristoph. Pax 1132 (chor.), as well as εκανσα Herod, viii. 33, κατέκαυσαν Thuc. vii. 25, καϋσαι PI. έσσευα Ε 208, σεϋα Υ 189, έσσευε Υ 325, έπέσσευε σ 256, σεϋα»/ £ 89,. εσσεύαντυ Λ 549 ; εσσενσα^ quoted only from Anth.vii.439 (έπισσεύσασα). έχευαν Γ 270, Δ 269, χενα^ # 436, χεύωσι Η 86, χευώ-ω Alcaeus fr. 36, 3 Be. 3 , χεΰαντων I 214, χεϋαι inf. λ 75 — with έ'χεαΐ' Σ 347, and the same form in Attic comic writers, conj. χε'77 Eur. Cycl. 329, εγχε'αι/ζ* Aristoph. Ach. 1055, έγχέαντος Plato Conv. 214 inf. σνγχέαι Isae. v. 18, ΰιαχέαι Herod, viii. 57, άμφιχέαι Hes. Opp. 65, χέασθαι Soph. O. C. 477. — χεύσας Metrodorus Anth. xiv. 124, 8. Tolerably numerous traces of these sigmatic forms from various M.SS. of Homer (e.g. α 291, β 354) are pointed out by La Roche Homer. Unters. 270. They evidently prove nothing but the uncertainty of the copyists. These five forms have been frequently explained by the assumption that σ has dropt out between the two vowels. But it is by no means probable that the dialects, which e.g. in έκλαυσα, έπλευσα and the numerous derived verbs in ει/ω e.g. in εβασίλευσα left the σ unchanged, rejected the same sound under precisely the same conditions from these 280 few. It might have occurred to us that the σ disappeared, not after . υ, but after the Ϊ which is closely connected with it, and that the process was : €-χ€£-σα €-χ€β£α εχευα e^ea so that the ? like the liquids and the nasals assimilated to itself the following σ, and then in some cases continued to exist as υ, in others dropt out altogether. But then we cannot understand how the /, which in similar forms is wont to appear only before vowels, got into its place before a. To make this explanation intelligible we should have to start, not from the roots χν, συ, but from χεΐ, σίϊ : — and as a matter of fact we cannot get further back than καΡ. But who could make up his mind to assume a root χε/ for the forms of χέω, while for κέχυται we are guided to χυ 1 κέχυται and the like would then have to be regarded as shortened, without our being able to see any reason for the shortening. Hence the purely phonetic explanation evidently has its special difficulties. I believe therefore that we cannot get beyond forms like *έκαΡ-α, *ϊχεϊ-α, *εσσεΡ-α etc. The η in έκηα may well be taken as compensatory lengthening. Where in Homer we find ει written instead, κίίαντες and the like, the more recent editors have for the most part rightly restored the η (cp. La Roche Homer. Unters. p. 159), which is the only form justified as a compensatory lengthening of a. A noteworthy parallel to έκηα is furnished by the Boeot. σεϊα. The gloss of Hesychius σεΐα ■ έΰίωζα Βοιωτοί is certainly not to be tampered with, for a Boeotian ει represents an Attic η. Hence we have before us the same formation as in κηα, έκηα. As the absence of the augment points to a poet, perhaps Corinna is the source of this gloss. On these considerations we must therefore decide that these five 2 aorists, among which Ιατεασθαι has 2 Joh. Schmidt Vocal, ii. 3,31 thinks he can add a sixth : i -πριά-μην, which he derives from *4-ιτρι-σά^μτην for an earlier *4-π*ρ-σά-υ.ην, regarding it as a sigmatic 460 THE SIGMATIC AORIST, ch. xvii. 281 absolutely no by-form with /or υ, rest upon a different analogy. We shall come back to them in speaking of είπα, ήνεγκα and the like. We may just mention here that the aorist without the σ εχευα εχεα is accompanied also by a future without the σ χέω ; and in this case there is still less probability of the loss of a υ and of ttj. The preference for the intensified vowel of the stem is characteristic of the sigmatic aorist as well as the future ; and in this respect Sanskrit (Delbriick Verbum 177 ff.) often agrees with Greek. Roots ending in a vowel regularly raise their vowel : εβησα, έστησα, εισατο (Ε 538), κΚηΙσαι, υισατο (ι 213), υποκύσας, those ending in a consonant generally follow the present: ερρηζε Γ 348, επληζα, κατέτηζε τ 206, έδειξα, ήλειψαν Σ 350, εείσαο Ι 645 (from ε'ίΰομαι), ήρειψα (Herodot. Pind.), λεΐψαι from λείβω (Η 481), πεϊσαι, κατέστεψας Soph. Ο. C. 467 — ζεΰζαι, ετενϊ,εν & 338, επικενσης ο 263. The exceptions from vocalic stems are discussed below : there are hardly any to be found from consonantal stems. I may mention λάξασθαι ■ κληρώσασθαι Hesych., by the side of which we have the Herodotean future λάζομαι (vii. 144), just as there is beside λελ^ -yyuai a by-form λέλαγμαι., though a late one, and beside λήζις λάξις. The remark of Uhle ' Sprachw. Abhandl.' p. 63 holds good here : ' the intensification attaches to the root.' Even more than in the case of the perfect the influence of the present makes itself felt in the sigmatic tense-forms. C) Irregularities. The sigmatic aorist shows a few phenomena which deviate from the prevalent laws of formation ; the first two of these are to a certain extent opposed to each other, but unite on the point that the two aorists, distinguished by the grammarians by means of different numbers, in spite of their great difference, sometimes coalesce. Buttmann i. 2 404 ff. says to the same effect, ' the terminations of the 2 aor. were in some unformed dialects interchanged with those of the 1 aor.' and note 10 i in the same way the converse is sometimes found to be the case.' Buttmann explains the fact by saying that ' the aorist forms in ο ν and a 282 originally differ only as dialects.' ' The ancient language formed the aorist sometimes with, sometimes without σ, and as regards the termina- tions, sometimes in ov etc., sometimes in α etc. The usage of the language settled upon the terminations σα and ο ν except in verbs in λ μ ν ρ, but retained remains of the formations in α and σοι.' Apart from the obscure phrase ' dialects ' [Mundarten] — for no dialectical difference can be demonstrated — this view seems to me thoroughly well established. The question is one of a crossing of the different analogies. It is remarkable here that the appearance of the vowels ο and ε, which as a rule are so much the more common, in place of the regular α belongs to the earlier period, but the extension of the heavier and rarer a, in place of the ordinary thematic vowel, to the later. But the riddle is solved simply by the fact that in the post- Attic period the so-called first aorists with their α were almost the only ones in ordinary use ; and that for this reason they drew the archaic forms of the so-called second aorists over to follow their analogy. aorist of -κίρνημι. But —to say nothing of other difficulties— the imperative of 4•κριάμ•ην is, as every one knows, irplcuro : if Joh. Schmidt were right it would have to be *irplai. Hence I adhere to the view stated on p. 120. ch. xvii. SIGMATIC AORISTS WITH ο AND ε FOR a. 461 We begin with the former and more easily intelligible case : a) Sigmatic aorists with the vowels e and ο for a. In view of the extraordinary number of forms which make use of this interchange between the e and the o, the extension of this phenomenon, and its penetration into the sphere of the old a, which is much less common as the vowel of inflexion, cannot at all surprise us. In the case of the perfect we met with processes quite corresponding on p. 393 ff. If we were right in what we said on p. 442 ff. about the origin of the sigmatic aorist, and in our conjecture as to a presumable ' primarium/ the shifting of the vowel becomes still more intelligible. A form like λέζεο is to a certain extent the natural imperative form from a *λέζομαι, which we might doubtless expect on the analogy of άλέζομαι. Hence the old grammarians were in a certain sense light in taking βήσετο, ΰυσετο as imperfects from βήσομαι, Ιυσομαι (Schol. A. on A 496). But they were wrong in identifying these latter forms with the futures spelt in the same way, and in talking of a μετάγειν εις ενεστώτα. (Herodian i. p. 447). The following instances, limited for the most part to the Epic 283 language, belong here. In most cases there are also variants with α : 1) imper. άείσευ Hymn. Homer, xvii. 1. Κάστορα και Πολνδ^υκβ' aeiveo Μονσα Xiyela. 2) imper. αίετε Γ 105 (η διπλή, οτι άντϊ τον άγετε) Ω 778, £ 414, αΐεσθε θ 505. Also άζέμεναι' ενεγκείν, αξεσθαι' άγαγέσθαι Hesych. 3) imper. καταβήσεο Ε 109, επιβήσεο Ε 221, βήσετο Γ 262, εβήσετο ψ 1 (προς- Β 48, άττ- Α 428, κατ- Ζ 288, εττ- θ 44). From the scholium on Γ 262 it appears that Aristarchus preferred these forms, where the meaning was intransitive, without however introducing any changes on that account (ου μετατίθησι). In Hymn, in Apoll. i. 141 έβήσαο appears as intransitive. Hes. Scut. 338 εβήσατο with the v. 1. εβήσετο. — In the transitive sense νώ άναβησάμενοι is well established at ο 475. 4) ΰνσετο Ζ 136, κατεΖνσετο Δ 86, νπ- ζ 127, αν- Α 496. Here too we have the evidence of Didymus on Β 578 that ή ετέρα των Άρισταρ- χείων had εΰνσετο, ' και εστί γαριεστερα. 1 The participle α 24 οί μ€ν δνσομενου Ύπβρΐονος, οι δ' ανιόντος and Hes. Ορρ. 384 Ιυσομενάων (Πληιάϊων), parallel with επιτελλομενάων, and therefore quite with the force of a present. 5) \Uv Β 667, ΪΕες Hymn. Horn. ii. 45, \lov (3 plur.) Ε 773, Κ 470, Ξ 433. 6) imper. λε£εο I 617, τ 598 ; at κ 320 Aristarchus wrote λε£ο εταίρων. One M.S. has λεξεο, others λέζαι. For λέζο see above p. 131. 7) imper. οίσε χ 106, 481, Aristoph. Ran. 482, οίσέτω Τ 173, θ 255, οίσετε Γ 103, Ο 718, ν 154, οισόντων Antimachus in Athen. xi. p. 468, and also οϊσέμεναι Γ 120 (κέλενεν)=οισέμεν γ 429, while the same form in Σ 191 may be a future. An otherwise identical aorist with α occurs, in άνόΐσαι Herod, i. 157 (M.SS. ανφσαι). 462 THE SIGMATIC AORIST. ch. xyii. 8) ορσεο Γ 250, ορσευ Δ 264. ερσεο' ΰιεγείρον (cp. εμετό' ώρμηθη 9 εριηι ' όρμήστι) Hesych. differs only dialectically, and must probably bo regarded with Mor. Schmidt as Boeotian (cp. 'Ερχόμενος). 9) Imper. νελάσσετον Κ 442. αλλ' εμε μεν νυν νηυσι ν. ώκυνόροισιν. 284 10) ενεσον common from Homer onwards in all its forms among Tonic and Attic writers in poetry and prose, while the Aeolians and Dorians used the regular thematic aorist ενετό ν noticed on p. 286. ενεσον cannot well have arisen out of ενετό ν by phonetic weakening, for it is only before ι (φησί), and sporadically before ν (συ) that r passes into σ. ενεσον is therefore for *ε-νετ-σο-ν and is to ένεσα i.e. *έ-νετ-σα just as εβησετο is to εβησατο. Thus in this case the formation, which is else- where accounted anomalous, has become the rule, while the normal formation has become the exception. Hence we cannot regard ένεσα, as Buttmann Ausf. Gr. ii. 2 278 rightly saw, as parallel to έλαβαν and other Alexandrian forms. M.S traces of the forms belong- ing here are found in νέσειε Eurip. Ale. 464 Dind. (cp. Kirchhoff), altered into νέσοι, νροςένεσα Eur. Troad. 292, altered into νροςένεσον (cp. Veitch p. 478), ένεσαν Herod, i. 21 (only the Aldine), ενενέσαμεν Aeschin. ii. 176. Considering the very frequent occurrence of the other formation it would certainly be rash to regard these as more than the blunders of copyists. Lobeck ad Phryn. 724 quotes forms of the kind from later writers (Sext. Empir., Achill. Tatius, N. Test.). 11) εχεσον, quoted by Herodian ii. p. 801 from the Ganymede of the comic poet Alcaeus : κατέχεσον [τής] Νηρηΐΰος. Cp. Meineke Comici ii. p. 826. Buttmann supports his view of ενεσον mainly by this parallel, and in fact: ενεσον (i.e. ε-νετ-σο-ν) : νεσονμαι ',', εχεσον (i.e. ε-χεΰ-σο -r) : χεσονμαι. Here there is abundant evidence for forms with α like έχεσα , χέσαψι, χέσαι, and no one attacks their normal character. b) Unsigmatic aorists with a. To these belong only two forms of the classic period : εΊνα and ήνεγκα, and from Homer onwards they are constantly interchanged with the regular formations είνον and ήνεγκον. The occurrence of the different formations is treated of thoroughly by La Roche Ztschr. f. d. osterr. Gymn. 1872 p. 125 ff. Here the following remarks will suffice : εϊνον (from ε -h-hvo-v, cp. above p. 291) has prevalently the regular vowels of the thematic aorist. By the side of these a establishes itself first in the 2 sing. ind. and 2 plur. imper. : εΤπ-ας A 106, 108, έΊνατε γ 427, φ 198. Herodotus, in whom middle forms like ανείνατο are 285 common, and the forms είνα (iv. 44), είναν, ε'ίνας, είπαί, elsewhere very rare, are well established, is the writer most fond of the α in this stem [cp. Bredow Dial. Herod, p. 324 sq., 353 sq.]. In the Attic winters the forms είπ-ας, είνατον, είνατε, ειπάτω may be regarded as the most common. Of course there is hardly any instance in which the M.SS. do not vary. The α has a much wider extension in ήνεγκον. Homer has the forms ανενεικας !£ 255, ενείκαμεν ω 43, ήνεικαν c 784, opt. ενείκαι Σ 147, imper.. ενείκατε θ 393, inf. ενεϊκαι Σ 334, partic. ένείκας Ρ 39, mid. άνενείκατο Τ 314, ηνείκαντο Ι 127. The α is also shown to be Doric by inscriptions ch. χνπ. AORISTS IN a WITHOUT THE σ. 463 ήνεγκα, εζενεγκαι, ενεγκάμενος (Ahrens p. 352). In Attic writers the α is prevalent in the indicative and imperative (ε νεγκάτω, ενέγκατε) of the active and in the whole middle. The greatest variation is shown in the optative, while in the infinitive and participle thematic forms are used. We find however in C. I. A. ii. 162, a, 4 ίνεγκασών. Herodotus here furnishes nothing remarkable with the exception of the diphthong ει, common to him with Homer and occurring also in Pindar and Theocritus. Joh. Schmidt is doubtless right in explaining this ει from the influence of the disappearing nasal (Yocal. i. 122 f.). In these two widely extended aorists we evidently cannot suppose the existence of a σ. Not only would it be quite unprecedented phonetically for forms like *εΙπ-σα, *ενεγκ-σα to change into είπα, ενεγκα, but there would not be the slightest internal probability for such forms in themselves, ειπον and ήνεγκον being reduplicated aorists, in which the notion could not occur to any one of conjecturing a smuggled σ. It is therefore especially clear here that the α is acting as substitute, so to speak, for ε and o. There is no alternative but to suppose that, at the time when the vowels were, so to speak, still in flux, in some thematic aorists the α resisted the general tendency to follow the ordinary change of vowels, as has regularly happened in the active perfect, and that είπα and ηνεγκα are relics of what we may conjecture to have been a large number of such archaic creations. The preservation of the α must have been favoured by the very large number of sigmatic aorists, and especially of those of 'suppletory' formation like ήγγειλα, ίνειμα, while conversely the phonetic changes of the stem which came about in είπον and ήνεγκον had made these aorists very much unlike their nearest kin, such as 286 ήγαγον, ελαβον. The confusion of the analogy of the two aorist forma- tions is a remarkable proof that the instinct of the Greeks regarded the aorist as a syntactic unity. Under these considerations it will be best to view in the same way the aorists with a υ or } in the root discussed on p. 459. Beside a present stem koFjo a second theme καΡο might establish itself, and this then, losing the character it had in common with a present indicative, became an aorist theme, and as such, on the analogy of the sigmatic aorist, became καΡα. The difference from forms like είπα, ήνεγκα is only this, that in the case of the latter we find by-forms with the ordinary vocalism, in the case of the former we do not. All other words belonging here rest either on doubtful or on late authority; thus άγάγας' αντί τον άγαγών, and ayayov ' αντί του αγαγε, ο^ηγησον 1 φέρε Hesych. εττανρασθαι, επηνρατο Aristot. Eth. Nic. v. p. 1163 a 20 ; similar forms in Hippocrates. είς-έΰρακα Orph. Argonaut, v. 133. είδα, πρώτα €>ε ε}<ϋα βίην Ήρακλήος θείοω ib.v. 119, ε'ίΰαμεν Ν. Τ. [and LXX; cf. Moulton's Winer p. 86]. άφ-έλαι C. I. 2557, 26, on "a Cretan inscription by no means very archaic, where Boeckh perhaps rightly writes άφελίν. On the other hand Hesych. gives ε'ίλατο (also Ν. T. [2 Thess. ii. 13]), εζείλατο [Acts vii. 10, xii. 11], άφείλαντο, forms which Phrynichus p. 183 [cp. Lobeck's note] rejects as un- Attic. In the late poem, which is full of strange and erroneous forms, Append. Anthol. 257=0. I. 3272 we actually find beside ε'ίλατο v. 5 ειλάμενος v. 9, which could only be a regular forma- 464 THE SIGMATIC AORIST. ch. xvit. tion under the hypothesis just stated, and must otherwise be a mis- formation. έλαβαν, εΚάβαμεν. έγκατεΧίπατε, έγκατέλπταν. ενραν. ενρασθαι is rejected by Phrynichus p. 139. εφάγαμεν, έφαναν. έφνγαν. 287 ήλθαμεν, έΧθάτω, εΧθατε (and also επηλνθα Anthol. P. xiv. 44) occur almost exclusively in the LXX and Ν. T., but with many variants. References are given by Sturz de dial. Alex. p. 60 sq. and Alex. Buttmann, Grammar of New Testament Greek p. 39 f. Three of these forms might possibly, on account of the liquid in the stem, be regular aorists of the 'suppletory ' formation, viz. έπανρασθαι, άφείΧατο, ενρασθαι The case of ώσφραντο (Herod, i. 80, v. 1. οσφραντο) beside ώσφροντο (Aristoph. Ach. 179) is a special one, because the whole verb is quite unique (cp. above p. 286). Here the form with α has earlier authority for it than the other. c) Aorists in -κα. These aorists in spite of their small number form an important link in the chain of the verbal forms. Evidently the vowel of these forma- tions is brought into quite a different light, now that we have met it elsewhere also ; and thus forms like ε-οωκα, έ-θηκα approach on the one hand nearer to the perfects with which we compared them on p. 410 f., and on the other to the unique aor. ε-πτα-κο -vivom the rt. τττα preserved νηε-πτή- την. These aorists are based upon a verbal stem characterised by the suffix -ha, with the retention of the ancient a. Setting aside the quantity of the middle syllable we can state the following equation of relations e -δω-κα : e -πτα-κο-ν '.'. ewra : euro-v '.*. δεδοικα : Syracus. δεδοικω. We know of five aorists in -κα, of which the three έδωκα (quite late εοωκάμην), ηκα (in Homer also εηκα : cp. above p. 80; (προς)ηκάμψ> first in Eur. El. 622), έθηκα (θηκατο Κ 31, other middle forms in Herod. Pind.) are common in all Greek, but chiefly in the singular and the 3 plur. of the indicative, though ενηκαμεν occurs as early as μ 401. The forms with κ therefore are interchanged with the primitive εΰομεν, έθετε, εΐην etc. precisely as in the perfect. There is further εττακαν έστησαν Hesych., which is certainly rightly regarded as a Boeotian or Laconian modification of ε-σπχ-κα-ν (Ahrens Dor. 103). A Boeotian analogy for ττ•=στ is supplied by εττε=εςτε until (Ahrens Aeol. 177), a Laconian by βεττόν=:εστόν. There is no reason for altering εττακαν into εστασαν, as Ahrens proposes to do, except the unique character of the form. — 288 There is also εφρηκα, preserved in Eurip. El. 1033 επειςέφρηκε, είς- έφρηκεν (M.S. είςέφρικεν)' εϊςεπέΰησεν, είσαφηκεν, εζέφρηκεν' άφηκεΐ' Hesych., a form in which Nauck finds the chief support for the view that the verbal stem φρε is based upon a coalescence of προ and ε. For the incorrectness of this view cp. Stud. viii. p. 327 ff. — Savelsberg Ztschr. xvi. 420 thinks he has discovered another instance in the Cretan άπέ- σταλκαΐ'. But on p. 385 we took this form as a perfect. And certainly ch. xvii. AORISTS IN tea. 465 the mere fact that in a similar passage in another Cretan inscription we find άπέστηλαν cannot suffice to prove that the other form is an aorist. It would be better to support this view by the participle άπεστύλκαιτες C. I. G. 3047, 2. But this form would be by reason of its ε such a marvellous hybrid between aorist and perfect that we hold, especially as a participle does not at all suit the context and as the copy is but poorly vouched for, that Boeckh was quite light in regarding it as a blunder for άπ(στΐί\καν. Savelsberg, following a conjecture of Bopp's, at the place referred to, in Ztschr. xvi. 54 ff. and 401 ff and before that in the Symbola phi- lologorum Bonnensium ii. 503 ff., tried with much learning to give a demonstration that the κ of these aorists has come from σ. I do not think that he has made many converts to his view, for a transition from the dental spirant σ into the guttural explosive, which is in its nature so absolutely unlike it, is as improbable in itself, in spite of the at- tempt to find intermediate forms, as it is unexampled j and besides, one can hardly conceive why the sibilant should have been preserved in many hundreds of common aorist forms, but in a few have been metamorphosed into r. But this zealous attempt has not been wholly without fruit, inasmuch as a number of little-noticed sigmatic aorist forms from the roots t)o, t, βε have been brought to light. It is true that much which Savelsberg brings in here is doubtful. For Homer especially I regard sigmatic aorists of these stems as not established. But in the Attic in- scription published by Rangabe Antiqu. Hell. no. 869, 17 we have άττο- οοσάιτωι, ib. 875, 5 άιαθέσαιτες. Again the unique form απυΰόας in the Arcadian inscription of Tegea 1. 13 can hardly be explained except as from άπυ-Ιόσας ; and we should have to assume an *ε£οα on the analogy of έΥ^α, εχευα. From Alexandrian and Byzantine Greek 289 Lobeck ad Phryn. p. 721 quotes forms like Μισής, ΰώσης, which occur also especially in scholiasts {κροσβησης, ίπιθήση) and accordingly we read in Coluthus v. 25 (Lennep) πωεα καλά μεβησας. From the rt. φρε be- sides the previously mentioned επειςέφυηκα the form with σ occurs even in Attic dramatists, e.g. Eur. Here. Fur. 1267έπέιςέφρησ(, though Nauck will not allow it to stared theie. Hesychius. gives καΰεσάμενο^' χαλάσας, βέσαι' θησαυρίσαι. d) Isolated forms. εισα (κάί)εσσαν Pind.), ε'ισάμην (ϊφέσσατο £ 529), mentioned because of the augment on p. 85, is unique only because the rt. el• from which it proceeded, does not occur elsewhere in the active. Beally ε\σα is not farther removed from ΐζρηψ than έστησα from 'ίσταμαι. Even Butt- mann recognized this i. 2 524, but he wrongly connected ημαι with these forms, for which see p. 103. τόσσαις Aeolic participle Pind. Pyth. 3, 27, επιτόσσαις ib. 10, 33, επέτασσε ib. 4, 25. As the meaning quite agrees with τνχεΊτ, it is pro- bable that the root of the two verbs is identical, and also that of the kindred forms τΰζον, τέκμαρ (Princ. i. 271). But the aorist form re- mains obscure. Finally in certain sigmatic aorists reduplication makes its appearance. There can hardly be more than two of them, and we may certainly assume that this strengthening of the stem, as comes out clearly in the II II 466 THE SIGMATIC AORIST. CH. XVII. second example, passed to these aorists from other forms of the verbs in question, and was not created for them : τεθησατο' εθηλάσατο Hesych. Also in π-θή-νη, τί-τθη (Princ. i. 312) we find reduplication. τέτρηνα X 396, φ 198, afterwards rerpara from the present rcrpairm (Herod. Aesch.). The reduplication attaches to the whole verb. c H xviii. THE FUTUKE. 467 CHAPTER XVIII. 290 THE FUTURE. The perfect has been discovered to be a kind of a present, individualised only by degrees, though long before the Greek language acquired its distinctive form. With still greater positiveness we may maintain that the future also is nothing but a present form. In the latter tense this view is pretty generally recognized, and it finds the most unmistakeable support in facts not merely of the cognate languages, but also of Greek itself. It is well known that in Gothic and Old High• German the in- dicative present is often used without any distinction with a future force. In the Slavonic languages ' the present of the verba perfectiva denotes the future. The present force is thrust into the background, and ap- pears in certain cases almost as an exception ' (Miklosich Vergl. Gr. der Slav.Sprachen iv. 77 2). 1 In the same way the Ch.-Sl. bada and the Anglo-Saxon beo, I shall be, take no distinctive sign to express the future. In Greek the employment of certain present forms, characterised by no distinctive mark, with the force of a future, is an uncontested fact, el μι has acquired its future meaning only through usage and by de- grees. . Even the indicative retains iu Homer occasionally e.g. Β 87 . (ήντε εθνεα εΤσι), II 160 (άγίΧηΰον "ιασιν), more rarely in Attic writers e.g. Thuc. iv. 61 (επίασιν) the original present meaning. In the other moods and in the verbal nouns, as every one knows, this never quite disappeared. We see therefore that the case is precisely the same with these Greek present-futures as with the Slavonic presents of the per- fectiva ; the present meaning is only to a certain extent ' thrust into the background.' While έ?ω occurs often enough in poets as a present, the similarly formed middle εΰομαι has from Homer onwards (e.g. Σ 271, ι 369 Qvriv εγώ πχψατον εΰομαι) exclusively the force of a future, νίομάϊ is found in Pind. 01. vi. 86 (τάς ερατεινόν νΰωρ πίαμαι) as decidedly a present. Hesychius gives the active of it καταπιεί' καταπίνει. 2 Else- 291 where πίομαι is future from Homer onwards : Ν 493 ώς ει τε μετά κτίΧον ϊΰπετο μήΧα πιόμεν εκ βυτάνης. A fourth precisely similar example Of the kind φάγομαι occurs first in the Hellenistic period. References to the LXX and Ν. T. are given by Veitch p. 246. With these we may place the two Homeric present-futures cY /ω and κείω (by-form κέω) e.g. I 685 επει ουκέτι Ζήετε τέκμωρ Ίλί'ου αιπειιής, Α 606 οι μεν κακκείοντες εβαν, η 342 ορσο κέων ώ Ιιϊνε. Buttmann wished to explain these forms as contracted from the regular futures, h)w from *ut £, and the assumed intermediate forms are wholly without analogy. Both are thematic present forms with an intensified stem-syllable. Cp. Princ. i. 178, 285. — For the quite similar Homeric βιίομαι with the by- i'orms βέομαι and βίομαι even Buttmann attempted no explanation of the kind. The future force is unmistakeable in passages like X 431 τι w βίίομαι αινα παθοΰσα Ο 194 τω pa και ου τι Διο? βέομαι φρατίν Hyinn. m Apoll. Py th. 350 πώς κα\ νυν βιόμεσθα ; Finally there is still the quite "unique αναΐράμεται in the epigram of Philippus Anthol. Pal. ix. 575 κα\ veKVS cis ζωών χώρον άναδράμ^ται. We should have a future perfect of a similar stamp in εκγεγάοιται Hymn, in Yen. 197, if it is not incorrectly recorded, as appeared to us probable on p. 417. Whether other formations, especially some Homeric ones, are to be placed with these, or whether we are rather to assume for them the loss of the sigma, will have to be considered hereafter. But in passing we may call to mind another way of denoting the future without any distinctive mark, that by means of the conjunctive. In Homer the use of the conjunctive, e.g. in υνπω 'i. The fol- lowing forms are on record : A) Active. 1 sing. βοαΟησίω C. I. no. 2554, 191, ί]πιτραψίω inscr. of Lyttus 13 (Hermes iv. 267), κακυτε~\χ>>ησίω ib. 12, σπευσίω inscr. of Drerus 42, all Cretan: we m;ty add Ιιτι•ψίω' ίντηάίω Hesych. — Of the uncontracted forms in -σέω we have only a few Delphic examples : ίκπραξέ* C. I. 1688, 5, (ιρκιίίω ib. 13. Contracted forms like οωσώ, ιυησώ are mentioned by the old grammarians as regular in Doric : οι Δωριείς τυνς bptwrua ύς με\- λοιτας περισπώσι Anecd. Oxon. iv. 198, hence Ahrens is certainly right in circumilexing the numerous futures on the (comparatively recent 1) ch. xviii. . THE SIGMATIC FUTURE. 469 Cretan inscription no. 2555 : εΐώ, ίπιτραψω etc., and so in Aristoph. Ach. 739 φασώ, 747 καμυζώ, Theocr. V. 142 καχαΕώ. 2 sing. For this person only forms like είεϊς, ΐωσεΊς are known, and similarly for the 3 sing., only those in εΐ, of which 11 are found on the Heracl ean 293 tables alone (Meister, Stud. iv. 430) άποτεισεΊ i. 109 (also Drer. 161), vrevffe~i 114, kiti 130 etc, 1 plur." Cretan ΐ.ίακαθεΐίομεν (Helhig de dial. Cret. 26), πραζίημεν C. I. 3048, 15, σνιδιαψυλαζίομεν ih. 31, 3058, 11. — οισεϋμες Theocr. xv. 133. 2 plur. For the severe Doric dialects forms in -σήτε would be ex- pected, but the only ones recorded are mild Doric like ΰοζεΊτε, ησεϊτε Ar. Ach. 741, 747. 3 plur. Cretan βοα\3ησίοντι βοαβησίοντι (inscr. edited by Bergmann 1. 15), Heracl. άπαζόντι i. 102, άποκαταστα σόντι i. 149, είόντι i. 120 etc. (Meister, Stud. iv. 430), mild Doric ΖιαΧυσεϋιτι, υπαρίενντι C. I. 2671, 1. 34, 49, and Ther. τταρ^Ιοϋντι ib. 2448, iv. 1. 32, νησοϋντι Sophron 19, ευρησοϋντι Epicharm. 92 Ahr., αυλησεϋντι Theocr. vii. 71. The active infinitives occur in the Cretan επιτραφήν, ν&εΐήν (Berg- mann's inscr. 12, 14, 70, 85), Delphic (C. I. 1688) άπογρα\Ιέν. — As participial forms we may quote the Megarian άγορασυϋντες Ar. Ach. 750, and the Cretan πρεσβευσόντας C. I. 2557 B. 4. B) Middle. For the 1 sing, we do not find the severe Doric -σίομαι which would be expected. We have forms like φσεϋμαι Theocr. iii. 38, βαιτεϋμαι ib. ii. 8, πεφασοϋμαι (Ahr. 217). 2 sing. βονκοΚιαίϊι Theocr. v. 44, Xaajj i. 4 etc. 3 sing. Heracl. εργαζηται i. 168, έγδικαίήται i. 130, καρττευσηται i. 159, Meister Stud. iv. 430. — Mild Doric ώνασείται Sophron 89, εσσείται Archimedes (Ahrens Dor. p. 203), Theocr. vii. 67. 1 plur. Cretan χηριζώμεθα C. I. 3048, 1. 36, Le Bas Inscriptions Grecques et Latines Tome iii. Partie v. no. 74, 1. 16. — ΰησεΰμεσΰ Theocr. viii. 13. — θωσοΰμεΟ' Epich. 167. 2 plur. Megarian πειρασεΐσθε Ar. Ach. 743, ΧωβασεΊσθε Theocr. v. 109. 3 plur. Ιψίοιτηι* άκοΧουθησουσιν Hesych. — Heracl. εργάζονται, εσ- <ϊόντ«ι i. 112, επιμεΧησόιται 119, υπογραψόιται 149. — βασεϋιται Theocr. iv. 26. — δείοΰνται Theraean inscription no. 2448, v. 12, έσσούιται Argive 294 and Laconian treaty in Thuc. v. 79. There are also infinitives like εσσείσθαι Sophr. 23. — In the accentua- tion of the forms I have substantially followed Ahrens : on this some stress must be laid, especially as regards the Heraclean third persons plural in -σόντι, -σόνται. For these forms are only recorded on inscrip- tions ; hence the accentuation of the penultimate is based exclusively on the hypothesis — not an improbable one — that the Dorians here still retained at any rate in the accentuation some remembrance of the vowel once present after the σ, which elsewhere they so carefully preserved. Even in syllables not long by position the short vowel sometimes appears: (}γ£&.~βι>α.ΰηηόμεν, ■χαριίόμεβα (πραίόμεν Helbig p. 27). We should in these cases assume without hesitation a transition into the 470 THE FUTURE. ch. xviii. Attic method of formation, did not τελόμαι=τε\ονμαι (Drer. 63) show us that in this dialect ο might be the remains of so. Outside the Doric dialect the futures of this stamp appear as the so-called Futura Dorica, exclusively with middle endings, and in only small numbers. The two Homeric instances are marked with *. 1) * έσσείται only Β 393, Ν 317, with απεσσεϊτίΐι τ 302, while else- where in Homer εσσεται, εσεται, εσται, εσπομένοισι etc. are common. There is also in Hesiod Opp. 503 ok αιεϊ θέρος εσηείται. 2) κληνσυύμεβα only Aristoph. Pax 1081, while κλαυσομαι etc. are common from Homer onwards (X 87) and established by the metre, e.g. Ar. Nub. 58. ' 3) νευσονμενοι only Xen. Anab. iv. 3, 12, where the more recent editors have adopted νευσόμενοι. In the gloss of Hesychius νευσημεθα * νηΖόμεβα there is certainly no reason for this alteration. 4) *πεσέυΐ'ται Λ 824, πεσέεται Herod, vii. 168, πεσονμαι the only future form of πίπτω in ordinary use from Aeschylus onwards. From what was said as to επεσον on p. 462. it results that the word must be divided πε(τ)-σέο-μαι, not, as some might think, τεσ-ίο-μαι. 5) ιτενσΓισθαι, the reading of the Med. and other M.SS. in Aesch. Prom. 988, retained by G. Hermann, but changed by Dindorf and Weil into πενσεσβαι., which is common from Homer onwards (ψ 262). πεύ- σεται Aesch. Choeph. 765. 6) πλενσονμεβα Thuc. i. 143, ττλενσΓισθαι viii. 1, and similar forms 295 also in the orators (Lys. xiii. 25, Demosth. lvi. 6), while by the side of it πλεΰσομαι everywhere occurs (even in μ 25). 7) πνενσεϊται found in the M.SS. in Aristoph. Ranae 1221, altered by Dindorf into πιευσεται, both being metrically possible. The con- tracted form in Aristotle Meteor, ii. 8 (p. 367, a, 13) πνευ'σε~ισ\)αι. έμτηεύσομαι Eurip. Andr. 555. 8) ρενσεΊται Aristot. Meteor, ii. 4 (p. 361, a, 33), pevaovvrat ib. 2 (p. 356, a, 16). On the other hand ρενσεται Theogn. 448, ρεΰσονται. Eurip. fr. 388 Dind. 9) φενζούμεθα established by the verse in Eur. Hel. 500, 1041, Aristoph. Plut. 447, ψευζοΰμενυν Ach. 1129, while Dindoif Eur. Bacch. 798 writes φενζεσθε for the φενζεΊσθε of the M.SS. In prose writers too εκφευζίϊσβαι and the like, e.g. Plato Rep. iv. 432 d., have been retained in our texts. By the side of it ψενξυμαι is in well-established use in Homer and Attic writers (Σ 307, Aesch. Suppl. 456). 10) χεσυϋμαι the only future form in use from χέζω, e.g. Ar. Vesp. 941. These ten remarkable exceptions 3 to a rule firmly based upon thousands of instances give us the impression of having maintained themselves in popular usage from an ancient date, all the more so that the verbs to which they belong are very common. We certainly cannot suppose that there was any borrowing from the Dorians. These forms show us rather that in the non-Dorian dialects it was only by degrees * An eleventh form of the kind, generally placed in the list with these, ταιξουνται Xen. Conv. 9, 2, is justly noted by Cobet Novae Lectiones p. 634 as nn- Attic, the words being spoken by a Syracusan. — A twelfth T«£«iea0« Arat. Phaen. ) 24 is very extraordinary, and for that reason suspected by Buttmann (Ausf. Gr. i.* H90) : if it is correctly recorded, we can only understand if as an imitative lengthening of *τφ'€σ06, τ«ξ«ίσ0«. ch.xviii. THE DORIC FUTURE. 471 that the shorter formation took its place by the side of the fuller. These exceptions support the view established by comparative grammar, according to which the Doric future preserved most completely the original elements of this tense. This view (Bopp Yergl. Gr. ii. § 648 ff., Schleicher Comp. 3 807, Joh. Schmidt ' La formation des future,' Revue de Linguistique 1870) is based • upon the comparison of Sanskrit, Zend, Lithuanian and Slavonic. The Sanskrit future in -sjd-mi e.g. dd-sjd-mi=Oor. Ιω-σίω agrees exactly 296 with the Doric form. The vocalisation of the j in other cases produced f, just as in the Homeric κενεό-ς from the primitive form Kper-jo -ς, which comes very near to the Skt. cunja-s for fyvan-ja-s. For these phonetic processes it is sufficient to refer to Princ. ii. 239 f. The future of Zend comes still nearer to the Greek. The termination -mi is here wanting in the 1 sing, vakh-shyd ( = Skt. vakshjdmi) from the rt. vac speak, would quite correspond to a Doric *ϊεπ-σιω (from εΊπον), the ace. sing, of the participle bushy ant-em to a *φν-σιοντ-α. In Zend there is even an example of the fut. middle participle, the gen. plur. zahyamnam (ca), where hy appears as the representative of shy. If we imagine a form *γενσομαι formed on the analogy of φύρσω as the future of the root γεν, the genitive plural might be translated into the ' rough Greek ' [Princ. i. 19] by *γενσομένων. Schleicher Comp. 3 806 speaks also of Zend futures ' with a dropped j,' which would answer in a still higher degree to the Attic futures. But these forms, e.g. άαοή/ιά=()ώσω, are better taken as conjunctives with the force of a future (cp. Jolly, Ein Capitel vergl. Syntax p. 38). The ordinary Lithuanian future in -siu, e.g. bu-siu (from bundu watch) has preserved the spirant before the u of the 1 sing, throughout in the form of the vowel i, while in other personal forms the syllable sja is shortened to si, by which e.g. bu-si-te becomes much like an Attic φΰ-σε-τε. Entirely isolated traces of a similar formation have been adduced from Church-Slavonic by Schleicher and Joh. Schmidt U.S. If we keep all these facts before our eyes, it becomes extremely probable, that the future in -σω is not a formation differing in principle from the Doric, but one proceeding from the same primitive form by the way of phonetic weakening. We shall be able to represent to ourselves the course of the phonetic change with most probability in the follow- ing way : From the Indo-Germanic primitive form e.g. dd-sjd-mi came as the Greek primitive form *£ώ-σ;ω. The σ in this form had the sharper pronunciation, which it possessed always before consonants. At the time when the spirant j began to be disappearing, it underwent a two- fold change, on the one hand being vocalised, the j becoming sometimes t, sometimes ε, which finally survived only in contraction, and on the 297 other being altogether lost. The former method of treatment was the prevalent one among the Dorians, though it was not wholly unknown to the other stocks, the latter in the remaining dialects. But throughout, even before vowels, the sibilant preserved the sharper pronunciation, which protected it to a large extent from passing into an aspirate. . The statement of the case here given differs somewhat from that which is to be found in my Tempora u. Modi p. 312. There with regard to a part of the forms here under consideration, I laid stress upon the double σ, which the Homeric dialect gives in forms like άγάσσεσθαι 472 THE FUTURE. CH. XVITT. 3 181, (πΐίσσομαι £ 388, ελάσσω Ψ 427, όλέσσω Μ 250, όνόσσεται Ι 55. I accepted the explanation of Bopp, who explains the double σ on numerous incontestable analogies, from assimilation, ελά-σσω would thus come from ελα-σ/ω, just as the Prakrit kar-i-ssadi=Skt. kar-i- shjati (he will make) has come from *kar-i-sja~ti. In following up this theory, which cannot be attacked from the point of view of the history of sounds alone, it was very natural to derive the simple σ of the future throughout from sj through the intermediate stage of σσ. We might even attempt to account in this way for the stubbornness with which the sibilant maintains itself even between vowels. But against this an insuperable objection is raised from the side of the Doric future, for the σ of the Doric Ιω-σίω, which certainly did not come from σσ, has just as much vital force as that of δώσω, which conceivably might have originated in σσ. Hence the reason for the vital force of the sibilant cannot be found in the swallowing up of the j. But there are also other objections to this view, raised by Leskien Stud. ii. 81 if. The double σ is even more common in aorists like άγάσσασβαι, όλέσσαι etc., where we cannot suppose the existence of sj, than in futures. It is evidently most closely connected with the n, which appears in the perfect middle, in the passive aorist in θ η, in the verbal adjectives, and in many nominal forms, and which will occupy us in a subsequent chapter. The j seems to have been lost not by assimilation, but through the intermediate stage of an irrational vowel, lying halfway between e and i. Bopp § 656 well 298 compares the 0. H. G. krefti-o, gen. plur. of the stem krefti {Kraft, strength) with its by forms krefteo and krefto. We may also compare Greek forms like the Homeric κενός by the side of κενεόν, ΰς beside εός, Ion. ορτή beside εορτή and other phenomena of the hyphaeresis discussed by Fritsch Stud. vi. 87. From this manner of considering the question it is perhaps still easier to understand how it comes about, that the ε is retained at least sporadically outside of the Doric dialect. Now that we have, while reserving for the present the so-called futurum secundum, established, as I believe, the unity of the whole Greek future formation, we must enter upon the origin of the form. The almost universally adopted doctrine of comparative grammar is that in the syllable sja, which characterises the future, the s belongs to the verb substantive, and hence is identical with the s of the sigmatic aorist, while the ja which remains is the exponent of the future meaning. Thus the future is held to be a doubly compounded tense, as compared with• the simply compounded sigmatic aorist. With respect to the manner of regarding these elements, and their original function, there are two different shades of the same fundamental view. Bopp in the future termination -sjd-mi brought out mainly its relationship with the potential of the rt. as, sjd-m extant in Sanskrit (Vergl. Gr. ii. § 648). Following him I went so far in the Tempora u. Modi p. 317 as actually to derive the termination -sjd-mifrom the optative potential {a)s-jd-m, and to regard the primary endings proper to the future as a later modification of the secondary endings belonging to the optative. The latter view is erroneous and cannot be supported by any analogy. Hence I have withdrawn it already in my ' Chronologic ' 2 p. 60, 63, and adopted the slightly differing analysis of the future, which has been put forth by. Benfey (Kurze Sktgr. § 304) and Schleicher (Compend. 3 803) and carried out by Joh. Schmidt 1. c. According to this the future is a compound ch. χνπι. ORIGIN OF THE SIGMATIC FORM. 473 present form, the first element of which is a verbal, root, while the second element is the present form (a)s-jd-mi, that is to say a present from the rt. as be, formed after the fourth or i-class. This explanation seems to be indubitable. But still with respect to the way of regarding the different elements here united, and their function, there are still some 999 differences between particular scholars. Schleicher holds the future force of the present form as-jdmi, which is added to the root, to be some- thing, so to speak, casual, though he identifies it, as had long been recognized, with the Lat. ero. He calls as-jd-mi 'a present form, which like so many present stems in Indo-Germanic has a future force.' Joh. Schmidt expresses himself still more positively to the same effect. Both consider futurity to be as little denoted in dd-sjd-mi as in έπομαι or πίομαι mentioned on p. 467. Benfey on the other hand takes as-jd-mi itself as a compound of the rt. as with the rt. jd go, so that he translates it by ' Ich gehe sein,' comparing the French use of je vais and the English of / am going to. The second view has decided advantages over the first. First of all, it replaces mere chance by a link of causation. According to Schleicher and Joh. Schmidt any other present formation might have been employed just as well as this to mark the future ; according to Benfey r whose view approaches that of Bopp, the reason for the choice of this present formation lies in the fact that its meaning was especially adapted to such an application. Hence it is no longer mere chance that two presents compounded with ja, ero=*esjo, and (though this is pre- served only in composition) -bo for *bio had a future force among the Romans. The view of Schleicher and Schmidt is really supported only by the fact that there are futures in which there is nothing to denote futurity. But from this fact nothing follows but that under some cir- cumstances futurity can remain undenoted, not that it always must so remain. The notion of the past is often undenoted in the historical present. Does it follow from this that the augment, the exponent of the notion of the past, acquires this function merely accidentally 1 ? But besides much weight must be attached to the fact that the termination of the future coincides with that of the optative. We thought above p. 325, that w.e could trace the optative syllable ja also back to the rt. ja go, and concluded from the extant traces of primary endings in this mood, that the optatives had once ended in -jd-mi in the 1 sing., and that consequently that from the rt. as was once as-jd-mi. Now the consistent carrying out of the view of Schleicher and Schmidt would 300 lead to this, that we should have to explain the optative force of this 1 present form ' as also something purely accidental. But we can hardly explain the multiplicity of the present formations otherwise than from the pressing need of denoting different sides of continuous action. The form in -jd-mi must therefore have also had originally some such special force, though this afterwards disappeared, and if we have before us in the rt. ja a verbal root, which was perfectly adapted to denote intended action, it is highly probable that we must recognize in composition with this root the starting point of the functions alike of the future and of the optative. Of course the future must then have arisen at a period in the life of language in which ja had not faded away into a mere present element to the extent to which it did afterwards. The distinction between this later-born tense and the mood which was probably earlier developed, was well provided for by employing for the future not the simple rt. ja, 474 . THE FUTITEE. CH. XVIII. but only the rt. compounded with as. Nothing hinders us from assuming, that at that period as-jd-mi was surviving also as used independently with the meaning ' I am going to be/ * I am becoming ' ; and that the Lat. ero with its firmly established future force is an inheritance from this ancient time. Sonne Ztschr. xii. 343 is the only scholar, so far as I know, who, within the sphere of comparative grammar, has attempted to give an explanation of the future differing' in principle from the analysis just stated : and this, after the fashion peculiar to this acute but somewhat audacious investigator, is supported with only a few words. Sonne finds it surprising that ' the future characteristic ja should have occurred originally only after the rt. as. 1 This objection is met, if we regard all presents in -jd-mi as parallel to as-jd-mi, and take the syllable ja, by no means as marked from the first with the character of the future, but only as a present expansive especially adapted to be employed for this purpose. He himself sees with Benfey, and in agreement with the view stated above, in the syllable -ja the rt. ja go, but is of the opinion that 301 this is compounded not with the rt. as be, but with nominal stems in -as, which he calls infinitives. He divides e.g. the Sanskrit form bhavish-jd-mi, I shall be, tracing it back to bhavas-jd-mi, into the * infinitive ' bhavas being, and jd-mi I go. According to this view all Greek and Lithuanian futures, and a very large portion of the Sanskrit futures also, e.g. dd-sjd-mi, dikshjd-mi, must have suffered syncope, and all the rest a weakening from a to *, and all many other phonetic changes besides. This is quite enough to upset this explanation, all the ' more so that we do not discover anywhere else in the neuters in as Gk. ες (nom. ος), Lat. es, os (nom. us) any tendency whatever to drop the vowel before the s. Besides, apart from the Latin infinitives in re, where however there are difficulties still remaining, there is no instance in which nouns of tlfis kind in as were used as infinitives. Sonne's view has therefore justly met with no assent. After determining the origin of the sigmatic future, it would be proper for us, as hitherto only the Doric forms have been separately quoted, to discuss more in detail the ordinary formations. But as the future is one of the tense- forms universally in use, to be expected from every verb, and is formed with the greatest regularity from the most different stems, there would be no sense in quoting here a multitude of examples. We may rather, as in the case of the sigmatic aorist, renounce altogether any such enumeration. What is otherwise noteworthy in the form which the stem takes as regards both consonants and vowels, the future shares almost altogether with this aorist. I may therefore refer on all these points to the preceding chapter. The relations of the con- sonants are there discussed p. 452 ff. where forms like βάζω, εγγναλίΕει, φΰρσω etc. find their explanation : those of the vowels p. 457 f. In the preference shown for intensified vowels of the stem the Greek and the Indian future coincide e.g. bhot-sjd mi (rt. budh awake) compared with Gk. πενσομαι (rt. πνθ), (je-shjd-mi (rt. , Ιερΰ>, τεμώ, <\π<>- κριι-οϋμαι, and from derived verbs like άγαλώ, καθαρά, ίσχί'πιώ, τεχμαρ- 304 οϋμαι, οικτερώ, need no examples. Such futures belonged to the stores of the Attic writers in daily use quite as much as the sigmatic. Hereby the so-called futura secunda distinguish themselves essentially from the other ' tempora secunda.' We proceed now from the demonstration of the facts to their explanation. The futures in -εω at first sight differ enough from those in -σω to justify the attempt to separate them completely one from the other. This attempt has been made by Hugo Weber in the Philologus Yol. xvi. (1860) p. 694 with the support of Yoretzsch de inscr. Cretensi p. 29. Weber's view, with which Benfey ' Entstehung des Optativs ' p.62 agrees, proceeds upon the notion that the futures in the Dor. -ι<•», Ion. -ε ω, without having lost a sigma, were formed by the addition of the rt. jd go, to a stem expanded by an f. The termination -ιω, -εω would thus be identical with the -jdmi from which came *a8-jd-mi, but also with the -jd-mi from which the numerous presents of the I-class proceeded. But whilst e.g. φαίνω comes from φαν-ιω, the future φαν-ίω or φαν-έιο would come from a *φαιε-ϊω or, translated into ante-Hellenic sounds, *bhana-jd-mi. I do not deny that this hypothesis is a possible one, and in harmony with many phenomena of the Indo-Germanic verbal formation. If the view of Weber were right, we should have in such futures the analogies to as-jd-mi of which Sonne so bitterly feels the want, as was noticed above p. 474. But to become probable, it would need definite and unmistakeable analogies from the cognate languages, of which it is entirely devoid, apart from the solitary Latin -bo, -bis. We hold it to be an essential principle of method that comparative grammar has to reckon as far as possible with given and really existing forms. As long as it is possible, without violating Greek phonetic laws and while keeping in mind other points of view here coming into consideration, to bring a widely extended future form into harmony with the prevailing sigmatic method of formation, this course seems to me to deserve the preference. • It is almost exclusively verbal stems of a perfectly definite phonetic character which form their future without σ : almost all others form theirs with a. It' is hence extremely natural to look for the reason of the difference not in the existence of two types originally quite 305 distinct, but rather in the phonetic character of the stems, and to assume essentially only one future formation. These reasons determine me to hold on the whole to the view which' Buttmann stated with an acuteness remarkable for his time, and like the Laconian υμιώμ(θα quoted above, which undoubtedly points to an 6μ(ό• μ*θα. όμηνμαι would then be the future of the expanded stem 6μο, ομΰται thai of the rt. 6μ which underlies the present ΰμννμι. ch. χνιπ. THE FUTUKE WITHOUT σ. 477 expressed in the following words (Ausf. Gr. i. 2 394) : ' We set down -σω as the proper and universal termination of the future every where : this was appended sometimes with, sometimes without the connecting vowel t.' The only doubt, I think, which can arise is whether the expression 'connecting vowel' is correct, and this will have to be discussed imme- diately. Bopp too Vergl. Gr. ii. § 656 maintained the unity of the Greek future formation. But he was• wrong in his explanation of the vowel appearing in forms like *στελ-ίω, στελ-έω, which lie regarded as identical with the j of the ending -sja-mi. According to Bopp's explana- tion the futurum secundum, to use the traditional fashion of denoting it, would be a degenerate form of the Doric future. But from *στελ-σιω, *μεΐ'-σιω we could never get στελ-ιω, μεν-ιω, as I showed in Tempora und Modi p. 315, but only *&rt \λ-ιω. *μεη-ιω, and further in Attic *στει\εω, *στειλώ, *μεα εω, *μειι>ώ, just as in the aorist from *'εστελσα came έστελλα, έστειλα. As in Greek it is only between two vowels that the sibilant is wont to disappear without leaving any trace, we should have to expect for *στελ-ίω .the previous stages *στε\ε-σιω, *στελε-ϊω. And this paves the way for the admission, on which Schleicher Comp. 3 807 bases his statement, that the Greek future formation goes back to the two primi- tive forms 1) ojio 2) εσ/ω ; and further it can hardly be doubted that this twofold form is not without connexion with the twofold formation to be recognized in Sanskrit, on the one hand -sja-mi e.g. vak-shja-mi from rt. vati speak on the other -ishja-mi e.g. tan-ishjd-mi from rt. tan stretch (cp. re*-*»). The only difficulty lies in the explanation of the vowel appearing in the second form, which in Sanskrit is i, in Greek ε. The different possibilities in the way of explaining this vowel have been so frequently stated of late, most recently by Clemm Stud. vii. 65, that I can deal with them briefly. There are three possible explanations. Either the ε belongs to the second part of the compound : *τεν-εσβω, or 306 it belongs to the first : *τενε-σβω, or thirdly it is a phonetic mediating element naturally produced between the two : φ τε¥-ί-ά}ω. The first explanation, supported by Benfey, Kurze Sanskritgr. p. 182, L. Hirzel Ztschr. xiii. 218 ff. and Schleicher Comp. 3 806 proceeds upon the notion that in this ε the initial vowel of the root ε'ς has been preserved. The chief reason against it is that the ε of the root is always lost in composition with this root, and that the t of the corresponding Sanskrit forms cannot possibly be so explained wherever it appears (Leskien Stud. ii. 79). The second view, that of Leskien and Joh. Schmidt, finds its chief support in the wide extension of e-stems as by-stems to shorter primitive forms, which we discussed on pp 258 f., 264. But we cannot fail to perceive a difference here. Forms like ενόήσω beside ενδω, ωη-θην beside βίομαι bear no relation whatever to particular sounds. On the other hand the so-called futurum secundum, with a few exceptions to be discussed hereafter, is only formed from stems in λ μ ν p, while we must confess that no rule has been discovered as yet to determine what Indian verbal stems have their future ending in -sja-mi, and what in ishja-mi. Against the third explanation, adopted by Bopp and by myself in the Tempora und Modi, according to which the ε has been developed purely phonetically, the 478 THE FUTURE. CH. XVIII. objection is made that we cannot see why the aorist e.g. l -μεν-σα (then ίμεννα, έμεινα) got on without the ε, while the future e.g. *τεν-ε-αβω (then *τενε^ω τενίω or τενεω) regularly took this vowel. But one cir- cumstance has been overlooked here. The presupposed primitive form •rev«rj«, from which *τενεσίω must have been developed by anaptyxis, is one degree harsher than *ε-τενσιι~ The three consonants ι aj could more easily produce a natural vocalic by-sound, than the two νσ. Considering everything, however, I incline to an explanation, which, if I am not mistaken, pretty nearly coincides with that of Leskien. There were, I believe, of old in many cases double stems : man and mana, tar and tara. At an early period in language the future was formed sometimes from the shorter, sometimes from the longer stems without any firm 307 distinction. This state of things continues in Sanskrit, only that here the a has sunk into i, and that the sigmatic aorist also by no means rejects the same vowel. In Greek, as in many other cases, so here too, a phonetic rule has been established, to which there are hardly any exceptions. The longer forms serve only to help out a difficulty where the future form without a vowel would become quite too harsh, and would hence lead to an obscuring of its origin. It still remains for us to trace out more precisely the path by which the presupposed -ε-σ^ω became the Doric -ιω, the ordinary Grek -εω, -ώ. We must, it seems, assume that the sound after the σ was vocalized in these forms very early, certainly before the separation of the Greek dialects, or; to express ourselves with more caution, — for it is hard to determine the priority of j and i — established itself as vocalic, in short, that there was a time at which men said * λβιπ-σ/ω but *τ€ν€-σίω. I do not, I confess, see any definite reason for this difference. Possibly the accumulated short vowels contributed to give a fuller intonation to the vowel at first irrational. From *τενε-σίω then came *τενε-Ίω, whence Dor. τενίω, in the remaining dialects ι-ειτω, τενω. From the form *τενε-ϊω everything goes on regularly. For it is well established in the case of the presents of contracted verbs also, that they proceeded (cp. p. 241) from ε-ϊω on the one hand to ιω, on the other to ε ω. Cp. Cret. κοσμίοντες, αδικίων. It is worth noticing the greater permanence of the t in these forms shown by the Heraclean dialect : αι-ανγελιοιτι, άνκοθκ- ρίοντι as distinguished from εζόιτι, κοφόντι. This greater persistency is occasioned by the fact that the t of the former forms has taken up an ε into itself. B) Contracted Futures without a Sigma from other Stems. A considerable number of stems with a short vowel, which are fol- lowed by a few consonantal stems with short accessory vowels, form a contracted future without sigma, which in its most extended application bears in our grammars the name futurum Atticum, again not at all in j the sense that this form was unknown to the other Greeks, but only ι 308 because the grammarians wished to recommend it as a good Attic form. ! We will first survey the facts of the case, and then proceed to explain them. The forms belonging here are of three kinds, according as the short vowel, after which the σ is rejected, is α, ε or t. CH.xYiii. CONTRACTED FUTURES. 479 1) From stems in a. The most common are the futures of stems whose present ends in -αζω, though, as Buttmann i. 2 391 notices, the Attic form here too is always only * a by-form' of the ordinary sigmatic formation, and in the case of many verbs is either altogether rejected 6 , or crops up only in the Hellenistic period, beyond the limits of correct prose. From presents in αζω we find the following futures : 1) άρπώμαι LXX, cp. Veitch p. 90. 2) βιβώ. βιβών Soph. Ο. C. 381, προςβιβ? Aristoph. Αν. 425, and the like in Plato, Xenophon, Demosth., with middle forms also : άναβι- βώμαι Amipsias (Com. ii. p. 713) Aeschinesii. 146, άναβιβάται Demosth. xix. 310. 3) ΰικ&ν. ουκ εφη ΰικάν en Herod, i. 97, while at i. 90 we read Ζικασόμενοι. The Attic writers in this instance used only the sigmatic form. On the other hand we find on the Cretan inscription C. I. 2554 1. 66 the strange form ΰικαώσι (ταλλα £ε Β. οι κριταί), which surprises us if only by the termination -σι besides numerous instances of -vti, and as the solitary witness to an Attic future among Dorians has the less support that it does not suit εΐίκαξαν and other forms of the kind. — The mid. Βικώμαι LXX. 4) άπο-ΰοκψφ only Herod, i. 199, unless we have there a present form for which there is no other evidence, for the context neither requires the future nor excludes it. 5) εζετώμεν Isocr. ix. 34, elsewhere always Ιξετάσω etc. 6) έργώμαι LXX. 7) κολφ Aristoph. Equ. 456, κολωμένονς ib. Vesp. 244. Often with the σ. 8) πελώ Aesch. Prom. 282 ; πελατε Soph. Phil. 1150: cp. πελάσω Eur. El. 1332. 9) κατα-σκεναν inscription of Olbia C. I. 2058 B. 1. 29, 53. . 309 10) κατα-σκιωσι Soph. O. C. 406> There are further the following futures from stems of the same kind, though, the present does not anywhere, or, at any rate, does not in the same writers end in -αζω : 11) αντίόω. κείσ ειμί, και άντιόω ποΧέμοιο Μ 368, άντιόων ταύρων α 25 beside άντιάσεις χ 28 (cp. άντιάσας and the like), άντιάζω Soph. Eur. Pind. Herod. 12) Ιαμψ. ει ΰη όμοϋ πόλεμος τε ΰαμφ και Χοιμος Άχαίούς Α 61, η ϊίομαι — ΐι Ιαμόωσι Ζ 368. The present was ΰάμνημι in Homer, but ΰαμάζω in the tragedians and elsewhere. 7 13) έλω. ελόωσι Ν 315, η 319 8 , ελώσι Herod, i. 207, ελάαν Ρ 496, β We cannot find any principle for the choice of one or the other form in Attic writers. Veitch p. 170 makes the good remark, ' What induced the Attics to spare xr in δικάζω, and expel it from others in the same category, βιβάζω etc., we leave for Uniformists to tell.' 7 Buttmann Ausf . Gr. i. 2 392 conjectures, with Dindorf 's conditional approval, for Aristoph. Eccl. 161 εκκληίπώσ', cp. Thesmoph. 90 with Dindorf 's note. 8 Cobet Mnemos. Nova series ii. 395 prefers in Ν 315 the reading quoted by Didymus from the υπομνήματα of Aristarchus, έόωσι and in e 290 idav, both as futures from the root a = sa (sa-tur) occurring in άμεναι, Ζομςν (Ζωμςν). But while we can easily understand the e in %ωμςν (cp. στέωμεν), we cannot understand it before an α in *€cur« for which we should have expected *ησω. 480 THE FUTURE. ch. xvn Aesch. Eumen. 75, ελ£ Soph. Aj. 504, ελ£ς Aristoph. Ran. 203 besides Horn, έλάσσω, παμελήσσεις Ψ 427, ordinary Greek ηΚασα etc. 14) κρεμάω II 83 (ο'ίσω και κρεμάω), κρεμώμεν Arist. Plut. 312. — κρεμάσω Alcaeus Com. (Com. ii. 827). — ε^ρέμασα, κρεμαστός etc. in ordinary Greek. 15) περάαι , future to πέριημι or πιπράσκω Φ 454 beside έττέρασσα. 16) άνα-πετώ Menander (Com. iv. 77), εκπετάσονσι Eurip. (Iph. T. 1135. — πετάσαι Homer, πετάπσαι) and the like, widely extended. 17) ΰια-σκεΰ^ς Herod, viii. 68,. 2, ΰιασκεδαν ib. i. 79. — σκέΰφ Aesch. Prom. 925, άποσκεόώ Soph. O. R. 138, συσκεΐάν Aristoph. Ran. 903. — σκευάσεις Theog. 883, and the like in later prose. — εσκέδασα from Homer onwards. 2) From stems in e. These futures are few in number. We can here again draw various 310 distinctions, especially that between stems which have ε throughout, and those where the ε is accessory. We place the former first. Among them the form άμφιώ (from άμφι-έ-σω, which is preserved in ε 167) takes a place of its own, inasmuch as the ε here represents the root itself. ■π-ρος-αμφιώ is found in Aristoph. Equ. 891, άπαμφιεϊ Menander in Meineke's Comici iv. p. 171. The remaining forms are : 1) γαμέω I 388, γαμεί Aesch. Prom. 764 etc. in poetry and prose. For this verb there is also the shorter stem γαμ, so that γαμέω as a future is related to εγημα as σφαλώ to εσφηλα. 9 2) καλέω. καλέουσ' 'ίε Γ 383, καλώ, καλούμαι in all Attic. But beside this there are εγκαλέσει Demosth. xix. 133, επικαλέσεται Lycurg. 17, έκκαλέσεσθαι Aesch. i. 174. 3) κορέει θ 379, κορεεις Ν 831, but κορέσω Herod, i. 212. 4) τελεΊ.» Ψ 20, εκ έέ και οάε τελΓι Δ 161, beside τελέσσω Ψ 559. At β 256 the M.SS. vary between τελε'εί and τελέσει, τελώ is quoted from dramatic poets and Plato, τελέσω from Pindar (Nem. iv. 43), Xenophon and Plato. There come now the few instances of a contracted future from stems ending in a mute, which according to the view formerly in favour were regarded as futura secunda, viz. : 5) καθεΐοϋμαι, Attic future to καβέζεσθαι καθεδεί Aristoph. Ran. 200, also in prose. The same stem-expansion is shown in" the post- Attic εδε-θλοι\ 6) μαΰίϋμαι, a very doubtful reading in Theocr. xi. 60 (cp. Ziegler ad loc.) ; Ahrens writes μασεϋμαι. The ε could only find a weak support in μεμάθηκα, μαβησομαι, as everywhere else the long e appears. 311 Ό μαχέοντΜ (cp. p. 269) decidedly as a future Β 366, cp. μαχεΊται Υ 26. The contracted forms are common in the Attic writers from Aeschylus onwards • until the Orators. We may compare the aorist μαχέσασ^αι in use from Homer onwards, and μαχετέον quoted from Plato. Beside these even Homer has μαχησεται Σ 265, μαχήσασθαι Ε 483. In • The case is quite the same with the rt. 3λ ; the fut. οκί-σω (ν 399) has heen formed from the s'era expanded by*, as well as ώλ*σα, όλώλίκα, 6\(dpos, u\to> in llenxlot us (άπολ€€ΐ i. 34), o\eW0e Φ 133; ολώ, 6\οΰμαι in the Attic writers are related to the shorter si em 6\ as βα\ώ is to /8ολ. We can see with especial olearnese here how the « in the future of verbs in λ μ ν p comes in contact with uuniistakeably slem-forming elements. Cp. above pp. 476, 477. ch. χνπι. NON-SIGMATIC FUTUKES FEOM STEMS IN ε AND ι. 481 Herodotus, where previously μαχέσυμαι was preferred, even by Bredow, p. 339, Stein now writes μαχήσομαι, e.g. vii. 102. In this verb the added e undoubtedly promoted the clearness of the tense-formation. 8) τεκέΐσθαι only Hymn, in Ven. 127, while τέζειν, τέξεσθαι are quoted from Homer onwards. Cp. τοκε-τό-ς. Cobet in the Mnemosyne, New Series ii. 392, has discussed a large number of the futures adduced here and to be adduced hereafter. He assumes that they have lost a σ, and believes he has discovered a well- established rule, at any rate for the stems in ε, with regard to this loss in Homer and Attic writers : * In quibus verbis ε non in η pro- ducitur, si est antepenultima longa, σ in futuro non eliditur, ut in ετταινέσομαι, αιΐέσομαι, αρκέσω, νεικέσω, et άχθέσομαι, quorum αϊΐέσομαι et νεικέσω et αρκέσει Homerica sunt. Contra ubi brevis est antepenultima, σ ubique summa constantia omittitur.' But the second part of this rule cannot be reconciled with our texts without violent emendations. We have given references above for εγκαλέσει, κορέσω, όλέσω, τελέσω ; αρέσεις appears Demosth. xxxix. 33, άρέσονται Aescb. Suppl. 655 ; έμέσω is quoted from Hippocrates, whom Cobet possibly might regard as not coming within the range of his assertion. But as the number of all the verbs of the kind is but small, and as we can find no intrinsic reason, i.e. no reason in the formation of the forms, for this rule, there is no sense in adopting numerous alterations to satisfy it, especially as the stems of the same kind in α would adapt themselves still less to such a rule (cp. δικάσω, κολάσω, αγοράσω). At most it is somewhat probable that we may assume in the case of the familiar verbs γαμέω, καλέω } τελέω that in Attic writers they formed their future only in this way. 3) From stems in t. The futures in -ιώ, -ιοϋμαι are especially often described by the ancient grammarians as Attic ; as by Apollonius Dyscolus de construct. p. 274 (Bekker) : τον άπο της Άτθίΰος σχηματισμοί/ κατά μέλλοντα φασι λνριώ, by Herodian on II. Λ 454 (Lentz ii. 806) κτεριονσι. η τοιαύτη 312 νφεσις του σ 'Αττική εστί και περισπφ τους μέλλοντας. But this formation too is not at all limited to Attic, but may be found also in Herodotus and Homer. There are four Homeric instances : άγλαϊεϊσθαι Κ 331 (ετταγλαϊεϊσβαί 2 133), άεικιώ Χ 256, κομιώ β 546, κτεριονσι Λ 454, in Herodotus, if I am not mistaken, eleven : άνασκολοπιείσθαι iii. 132, άτρεμιείν viii. 68 (cp. άτρεμιεΊσθαι Theogn. 47), έναγωνιενμαι iii. 83, έζανΊραποοΊεννται vi. 9, επισιτιεύμενοι ix. 50, θεσπιέειν viii. 135, καταγιέϊν i. 86, κομιεΊ ii. 121, νομιενμεν (Stein νομιοϋμεν) ii. 17, όπωριενντες iv. 172, χαριείσθαι i. 158. More than twenty such forms can be quoted from Attic literature, and it seems superfluous to cite them separately ; many, e.g. έλπιώ, σεβιώ only occur much later, but this may often be due to the accidental nature of our authorities. The following may serve as examples from the best Attic period of verbs originating in dif- ferent ways : βαΰιεΊ Aristoph. Thesm. 617, Ιειπνιεϊν Diphilus Com. iv. 405, έθιοϋσι Xen. Cyr. iii. 3, 53, ισχνριείται Lys. vi. 35, κονψιεϊς Soph. Ant. 43, μεταχειριεΊται Plato Rep. 410, νοσφιεΊς Eurip. Ale. 43, κατοικτιεϊ Aesch. Suppl. 903, οικιοΰντες Thuc. i. 100, προφασιοννται Aeschin. iii. 24, ώστιοϋνται Aristoph. Ach. 24. Futures in -ίσω are not common in the good Attic period : but we have no trustworthy collections on this ι ι 482 THE FUTUKE. CH. XVIII. point. Two forms of a special kind have still to be mentioned here : καθιώ and κλιώ. The former is sufficiently established from Xenophon and Demosthenes : Xen. Anab. ii. 1. 4 καθιείν (the better M.SS., how- ever, have καθίσει»'), Demosth. xxxix. 11, καθιεΊ, xxiv. 25 καθιέΐτε. καθίζω, in spite of its origin in the rt. !(>, has evidently been regarded as a verb in -ιζω, like κομίζω, ελπίζω, and treated accordingly. The form κατακλιεΊ is more difficult : we know it only from a fragment quoted by Choeroboscus from Eupolis χρνσονν γένος (Meineke Com. ii. 544 : cp. Herodian ed. Lentz ii. 809) : εΐ μη τις αυτήν κατακλιεϊ. The word is explained by the grammarian in Cramer's Anecdota Oxon. iv. 195 by άσφαλιεί, and treated by all grammarians as a rare future form from κλείω, of which the Old Attic present is κλήω (cp. p. 249). Joh. Schmidt Vocal, ii. 254 f. has conjectured that this κατακλιεϊ belongs rather to κατακλίνω with the force ' veil, cover up,' attested here and 313 there for κεκλιμένος and έκέκλιτο. But though with much acuteness and learning he establishes this meaning for the stems Mi, kal from different regions of speech, and also for Homer (E 356), yet there is no trace whatever in Attic Greek of any such use of κατακλίνω, and in the passage of Eupolis, according to all appearances, the meaning required is that of ' shutting up.' Hence scarcely any course remains except with Meineke to put κατακλιεϊ side by side with Ιανιώ or δανειώ from δανείζω, against which Photius in his Lexicon p. 85, 21 warns us, and to conjecture that the comedian put the form into the mouth of a barbarian. The Doric forms κλαζώ (Theocr.), άποκλα"ζας have been formed on the analogy of the verbs in ζω. Perhaps there was really a present κλτ}ζω by the side of κλτίω (cp. Veitch p. 332), and the presumed barbarian meant to say κατακληεΐ, which would have some analogy in its favour, but said by itacism κατακλιεϊ. Now that we have taken a survey of the stock of forms, we have to explain them. Two possibilities here present themselves. Either there never was a σ here, so that all these futures would properly be presents used as futures, or the sign of the future, the sibilant, has been dropped. The first view might seem admissible at any rate for the first two groups ; then, as Herodian ii. 809 says, γαμέω, τελέω in their use as futures would be really an ενεστώς αντί τον μέλλοντος. But such a view is impossible for the third group. Presents like *κομιέω, *άγωνιοϋμαι are unknown and without any analogy. Hence no other course remains in the case of the third group but to assume the loss of the σ. And this makes it very probable that the same phonetic process took place also in the verbs of the first and second groups. While, therefore, in my opinion all three groups agree in changing the σ into a breathing, the first two differ from the third in one essential point, βιβώ has come from βιβάσω, καλέει as a future from καλέσει, that is, both from the usual future form, but κομιέω cannot be explained from κομίσω. It rather presupposes *κομι-σεω. Kiihner Ausf. Gr. i 2 . 570 is of a different opinion. He thinks that κομι-έω comes rather from *κομι-εσω. But this assumed intermediate 314 f orm would find a weak support only in the rare future forms like μαχέ- σομαι cited on p. 481 ; and it is still less admissible to presuppose for an intermediate form so deduced the loss of a £, as Kiihner does : *κομι$- χσω, *κομι εσω } for a loss like this is quite unknown. On the other hand ch. χνιπ. OTHER SON-SIGMATIC FUTURES. 483 -everything is quite right, as soon as we start from a primitive form *κομι-σεω formed on the Doric analogy. We saw on p. 470 that in Attic writers this formation was not altogether rare, at any rate with middle terminations. From κομι-σέω there came κομι-έω as from τελέ-σω τελέ-ω. This extremely simple explanation is confirmed, I think, by the existing Doric forms like όρκιζέω (C. I. i. 1688, I. 13), χαριζιόμεθα (p. 469). The Attic χαριοϋμαι comes therefore from the same primitive form as this χαρίζω μέθα. As to the loss of the σ, it may surprise some that we have assumed this here without further discussion, while in the case of the aorist a similar procedure appeared improbable. But there another, and, I think, an easier means of explanation presented itself. Besides the aorist is to a much greater extent without duplicate forms with and without σ, which here, especially if we include the Doric dialect, are found in abundance. We cannot deny that there is elsewhere too some fluctuation with regard to σ between vowels, e.g. in the formation of the 2 sing. mid. (-σα«, -σο). By assuming the same phenomenon in the future we can at once understand how even in Homer τελέω can act sometimes as a present, as the product of an earlier τε\ε(σ)^ω, sometimes as a future coming from τελε(σ)-σω. The loss of the σ in contracted forms was evidently favoured by the analogy of the common futures of the verbs with λ μ ν p. C) OTHER FUTURES WITHOUT σ . We have learnt above (p. 467 ff.) to recognise a series of future forms which were characterised as such by nothing, but which did not, how- ever, coincide with present forms, because the presents of these verbs were on their side distinguished by stem -expansions. Our investigation now brings us back to formations, which externally are extremely like those forms, but which are distinguished from them by the fact that the future is here often completely identical in form with a present in more or less frequent use. There are not many verbs which come under 315 this head, and it is a question whether all can be explained in the same way. In Homer there are three futures, resembling presents, in -νω, viz. : άνύω, which may indeed in Δ 56 ουκ άνύω φθονεονσ be very well taken as a genuine present, but in the compound είανΰω A 365 η θήν σ εζανΰω γε και ύστερον άντιβολήσας (=Υ 452) points more decidedly to the future, though this might have been denoted as in π 373 by ανΰσ- σεσθαι from the same stem. έρνονσι A 454 α δβιλ', ov μεν σοι ye πατήρ KOI πότνια μητηρ οσσε καθαφήσονσι θανοντι περ } αλλ' οϊωνοί ώμησταϊ ip -υουσι . . . Cp. Ο 351 άλλα κΰνες έρνουο*ι, Χ 67. The best ancient grammarians assumed here present forms with a future force : thus Aristonicus on X 67 : ή ΰιπλή οτι χρόνος ήλλακται, αντί του ερύσουσιν και εν ΌΖυσσεία. νευρην εντανύειν (^ 97, 127) αντί τοϋ λντανϋσειν and Herodian on Λ 454 ερύουσι. προπαροζυνεται' ένήλΧακται ιι 2 484 THE FUTUKE. ch. xvnr.. γαρ 6 ενεστώς 'χρόνος άντι μεΧΧοντος. όντως και Άρίσταρχος" 6 ΰε'ΑΧεζίων περισπά, ονκ εν. ενταννειν in the two passages already cited from φ and also 174. Why La Roche φ 97 and 127 (after ελπομαι) against Aristarchus and good M.SS. writes ενταννσειν, and in the third passage ταννονσι (άλλ* άλλοι ταννονσι τάχα) I cannot understand. There is also νέομαι Σ 101 [=Ψ 150] ννν δ' επει ον νεομαί γε φίΧην ες πατρίδα γαΊαν, Σ 136 ήώθεν γαρ νενμαι, ΰ 633 όππότε Τηλέμαχος νε'ιται, ξ 152 ώς νείται ΌΙνσενς. [Cp. La Roche on Ξ 505.] The Attic dialect has a similar instance in the future use of χέω and' χέομαι (cp. Elmsley Review of Hermann's Snpplices on v. 772, Eurip.. Supplices Lips. 1822, p. 254). Certain instances are found in Eur. Suppl. 773 αλλ' ei/z', έπαρω χβίρ' άπαντησας νςκροϊς "Αιδον be μοΧπάς ΐκχίω δακρνρρόονς Fragm. 388 Dind. κάρα Τ€ γάρ σον σνγχ4ω κόμαις ομού ράνω τε ττ^δόσ' έγκεφαλον 316 Aristoph. Pax 169 κάπιφντ€νσ€ΐς ΖρπνΧλον άνω κα\ μνρον βπιχΰς Plato Com. (Meineke Com. ii. 637) εγώ δε Χίτρον (?) παραχέων Έρχομαι, κάγω de παρακορησων Isaeus vi. 51 επι τα μνήματα ιέναι χεόμενον και εναγιονντα. Whether the language of Homer possesses a future-present χενω corresponding to this is doubtful, for χενομεν Η 336, and of course also χενω β 222, may be the conjunctive of the aorist. A very unique Cretan form of the same kind is found in the 3 plur. αναγνωοντι C. I. no. 2554, 1. 39, ει δε μη εζορκι[ζ]ό ντι οι Αάτιοι κόσμοι η μη παραγγε\[ε\οντι επι ταν ανάγνωσιν τάς σννθήκας, η μη αναγ- νωοντι, αποτεισάντων . . . and again in 1. 45. Boeckh took the form to be a future with the σ lost, Ahrens Dor. 339 to be an aorist conjunctive, but this does not fit in with the futures of the context. We must not indeed forget that we have no very satisfactory guarantee for the exact reproduction of the whole inscription. There remains finally the Old Attic σώω : C. I. A. i. 2 B. 1. 7 και τα. κοινά τα Σκαμβωνιδών ΣΟΟ και αποδώσω. On this form, which is un- doubtedly used in a future sense, I can now refer to the well-considered discussion by Cauer Stud. viii. 416 if. Now what opinion are we to form upon these cases 1 For the four Homeric forms the rejection of a σ is just as possible, as for those pre- viously quoted, but who can believe it possible that the same loss has happened to αναγνωοντι and σώω. Even if on the analogy of εζ,όντι, Ko\j,ovTi we should adopt the accentuation αναγνωοντι, there would still be no analogy for the rejection of the σ in the Doric form. Buttmann Ausf. Gr. ii. 2 296 does indeed maintain this for σώω, but Lobeck on this passage justly describes this as ' very improbable.' It would be easier ch. χνιιι. OTHER NON-SIGMATIC FUTURES. 485 to approve the attempt (v. Bamberg Ztschr. f. Gymnasialwesen 1874 p. 619) to take σωώ with this accentuation as an Attic future, as we have •clear traces of the spelling σώζω, if this very ι, which was all that was ca- pable of bringing in the analogy of the presents in -ιζω, was not absolutely wanting in the form σώω. But Cauer 1. c. has now shown that σώω in 317 Homer several times approximates to the force of a future, for which ι 430 α ην σώοίτες εταίρους is especially to be noticed. Hence I regard it as proved that σώω belongs to the present forms used as futures. The same must undoubtedly be assumed in the case of χέω. It is •quite astonishing to find that an investigator like Buttmann (Ausf. Gr. ii. 2 325) thinks it possible that χέω may have originated in *χέσω. The ευ of χευω as a conj. aor. and the υ of κέχυμαι etc. ought at least to have made him hesitate. The parallel with τελε'ω is quite erroneous and •cannot be at all supported by late and possibly not even well-established formations like έχέθην. If χέω had lost an σ, there would be no alter- native but to assume the following as the successive stages of the corrup- tion : *χεύ-σω *χεύ-ω *χίΐ-ω, χέω, and a fifth stage would be furnished by -contracted forms like επιχεΊς. ■ Who can accept this as credible 1 I may refer rather to what was said on p. 461 on the occasion presented by the aorist έχεα. In its formation χέω for *χΛ is certainly a present. The coincidence of the two tenses is, as I suppose, to be explained, much as in the case of τελέω and καλά*, by assuming that the form χέω destined exclusively to express the present goes back to an earlier χείω (εγχείτι ι 10), the future χέω to *χεϊω. This χείω is in any case parallel to the forms πλείω and πνείω mentioned on p. 156. The less expanded present form was retained for use as a future. If we now look back from these cases to those which remain, we may really doubt whether in these there has been anywhere a loss of the σ. I believe that there is no certain criterion in ανΰω, ερΰουσι, εντανύειν, τέομαι (cp. νίσσομαι i.e. νεσ-^ο-μαι) to decide the question definitely in one way or the other. Even for the futures with «-stems cited on p. 479 f. it would be possible in some instances with similar probability to assert & purely present formation. For instance, ελώ as a future might be related to έ\αύνω, κορέει to κορέννυμι, αναπετώ to αναπετάννυμι much as πίομαι is to -πίνω. But as forms like τενώ and κομιώ, which can be understood only as from *τενέσω, *κομισέω prove the loss of the σ within a -certain circle to be a fact, we preferred on p. 482 the same explanation 318 for these too. In conclusion it is best to lay it down that there are three kinds of futures with a formation resembling that of the present : 1) Such as, by the side of an actual use as presents, take also the force of futures e.g. είμι, σώω, 2) Such as have ceased to be used as. presents, and only act as futures : εΰομαι, πίομαι, 3) Such as, in consequence of the rejection of spirants, either like καλέ» coincide with the present form at a relatively not very ancient -date, or like τενέω, βφώ have exclusively the force of a future by the side of a present characterised in a quite different manner. ΙΠ. MOODS AND VERBAL NOUNS OF THE FUTURE. The optative of the future, which is employed only in a subordinate clause depending on a past tense, is, as we saw on p. 6, quite unknown 486 THE FUTUKE. ch. xviii. to the language of Homer, where scarcely any opportunity for its use presented itself. Perhaps Pind. Pyth. ix. 116 συν δ' άέθΧοις eictXtvaev ΰιακρίναι σε'θεν αντινα σχησοι τις ηρώων is the earliest instance of the use of this mood, which we find afterwards in the tragedians, e.g. Aesch. Pers. 369 (ψενξοίατ), Soph. Ο. T. 1274 ff. (οφοιντο, όφοίαθ\ γνωσοίατο), Philoct. 612 (περσοιεν), in Herodotus e.g. i. 127 εκέΧενε άπαγγέΧΧειν οτι πρότερον ήζοι (v. 1. ϊ/'ξε*), more commonly first in Attic prose, in Thucydides e.g. vi. 30 (κτήσοιντο), 74 (φενζοιτο), Isocrates (εΙςπΧενσοίμην xvii. 9), Plato, Xenophon, and Demosthenes (xxxi. 2). This mood-form never became very usual, owing to its re- stricted application. But the verbal nouns are on the contrary very common. Like the optative they are evidently formed entirely on the analogy of the present. It will be sufficient here to quote Homeric forms, which occur in great abundance, e.g. αΐέμεναί Ψ 50, αιρησέμεν Ρ 488, χραισμησέμεν Φ 316, γηθήσειν Ν 416, ενφρανέειν Ε 688, εκτεΧέειν κ 27 — αΐτησων ρ 365, παύσονσα Α 207, άγγεΧέοντα Ρ 701, αποστρέφοντας Κ 355, ερεοντε π 334 — άφαιρησεσθαι Α 161, ΰείσεσθαι Ο 299, κρανεεσθαι Ι 626, άποΧεϊσθαι 319 θ 246 — Χυσόμενος Α 13, όφόμεναι 2 141. Future infinitives and parti- ciples belong to the indispensable stores of the Greeks from the earliest times to the latest. Finally we may say two words upon the familiar idiom, that so many active verbs form their future in the middle voice without any difference of meaning. Lists of such futures are given by Buttmann Ausf. Gr. ii. 2 85, Kiihner Ausf. Gr. i. 2 684. 1 The latter in ii. 2 91 attempts an explanation of this strange proceeding. Yerbs, which denote intellectual or physical perception, had (he says) not uncommonly by the side of the active a middle inflexion also, without any essential alteration of the meaning, e.g. άκούετο Δ 331, ορατό A 56. From this he thinks we can explain the usage referred to, especially as it is a question almost exclusively of verbs ' which denote the expression of a physical or intellec- tual activity.' Kiihner goes on to say ' the notion of futurity is really only something subjective, existing only in imagination.' But unfortu- nately the attempted limitation of the meaning of the verbs which are in question is much too elastic, and by the addition of ' almost ' it is made still more indefinite. It would indeed be very diflicult to find any meaning shared in common by futures like άπαντήσομαι, αποΧαΰσομαι, γεΧάσομαι, γηράσομαι, εΰομαι, τενζυμαι, γεσοϋμαι. The question is trans- ferred into an entirely different region, when Kiihner thinks he can further find something subjective in the category of future action gene- rally, which has some internal connexion with the middle action which is ruled in a higher degree by the subject. But the latter notion is evidently based upon a confusion of very different things. The speaker cannot of course speak of the future with objective certainty, but only in the form of subjective conjecture ; and that is why the potential approximates to 1 Kriiger Sprachlehre § 39, 12 gives a list of verbs which in the usage of the Attic writers more or less positively form their futures in the middle voice. [Farrar, Brief Greek Syntax § 90, points out how several of these middle futures correspond to reflexive verbs in French.] ch. χνπι. MIDDLE FUTUKES. 487 the future. But the middle is just as definite a form of expression as the active. Hence while it is quite right to compare άκουσομαι with the Homeric use of ηκονετο, e.g. ακουετο λαός αντής, we cannot speak in 320 either case of subjectivity or ' mere imagination.' The future can be * mere imagination ' at most for the speaker, that is for myself, if I say άκούσεται, while the middle form expresses a fuller participation of the third person referred to as the subject of the action. Hence I doubt whether we can find any sufficient reason for the ' elective affinity ' between the future and the middle : and I am afraid that we must content ourselves with the fact that the active and the middle are separated by very slight boundary lines. In consequence of this slight distinction, which often is so delicate that it can only be detected by instinctive feeling, at a very early period active and middle divided themselves upon different tenses in a number of very common verbs ; and the middle form especially settled on the future. A number of others afterwards followed the example of these. It is no more possible to determine them by their meaning, than e.g. to lay down any distinction of meaning in modern languages for the use to ' be ' and ' have ' as auxiliary verbs. Naturally there is here too no lack of deviations from the rule, upon which we cannot enter here. Perhaps the one fact, that εΙμί has only the middle future εσομαι, to which we have referred on p. 435 under the head of the future perfect, has had greater influence on the other verbs, than might have been supposed. 488 THE PASSIVE STEMS. CHAPTER XIX. THE PASSIVE STEMS. As the two tense-steins discussed in Chaps. XVII. and XVIII. stand unmistakeably in a closer affinity one to the other, so the two groups of passiye stems again form a common division in the great whole of the Greek verb. They share in the first place the negative characteristic, that they have nothing quite corresponding in any one of the cognate 321 languages, and are therefore undoubtedly to be regarded as essentially a recent formation of the Greek language. They are further united by their identical force, that of the passive, which in all other instances attaches only to the middle endings, but here is in part united with active per- sonal endings. Finally we must notice the inflected e which runs through both after the fashion of the unthematic conjugation, and by which the passive aorists receive their peculiar stamp. Hence we have to do unmistakeably with a pair of stems, the parallelism of which could not escape even the Greeks, and the development of which had a mutual influence on both sides. It is in the case of this pair of equivalent creations that there is most sense in the old terminology which denoted tenses by means of numbers. For it is not very uncommon here to find both in use at the same time, e.g. εμίγην and εμίχθην ; and it is not altogether preposterous here to denote the rarer form as the second, the more common as the first, as the former cannot by any means always lay claim to greater antiquity. Still it is better here too to replace numbers which signify nothing by definite names. Hence, reserving the ex- pressions ' strong ■ and * weak ' for the grammar of schools, I call the one passive stem that in -??, the other that in -θη. But as shorter terms are sometimes indispensable, I occasionally describe the former as the lighter, the latter as the heavier. We shall in the first place discuss the peculiarities common to both passive-stems, and then examine the forms of each stem according to their occurrence and origin. The e at the end of both passive stems is treated throughout just as in the so-called Aeolic inflexion of the derived verbs in -η-μι (=-ε-ω). Compare Horn, φανή-την with ομαρτητην^ ΰαήμεναι with καΧημεναι, ΰμηθείς with Aeol. ψίλεις. In discussing the moods above pp. 314, 319, 329 etc. we were therefore able to examine at the same time forms like ΰαμήετε, τραπείομεν, μιγείης etc. It results from this that the ε, originating from contraction (cp. p. 247) is essentially long, and is shortened only under the conditions examined on p. 135. In this respect the iterative formation φάνε-σκε-ν Λ 64, belonging to έ-φάνη-ρ, is instructive. The two consonants here gave occasion for the shortening. 322 Hence the € of ψανειήν, Ζμηΰέντος has no greater claim to be considered original than that of ψιλείην, φίλεντος, and we must regard, not ψανι •ch. xix. PECULIARITIES OF THE TWO STEMS. 489 φίλε μιχθε, but φανη, ψιλή, μιχθη as the true stems. Compared with primitive forms like θες, ιός, the long vowel in the passive stems holds its place more obstinately, as is shown by ψάνηθι, ψανητω, λνθήναι. The very close resemblance of the passive stems to the derived ε -stems of the more archaic inflexion makes it probable that both belong, so far -as the form impressed upon them goes, to the same period of language, i.e. to that in which the thematic vowel in its ordinary form had not as yet been united with the contracted theme in η which had been inherited from an earlier period. We saw on p. 246 f. that to understand ψιλέω we must start from an earlier *φι\ήεμι, ψίλη-μι. At a time when ψίλημι, έφίλην 3 plur. εψιλεν (or εφίλειϊ) φίλημεναι, φίλεις were not yet limited to the Aeolic dialect, εφάνην and ετέθη ν with their moods and verbal nouns must have been formed. We shall come back to this point in the course of the present chapter and try to determine its importance for the history of the developement of these forms. For the indicative of the two passive stems we have first to take into consideration the quantity of the e, and the different formation of the 3 plur. It might appear at first sight that a trace of the long e before the termination -v(t) of the 3 plur. had been preserved in the entirely unique μιάνθην αΐματι μηροί (Δ 146). The scholiasts Β L took μιάνθην to be a ' Ιυικον εκ συγκοπής,'' a dual form shortened from μιανθητην ; but no one will be willing to accept this view. Buttmann Ausf. Gr. ii. 2 244 saw in it an aoristic dual form of the middle from a stem μιαν formed according to the rules of the primitive inflexion. As Ιεκ-το comes from the rt. ΰεκ, so we might imagine a form *μίαν-το as a 3 sing. aor. mid., and μιάν-θην (for *μιαν-σθην, cp. ΰέχ-θαι) would be the correspond- ing second person dual. But there are several objections to this too, and it is hard to make up our minds to regard this form as differing in principle from μιάνθησαν which occurs in exactly the same sense at Π 795. Hence I consider it best with Ahrens (Conjug. auf μι p. 36) to write μί ανθέ ν, 1 but not as though the 'bucolic caesura' produced the 323 length of the syllable, for what was so regarded is better explained as original length ; I suppose rather that μίανθεν retained the old length of position in the final syllable even in thesis. This principle of explana- tion is recognized by Hartel Horn. Stud, i. 2 Ill, at any rate in cases of arsis, with reference to εφαν and certain other forms of the kind. For we cannot really believe that the original η retained its natural length unimpaired in this isolated instance before vr and the later vv, v. The regular forms of the 3 plur. in -εν are far more common in Homer than the longer ones in -ήσαν. Against 46 forms in -εν, e.g. άγεν, ΰάμεν, ΰιέτμαγεν, κόσμηθεν, πήχθεν, τάννσθεν, there are only 15 in -ήσαν, e.g. μίγησαν (beside μίγεν), τάρπησαν, εχάρησαν, ήίχθησαν, θωρηχ- θησαν. On the other hand in the language of later poets the shorter formation is an archaism, used here and there. From Pindar Peter de •dial. Pindari p. 59 cites 16 forms of the kind, e.g. βλάβεν Nem. vii. 18, εφθαρεν Pyth. iii. 36, όνόμασθεν 01. ix. 46, εμιχθεν Tsthm. ii. 29. For the dramatic poets I may refer to Gerth Stud. i. 2, 257. There are well-established instances in Eur. Hippol. 1247 εκρνψθεν, Arist. Pax 1283 έκόρεσθεν, Vesp. 662 κατένασθεν : hence in Soph. Antig. 973 for [ 1 μίανθεν is actually found in one M.S. Cp. La Roche ad loc. who reminds us cf the fact that the early alphabet had only one sign for Ε and H.] 490 THE PASSIVE STEMS. ch. xix. the τυψλωθέν of the M.SS. I have conjectured τυφλωθεί'. We may quote also έφίληθεν (or εψίλαβεν) Theocr. vii. 60. For the shorter forms it is of importance to notice that they occur also on Doric inscriptions, where we may give them the Doric accentuation : ΰιελέγεν C. I. G. 3050 1. 7, 3052 1. 10, for which in 3048, 1. 8, certainly only from oversight, ΰιελέγην has been written, which Boeckh with Buttmann alters into ΰιελέγεν* κατεΰικάσθεν Tab. Heracl. i. 122, 143, ΰιελέχθε ν treaty between the Cretan towns Hierapytna and Lyttus (Naber Mnemos. i. 105 1. 13). From Archimedes Ahrens Dor. 317 quotes κατέγνωσθεν, σννεζέΰοθεν, ετεθεν. It is therefore well established that the Dorians made use of such forms even in prose. — Forms like ερητνθεν, κοσμηθεί are quoted as Boeotian, and occasionally also as Aeolic, i.e. Lesbian Aeolic by grammarians (Ahrens Aeol. 211). That the longer forms were not unknown to the Asiatio Aeolians is proved by εστάθησαν Sappho fr. 53 Be. 3 We have an instance of a Doric form of the same kind from inscriptions in Ιιελέγησαν C. I. 3047, 1. 7. Evidently the two formations were both in use before 324 the dialects parted off. In the Attic writers and in Herodotus it was only the longer ones which became established. Except in the 3 plur. the traces of shortening are very uncertain. The ancient grammarians, as Boeckh conjectures in the notae criticae on Pind. Pyth. iv. 115, regarded the Homeric forms έτραψέτην Ε 555, τραψέμεν Η 199, Σ 436, γ 28 as shortened. But the intransitive use of the active aorist ετράψον is completely established (cp. above p. 287), so that we are even tempted to doubt with Buttmann (Ausf. Gr. ii. 2 307) whether the form ετράφην later in use is not to be denied to the language of Homer. For the active forms may always be restored by slight alterations, sometimes supported by traces in our authorities (e.g. Ψ 84 supported by Aeschines c. Tim. § 149). — Besides these probably the only word that can be quoted is στερθέμεν στερηθήναι Hesych. But the authority for this is too doubtful to carry much weight. Of the moods the imperative of the lighter passive stem is represented in Homer only by φάνηθι Σ 198, ψανήτω υ 101, while of imperatives of the other form there are 7 examples, as αίΰέσθητε β 65, κατακοιμηθήτω I 427 y σαωθήτω Ρ 228. — Conjunctives like ΰαείω (or ΰαήω) Κ 425, ψανήτι Τ 375, ΰαώμεν Β 299, χολωθ^ς Ι 33, Ιανθϊ) χ 59, πειρηθώμεν Χ 381, πειρηθήτον Κ 444, optatives like τνπείης Ν 288, πειρηθείης Λ 386, ΰιακρινθε'ιτε Γ 102, and infinitives like ΰαημεναι Ζ 150 and ΰαμήναι Ν 98, άριθμηθή- μεναι Β 124, μνησθήναι δ 118 hardly need any further discussion, still less do the extremely numerous participial forms like άλείς Π 403, πληγέντε θ 455, μιχθείς Γ 48, ρεχθέντος Ι 250. The lighter passive future is represented in Homer only by ΰαήσεαι 2 γ 187, r 325, μιγήσεϊθαι Κ 365. Instances of the heavier are entirely wanting. It is noteworthy that the Dorians do not exclude the active personal endings for these futures (Ahrens Dor. 289), e.g. συναχθησοϋντι C. I. 2448, i. 25, ωατωθησώ ' ακοΰσομαι Hesych. [Phot, and Suid. add Δωριείς], ψανησεϊν Archimed. beside ΰειχβήσεται and the like. This 325 peculiarity is connected with the similar phenomenon in the case of the futurum exactum discussed above on p. 436. Finally we may mention a pair of remarkable deviations in the Dorian and Aeolian dialects. To these belongs the Heraclean conjunctive 3 This form was not included on p. 5 because the meaning is not at all passive,, but in formation it decidedly belongs to this category. ch. xix. PECULIARITIES OF THE TWO STEMS. 491 εγ-ϊηληθίωντι=έζειληθώσι (Tab. Heracl. i. 152) where e as often (cp. Cret. ϊ'ωμες=ΗοπιβΓ. εωμεν) is replaced by t. But it was very unex- pected to find in the decree of the honours of Damokrater, dug up at Olympia (Archaol. Ztg. 1876 p. 1 &.) the three passive forms with d : άνατεθφ, ΰοθφ, άποσταλάμεν 1. 32, 35, 37 by the side of γραψέν 1. 31. There can be no doubt therefore that the ά as an older phase of η was not rejected in this place either. This fact throws new light upon isolated forms with a, which crop up elsewhere, e.g. ετύπάν, which is found in some M.SS. in Theocr. iv. 53, but is generally replaced by irwnjy, and άπεσσούα given as Laconian [Xen. Hell. i. 1, 23] = *άπεσσύη (Ahrens. Dor. 147). I. THE PASSIVE STEM ΓΝ -η. It will be necessary to state somewhat more exactly what the language possessed in the way of these stems. For this purpose we make three divisions, Homeric, Attic and non- Attic forms. Of Homeric passive stems of this kind there are 22 or 23. I give some references for each of them : αγη (α) Γ 367, εάγη Λ 559, 3 plur. αγεν Δ 214 (κατεάγη Arist. Vesp. 1428). εάλη (α) Ν 408, 3 plur. &Χεν Χ 12. εβλαβεν Ψ 461, βΧάβεν Ψ 545, also Attic (Aesch. Thuc. Aristoph.). άνα-βροχέν only λ 586. k -Ιάην Γ 208, Ζαώμεν Β 299, ΰαηναι Ι 493, also in Theogn. Pind. and the tragedians. Ιάμη I 545, εΰάμημεν Ν 812, Ιάμεν θ 344 etc. also in Pindar and Attic poets. θερέω only ρ 23. εκάη A 464, Β 427, καήμεναι Ψ 210, κατακαήναι also ill Herodotus ii. 107. λ/πε^ Π 507, 67τεί Χίπεν αρματ ανάκτων, the reading of Aristarchus 1 avakoyov τοϋ έλείφθησαν,' while Zenodotus read Χίπον, which presents 32(> difiiculties. άποΧιπήναι does not make its apppearance before Dio Cassius. μίγη Ε 143, εμίγην Γ 445, μιγέωσι. Β 475, μιγήμεναι Ζ 161 etc. also in Herodotus and Attic writers. πάγη Δ 185, πάγεν Λ 572, irayiv Plat. Tim. 59. κατ-επΧήγη Γ 31, πΧηγείς θ 12, also in Herodotus and Attic writers. νπ-ερράγη θ 558, Π 300, ερράγη also in Herodotus and Attic poets. ρνη γ 455, also Attic. σαπήτ) Τ 27, σα -iry Herod., other forms of the stem in Attic writers. τάρπημεν A 780, έταρπήτην \p 300, τραπείομεν V 441, ταρπήμεναι Ω 3. τερσημεναι ζ 98, τερσήναι Π 519. τμάγεν Π 374, di -έτμαγεν Α 531. τράφη Λ 222, cp. above p. 488, also in Pindar, Herodotus, and Attic writers. ha -τρνφέν Γ 363. 492 THE PASSIVE STEMS. ch. xix. ε-τύπη Ω 421, τνπείης Ν 288, τνπείς Λ 191, also in Pindar and Attic poets. φάνη A 477, φάνημεν t 4:66, φανήτην Η 7 etc., common in later times. ε-χάρη Γ 23, εχάρησαν Γ 111, χαρέντες Κ 541, also in Pindar, Herodotus, Euripides, Aristophanes, Plato Rep. 606. There are further the following 35 additional from the Attic period, many of which make their appearance as early as Herodotus : tf-aXifrj Plat. Phaedr. 258. άλλαγήναι common Attic, εζαπαλλαγ^ ThllC. iv. 28, απαΧΚαγησομαι Herod, ii. 120, Thuc. This is the only Attic instance of a lighter passive formation from a stem decidedly derived. βαφ^ Plato Rep. 429. βραχε'ισα Aristot. Probl. 12, 3. γραψήναι Herod, iv. 91, Thuc. Plat. Dem. Cp. αναγραφή μεν Cret. inscr. Naber Mnemos. i. 114 ff. έκΰαρέντα Herod, vii. 26, άποΰαρέντα Xen. Anab. iii. 5, 9. 327 εζύγην Pindar, Tragedians, Plato. θλφήναι Aristot. κλαπέντες Plato Rep. 413, το ΰιακλαπέν Thuc. vii. 85, εζεκλάπησαν Xen. Hell. v. 4, 12. κατα-κλιιήναί Aristoph. Lys. 904, Plato, Xenophon. κοπείσαν Aesch. Ag. 1278, εϊ,εκόπη Aristoph. Nub. 24, Herod. Thucyd., συγκοπησεται Lys. iii. 34 κρύψεις Soph. Aj. 1145, κρυφή σο νται (M.SS. κρνβήσονται) Eur. Suppl. 543 Elmsley. έκ-\απήναι Aristoph. fr. 211 Dind. ζνν-ε-λέγημεν Aristoph. Eccl. 116, καταλεγήναι Lys. xxx. 8, also Thuc. Plat. Isocr. ΰαλεγήναι corresponding to ΰιαλέγεσβαι first in Aristot. Eth. M. i. 29. εκ-μαγήναι Plat. Thaeaet. 191. ε-μάνητε Eurip. Bacch. 1296 and corresponding forms elsewhere in poetry and prose. κατ-ορνχησόμεθα Aristoph. Av. 394. συμ-πλακτ] Demosth. ii. 21, συμπλακείς Soph. fr. 548 etc. άπο-πνιγεΊεν Xen. Cyr. viii. 2, 21, άπο-πνιγήσομαι Aristoph. Nub. 1504. ραψήναι Demosth. liv. 41. ριψήναι Plato, εζερρίφη Aeschines ii. 153. ε-σττάρην Soph. 0. R. 1498. ε-στερην στερείς Eur. Ale. 622, Hec. 623, with στερησομαι Soph. El. 1210, Thuc. iii. 2. εστράφην Solon fr. 37, 6, Hdt. Soph. Attic prose. σφαγείς Aesch. Eumen. 305, κατεσφάγη Xen. Anab. iv. 1, 23 etc. σφαλήναι from Aeschylus onwards in poetry and prose : Soph. Aj. 1136, σφαλήσεσβε Thuc. v. 113. νπο-ταγείς Phrynichus Com. ii. 603, then in post-Attic prose {Plutarch); cp. Nauck Bulletin de l'acad. de St. Petersb. xx. p. 506. τακήναι in tragedians (Soph. Tr. 463) and Plato (Phaedr. 251). ταφή ναι from Aesch. onwards (Sept. 1021). i -τράπην ditto (Aesch. Pera. 1027). ch. xix. THE PASSIVE STEM IN η. 493 επι-τρφήναι Aristoph. Nub. 1407, cp. Demosth. xviii. 194 etc. επι-τυφτ) Aristoph. Lys. 221, εκτυφηΰομαι in other comic writers. 328 ε-φθάρην from Pindar (Pyth. iii. 36) onwards in tragedians, Herodotus and Thucydides. vrj Plato Pep. 494. φνχήναι Aristoph. Nub. 151, Plat. Phaedr. 242. Besides these there are 32 more passive stems of this kind, which occur either only as quite unique in older non- Attic writers, or like άπολιπήναι in Dio Cassius, only in the post- Attic literature, or are quoted only by Hesychius. Of these we may notice the following : γηρέντος Xenophanes (fr. 8 Be. 3 ) according to Herodian ii. 829, and Ε. M. 230, 50 avbpbs γηρέντος πο\\6ν άφαυρότερος. γναφήναι' κλασθήναι, κναμφθηναι Hesych. Cp. Homer, επι-γναμφθή- ναι. Ζρακεντες Pind. Nem. vii. 3, Ιρακέϊσα Pind. Pyth. ii. 20. εριπέντι Pind. 01. ii. 43. icaprj Herod, iv. 127. άνα-παρείς ib. iv. 94 : cp. επάρη' εκεντηθη Hesych. And as instances of later imitative formations άγγελήναι Plut., now banished from Eur. Iph. Taur. 932. επάην και άνεπάην εν ry συνήθεια Choeroboscus, Lentz Herodian ii. 800, άιαπαήοΌνται Ν. Τ. εφλέγην Dion; Hal., Luc. etc. The total number therefore amounts to 89. We can plainly see the gradual developement of this formation, which however by its very nature could never attain to the wide extension of the second. A glance at our lists of the thematic aorist forms (above p. "283 ff.) shows con- siderable chronological differences. It may be further noticed that the future in the post-Homeric time is tolerably extensively formed from the stock of these stems once in existence. As Attic futures of the kind in addition to those already incidentally mentioned we find : βλαβησομαι Plat., Isocr., μετεγγραφησεται Aristoph. Equ. 1370, κατακΧινησομαι Aristoph. Plat., συλλεγησύμενος Aeschin. iii. 100, παγήσεται Ar. Vesp. 437, έκπλαγήσομαι Herod., Thuc. the Orators, εκραγ η σονται Aesch. Prom. 367, ειςρνηεσθαι Isocr. viii. 140, κατασαπησεται Plat., άποσφαγήσοιντο 329 Xenoph. Hell. iii. 1, 27, έκτριβήσεται Soph. Ο. T. 428, φανήσομαι, Ιιαφθαρησομαι. As to the vocalism, the stem-syllable of these passive stems shows a preference like that which we noticed on p. 278 above in the case of the thematic aorist, for the vowel a. 28 of the lighter passive stems have an α belonging to the root, e.g. άγψ, βα<ρη > γραφή, μανή, χάρη, 14 have α by the side of an ε in the root, e.g. άλη, ΰαρη, κλαπη, πλα^τ?, 3 ταρπη, only 5 retain the ε : άγγελη, θέρη, λεγη, τερση, φλεγη, ο appears only in άναβροχέν, εκόπην, ι is represented by 12 instances e.g. άλιφη, εριπη, κλίνη, μιγη, υ by 16 e.g. γλνψη, ζνγη, ρυη, φνχη. The long vowel in the above-mentioned γηρείς is quite abnormal : so are the diphthongs of the un-Attic forms : ήνοίγην, Christ, pat. 996, άνοιγήσομαι LXX, and of 3 ίΐτ\4κην occurs very often in the M.SS. as a variant for έπλάκηι/. In Polyb. iii. 73 συκίΓλίκτησαν has been accepted by Bekker and Hultsch. Cp. καταπλεκεΤσ-ι • ffvvHeCuffi, τΓ€ριπ€ΐτ\€'γμ4νοΐ5 Hesych. 494 THE PASSIVE STEMS. ch. xix. ξννεξερευθείην quoted from Hippocrates, though the authority for the last is but weak. The short vowel in the stem-syllable is as a rule as much liked in the lighter passive stem as in the thematic aorists. Roots ending in a vowel do not form stems of the kind, as is easily intelligible. Four roots in ν are exceptions : ΰυ (ΰιεκΰνήναι Hippocr.) πτυ (πτνήναι Hippocr.) pv, φν, and also the poetical la (Ζαηναι) where perhaps as in καήναι (stem καν, καΐ Α ) a spirant has been lost. The most various consonants are found before the η, but, curiously enough, never £. The most common is y, e.g. in άγη, ζνγη, λεγη, ραγή. In later Greek this intruded itself also into the place of χ in όρνγήναι, άναπτυγήναι (Hippocr.), άυγήναι, and corresponding to this we find at this late date κρνβή ναι beside κρυφηναι. Lobeck discusses these cases on Soph. Aj. v. 1145 : cp. Principles ii. 141. We come now to the difficult question of the origin of this stem, for which, as already mentioned, we are quite without the help of a clearly corresponding form in any other language. In my Tempora und Modi p. 330 I explained the lighter passive stem as from the rt. ja (Skt. ja) 330 go, the same as that which I considered to be the source of the present formation in -ja (p. 206 f.) and the Sanskrit passive suffix -ja. The close connexion of the Indian passive with the present expansion, which even in the case of active terminations is not unfrequently associated with an intransitive or even a passive force, has been confirmed by Delbruck's Altindisches Yerbum p. 166 ff. In the same way the use of the Latin ire in a passive sense (e.g. venum ire) serves to strengthen the hypothesis that a verb of going is at the bottom of this passive formation. My explanation, though put forward even by myself with some reserve, has found acceptance with several writers. Yoretzsch de inscriptione Cretensi (Hal is 1862) thought he had found a confirmation of it in the gloss of Hesychius* ελαχία* εΰάρη Κρήτες. He takes ε-λαχ-ία as the 3 sing, of a passive aorist from the rt. λαχ, which he regards as identical with that of the common Greek ράσσειν strike, smite, and he thinks that the assumed j of the rt. ja has been preserved here in the form of the vowel t. Such a confirmation of my previous view, based on original authorities, would be very attractive to me ; but although the ά of this form has now found support in the Elean form quoted on p. 491, I cannot make any use of it. Even if the explanation of this completely isolated ελαχία as a passive aorist should be admitted to be correct, it would not be by any means certain that ι here represented a j, for in the Cretan dialect this vowel before other vowels has not uncommonly originated from f, e.g. in 'ίωμες= Homer, εωμεν, in τίρως=. A tt. θέρεος (Helbig de dial. Cret. p. 25). Hence ελαχία might very well have come from *ε'λαχε'α. As a matter of fact there are Various objections to my previous explanation, which now prevent me from regarding it as correct. The j of the added root ja would have been treated in a completely different way in the passive stem and in the present stems. For the latter j gives rise to all those changes in the preceding consonants mentioned on p. 211 ff. I cannot see any reason why e.g. the hypothetical φρικ^ά-μι should become φρίσσω, but hpaK-ja ΰρακ-η, why φαν^ω should become φαίνω but φαν -ja φανη. The length of the vowel in ε-ΰράκη-μεν ε-Ιράκη-τε and in the 4 Compare iitSafifj • ίκκαυθη Λάκωνα in Hesych. according to the convincing justification of the M.S. reading by Ahrens Dor. p. 49, and δα/3ρ • καυθί}. "> W™^ στερέω, έστέρησα (p. 269) beside έστέρην f -^τ τραπέω, in a different application (p. 269) beside Ιτράπην ^Ο 1 * V %ρρνηκα (p. 270) beside έρρνην, ρνησομαι Ky τνπησβι ,, ,, βτνπην, τνπησομαι Λ^ 5 A somewhat differing attempt to explain these forms on the same >rincipk3-r may be found in Westphal's Formenlehre der Gr. Spracheii. 1, 290. ^He explains ϊ-μί-γη-ν to be for ς-μη^αν, regarding the e as the representative oX.theJ. Most ^^ of the arguments developed in the text hold good also as against this e^piana- tion • ΑΛ Ai> 496 THE PASSIVE STEMS. ch. xix. «χαρηότα, «χάρηκα j beside . . ' κεχαρηντο, κ€χαρησβμ€ν J A r ' In cases where there is no difference of meaning as in the forms of the roots δα, ρυ, χαρ it is the most impossible to deny the common character. For &οαη-κα and i -Βάψν, κεχαρηότα and ίχάρην we must set down the same stems δα??, χάρη, and we have to decide between two equally dubious hypotheses, viz. the one, that this agreement is based upon chance, and the other, that other tense forms beside the aorist and future can occasionally be formed from a passive stem. Others how- ever of the forms cited above, like στυγέω, γεγράφηκα, have an active meaning, so that there is not the slightest reason for deriving them from a passive stem. Hence we should only have pure accident left. After what has been said there can, I imagine, be no doubt about the true state of the case. We may say with confidence : the lighter passive stems are nothing but stems expanded by the addition of e, and inflected 333 in the Aeolian fashion. The e-stems have in fact repeatedly presented themselves as important new formations at different points in the sphere of the European languages, and they often appear, as we have seen especially on p. 259, and in the case of the future on p. 477, acting as i second' or vicarious stems by the side of shorter ones, especially, outside of Greek, in the Latin, Lithuanian and Slavonic present. The same stem-expansion we recognized in the case of the perfect as an extremely important contribution towards understanding the Greek as well as the Latin perfects. Here our investigation brings us back once more to the same phenomenon. If any one chose to quote the Elean forms in ά cited on p. 491 as an objection to our representation, we could answer that the Eleans too furnish the e in the neuter participle γραφέν, and that there- fore the ά in £o0£, άποσταλάμεν is based upon a very extensive prefer- ence of this dialect for ά, upon which we cannot form a more definite opinion, until more materials are at our command. But one circumstance, we must admit, seems to stand seriously in the way of our view : — the meaning. Whence comes the passive mean- ing for this passive stem. I cannot get over this point as easily as Schleicher does with what he says about 'function.' It is one of Schleicher's weak points that he withdraws into a cold nescience before all questions of meaning, with the timidity all his own. But it is true that transitive and intransitive, active and passive meaning are not unfrequently interchanged in an extremely capricious fashion. For instance it would not be easy to find any reason why the aorist άλωνα c along with άΧώσομαι reached its sharply passive force ; and some might wish to employ such examples in order to throw overboard any question as to internal reasons in the case of our passive stems. But the case is different here. It has been already pointed out more than once that by no means all passive aorists of this formation have a rigorously passive force. Kiihner Ausf. Gr. i. 2 p. 560 says ' the so-called aor. ii. pass, is nothing but an aor. ii. act. constructed according to the formation in μι with an intransitive force.' The meaning of very many of these forms is intransitive, not passive. Thus the category of the passive is 334 completely inapplicable to δούναι learn, γηρείς grown old, βερήναι warm one's self (επεί κε πυρός θερέω ρ 23), ρυηναι flow, μανηναι grow mad, σαπήναι rot, τακηναι melt, τερσημεναι dry {είματα δ' ήεΧίοιο μενον τερσί}- ch. xix. ORIGIN OF THEIR FORCE. 497 μεναι αυγή ζ 98, ονΰέ μοι αίμα τερβήναι δύναται Π 519), ψανήναι appear, χαρήναι rejoice; εξεκλάπησαν in Xenophon Hell. v. 4, 12 and elsewhere means not 'they were stolen' but 'they stole away/ ξυνελέγημεν in Arist. Eccl. 116 not ' we were assembled' but \ we assembled ourselves/ In short there can hardly be a doubt that here as elsewhere the passive force is only a special developement from the earlier, partly intransitive, partly reflexive, force. In this way we obtain for comparison a tolerably large number of formations phonetically similar, and as we shall now with more con- fidence assert, comparable with these passive stems in meaning and not merely in external form. "We saw on p. 244 f. that the intransitive force attaches to the verbs in -ε ω, though not exclusively, yet to a large extent, and much more commonly than to the kindred verbs in αω, οω. We there compared the prevalent intransitive usage of the Latin verbs in ere, 6 and the Church-Slavonic verbs in ejeti. Our Greek passive stems in e may be very well connected with these. At least in the case of some passive aorists Latin and Greek completely agree. We may compare torrere with τερσήναι. If the participle answering to the latter occurred, its stem would necessarily be *τερσεντ, which would exactly coincide with the Latin torrent : in the same way fulgere answers to φλεγήναι, and there is more justice in the old comparison of car ere with καρήναι than one might think at first sight, the two meeting in the notion of to be shorn, bereft. The late licet is to linquit much as ελίπη to Χιμπάνει or λείπει (Princ. ii. 61). Frequens presupposes a verb *frequere to be close, crowded, which agrees with φραγείς, for which there is but late authority (Princ. i. 376). The difference between the proceeding of the Greeks and that of the Romans lies mainly in this, that the former brought together transitive and intransitive forms into the unity of one verbal system, as indeed happens frequently elsewhere, 335 and even in the case of the most primitive verbs. Compare ϊ&τημ* — εστην, φνω — εφϋν. The Romans on the other hand treat pendere and pendere, jacere and jacere as distinct verbs, and carried each out through all the forms, though these were far less numerous. Hereby the e-stem inflected in the Aeolic fashion becomes one of the different valuable members in the body of the same verb, and thus acquires a very different appearance from that of the Latin e-stem. In Greek the intransitive usage may have been favoured also by the resemblance of the termina- tions -ην, -ημεν, -ητε, -ήσαν, -ητην to the past tense of the rt. ες*, a resem- blance indeed which in the case of several personal forms could only make itself felt with time. For we have learnt to recognize traces of ήσμεν, ή(Ττε, ήστην, hence it would be wrong to conjecture that such a resemblance was the proper source of the passive usage. If we survey once more the course which in our view the language struck out for itself in giving a stamp to these forms, we may lay down the successive steps somewhat as follows : 1) At a time when the European languages, and still more probably the two South-European languages, had not separated from each other, by the side of monosyllabic verbal stems, there came into use also in many cases disyllabic stems characterised by the addition of an e, inter- changing with the shorter ones. 6 Friedr. Haase in his < Vorlesungen uber lat. Sprachwissenschaf t ' i. 97 describes the characteristic of the verbs in ere as ' quiet reat in a state.' Κ Κ 498 THE PASSIVE STEMS. ch. xix. 2) These stems were inflected quite after the pattern of the derived stems proceeding from a-ja, and hence fell to the e-conjugation. 3) In this conjugation, as distinguished from others proceeding from the same primitive form, the intransitive meaning prevalently formed itself. 4) At a time when the process of creating these e-stems was at its height, the Aeolic method of inflexion, characterised by the long e, pre- vailed in Greece. 5) Thus were formed in imitation of the imperfects the indicative, in imitation of present imperatives, conjunctives, optatives, and verbal nouns the corresponding forms of the lighter passive stems. 6) Developing further the tendency pointed out under 3) these stems found their employment only with an intransitive and passive force. 336 7) It was only later on that the corresponding futures were formed from the same stems on the analogy of the numerous futures in -ησομαι. II. THE PASSIVE STEM IN -θη. Here it is unnecessary to enumerate the individual forms. From Homer onwards the passive formation in θ is very common. According to my collections there are in that writer 130 aorists of this sort from stems of the most various kinds, a remarkable excess over the 22 of the lighter formation. In later Greek it is not worth while to do more than count the forms proceeding from consonantal stems. There are 251 of these. From vowel stems, and from all denominative stems this passive stem is properly to be expected in every case, and it is a matter of no impor- tance whether it actually occurs or not. Hence I content myself with citing Homeric examples of verbal stems of different kinds. Homeric aorists of this kind are Ιιληθην £ 120, γνμνώθη χ 1, λνθη Ε 296, εκίνηθεν Π 280— εβέλχθης κ 326, ήΐχθη Γ 368, θρνλίχθη Ψ 396— ίρείσθη Η 145, ΰιεσχίσθη Π 316, πέλασθεν Μ 420, κρύφθη Ν 405, εβλάφθησαν Ψ 387, ενιχριμφβείς Η 272— άρτύ,θη Λ 216, όρίνθη Ε 29, φάανθεν Α 200— άγέρθη Δ 152, αερθεν θ 74— αιοεσβητε β 65.— Of peculiar forms I may mention άχθητι- λνπήθητι Hesych. as Mor. Schmidt is un- doubtedly right in reading in accordance with the alphabetical order, in the place of the αχητι of the MS. This aorist to the present αχνυμαι is especially noteworthy because of the present form άχθομαι. Also εφ-έ-ασθεν' ε'γελασαν, ΰιεχύθησαν, quoted before on p. 79 because of the augment, μερθείσα ' στερηβεϊσα, αμερθείσα placed already by Lobeck El. i. 37 along with μείρεται ' στέρεται and άμέρΰω. Another word μορθήναι ' πειραθηναι, γενέσθαι is less clear; still it must certainly belong to εμμορε, εϊμαρται. Of όΰυσθήιαι ' όΰύσασθαι, χολωθήναι nothing else is known. With regard to the vocalism of the stem-syllable we have to notice two points, first the a, which, as in the middle perfect and in the lighter passive stem (cp. p. 493), so here too sometimes answers to the ε of other forms, and secondly the intensified vowels. The appearance of the α we 337 touched upon on p. 82 in speaking of the Homeric εάφθη which with Aristarchus we derived from έπομαι. There is good authority for τραφθηναι ο 80, τάρφθεν ζ 99, τάρφθη γ 213, 251, φ 57, according to which I. Bekker was certainly right in writing ταρφθείη in ε 74, while ch. xix. THE PASSIVE STEM IN -θη. 499 κατεστράφΰησαν in Herodotus i. 130 (Stein, V. 1. κατ εστράφη σαν), εστράφθη in Sophron fr. 78 Ahrens, στραφθέντες in Theocr. vii. 132 make their appearance beside the Homeric στρεφθέ ντ ε, though for this in Ε 575 στραφθέντε is given as the variant of the cod. Μ. Ιαρθείς was used for the usual ΰαρείς by the comic writer Nicocharis (Mein. Com. ii. 844), εσπάρθηΐ', σπαρθί]σομαι have no authority according to Yeitch p. 529, άττοσταλθέΐ'τες is read on the inscription of the Cnossians C. I. 3053 1. 4, άπ-οσταλθϊί in the Schol. Τ on θ 21. — While in the cases mentioned a liquid was usually present, the α is produced by a nasal in φαάνθη Ρ 650 etc., φάανθεν A 200, εξεφάανβεν Τ 17, with which compare φαάν- τατος (ν 93). The similar εκτάνθη belongs only to late Greek (LXX). It is only, as is shown by άγέρθη Δ 152, άερθεν Θ 74, κερθέντες Pind. Pyth. iv. 82, a phonetic tendency appearing quite sporadically which is here in question. "We shall come by and bye to the a of ετάθη etc. An intensified stem- vowel, mostly in imitation of the present, meets us in this passive stem under exactly similar conditions as in the middle perfect and in the sigmatic aorist. We may take as examples : άλειφθηναι Lys. Plat., άπημείφθη Xen. An. ii. 5, 15, εδείχθη (Herod. εδέχθη), δειχθησομαι common Attic, επειχθήναι Thuc. i. 80, ζενχθήναι Pind. Trag., τενχθηναι Hippocr. beside Horn, ετνχθη (Δ 470), επεισθην common Attic, and so ελείφθην (as early as Hymn, in Merc. 195 7 ), εκπληχθήναι Eur. Tro. 183, σκηφθηναι Plato, συντηχθήναι Eur. Suppl. 1029. The ν before θ is not fixed in primitive verbs, a circumstance which again reminds us of the perfect (cp. above p. 419 f.). Instead of ε we find in such cases α: κατ-ε-κτα-θεν Ε -558 (cp. εκταν, έκτατο, κτάσθαι), έτάθηΐ' Homeric and Attic (τάθη Ψ 375 etc.). Perhaps εβάθη- εγεννηθη belongs here, supposing we are to assume that the gloss is Boeotian— in which case we must write εβάθει — from the root βα=γα, γεν recorded 333 in the Boeot. βανά=γννή. έβάθη would then be connected with γε- γάασι, γεγαώς, εκγεγάασθε. Still βαίνειν might also be intended in the sense of cover, beget. In the case of κλίνω and κρίιω the want of a nasal in the passive stems κλιθη, κριθή, may doubtless better be explained from the interchange of the stems k\l and κλιν, κρι and κ pi v. Hence the fluctuation in Homer : εκλίνθη Γ 360, κλινθητην Κ 350, but εκλίθη τ 470, κΧιθήναι α 366, ΐιακρινθημεναι Γ 98, κρινθέντες Ν 129, but ΰιεκριθεν Β 815, while subsequently the forms without ν are alone in use. The fluctuation between stems with and without ν met us in another way in Ζηρινθητην Π 756 beside Ιηρίομαι (Pind.), ΙΙρΰνθησαν Γ 78, Η 56 beside Ίδρυε Β 191, afterwards almost exclusively ιϊρνθήναί : άρτύνθη has the two present forms άρτΰνω and άρτνω. Forms like ώρίιθη Π 509, έμαράνβη I 212, εΐηοάνθη Φ 348, μιάνθησαν Π 795, where the ν is generally an in- tegral and indispensable part of the verbal stem, show how little there was any phonetic tendency to suppress the ν before Θ. Finally it is worth while noticing how the nasal makes its way in. from the present stem in the Herodotean ελάμφθην (vi. 92, ix. 119), for which elsewhere we have έλήφθην following the precedent of έίληφα, in late Greek ελήμφ- θην. Cp. p. 174. We have mentioned incidentally above that the two passive stems are in use side by side incomparably more commonly than the sigmatic 7 Ζλιφθεν Callim. Hymn, in Cer. 94 is certainly properly corrected by O. Schneider after Blomfield into ^Aci^e/. κ κ 2 500 THE PASSIVE STEMS. ch. xir. aorist by the side of the primitive or thematic. Even in Homer we find the pairs : βλάβεν and εβλάφθησαν (Ψ 387). Ζάμεν and Ιμηθεντα (Δ 99). μιγηναι and μιχθήμεναι (Λ 438). τάρπημεν and τάρφθη (φ 57). In Attic the following are in use : εζαλιφτ} and αλειφθεί' (both in Plato). απηλλάγη ν Aesch. Pr. 750 and άλλάχθ^ Eurip. I. A. 798 (ch.)' βαφηναι (Plato) and απεβάφθη (Aristoph. fr. 366 Dind.) εζνγησαν Plato Rep. 508 and ζενχθε'ισα Plat. Polit. 302. εκλίνην (cp. above p. 492) and έκλίθην (Plat. Xen.) κρνφείς Soph. Aj. 1145 and κρνφθεντα Soph. El. 837. ριφέντα Eur. fr. 486 D and ριφθέι-τες Eur. Hec. 335. Ιιετρίβη Thuc. i. 125 and τριφθεϊ&α Thuc. ii. 77. τακτ) Eur. fr. 230 and ξνντηχθείς Eur. Suppl. 1029. 339 For the usage of the tragedians Porson on Eur. Phoen. 986 (=972 Dind.) thought he had discovered the rule, 'asperas et antiquas formas adamarunt Tragici, ideoque aoristos priores praetulere.' But apart from the fact that many of the lighter aorists are just as old as the heavier ones, the rule is not observed, as Yeitch has shown under άλλάσσω p. 45 by a number of examples. The unrestricted choice between two equiva- lent forms still seems to many a scholar something unworthy of a formed language, though our own mother-tongue gives a precisely similar instance in the parallel usage of wob and webte, frug and fragte. [Cp. our own use of my and mine, loves and loveth, kilVd and killed : and see D. B. Monro on Homeric Grammar § 57.] In this case fortunately the metre not uncommonly presents insuperable obstacles to the passion for alteration. If we now attempt to arrive at the origin of the passive stems with 0, it is in the first place quite certain that these stems find their place in a larger group of forms which, though very differently used, are con- nected together by the addition of the same consonant. We must there- fore first take a survey of these forms. θ elsewhere than in the passive stem. The present forms in -θω and the past tenses in -θο-ν have been actively discussed and diligently collected by modern grammarians, not as yet following in the lines of comparative philology. The discussion was set on foot by the question raised by Elmsley on Eur. Med. 186 and Soph. O. C. 1015, whether the past tenses in -αθο-ν provided with this termination are, as the English critic maintained, aorists, or imperfects. Buttmann Ausf. Gr. ii. 2 p. 61 ff., Lobeck in his note on this passage, but especially Immanuel Herrmann in the Erfurt programme of 1832 'de verbis Graecorum in αθειν, εθειν exeuntibus' and still more thoroughly Wentzel in the Oppeln programme of 1836 'qua vi posuit Homerus verba quae in θω caduntT have discussed actively this question of meaning and accentuation — for the latter concerned the critics especially. G. Hermann has incidentally (on Soph. O. C. 1015, and on Eur. Phoen. 1184) expressed himself as against Elmsley. Valuable investigations in 340 a different direction are furnished by Lobeck Rhem. 92 ff. We can ch. xix. ANALOGOUS FORMATIONS ELSEWHERE. 501 therefore draw upon the collections of these predecessors in exhibiting here the stock of forms. We give past and present tenses side by side as our present question is only that of formation. A) Presents in -θω and Past Tenses in -θο-ν. 1) From monosyllabic vowel-stems. a) With a short stem-vowel. ε-σχε-θο-ν, poetical from Homer onwards (εσχεθε Μ 184, ανεσχεθομεν t 294, σχεθέτω θ 537, σχεθέμεν Pind. 01. i. 71, σχεθεϊν Aesch. Prom. 16, κατασχεθόντες Soph. El. 754). The other forms of the same kind ε-μαθο -r, ε-παθο-ν and ε-ΰραθο-ν (by the side of ε-Ιαρθο-ν) have been discussed above p. 280 and 284 f. They are distinguished from ε-σχεθο-ν by the fact that the θ is not limited to this one tense-stem. b) With a long stem- vowel. βρίθω from Homer onwards (βρίθει r 1 1 2, βρΊθομένη θ 307, βρίθεις Soph. Aj. 130, βρίθει Plato Phaedr/p. 247) with the perfect βέβριθε (Π 384). The accessory nature of the θ is shown by βρι-αρό-ς (Princ. ii. 77). The comparison with βρενθΰομαι and Lith. brqsti to fill (intrans.), suggested by Joh. Schmidt Yocal. i. 124, according to which the θ would belong to the root, can hardly be reconciled with the use of βρίθειν. γηθο-μένων Quint. Smyrn. xiv. 92, γηθομένΎ} Anthol. Pal. vi. 261, γηθονται Sext. Empir. p. 567, 11 ed. Bekker. There are further the perfect γέγηθα cited above p. 401, and the present γηθέω to be discussed hereafter. The rt. yaf underlies the shorter γαίω (Princ. i. 211). κνηθω, κνήθομαι, a by-form to κνάω, which can be quoted from Aris- totle onwards. νήθω Plat. Polit. 289, shorter present form v€w=Lat. neo. πλήθω from Homer onwards (πλήθει Φ 218, πλήθεν θ 214) Dor. πλάθω (πλάθονσι Aesch. Choeph. 589 chor.), in prose πληθούσης αγοράς. There is also πέπληθα, cp. above p. 401. ενέπρηθον only I 589. A present πρηθω is nowhere found. ττνθω poetical from Homer onwards : πνθεται Λ 395, and in later prose. 7ri/-o-r=Lat. pus, -πνέω show that the θ is an accretion (Princ. i. 341 356). άποσήθειν only quoted by Athenaeus xiii. 591 from the grammarian Herodicus, a pupil of Crates. c) With preceding consonants. αχθο-μαι. Cp. p. 265, and for the origin of the root άχ (άχ-νυ-μαι) Princ. i. 234. ε σθω poetical and late prose present from the rt. έ£, beside the usual εσθίω: εσθοντες θ 231, πίνε και ήσθ' π 141, εσθει Aesch. Ag. 1597, εσθυντας Philippides Com. Meineke iv. p. 469. 502 THE PASSIVE STEMS. ch. xix. 2) From disyllabic stems. a) In a. » άλκά-θω και άλκάθειν 8 Σοφοκλής καϊ Αισχύλος, σημαίνει δε βοηθεΐν Bekker Anecd. p. 383, 31. αμννάθετε Aristoph. Nub. 1322, άμννάθειν Soph. 0. C. 1015, Eur. Iph. A 910, αμυνάθου (imper.) Aesch. Eumen. 438, Ιιωκά-βω Aristopb. Nub. 1482, ΰιωκάθειν Eurip. fr. 364, 25, Plato Euthyphr. p. 15, έΰιώκαθες Ar. Yesp. 1203. νπ-εικάθοιμι Soph. El. 361, παρεικάθη Plato Sophist. 254, εΐκάθοντα Soph. Trach. 1177. ε-έργα-θεν Ε 147, άποέργαθε Φ 599, ζννεέργαθον & 36, άπειργάθρ Soph. Ο. C. 862, κατειργάθον (imper.) Aesch. Eumen. 566. κατ-ε-κείαθε • κατεκοιμήΟη Hesych. with which compare κίασθαι ' κει- σθαι cited p. 120. μετ-ε-κίαθε Π 685, μετ-ε-κίαθο-ν Λ 52, Σ 581, then in Apollon. Rhod. and Callim. Hymn, in Dian. 46. πελάθεις Aesch. fr. 131, πελάθει Rhes. 556, Aristoph. Thesmoph. 58- b) In e. ήγερέθονται to άγείρω Γ 231, -ντο Μ 82 etc., ήγερέθεσθαι Κ 127. Cp. άγράθεν (Doric?)• σννάγειν Hesych. on the analogy of έέργαθον. ήερέθονται Γ 108, Φ 12, then in Apollon. Rhod. 342 'Αρέθουσα, doubtless belonging to αρέσκω. εμέθω το εμώ Cramer Anecd. Oxon. i. 87, 7. έρέθονσι τ 517, δ 813, έρέθησι A 519, ερεθε Γ 414, ηρεθον Theocr. xxL 21, expanded ερεθίζω A 32. θαλέθων ψ 191, θαλέθοντες ζ 63, Ι 467, θαλέθονσιν Theocr. XXV. 16. νεμέθοντο Α 635, νεμέθων Nicand. Ther. 430. ε-σττερέθοντο ' εσπειρον Hesych. τελέθει Η 293, τελέθουσι Μ 347, τελέθοντες ρ 486, then in Pind. (Pyth. ii. 78), and the tragedians (Aesch. Suppl. 1040, Eur. Med. 1096) τελέθει Tab. Heracl. i. 111. ψαέθων A 735, Soph. Eur. Φαέθονσα μ 132. φλεγέθει Ρ 738, φλεγέθοντι Φ 358, ψλεγεθοίατο Ψ 197, Aesch. SuppL 87, φλεγέθων Soph. Trach. 99. χρεμέθωσι Oj)pian Cyneg. i. 163, έπιχρεμέθων Apoll. Rhod. iii. 1260. There is also with a long e άλήθω quoted from Hippocrates and Theophrastus. Babr. 131, 5 ηληθον. c) In v. βαρύθει Π 519, Hes. Opp. 215, βαρύθοντο Quint. Smyrn. xiii. 6. ηλνθο-ν quoted above p. 284. It is best to refer it to a rt. έλ with the expanded by-form ελυ, which occurs in προς-ηλυ-τη-ς, προς-ήλυ-το-ς and in έλι)λν-τε (above p. 387, Princ. ii. 179; cp. rt. Μλ beside αλυ-σι-ς, rt. hp beside hpv p. 122), especially as the aorist is thus brought into' 8 Elmsley and after him Dindorf and others write άλκαθ^ν and regard άλκάθω as an 'invention of the grammarians,' and so in the rest of the verbs, though I do not in every case mention it, in all corresponding instances. ch. xix. ANALOGOUS FOKMATIONS. 503 connexion with the present έρχομαι (p. 197). Fick's attempt to start from a root Xu0=Skt. rudh (Ztschr. xix. 250, Worterb. i. 3 200) is not at all borne out by the meaning, and would lead to the separation of 'έρχομαι and ήλνθυ-ν. The syncope of a υ in ηλθον would also be hard to explain. From ελ ηλ-θο-ν is formed independently, just as ήλν-θο-ν from ελν. For the perfect form cp. p. 398. But perhaps the Skt. rudh is itself only a variant of ardh, and comes like this from ar. μινύθει Π 392, μινΰβονσι Ρ 738, Hes. Ορρ. 244, Aesch. Eum. 374, Soph. Ο. C. 686. φθινύθουσι Ζ 327, φθιννθειν Β 346, φβίινθον Ρ 364, άποφθινύθωσι 343 Apoll. Bhod. i. 683. d) In consonants. A special group is formed by the presence of σ before θ : αίσθω ' θνμον αΐσθων Π 468, θνμον αίσθε Υ 403. The connexion with αημι, αί'ω (επεϊ φίλο ν αιον ητορ Ο 252), άάζω etc. from the rt. να breathe (Princ. i. 483 f.) cannot be doubted, άΐσθω is probably for ά/ετ-θω and is most closely connected with αετμόν ■ τηενμα Hesych. βιβάσθων, only in this form Ν 809, Ο 676, Π 534. The σ here probably comes from a c), so that βιβάσϋων is to be explained from the expanded rt. βαδ, occurring in βάΐ-ο-ς • βα^-ίζω. ώλισθον from Homer (Υ 470) onwards, όλισθάιω from Sophocles onwards, perhaps from a stem γλιτ (γλισ-χρό -c) : Princ. i. 458. ερέχθων ε 83, ερεχβομένην Ψ 317, Hymn. Apoll. Pyth. 180 is probably rightly grouped by Spitzner (Exc. 34 ad Iliadem) and others with the almost equivalent ερείκω, to which it is related on the ground of its second ε much as ερηρίΐατ to ερείΰω (above p. 418). Cp. Έρεχθενς and Έριχθόΐ'ίος. Β) Formations further derived. Besides θ we find other expanding elements in the following forms : γηθέω beside γηθόμενος mentioned above p. 501, is not uncommon in poets from Homer (Λ7 140, εγήθεε II 127) onwards, as well as γηθήσω 7 γήθησε. "^ηβέω is quite parallel to gaudeo, and is related to the shorter formation as ήθέω to σηΒω (Lobeck Bhem. 93 f.). νττ-εικαθεων only in Oppian Halieut. v. 500. όρεχθεον only Ψ 30 πολλο\ μεν βόες άργοι όμέχθεον άμψι σιΰήρφ : Spitzner and others are certainly right in taking it in the sense of ώρέχθησαν. The view of Wentzel, that the Homeric forms άνσχεθέειρ (ε 320) and εν σχεθέειν περί τέρμα Ψ 466 are not aorist infinitives but present forms of the same formation as the three just quoted, is perhaps correct. An ι forming the present meets us in εσθίω (cp. p. 207). The following again are expanded in a different way : 344 κιναθ-ίζω, which occurs only in Hesych., 9 while the substantive κινάβισμα thence derived is found in Aesch. Prom. 124. όροθύνω. όρόθυιε Ν 351, Ο 595, ε 292, όρόθννον Φ 312, ώροθννετο Aesch. Prom. 200, evidently belonging to όρίνω, ορνυμι. 9 Of the different meanings of this verb Ιδιάζειρ, αποθησαυρίζει? κατά μικρόν συλλβγοι/τα, %νιοι μινυρίζζιν κάϊ κινέΐν it is only the last which comes into considera- tion here. 504 THE PASSIVE STEMS. ch. xix. It is worth while noticing how various the stems are from which these formations are produced. By the side of stems little differing from the root, such as those cited under 1), we find stems which we are justi- fied in regarding as present stems. This comes out most plainly in the case of φθινν-θω and μινΰ-θω, from which we can restore the present stems in -νυ (p. 108 ff.). As φθι-νύ-θω is to τί-νυ-μαι, SO is ήμύ-να-θο-ν (cp. Fritzsche Stud. vii. 386) to a hypothetical *άμύ-να-μαι, and i -κί-α-θο-ν to the i -κί-α-το ' εκινέίτο (cp. p. 120) actually preserved in Hesychius, κατεκείαθε to the κία-σθαι ' κεϊσθαι also quoted by him. Hence we shall be inclined to compare also the α of πεΧά-θειν, άλκά-θειν with that in άγα-μαι, ερα-μαι (p. 118 ff.). C) Meaning. This is to be discussed in two directions. In the one case it is a question of the interchange between active or transitive and passive or intransitive usage, — relations which have to be taken into considera- tion especially for the understanding and the correct analysis of the kindred passive aorists, — and in the other of the dispute, which has been carried on with more zeal than insight, whether the past tenses belong- ing here are past imperfects or aorists. With regard to the first point, we meet with an extraordinary variety. We can distinguish three cases : 1) Decidedly transitive forms. To these belong άίσθων (θυμόν), άΧηθειν grind, εσθω and εσθίω, ερέθω (οζεϊαι μεΧεΰώνες όΐυρομένην ερέθουσιν τ 517), £ι ω κάθε ι ν (αισχρούς έρωτας οημοτών" ΰιωκάθειν Eur. fr. 364, 25, κάπρον εΐιωκαθές ποτ* Aristoph. Vesp. 1203), ενέπρηθον μέγα άστυ Ι 589, νήθειν spin, σηθειν sift, εσπερέθοντο ' εσπειρον, ΙΙηΧείωνα — άποέργαθ ε Χαοΰ Φ 599, εμέθειν =εμεΊν, Άργείους δε ΤΙοσειΰάων όρόθυνε Ν 351. 2) Decidedly intransitive. 345 ε δρα 00)', εκίαθον, ει κάθε ι ν, πελάθειν, μακρά βιβάσθων Ν 809, βρίθειν, ήερέθονται, ήγερέθοντο, νεμέθοντο (they were feeding — pascebantur), βαρΰθει 3ε μοι ώμος υπ' αυτόν Π 519, Hes. Ορρ. 215, κατακείαθε ν' κατεκοιμηθη, ηιθεοι θαΧέθο ντες ζ 63, ήεΧιος φαέθων Λ 735, νυξ V ήΰη τελέθει II 293, χρεμέθειν, πυθεσθαι, άχθεσθαι, όρέχθεον. 3) Fluctuating instances in Homer. πόΧιν ψΧεγέθει Ρ 738, πυρσοί φΧεγεθουσι Σ 211, ή& οτινας μι- νύθη Ο 492, μινύθει ΰέ τε εργ* ανθρώπων Π 392, <ρθι ν ύθουσι ψίΧον κήρ κ 485, τούςΰε δ' εα φθινύθειν Β 346, άνεσχέθομεν Αιι χείρας ι 294, ουΰε Ζυνάσθη αίφα μάΧ' άνσχεθέειν μεγάΧον υπό κύματος ορμής ε 320; πΧηθειν in the earlier time is only intransitive : ποταμψ πΧηθοντι εοικώς Ε 87, πΧηθυΰσης αγοράς (Attic), but in later poets πΧηβειν is tran- sitive also : πΧηθει £' αυτέ κύπεΧΧα βοών γΧάγος ήΰε καϊ οιών Quint. Smyrn. vi. 345. ch. xix. FORCE OF ANALOGOUS FORMATIONS. 505 From these groupings it becomes plain that we cannot talk of any- definite meaning for this class of verbs. Still the intransitive force is the prevalent one. The other question, whether the forms in -θυν, -Οομην are of an aoristic or a present kind, has, as we saw, raised much dust. After all that has been said in this book as to the nature of the aorist, it hardly needs any further explanation that here as everywhere the aoristic force u, which occur as intransitive in the most different verbal and nominal forms, it would be hard to establish this clearly for any form whatever belonging to the rt. θε. Hence I hold it to be idle to ascribe to the syllable θη itself the intransitive force which comes out so clearly in the passive aorists. Besides, this would create a separation between the forms in θη-ν on the one side and those in -θο-ν, -θω on the other along with the weak past tense in the Teutonic languages, which we are hardly entitled to assume. The correct view seems rather to be the following. We must give up treating the appended syllable as the immediate sign of the force which appears in the usage of these forms. Scholars were not previously 352 sufficiently aware of the interval which exists between the time at which a category of forms originated, and the time, often long subsequent, at which their employment became fixed. The agreement of so many languages proves that even before their separation the composition of more significant roots with the root dha * do ' was carried out tolerably extensively. We cannot talk of a special force for formations of such a kind, for every verb in itself denotes an activity, and it is tolerably unimportant whether the exponent of the conception of activity is expressed or omitted. The periphrastic use of the verb ' do ' in German dialects, e.g. * er that kommen ■ and in English e.g. ' did you come 1 ' * he did not come/ presents us with a parallel from a later period of language to what we assert for the earlier period. Such compositions with dha were then, I believe, used at a much later period, during which men were no longer conscious of their origin, in order to complete in various ways the system of verbal forms which was gradually becoming more widely ramified, and to fill up deficiencies, which had originated owing to phonetic losses. In this way the Teutonic weak verbs got their past tense, though in the rt. do, which in the French actuel expresses rather the present time, certainly nothing is contained which could point to the past. Thus the aorist stems in -η and -θη, which in their form were active, came to be used as intransitive and passive. The former formation has been thoroughly discussed above. We were able in that case to discover many intermediate links and stages for the developement of meaning. A similar service is done for us here by the forms in which is connected with the ordinary thematic vowels. As to the phonetic relation between these forms in -θω and -θο-ν and those in -θη-ν, we can hardly feel any doubt. In discussing the verbs which follow the older method of inflexion we have seen repeatedly that all the so-called verbs in -μι have a tendency to follow the stream of the verbs in -εω, which in time overspread everything, and that in two ways, the thematic vowel on the one hand being added to the final letter of the primitive stem, or the final primitive vowel on the other changing into the thematic vowel. In the first way from -ιά-μι (νη-μι) came 353 510 THE PASSIVE STEMS. ch. xix. -νά-ω, from νϋ-μι -νν-ω, from *βά-μι */3ά-ω, from *στά-μι στάω (Lat. sto), from *θη-μι the *θέω occurring in the Homeric προ-θέουσι, in the other from -να-μεν νο-μεν, from ε-θε-ν-το ε-θο-ντο, from Ιε-ίη-ν ιο-ίη -r. In the same way we have in the case of the formations in θ three stages : 1) those with the final letter of the root retained -θη-ν, -θη-μεν, e.g. εσγέβην : 2) those with an added thematic vowel : όρέχθεον, γηθέώ : 3) those with a thematic vowel taking the place of the final vowel of the root : 7τλ?;-0ω, ε-σγε -do-v. The second class of forms is quite small in number \ the third was limited to some tentative forms of the earlier time, which were after- wards almost entirely lost. There was no definite force attached, as we saw above, but the intransitive usage was prevalent. It was only the first class with its archaic inflexion, which became an important part of the verbal system, and which adopted throughout the intransitive and passive force. I do not consider it impossible that aorists of the lighter formation like εγάρην, εμίγην, εΰάην, ετράπην preceded them in this course, and that the completely similar terminations of the two groups of forms contributed to bring them near to each other in meaning also, or, to put it more exactly, that the forms with -θη, originally less differentiated, by degrees under this influence practically dropped alto- gether the active force, and established themselves for the most part only in an intransitive or passive sense. As every one knows, there was no period of the Greek language which was entirely without passive aorists in θ which were used in connexion with a middle present quite as active, and in fact sometimes as transitive, as in Homer άλήθην ξ 120, αΙΙέσθητε β 65, νεμεσσήθητε Π 544, πεφηθήναι Ε 220, later εβονλήθην, ωήθην, εΐυνηθην, ΰιαλεχθήναι, ττορενθηναι. We may recognize in these traces of an indefiniteness originally much more extensive. 511 CHAPTER XX. 354 THE VERBAL ADJECTIVES. All participles may properly be called verbal adjectives in the wider sense, as we saw on p. 2, inasmuch as their nature is essentially that of adjectives, while they share the various characteristics of the verb. But the name is ' restricted in practice to two Greek formations which are distinguished from participles by the fact that they share to a much less degree than these do the specifically verbal varieties of meaning. The participles, in spite of their adjectival character, are capable of denoting the ' kind of time/ and in part also (partic. f uturi) the ' grade of time ' ; and are distinguished from each other by a definitely regulated division into active, middle and passive. Hence, in respect of their form, the difference between the unstrengthened verbal stem and the expanded present stems, and the whole multiplicity of the tense-forma- tion are expressed in the participles. In consequence of this every participle belongs to some one definite tense-stem. On the other hand from each verb there is only one pair• of verbal adjectives, which for this very reason are derived from the verbal stem, and only here and there adapt themselves exceptionally in particular cases to the present stem. The only power shared by the verbal adjective with the verb is that of serving for predication in a higher degree than any ordinary adjective, and of being capable of the distinction between active and passive, though with a decided preference for the latter. Considering the close affinity of meaning between the most usual verbal adjective and the participle, we can easily understand how in those languages in which the variety of the tense-stems falls quite into the background, e.g. in Latin, the verbal adjective is reckoned among the participles. Of the two verbal adjectives, which were at the command of the Attic language in the case of every verbal stem, one, that in -το, which 355 is most extensively in use from Homer onwards, finds a parallel in all the other Indo-Germanic languages. We can hardly doubt that an adjective with the suffix -ta, used essentially with a passive force, belonged to the stock of nominal forms closely connected with the verb, which we may assume as existing at the time preceding the separation of the languages. Referring for details to Bopp Yergl. Gr. ii. § 818, and Schleicher Comp. 3 p. 421 ff., I content myself here with setting forth the entire agreement of the Greek formation with that of the cognate languages in the case of a number of simple and evidently very ancient forms. /3a-ro-c=Skt. ga-td-s γνη-τό-ς (cp. ΰώ-γνη-ο-ς), Lat. gna-tu-s Goth, airtha-kun-th-s ' earth- born.' y>'w-ro-c=Skt. -ro-c=Skt. gru-td-s Zd. grii-td Lat. (in)-clu-tu-s πεπ-τό-ς cp. Skt. pak' -a-ta-m (cooked) Lat. coc-tu-s Lith. kep-ta-s ra-ro-e= Skt. ta-td-s φερ- τό-ς = Skt. bhr-td-s. The verbal adjective in -τέο-ς is quite unknown to the language of Homer as denoting necessity, — the force which it afterwards usually has — as has been noticed by Kiihner Ausf. Gr. i. 716 and Leo Meyer Vergl. Gr. ii. 383. But we may find a precursor of this formation in the Homeric νη-γάτεο-ς, used indeed only of articles of clothing (B 43, Ξ 185), but still doubtless rightly referred to *νεή-γα-το-ς as the verbal adjective from the rt. γεν (cp. γέ-γα-μεν). But *γα-τεο-ς is to the pre- sumable *ya-ro-c (cp. τηλν-γετο-ς) as ϊο-τέο-ς is to δο-τό-ς. In Hesiod we find the form φατειός, only in the thrice recurring formula ούτι φατειός -^infandus, and in usage not differing much from the Homeric- άθέσφατος. The passages are Theog. 310, Scut. 144, 161. It is only in Herodotus and onwards (διωκτέος, ΰοτέος) that we find the forms in -τέο-ς with their well-known force firmly established. From Aeschylus I have one solitary instance in my collections : Choeph. 298 έργαστέον : there 356 are far more in Sophocles and Euripides. But Plato, Xenophon, and Aristophanes are the earliest writers who supply them in great abun- dance. 1 Evidently this second verbal adjective, as Kiihner i. 716 remarks, belongs quite especially to the colloquial Attic. These facts are of some importance in the enquiry into the origin of the suffix -τέο-. This is commonly compared with the equivalent Skt. -tavja. So Bopp- Vergl. Gr. iii. § 902, Schleicher Comp. 3 p. 382. The phonetic possibi- lity of the agreement of a ddtavja-s with the Gk. ΰοτέο-ς cannot be denied. The suffix -tavja might in Greek become first -τεΐβο, then -τειο, and finally -τέο. The second stage, to which φατειό-ς quoted above bear» witness, would be a parallel to άστεΊο-ς, which has certainly come from αστεϊ-ιο-ς, though it does not undergo a further reduction from ει to ε. An analogy for the latter might perhaps be sought in the Homeric βαθέη=βαθεϊα for βαθεΐ-ια, ώκέα=ώκεία for ώκεΐ-ια. The only question is whether the comparison of the two suffixes has on other grounds so much in its favour, that we are compelled by overpowering reasons of probability to assume such a considerable mutilation. Now these can scarcely be said to exist. As we have seen, -τέο-ς in its ordinary force is unknown to the Homeric poems : -tavja-s according to Delbriick Altind. Verb. p. 238 is unknown to the Bigveda. This fact of itself is of some weight as against the identification. In the cognate languages no trace of this suffix has been anywhere pointed out. For though some scholars were formerly inclined with Bopp u. s. to refer the Latin suffix -two, e.g. datwu-s, captivu-s, to the same source, this view, attacked already by Schleicher Comp. 3 382, must be unconditionally sur- rendered. From -tavjas we could at most get to -tiviu-s, never to tivu-sr. 1 The industrious • Quaestiones de adjectivis graecis quae verbalia dicuntur r h Moiszisstzig, which are buried in a series of programmes of the gymnasium a; fvonitz, I have only been able to hunt up in part. According to the ' Particula' which appeared in 1868 of verbals in -rtos there are ' non ita multa apud lyricos vales et Aeschylum, perpauca apud Herodotum, nee plura apud Thucydidem> apud posteriores innumerabilia fere.* ch. xx. OKIGIN OF THE SUFFIX -τεο-, 513 Besides, the Skt. suffix -tav-ja has evidently arisen out of the suffix -tu- by adjectival expansion. Hence -tav-ja attaches itself to the infinitives 357 in -tave (e.g. gatave, dcitave) and -tavdi (e.g. ddtavdi), so common in the Yeda, and to the later accusatival infinitives in -tu-m. Now the suffix -tu in the form of -τυ is by no means unknown to Greek (e.g. Ιτύ-ς, έΰη-τύ-ς) but it is not used with an infinitive force, nor does it anywhere show a trace of the by-form -tav, produced by intensification, which must be presupposed for the further formation -tav-ja. Again the support, which some have thought they could find in the similarity of meaning, is not so strong as it seems to be. For modal applications of adjectives of this kind are demonstrably not very ancient, as we can see by comparing the Latin datu-s with the Greek Ζοτό-ς. If there had been a verbal adjective provided with the definite function of the participium necessitatis in the time before the separation of lan- guages, we may be sure that this would not have been lacking in the Rigveda. Besides, this formation was at no time the only one in use for such a purpose in Sanskrit. Taking all together, I believe that the correspondence of -tavja-s and -τέο-ς is merely apparent, and that we must rather regard both formations as products of the separate life of the two languages. The suffix -τεο I regard as an expansion of -το. ΰο-τέο-ς is related then to ΰο-τό-ς as ήγάθεο-ς to άγαθό -c, ΰαίΰάλεο-ς to ΰαίΰαλο-ς, ΰαφοινεό-ς (Σ 538) to ΰαφοινό-ς (Β 308), κυάνεο-ς (Λ 39) to κνανό-ς, which we may infer from κνανό-πρωρο-ς. κνανο-χαίτη-ς, though it occurs uncompounded only in late poets, λαΐ^ο-ς (X 154) to Χάίνο-ς (Γ 57) and much as the Hesiodic λοχεό-ς (Theog. 178) to the equivalent λόχο-ς. The suffix -εο, used to form adjectives from adjectives with little or sometimes abso- lutely no change of meaning, can hardly be very different from the suffix -io, by the addition of which άπατήλω-ς (ξ 288) differs from άπατηλό-ς (A 526), έλενθεριο-ς from έλενθερο-ς, and by which άεικέλω-ς is fuller than έίκελο-ς. The close connexion between -εο and -io is made plain by pairs like γηράλιο-ς (Hesych.) and γηραλέος, νηφάλιο-ς (Aesch.) and the later νηφαλέο-ς. That -εο in the cases quoted comes from -εω is made probable by βρότεο-ς (r 545) beside βρότεω-ς (Aesch.) βόεο-ς (Ρ 492) beside βόειο-ς (Ρ 389), χρνσεο-ς beside χρύσειο-ς, both Homeric, and other instances of the kind. Various useful collections to 358 the same effect are contained in the doctoral dissertation of Aly ' de nominibus w suffixi ope formatis' (Leipzig 1873). I believe we may form a conception of the course of developement in the following way. From the earlier and shorter verbal adjectives in -το by-forms in -τειο (φατειό-ς) were produced in accordance with numerous precedents in the case of other adjectives : and this -τειο was afterwards shortened into -τεο. Originally there was only an imperceptible difference in meaning between this suffix and -το. But as by degrees a definite category of meaning arose for this formation, it became, first in Attic, more and more common and gradually a distinct verbal adjective to be expected from every verbal stem. In a precisely corresponding manner from the shorter suffix -αλο (χθαμα\ό-ς, α/θαλο-ς, ομαλό-ς) the much more usual suffix -αλεο, which Aly op. cit. points out in about 80 words, branched off by degrees. It was only during the developement of such a new category of meaning that the accent became fixed, and that not only in the verbal adjectives but also in the adjectives in -αλέο, on the penulti- L L 514 THE VERBAL ADJECTIVES. ch. xx. mate. How easily subsidiary modal meanings find their way into the usage of adjectives may be shown by the example of the adjectives in -t/io e.g. αγώγιμος, φύζψος (ε 359), αλώσιμος (Aesch. Ag. 10), εκπετησιμος (fledged, Ar. Αν. 1355), while the Romans derive from the formations with I their adjectives in -ili-s, -tUi-s, -bili-s with cognate meanings. In their formation the forms in -to and -τεο are most closely parallel, and were evidently felt by the instinct of language to be quite of the same nature. The verbal adjectives in -το show even in Homer the same characteristics, as compared with the verbal stems, as later on : άγητό-ς, γνωτό-ς, τρητό-ς, τρώτο-ς, — εράτό-ς, κριτό-ς, χϋτό-ς, — υφαντό-ς, άσπαστό-ς, απρηκτο-ς, ραπτό-ς. The quantity in the vowel-stems and the accessory σ will occupy us in the next chapter. In the case of con- sonantal stems, just as in the perfect middle, there is a fluctuation between the short root- vowel and the intensified vowel of the present : on the one hand τυκ-τό-ς (τυκτην ρ 206, ευτυκτον Γ 336, cp. τέτνκται), on the other hand νεότευκτου Φ 592 (cp. τετεύχαται), on the one hand πκττό-ς from Homer onwards (O 331, cp. επέπιθμεν), on the other πει- 359 στεον, from Sophocles onwards, εύπειστος, the latter forms in meaning too coming nearer to the present stem, ρηκτό-ς (Horn.), τηκτό-ς (Soph.), \ηπτό-ς (Plato), Ζιάμειπτος (Sappho), εζάλειπτέον (Lysias), φευκτό-ς (Soph.) beside άλαστος (Homer), άπυστος (Homer), πυΰτεον (Plato). There is a very strange form ειμαρτό-ς (Plut. Alexand. 30), which may serve to show us how strongly was felt the analogy of the verbal adjec- tives to the 3 sing. perf. mid. On the other hand στορνυτέα ' καταστρω- τέα preserved by Hesychius is due to the intrusion of the present stem. The distinction between the dialectic βελτός, also preserved by him, and βλητός is purely phonetic. In Sanskrit and in Latin two ways of attaching the suffix to the stem are in use : viz. either immediately : Skt. juk-ta-s Lat. junc-tu-s, or by means of an i : Skt. kup-ita-s, Lat. gen-i-tu-s. In Greek this two-fold character is known only to a small extent. Sometimes ε acts as an apparently inserted vowel here ; but upon closer consideration we can see that it is a stem-expansion, as in γαμί-τη (Plato), ευρετό-ς (Xen.), ευρετεος (Thuc), άμάχετο-ς (Aesch.), μαχετέο-ν (Plato) beside μαχητός (μ 119), μενετό-ς (Thuc), that is in verbs ofthee-class alone, which in γαμέω, ευρεσις, μαχεσομαι, μεμενηκα show other traces of an • appended e. There is an unique instance in ελετό-ς I 409 formed from the aorist-stem with the thematic vowel; to which we may add υπελθετέον quoted from Strabo xiii. 622. The same e appears in some adjective and substantive forms, which, though not felt any longer to be proper verbal adjectives, yet in their origin can hardly have been different, like άρι-ΰείκετο-ς (Homer), ά-μαιμάκε-το-ς (Homer), σκελε -ro-c, ττάχε-το-ς (θ 187), Ια<ε-τό-ν, ερττε-τό-ν. The masculine adjectives in -ro-c and the feminine in -τη which have become substantives and denote actions, like εμε-το-ς (cp. vomi-tu-s), νιφε-τό-ς, υε-τό-ς, κάπ-ε-τ-ο-ς, άρε-τι), (cp. άρέ-σκω), γενε-τη (cp. geni-tu-s), τελε-τή and others serve to confirm this view. In such substantives we sometimes find in Sanskrit an a instead of the *', e.g. paJca-td-s fire (pale cook), mara-td-s death, which makes it very probable that there was once an a in this place, just as in the Gk. ακάμα-το-ς, αΐάμα-το-ς (cp. domi-tus), θάνα-το-ς, and that the ε as well as the Indian and Latin i are only phases of this a, in which 360 perhaps we ought to see nothing but the thematic vowel. As this •ch. xx. THE MEANING OF THE VERBAL ADJECTIVES. 515 vowel elsewhere characterises the present or aorist stem, we may regard the stems appealing before the suffix -ta to a certain extent as present or aorist themes, a view which has forced itself upon us irre- sistibly already in the case of έλθετέον. In the vocalism of the root- syllable also we found clear traces of the extension of the present stem to this place. Finally we have still to say a word as to the meaning of these forms. A large number of the forms in -ro, e.g. βροτός, γνωτός, θνητός, κεστός, κλντός, πονητός always continued to be purely participial; and here, «specially in the case of compounds, we may notice the freest inter- change between the more usual passive and the rarer active force. Thus αναίσθητος means insensible, άπρακτος sometimes ineffectual, ανέλπιστος not merely unexpected, but also hopeless, πάμφθαρτος all-destructive. But even the simple μεμπτός in Soph. Trach. 446 means blaming. Cp. Kiihner Ausf. Gr. i. 2 715. A glance at the active participles of the Latin deponents like nactu-s, usu-s, locutu-s, potitu-s is sufficient to make this interchange much less surprising than it might seem to be at first sight. Even in the suffix -τεο we find an interchange between the personal passive application, η πόλις ώφελητέα and the neuter active την πόλιν ώψελητεον. The modal force of possibility probably established itself quite as imperceptibly in the verbal adjectives in -ro, e.g. in άίστος, άπνστος, as in the Latin participles invictu-s, acceptu-s and in the adjec- tives in -ιμο- mentioned above. Still even in Homer there are unmistakeable cases of it : e.g. Β 361 οντοι άπόβλητον έπος εσσεται οττι κεν ε'ίπω, Ι 526 2ωρητοί τ επέλοντο παράρρητοι τ έπέεσσιν, Α 573 ονΰ' ετ ανεκτά, Ι 409 άνΰρος £ε ψυχή πάλιν έλθέμεν οϊιτε λεϊση) ονθ' ελετή, Ζ 434 'ένθα μάλιστα άμβατός εστί πόλις. Indeed in the lack of a verbal adjective of necessity the Homeric language sometimes applies the verbal adjectives in -το in a manner which approximates closely to the later usage of those in -τεο, e.g. r 260 (=r 597, ψ 19) Κακοίλιον, ουκ όνομαστην, θ 307 έργα γελαστά. Afterwards the modal application evidently constantly grew more common. But in time, in the case of a large number of the nume- 361 rous adjectives compounded with prepositions, the delicate distinc- tion pointed out by Lobeck Paralipomena p. 477 sqq. established itself, that the barytones e.g. ΰιάλυτος = solutus were used purely partici- pially, the oxytones e.g. ΰιαλντός -^dissoluhilis with a modal force. Evidently the accent in the case of the latter raises the syllable, on which it is placed, to a more decided force (cp. Kiihner Ausf. Gr. i. a 415). L L 2 516 IRREGULARITIES OF THE VOWEL STEMS, ETC. ch. xxi. CHAPTER XXI. IRREGULARITIES OF THE TOWEL STEMS IN THE FORMATION OF THE PERFECTS, FUTURES, PASSIVE AORISTS AND VERBAL ADJECTIVES. Now that we have discussed all the essential groups of the Greek verbal system which can be shown to be in frequent use, we must enter upon an irregularity which extends tolerably deep into the structure of the verb. We have already repeatedly touched upon the abnormal phenomena, which may be detected, in the case of vowel stems, in the formation of tenses, consisting partly in the varying quantity of the stem-vowel, partly in the apparently very capricious insertion of a σ. The sphere over which this irregularity extends, includes the perfect stem, especially the perfect middle, the future stem, the sigmatic aorist, the passive aorist with θ and the verbal adjectives. This was the reason why we reserved this phenomenon for a general discussion at the end; though this cannot claim to be exhaustive, for that would require very extensive material derived from nominal formation, which is foreign to our present purpose. My aim at present is chiefly to bring the processes belonging here into the correct points of view, of which some, I hope, may be clearly established. In sharp contrast to the stems ending in consonants, the union of 3 β 2 which with the appended tense-forming syllables hardly ever shows any variation, the final vowel of a stem in the groups mentioned is some- times long, (and this is the rule) e.g. βήσω, ένησα, ΰεΰάκρΰσαι, ορθωθείς, αδήριτος, sometimes short, e.g. γεΧάσομαι, αιΐίσασθαι, κέχνται, ΰοθήναι, φθιτός, and sometimes after short, sometimes after long vowels in the perfect middle, in the passive aorist, and in the verbal adjectives there appears what seems an extremely strange, and hitherto insufficiently explained sigma, e.g. εσπασται, επρίσθη, γνωστός. Nominal formations like βήμα, ορθωσις beside ΐότης, χυσις, ψθίσις and σπασμός, πρΊσμα, αναγνώστης show the same variation in the stem. These different phenomena are unmistakeably connected with each other to a certain extent, but it is by no means possible to explain them on any single principle. We must for the present be satisfied with dividing them into two groups clearly distinct one from the other, and only touching each other here and there : viz. I. FORMS WITH A SHORT VOWEL WITHOUT σ. Roots which follow the primitive conjugation generally leave their vowel short in tense-formation as in nominal formation, e.g. dcdorcu, ετέθη — δόσις, θέτης, θέμα, φάτις. Here the short vowel is properly in no way surprising. For why should not the. root, which shows itself to have a short vowel also e.g. in είομεν, έθετε, εφασαν, do the same in the CH. xxi. SHORT VOWELS IN THE STEM, 517 cases mentioned Ί If side by side with these the long vowel now and then appears, e.g. in δώσω, έστηνα (beside εστάμεν) in δώτορ, θήκη, we may compare this difference with the interchange which exists between ψαμεν and φημί, 'ίμεν and ειμί (cp. p. 96 ff.). A very uncertain vocalism was to be seen also in the cognate primitive aorists like βή-την and βά-την, εθεμεν beside Skt. a-dhd-ma, as was pointed out on p. 135. The proper origin of this interchange of quantity has not indeed been discovered by any means as a rule, but it extends over a wide circle of similar forms, and may be summarily expressed in the statement that in the case of vowel-roots where there was immediate affixing the language had at its command not only the original short vowel, as we take it to be, but also the long one proceeding from it. 363 Why indeed in δώσω, έστησα, εθηκα, δέδωκα the long vowel prevails, in εδόθην, στάτύς, cicorai the short one, might be hard to determine. Only so much may be noticed, that the middle perfect, the passive aorist and the verbal adjective incline most to the short syllable. Now the primi- tive inflexion is not separated by an abyss from the thematic. On the contrary, from the earliest times there were transitions from the one into the other, so that the primitive verbs are everywhere interlaced with thematic forms. On the other hand from the so-called verbs in ω there are several by-forms of primitive stamp, e.g. βήναι, βιβάς, βέβαμεν beside βαίνω, Χντο beside λύω. Hence it is certainly allowable to bring short vocalic forms in such cases too into connexion with the primitive method of inflexion, and so to put e.g. βέβά-μα;, ε-βά-θη-ν beside δέδο-μαι, ε-δό-θη-ν, in this way explaining the difference of the former from τετίμη-μαι, ε-τιμη-θη-ν. In other words, we take this first class of forms to be remains of the older method of formation, which often united with others following a later rule into a unity of verbal usage. In this way we may explain the following 15 cases : 1) Rt. βα. The extant primitive forms have been discussed more in detail on pp. 126 f. and 387. Hence come παραβεβάσθαι Thuc. i. 123, τταραβαθη Thuc. iv. 23, άμβατος (Homer), and also βάσις, βάθρον, βαθμός, but βήσω, εβησα, βέβηκα — βήμα. 2) Rt. δε bind. δίδη p. 105. Hence come δέδεκα Demosth., δέδεντο κ 92, δεθείς Soph. Aj. 108, συνδετεον Aristoph. Eccl. 785, and also δεσις, but ανυπόδητος, διάδημα. 3) Rt. δο needs no references. δέδοται, εδόθη, δοτός, δοτέος are in use in all Greek, as well as δοτηρ, δότης, δόσις. On the other hand δώσω, έδωκα, δέδωκα and δώτορ εάων (θ 335), δώτις, δωτίνη. 4) Rt. δν. For εδϋν etc. cp. p. 129. Hence ειδεδνμενην Menand. •Com. iv. 199, άποδϋθτ} Aristoph. Ran. 715, άδυτος Hymn, in Merc. 247, δνσις. The length of the ν in the present is certainly due to a *δνιω after the I-class, formed on the analogy of <\>υίω quoted as Aeolic. Cp. p. 147. Also δύσω, δύσομαι, δύσετο, δέδϋκε (I 239), not before the Antho- logy (v. 73) εκδέδνκας. 5) Rt. ε. 'ίημι etc. εθί/ναι (Plat. Pol. 270 ανεθϊι), άνεθησομαι Thuc, 364 άνετος Plat., also εσις, αφέτης, but on the other hand ηκα, άώήτωρ (I 404). 6) Rt. θε. τίθημι, ετέθην, τεθησομαι, θετός, also θέμα, θεσις, άγωνο- Βετης, on the other hand θήσω, εθηκα — θήκη. 1) Rt. Ι, ει μι, ϊμεν — πρόςιτος, Ιτέον, also 'ίτης, ιταμός — on the other hand ε'Ίσομαι Ξ 8, εισάμην Ε 538. — οϊτος, οϊμος, οιμη. 518 IRREGULARITIES OF THE VOWEL STEMS, ETC. ch. xxt. 8) Rt. λν. λντο Φ 114, cp. p. 129. λέλύμαι Aesch. Pers. 592, λνθεν θ 360, λϋθέντων Eurip. Hel. 860, λυτός. In λέλυκα it is generally assumed that the υ is short, but the only passage quoted for it by Yeiteh (Aristoph. Yesp. 992 εζ,ηπάτηται κάπολέλυκεν ούχ εκών) decides nothing. But Cboeroboscus, ed. Gaisford ii. p. 548, 26 (cp. Buttmann Ausf. Gr. i. 2 388) bears witness to its short vowel. Also λύσις, λύτήρ, λΰτρον. On the other hand λύσω, ελΰσα and in the present λύω from *λυίω } but also with a short ν ; cp. p. 148. 9) Rt. 7Γο. σύμπωθι and πώ Aeolic imperatives : cp. Ahrens AeoL 140. εκπεποται χ 56, εκποθεντ Aesch. Choeph. 66, καταποθήσομαι Ar. Yesp. 1502, ποτός, ποτέος, and also πότος, ποτήριον, but πόμα (Pind.) and πώμα (Eurip. Plat.), πεπωκα like Lat. potus, poculum. 10) Rt. στα. ΐστημι etc. ΰιεσταμένος (Plat.), εστάθη ρ 463 (also Attic), στατός. Also στάσις, στατήρ, επιστάτης, σταθμός, but στήσω,, έστησα, διάστημα, στημων. 11) Rt. συ. σύτο, σύμεναι ρ. 130. εσσύμαι Ν 79, εσσΰτο & 519,. εσσυθη Soph. Aj. 294, επίσσυτος (Aesch.), on the other hand έσσευε > εσσευα. 12) Rt. σεχ, by metathesis ΰχε, σχες p. 132, εσχεθον (p. 501),. εσχέθην (post- Attic), σχετός, also σχέσις. σχέτλιος, but εσχηκα, εσχημαι (Attic), and σχήμα. 13) Rt. ψθι. εφθίμην p. 129. εψϊμαι υ 340, εφθϊσο Aesch. Sept. 970, εφθϊθεν φ 331, <ρθϊτός Aesch. Pers. 523, άποψθίσειν Soph. Aj. 1027 t also φθίσις. But ώθίσονται Λ 821. 14) Rt. <\>υ. εψυν p. 130. The only evidence for the short vowel is> ψϋτόν (e.g. Ξ 123) which has become a substantive, with its numerous derivatives {φυτεύω etc.) and φύσις. The long vowel prevails in ψύσω> εφυσα, πέφυκα. 15) Rt. χυ. εχϋτο p. 130. συγκέχΰκε Menander Com. iv. 294, κέχϋτ(ο) Υ 421, χυθείη τ 590, χυτή Ζ 464. Also χύσις. On the other hand χέω, εχευα. 365 Besides these 15 instances, for which we have evidence of the primi- tive inflexion, the short vowel without σ appears in a number of stems, to which forms in ν correspond. To these belong κλι κ4κλικα €κλίθην κλίτος but κλΐτύς {κλίνω) κεκλιμαι κλίσις κλίμα κλϊμαξ κρι κ€κρικα εκρίθην κριτός but κρΐμα {κρίνω) κίκριμαι κρίσις κριτής πλυ πίπλνται πλυθησομαι νεόπλντα {πλύνω) Com. iv. 482, 3 Com. iv. 647 £64 τα τετακα ετάθην τατός {τείνω) τέταμαι τι ατϊτος but (ITITOS {τίω and τίνω) Ν. 414 3 484 It is hard to decide whether here the shorter stems are treated on the analogy of those inflected primitively, among which the rt. ψθι most resembles them, or whether a loss of the nasal leaving the vowel short has taken place. The form κατέκτάθεν Ε 558 beside άπέκτάν is in favour of the latter view. The conjecturally Boeotian έβάθη mentioned on p. 499 is exactly like it. ch. xxi. FORMS WITH σ. 519 According to the view just proposed there is no place for a σ in these verbs, and as a matter of fact forms like *βαστός, *εστάσθη, *Ιοστός and the like are quite unknown, and even if solitary instances of the kind like έβάσθη do occur at a very late date, this is to be regarded only as an aberration of the failing instinct of the language. But neverthe- less some few sigmatised nominal forms belonging to such stems occur in the best period. They are the following : θεσμός (in Homer only in ψ 296 beside θέμα, θέμις, θέσις), which we may suppose to come from the reduplicated root θε-θ(ε), much as the Skt. participle datta for da-da-ta from the rt. da, hence the Doric τε-θ-μό-ς, e.g. Pind. 01. vi. 69. The earlier form for both the dialectically differing formations would thus be *θε-θ-μό-ς. The θ might become σ, as in the perfect forms λελα- 3ββ σμένος, πέπνσμαι quoted on p. 420, which occur as early as Homer. On the same principle we might be disposed to refer the forms ΰ εσμός, Πσματα, which are common even in Homer, to h -ΰ-μό-ς. But there is a lack of positive support for this, such as was supplied before in τεθμός. Hence I conjecture that the rt. δε was expanded to ΰε-θ, as the root βα was in βα-θ-μό-ς, and that then the θ was changed before μ into σ. Herodotus's Ιυσμή beside Ιυθμί] (Callimachus fr. 539 ed. Schneider) is especially in favour of this explanation. Perhaps we must take in the same way the Homeric κλισμός beside κλίμα. The form ψνστις beside φύσις would be extremely strange. But the more recent editors are cer- tainly right in banishing it from the only passage in which it was formerly read (Aesch. Pers. 926), in favour of ταρφύς ης (for γαρ φύστις), the emendation of Franz. II. FORMS WITH σ. To this second division belong forms like ζέσω, ζέσις, τετέλεσμαι, άτέλεστος. The language of Homer has here often retained in the future and the sigmatic aorist a double σ, e.g. εσσω (ενννμι), ζέσσα, ετέλεσσα. Evidently the first of these two sibilants is of the same origin as that which was always retained in forms like έτελέσθην, άσβεστος, and which has often been regarded as ' inserted.' The short- ness of the penultimate in 'έζεσα, έτέλεσα is due to the reduction of the double σ to a single one, which begins, even in Homer, and has become the rule in the Attic ίρέσαι beside the Homeric έρέσσαι, κυμίσαι beside κομί&σαι, but also in υρεσι beside ορεσσι, 6σος beside οσσος. The explana- tion for the verbal forms mentioned is based upon the proof that the stem originally ended in a dental consonant, which before τ,θ, μ often maintained itself as σ, but was assimilated to a following σ, and in this latter case finally disappeared altogether. The most comprehensive investigation for the purpose of clearing up the phenomenon on this principle is that of Leskien Stud. ii. 68 ff. Earlier grammarians contented themselves with the assumption of a ' strengthening ' sigma. But how little the short vowels needed such strengthening is proved incontrovertibly from the formations which we have discussed above under the first division. Why έΐόθην but ώιόσθην, why ΰέΰεται but 367 τετέλεσταιΐ In the case of long vowels and diphthongs the assumption of a σ on purely phonetic inducements would be quite inconceivable, and yet we find πέπλησμαι, ήκονσθη, σειστός. The assertion that the doubling took place ' metri gratia/ often repeated for certain forms, might have 520 1KREGHTLAKITIES OF THE VOWEL STEMS, ETC. en. xxi. been easily disproved, if only by the well-established distinction between δώσω, φϋίσω, έβησα on the one hand, and ομοσσα, ζέσσα, κλάσσε on the other. However scholars have not succeeded hitherto in giving the desired proof in each particular case. We also must be content with establishing the principle by a number of indubitable instances. It will be proper in doing so to treat the radical verbs and the derived ones separately. A) Radical Verbs. a) Boots originally ending with a sigma. In the case of these the σ apparently inserted is really retained as the final letter of the root. The shifting of the original relation does not lie with the future, aorist, perfect or passive stem, but rather with the present, where however it is easily explained from the most familiar phonetic laws. Here belong 14 verbs, viz. 1) St. uhc, from /cc—Skfc. vas dwell, to stay, live. In Princ. i. 484, following Lobeck, I referred the Homeiic aorist αεσα, plur. άέσαμεν e.g. τ 342, y 151 to the rt. *ν)'ω αω €ω να> αζω (ζω (ιζω) νζω The hypothesis that at an earlier period the language fluctuated, even more frequently than we can precisely prove to have been the case, between the loss of the j and its assibilation after it had been preceded by an inserted δ, is certainly not an unjustified one. Now later on the same principle of formation did not always hold good in the present stem as in the remaining tenses, hence -γελάω but γελάσω (as if from *γελάζω), ελκύω but ελκνσθήναι (as if from *ελκύζω). Pott Etym. Forsch. ii. 2 970 if. has already discussed a large number of the anomalous forms in question from the same point of view, and in the Elucidations p. 133 I have followed him. Particular verbs are especially adapted to throw light upon this process. Thus we have άγαμαι, while forms like άγάσσατο Ρ 71, άγάσ- σεσθαί δ 181, άγάσησθε 3 111, άγασθήναι from Hesiod onwards, αγα- στός common from Xenophon onwards, follow άγάζω, άγάζομαι, which ^ΊΑ occurs in Pind. Nem. xi. 6 (άγαζόμενοι) and in Aeschylus (Suppl. 1062). The difference of meaning was discussed on p. 118. As in the case of this stem we have authority also for άγαίομαι with a vocalised j, and άγάασθε, we have in this case the different phases of the primitive form unusually clearly before our eyes. — £a//aw=Lat. domo has been pre- served in the older literature only with a future force, and hence it was discussed on p. 479. Homer uses for the present Ιάμνημι and Ιαμνάω. ΰάμασσε Ε 106, δα/ιάσσας Pind. Pyth. viii. 80, Ιάμασον I 496, εΐαμάσθην θ 231, Ιαμασθεν Eur. Phoen. 563 belong to the rarer present ΰαμάζω, which crops up first in Hesiod (Theog. 865), then in Pindar (Ιαμαζο- μίναν Pyth. xi. 24) and Aeschylus (δαμάζει Choeph. 323). — To the present αινίω belong properly only forms like αινησονσι (π 380), αινη- σωσι Ψ 552, while τίνεσα (from Aeschylus onwards), αίνεθείς (from Herodotus onwards) suit rather αινίζομαι (Ν 374) or the presumable earlier form *αη>έζομαι. — προκαλίσσατο (Η 218) is hence only a more archaic aorist form to προκαλίζετο (Γ 19) than *προνκαλί(Τατο would be, not an irregular one. And doubtless the short vowel, which always held its ground in έκάλεσα is due to this. — The variable σ in the inflexion of σώζω finds its explanation, as Buttmann Ausf. Gr. ii. 2 295 correctly taught, in the confusion of a verb σώω (from σαόω) with σώζω. In Homer we have no trace of the σ, and the ζ of the present appears only in ε 490 (σώζων), where however Didymus read σώων : and by the side of this εσάω Φ 238, σαώσει, σαώσειαν, σαωθήτω etc. ; cp. Mangold Stud. vi. 199. In Aeschylus first (Sept. 820) we find σέσωσται, in Euripides (Here. F. 1385) σωστέον, while εσώθην, σωτηρ never take a σ. The anomaly becomes more confused here, because the form with ι σώζω, thoroughly discussed by Usener in Fleckeisen's Jahrb. 1865 p. 238 f., is established by the Heraclean Tables (κατεσωζαμες Stud. iv. 428), by Attic inscriptions in part of a very early date (Wecklein Curae epigraphicae p. 45, Cauer Stud. viii. 416) and by grammarians (Herodian e . Lentz i. 444, 6). Of course σώζω can only have come from σωίζω. But the statement that Didymus wrote no ι (Herodian ed. Lentz ii. 586)• 524 IRREGUXAKITIES Ol•' THE VOWEL STEMS, ETC. ch. xxi. and the fluctuation of the grammarians' theory between the derivation 373 from % σαόζω and that from σωίζω are very noteworthy. 1 A form so isolated as σαόζώ perhaps is not due to the invention of the grammarians. It might well attach itself to δεσπόζω and αρμόζω according to the view stated on p. 238. σαόζω as a derivative from σάος is related to σαόω just as δαμάζω is to ΰαμάω. The existence of a σαόζω, σώζω formed from σάος would naturally by no means exclude the later formation of a σωίζω from σώος. σαόζω would be to σωίζω much as the Cyprian καλι'ιζω (Herodian ii. 332) to κληίζω. In this way we arrive at a system of equivalent forms, which is certainly somewhat complicated, viz. ) σαοω { ^ er j ye( j f rom σ £ ος whence έσώθην σωω J ' -,) σαοςω ι s j m jj ar iy derived from σάος, whence σεσωσται ) σωιςω Ί ( j er j ve< j f rom σωος whence κατ€σώξαμ€ς. σώζω J ' * r Some radical verbs too have apparently a moveable £, thus βλύζω (άπο- βλΰζων I 491) with εβλνσα (Apollon. Rhod.) which belongs to it, and βλύω (αναβΧνειν Hippocr.), κτίζω with εκτισσα, εκτισται etc. beside περικ- τίονες, κτίμενος, σχάζω Xen. Hellen. v. 4, 58, whence άποσχάσω (Crates Com. ii. 249), σχάσας Eur. Phoen. 960, and εσχων Ar. Nub. 409. On p. 251 we saw that the source, from which so often the ζ sprang, the spirant j both after a simple ν and after the diphthongs αυ, ευ, ου had its place in the present. As a μεθνίω is actually on record, so we deduced previously a *λνίω. Now as soon as a 2 developed before the /, *μεθύζω was produced, which is not itself on record, but we may venture to assert that it survives in μεθυσθηιαι (Herod, ii. 121). In this way we may perhaps explain θραυσθέντα (Soph. Antig. 476), κεκέλενστο (Herod.), εκελεύσβην (Soph. Thuc), λενσβήναι (Soph.). For the rts. καν 3 » 4r an( j K \ av we find a present formation in i actually existing in καίω, κλαίω i.e. κα^ω, κλα^ω, and so for the intensified stems πλεν, πνευ from 7τλν, πνυ in the Homeric πλείω, πιείω (cp. pp. 156, 210). These are followed by κανστός (Eurip.), κλανστός (Soph.). The use of the σ is especially extensive after diphthongs in particular. The form ΰυνάσθη which occurs as early as Homer (Ψ 465, ε 319) points to an obso- lete *ΰυνάζομαι, which in its formation would be related to δΰναμαι much as Ιακνάζομαι (Aesch. Pers. 571)ίο£ά«;ΐ'ω. πεφασμένος perf. part, to φημί (S 127) leads us to conjecture a present *φάζω as a by-form to φάω (p. 148)=Lat./ari. The dental presupposed in the verbs mentioned, as a predecessor of the σ, actually occurs in a few instances, as in εληλέδατ η 86, discussed on p. 242, from which it is not too venturesome to deduce a stem έλαΰ, which presupposes a present *ελάζω beside the actually extant ελάω {ελών ίϊ 696). This would justify ί/λασσα beside ήλασα, ήλάσσατο beside ήλασάμην, ελάσσομαι etc. — κλείω Old Attic κλήω has most probably come from the nominal stem κλ r\h (=Lat. clavi) with the by-form κληϊιϊ. The two stems interchange in nominal as well as in verbal forms. In 1 The words recorded in Et. Magn. p. 741, 25, which according to Lentz are drawn from Herodian run thus : άλλ' η παράοΌσις ϊχα rb ι. rb ii σφζω, 8τβ μϊ? ylvtrat birb rov σωο5 σωίζω is \4iroi λατίζω κα\ κατά συναίρτσιν σψζω, £χ€ΐ rb ι. η νίκα δ* awb rod σάο$ σαόζω κα\ κρασί ι σώζω, ουκ ίχα "κρο^^ραμμένον rb ι. ch. xxi. FORMS WITH σ. 525 Homer we have the nominal forms κλη~ώα Ε 146, κληΊσιν σ 294, κλη'ι- ΰεσσιν yu 215, and corresponding to these κλψσσεν ΰε θύρας τ 30, κληϊσται σα νίΰες β 344. In Herodotus iii. 58 we read άπεκληίσθησαν; on the other hand the Attic writers use the contracted form, and that partly in later συνεκέκλειστο Andoc. i. 48. In the tragedians and Aristophanes according to Nauck, Dindorf and Wecklein (Curae epigr. 66) we should write η throughout. Besides the present stem the perfect middle also attaches itself to the form without c, in Herodotus ii. 121 κεκλήιμένον, in Attic writers here and there κέκλημαι, κέκλειμαι, κέκλεισμαι. The Doric κλαζώ, κατεκλάζατο (Theocr. vi. 32, xviii. 5) also deserve mention, inasmuch as this ξ too as the correlate of the present properly requires Λ ζ. In this whole question we must not leave out of sight the fact that the usage was extremely fluctuating, and that on this account the authority of our texts cannot be regarded as very certain, and that evidently the 375 σ became more widespread in course of time. Buttmann noticed this in Ausf. Gr. i. 2 424, where he says ' So that we see that this σ has made its way in by degrees, for which the formation of the words in ζω, θω, δω supplied the analogy.' He was only wrong — with Lobeck (* assumto sigma corro- boravit' Paralipp. 320 sqq.) — in assuming the operation of a ' euphonic principle,' for we cannot possibly speak of this. Forms with a short vowel like ΰέΐοται, ετέθην, λέλνται etc. always remained unaffected, while on the other hand έμνήσθην {έπιμνησθείς δ 189, πολνμνηστη, μνηστήρες), επλήσθη, έρραίσθη (Π 339) όρχηστηρ, παλαιστής, θρωσμός in spite of their long vowel show the σ even in Homer. For the variations of our authorities I content myself here with referring to the abundant material collected by Lobeck ad Ajacem v. 704, and supplemented by Wecklein Curae epigraphicae p. 61. Elmsley regarded ενγνωτος as the only correct Attic form, but even in Homer φ 175, ν 191, 397) we find άγνωστος, which is known also to Pindar. Bigour is here by no means in place. We cannot form a more definite judgment as to the authority for many forms in the Attic writers, until the inscriptions of the best Attic period have been carefully examined with this in view. We can see that it would be in vain to seek any special inducement for each particular form, from the fact that while the σ* makes its way into forms to which it did not originally belong, on the other hand the same sound elsewhere lost its original place. We can hardly explain otherwise εργάτης (as early as Archilochus fr. 39 Be. 3 βονς εστίν ημ~ίν εργάτης εν οικίη), with the derivatives εργάτις, εργατικός, εργατίνης, which in its formation reminds us of δεσπότης, τοξότης, οικέτης, φνλέτης, and further βανματα έργα Hymn, in Merc. 80, 440, Hes. Scut. 165 (cp. Pind. 01. i. 28), κέκεντ((ΐ' κεκενσται Hesych , γέγενμαι in spite of the rt. γνς. Evidently such forms point to an early obscuring of the instinct of the language. Hence we must be always very cautious here with regard to bold etymological combinations. The possibility of such presents itself often enough. Thus we might conjecture that under the aor. ίκλάσθη lies a root ' κλας expanded from καλ (Lat. cettere); but it is just as con- ceivable that the present form *κλά/ω arising from metathesis was 376 expanded by a c and produced a *κλάζω, which though afterwards 526 IRREGULARITIES OF THE VOWEL STEMS, ETC. ch. xxi. obsolete, was the source of the sigmatic forms. For γνωστός we might suppose a stem γνως expanded from γι>ω, but also a present *γνω;ω, which might be compared with Ο. H. G. knau. The same alternative recurs in the case of μνάομαι, χράομαι. Thus elsewhere too different paths present themselves, between which it is hard to decide. But speaking generally, the whole phenomenon here in question does not belong to quite the earliest period of the language, so that the spread of an analogy, not very sharply denned, on the strength of a number of old precedents can hardly be denied. I doubt whether it will ever be easy to arrive at the proof of the principal causes for the whole phenomenon, which we have here endeavoured to complete. 52; CHAPTER XXII. THE ITEBATIVES. The essential verbal forms, which are carried out to a large extent in the case of all Greek stems, have now been discussed by us. There remain only some formations which are, so to speak, more tentative than complete, and which are limited to particular portions of the language. Among these the iteratives deserve the first place, if only for their abundant developement in the period of the language concerning us, and the manifold forms they take. They are tolerably numerous in the language of Homer. I have noted about 130 altogether, in round numbers. But it is very noteworthy that only a small number of these forms occur frequently, e.g. ϊώασκε, φιλέεσκε, ε'ίπεσκε ; by far the majority have the stamp of formations which were ventured once or twice, and which therefore the poet certainly had at his command, if he needed them, without their having attained the full franchise by the usage of the living speech. The number of the απαζ εφημένα among them is 377 very large. With regard to the particular personal forms too a restric- tion appears. It is only the third person singular and plural which is at all common. On the other hand the first and second person plural are only represented each by a single example in Homer : νικάσκομεν λ 512, έφάσκετε χ 35, and so the 2 sing, of the middle πελέσκεο X 433. After Homer indeed about 90 new forms are essayed. But there are but few which we can get from the older poets. Hesiod, according to Eorstemann de dial. Hesiodea p. 34, has only 10 such forms, Pindar according to Peter de dial. Pindari p. 58 only three. On the other hand the iterative is quite common in Herodotus. Bredow de dial. Herod, p. 285 ff. gives a list of 24 forms of the kind from different verbs. The tragedians have left us only 4 such (Gerth Stud. i. 2, 259), of which one already occurs in Homer. All clearly bear the mark of the intentional imitation of earlier poetry. In Aristophanes we find probably only βινεσκόμην Equ. 1242, and in the imitation of an oracle ϋαπάτασκον Pax 1 070. On the other hand the later Epic poets, espe- cially Apollonius Bhodius and Quintus Smyrnaeus delight in employing and multiplying Homeric iteratives, and also Theocritus and Moschus have some new instances. The whole form was therefore living only among the Ionians, and hence it is not unknown to Hipponax, who is thoroughly popular in his style : fr. 37 Be. 3 Θύεσκε. So far as we can judge, it owes its wider extension only to the influence of the Homeric Epos. It is very significant that Attic prose let this form altogether drop. The precise usage of the durative forms as distinguished from the aoristic made the iteratives in point of fact unnecessary, especially as auxiliary verbs like ειωθέναι, φιλεΊν, and above all the use of άν with the indicative of the past tenses with the force of a frequently occurring 528 THE ITEEATIVES. ch. xxuv action (Grammar § 507 note 3) were at command, if it was needful te express the specific force of repetition. In Herodotus, as a kind of inter- mediate stage, an av of this kind is sometimes added even to iterative forms. The iteratives share the group σκ with the inchoatives discussed in Chap. X. ; and we cannot fail to see that the force of the two is cognate. 378 The iterative action forms a kind of contrast to the regularly continuous, just as much as the inchoative. Every repeated action is on the other hand just as distinct from that which is entirely complete as that which is coming to completion, which realises itself in incipient attempts Still it must be admitted that starting from this common basis the usage of the past tenses cut loose from an indicative present diverged on a line of its own, and thus became an excellent means, especially for narrative statement, to distinguish with the greatest brevity frequently repeated action from the single act. The iterative past tenses, e.g. Ιόσκον are related to the present Stems of the inchoative class, e.g. βόσκω much as- the isolated past tenses in -θο-ν, which occupied us on p. 501 ff. e.g. εσχεθον are to present forms like πλήθω. It is occasionally difficult to decide whether a preterite form is to be regarded as an imperfect of the inchoative class or as an isolated iterative form. Thus on p. 192 we decided to take τζαρέβασκε Λ 104 as an imperfect, because βάσκε occurs as an imperative. On the other hand we shall be obliged to allow εφασκες Τ 297, φύνχ 2 191 (Νέστωρ φάσχ 6 γέρων οτ έτημνησαίμεθα σεΊο) as iteratives. For ψάσκω as a present occurs first in Attic writers (cp. p. 193). In this instance therefore in the course of the history of the language first the past tense established itself, and afterwards the present, though only with a greatly faded meaning. Again έσκον admits of considerable doubt. An iterative force does not come out clearly in any of the Homeric passages, and is more than once decidedly excluded, e.g. Η 153 γενέη ΰέ νεώτατος εσκον απάντων, Γ 180 Βαήρ αντ έμυς εσκε κννώπιΰος. Cp. JPind. Nem. v. 31, Aesch. Pers. 656. On the other hand in several of the passages in Herodotus referred to by Bredow p. 285, we must recognize the iterative force, e.g. vi. 133, vii. 119. Hence we shall perhaps do best to place the form with the others formed by the suffix -Ko, of which we spoke on p. 411 ff., and to regard the iterative usage as a later developement which easily resulted from following the forms in -σκο-ν. The Old Latin escit (Neue Lat. Formenlehre ii. 2 596) acts as a present with a future force formed from the same stem . The description of these forms as iteratives and the right apprehension of their usage is a matter of quite recent date. The ancient grammarians 379 seem to have taken all the forms of this kind as merely Ionic expansions of the current past tenses : τά δια τοϋ σκε παρηγμένα Ίακώς, as Herodian says (ed. Lentz ii. 792). Buttmann (Ausf. Gr. i. 2 382 note) even believed that he was the first to notice the force of these forms, after his attention had been called to it by Grotefend. He did not fail to see however that the repetition of the action did not come out with the same sharpness in all the Homeric passages. This is still more true of the later poets, who use the forms more as an archaic garnish. That the augment was not used with the Ionic past tenses in -σκαν was taught even by the ancient grammarians. Cp. Et. M. p. 295, 14 : τά yap τοιαύτα άποβάλλουσι την εν apxfj κΧιτικην 'έκτασιν. This is true of the great majority. But the augment is often used in Homer with the ch. χχπ. ITEKATIVES FROM PRESENT STEMS. 529 iterative of φημί : εφασκες (αλλά μ εφασκες Τ 297), εφασκε {υς εφασκε β 565, ν 173, υν-ηοτ εφασκε ρ 114). Any one who chose to apply the critical knife here, to restore the uniformity of which many are so fond, would find many a passage not easy to deal with, e.g. μ 275 at the end of the verse άμμιν εφασκον, ζ 321 κείνος γαρ εφασκεν. Other augmented iterative forms are εμισγέσκοντο ν 7, άνεμορμνρεσκε μ 238 (La Roche with the first hand of Μ. άναμ.), παρεκέσκετ ζ 521, ήγίνεσκον Aratus 111. In Callimachus Hymn, in Dian. 123 O. Schneider writes with the best M.SS. πολλά τέλεσκον not νό\\' (τέλεσκον. For Herodotus Bredow denies altogether the use of the augment in these forms. Stein too writes e.g. Herod, i. 100, where some M.SS. have είςεπέμπεσκον, έσπέμπεσκον, iv. 130 λάβεσκον, not as it is in earlier editions, ελάβεσκον. And certainly the number of the non-augmented forms here so greatly preponderates that this critical proceeding is justified. "It is difficult to find any reason for the suppression of the exponent of the past character of the action in the iterative forms. Buttmann Ausf. Gr. i. 2 382 says it is sufficient that the forms were Ionic, since the Ionians always inclined to omit the augment. But that does not suffice. For in Herodotus, as we saw on p. 92, the syllabic augment is not rejected in any other case. Buttmann is more correct in pointing to the length of the forms, itself already considerable. Something of the same kind we 380 saw to be the case in the pluperfect. A certain dread of burdening the beginning and the end of a form too much with formative elements can be detected elsewhere too. The speech-forming art also of the Greeks knows a μηΰεν άγαν. In their origin the iteratives are extremely various. We can distin- guish four principal kinds : A) Iteratives from present Stems. These are by far the most numerous. It will suffice to quote some characteristic instances from each kind of present stems : thus 1) from those formed primitively on the one hand φάσχ* 3 191, κεσκετ (for *κείσκετο) φ 41, ρύσκευ Ω 730, belonging to ειρναται ρ. 122, on the other 'ίστασχ τ 574, άνίεσκε Hes. Theog. 157,7-/0f(f/v-£Hes. fr. 96 Gottl., ρήγννσκε Η 141, ζωνννσκετο Ε 857, Ιάμνασκε Hymn, in Ven. 251. 2) from thematic stems without any further strengthening εχεσκες Ε 472, εΰεσκε Χ 501, θέλγεσκε γ 264, άγεΰκον Herod, i. 148, πέμπεσκε Herod, vii. 106, πελέσκεο Χ 433; 3) from verbs which lengthen the vowel Φευγεσκεν Ρ 461, ληβεσκε Ω 13 ; 4) from verbs of the T-class κλεπτεσκε αν Herod, ii. 174, άστράπτεσκε Moschus ii. 88 (Hermann, Meineke) ; Μ Μ 530 THE ITERATIVES. ch. χχπ. 5) from the nasal class πίνεσκε Π 226, βλαστάνεσκε Soph. fr. 491 ; 6) from the inchoative class βοσκέσκοντο μ 355, μισγεσκετο σ 325 ; 7) from the I-class κλαίεσκε θ 364, κτείνεσκε Ω 393, ποιμαίνεσκε ι 188, αποπλννεσκε ζ 95, σπείρεσκον Herod, iv. 42, απαγγέλλεστε Ρ 409, βαλλέσκετο (Stein : βαλέ- σκετο) Herod, ix. 74, πρησσεσκον θ 259, λισσέσαετο Ι 451, ρνστάζ εσκεν Ω 755, ερίζεσκον θ 225, ρεζεσκον χ 209 ; 8) from formations with θ βαρνβεσκε Apoll. Rhod. i. 43, μινΰβεσκον ξ 17, ψΘιννθεσκε Α 491, τελέθεσκε Hymn, in Cerer. 241 ; 9) from a stem expanded by κ the isolated όλέκεσκεν Τ 135 : 10) from denominative verbs with vowel stems. 381 No difficulty is presented by forms like Ζινενεσκ Ω 12, άριστενεσκε Ζ 460, ταμιεΰεσκε Soph. Antig. 950. But from the stems of the contracted verbs the iteratives are formed in three ways, either a) without any important alteration νεικείεσκε Β 221, βουκολέεσκες Φ 448, καλέεσκε Ζ 402, άπαιρέεσκον, Herod, i. 186, ποιέεσκε Herod, iv. 78, ψιλέεσκε Ζ 15, υπνωεσαον Quint. Smyrn. Η 503; b) with assimilation γοάασκε θ 92, περάασκε ε 480, ισχανάασκον Ο 723, πε&άασκον ψ 353, άμφαφάασκε Moschus π. 97, μεώιάασκε Quint. Smyrn. θ 117; c) with loss of one of the vowels άθεσκε λ 596, καλέσκετο Ο 338, πωλεσκετο Ε 788, μυθεσκοντο Σ 289, οίχνεσκε Ε 790, κράτεσκε Pind. Nem. iii. 52, βινεσκόμην Ar. Equ. 1242. — εϊασκον Ε 802, μνάσκετ υ 290, τρωπάσκετο Α 568, έξαπάτασκον Ar. Pax 1070, συλασκε Hes. Scut. 480. The ancient grammarians expressly teach that the vowel before σκ is always short (Herodian ii. p. 792). This excludes e.g. in the passage cited above from Hesiod's Theogony (v. 157) the reading of the M.SS. άνίησκε. This third method was followed also by some forms at first sight surprising, viz. ρίπτασκον Ο 23, ρίπτασκε Ψ 827, θ 374, λ 592, ΰιαρρίπτα- σκεν τ 575 κρυτττασκε θ 272 (κρύπτεσκε Hymn. Cer. 239), which Kiihner (i. 550) disposes of by saying that here ασκον took the place of εσκον. But we can hardly be contented with that. If I am not mistaken we have a fresh instance here of the intermixture of verbs in αω and αζω ch, χχπ. ITERATIVES FROM AORIST STEMS. 531 noticed more than once, and to which reference was made on p. 235. This is seen most clearly in Ισάσκετο Ω 607, which is formed like τρωπά- σκετο. The only present known is Ισάζω. But the iterative has come from *ισάω. I. Bekker thought this form so strange, that on the strength of the reading of the papyrus ' είσάσκετο' he adopted είσάσκετο, but it is certainly better to say of Niobe ovvck άρα λητοϊ Ισάσκετο καλΧιπαρ^ω and further είσάσκετο in the sense of 'compared herself would be quite unique. Now as ισάσκετο is to Ισάζω, SO is ρίπτασκον to ριπτάζω, the 382 intensive of ρίπτω. At S 257 the god of sleep says of the angry Zeus ριπτάζων κατά ΰώμα θεούς, εμε δ* έξοχα πάντων ζητεί, and at Ο 23 the same Zeus boasts bv Se Χάβοιμι ρίπτασκον τεταγων άπο βηΧον. For κρΰπτασκε the corresponding κρνπτάζω occurs first in Diodorus and ecclesiastical writers. Still it would not be too bold, considering the complete analogy of the two forms ρίπτασκον and κρύπτασκον, to con- jecture the same origin for the latter also. How well the iterative form adapts itself to frequentatives may be seen from ρυστάζεσκεν (Ω 755) the regular formation from ρυστάζω, which according to our view stands in the like relation to ρίπτασκον as σύΧασκε to γοάασκε. It is true that two rare forms with a strange α still remain, ροίζασχ Hes. Theog. 835, which however recent editors replace by ροίζεσχ for which there is good authority, and πάντο& ανασσείασκε Hymn, in Apoll. Pyth. 225, where it is very easy to read, following most M.SS. άνασείσασκε, άνασσείσασκε (only Μ. ανασσείασκε) whereby we get a regular iterative from the sig- matic aorist. B) Iteratives from Primitive and Thematic Aorists. The need to render possible the expression of repetition not only of the continuous action denoted by the present stem, but also of the momentary action, evidently led to the extension of this formation more widely beyond its original sphere. Thus arose the various aorist itera- tives, which we have now to point out. There are but few primitive formations of the kind: οόσκον I 331, στάσκεν Γ 217, ΰύσκε θ 271, οντασκε Ο 745 belonging to οντά (cp. p. 134). Thematic aorist iteratives are : προβάΧεσκε ε 331, γενέσκετο Χ 208, ε'ίπεσκε Β 271 etc., εΧεσκον ξ 220, μεθέΧεσκε θ 376, 'ίΰεσκε Γ 217, έςίΰεσκε -φ 94, άποΧέσκετο ι Χ 586, φΰγεσκε 383 ρ 316, καταΧίπεσκε, Χάβεσκε Herod, iv. 78, εξεΧάθεσκε Or. Sibyll. i. 44. C) Iteratives from Sigmatic Aorists. άγνώσασκε, as is now read after Apollonius in the Lexicon p. 8, 18 in \p 25 instead of the hardly explicable άγνώσσασκε or άγνώσσεσκε of 1 The active απ-όλεσκ^ν, which Bekker has adopted Θ 270, is only supported by the Harleian M.S., hence La Roche is doubtless right in reading with the best M.SS. απ-6\ζσσ€ν. Μ Μ 2 532 THE ITEKATIVES. οη,*χχιι•. the M.SS., contracted from άγνοήσασκε, like όγδώκοντα Β 568, 652 from όγΰοήκορτα, κατ-αζήνασκε λ 587, άίξασκε Ψ 369, (επ- Ρ 462), αυΰήσασκε Ε 786, Ιασάσκετο Ι 333, Ιησάσκετο Ω 15, ε'ίζασκε ε 332, ε\άσασκε Β 199,. ερητνσασκε Β 189, εζερνσασκε Κ 490, θρέζασκον Σ 599, ά7Γο*,ινι/σασί:ε Λ 636, μνησάσκετο Λ 566, όμοκλήσασκε Β 199, ορσασκε Ρ 423, ον-ησασκε Χ 375, σπείσασκε θ 89, στρέψασκον Σ 546 (άπο- Χ 197), ωσασκε λ 599. I do not find anything of the kind cited from post-Homeric literature with the exception of άκίησασκε Orph. Lith. 364. It is very remarkable that the New Ionic prose, which, as we saw, elsewhere made such abundant use of the iteratives, rejects these somewhat awkward forma- tions. D) Iterative from the Passive- Stem. Here we have only to mention the one form φάνεσκε : Λ 64, λ 587, μ 241, 242, Hes. fr. 44, 3, which has evidently come from the passive- stem φα νη with the same shortening of the final vowel, as occurs in εψανεν, φανέντος, φανείηΐ', and which here finds its complete explanation in the general rule that only short vowels are admitted before the σκ of the iteratives. We may also compare κέσκετο cited above. The notion of assuming an εψανον for ψάνεσκε is altogether to be rejected. How little authority there is for such an aorist is shown by the collections of Veitch p. 588. The whole category of the iteratives, which so to speak unfolds itself before our eyes out of a not inconsiderably different formation with essentially the same formative elements, and afterwards disappears again, is an extremely significant phenomenon for the processes in the history of language, instructive with respect to the fact which we have so often assumed, and which we denoted by the name ' branching off.' If it was 384 incontestibly possible at a tolerably late date for an offshoot of the inchoative class to attain to individual life, and extensively propagated by a spread- ing analogy to become an independent species, this fact gives additional support to the hypotheses which we ventured to advance for an immea- surably earlier period in respect of other formations, more deeply rooted in the life of the language, e.g. as to the origin of the conjunctive and the optative. 533 CHAPTER XXIII. DESIDERATIVES, INTENSIVES AND FREQUENTATIVES. We have only a small gleaning of forms left. These are those which are proportionally rarely employed, but which still cannot be passed over as parts of the great whole, because they also belonged to the system of the Greek language. Compared with the main pile of the edifice they form, if we may say so, small side-buildings, like the pleasure-houses or pavilions of a palace, which have been created for special subordinate needs, and might well be dispensed with altogether, without anything essential being felt to be wanting, but which hold their modest place, and bear witness to the inventive power of the architect. I. DESIDERATIVES. "While the iteratives, as we saw, were treated by the ancient gram- marians without any regard to their meaning simply as past tenses with an Ionic expansion, the desideratives used in good Attic could not be passed over by them without violating their principle, even as a matter of usage. The technical name for these forms seems to have been εφετικά, as Lobeck ad Soph. Aj. v. 325 saw, writing εψετικά in the words of the scholiast καλούνται Μ εφεκτικά, and in the same way in Theodosius Grammat. p. 67, 18 Gottling writes ( — ο'ιον οφω όψείω, βρώσω βρωσείω, άτινα και εψετικα λέγονται). Elsewhere e.g. in the scholium on S 37 we are told that έπιθυμητικώς εχειν is the force of these forms, or an 385 adverb in -τικώς formed from the same stem is employed to reproduce the sense; thus in the Lexicon of Apollonius p. 125, 32 ό-φείοντες, οπτικώς έχοντες ' ό ί)ε τύπος της λέξεως 'Αττικός ' κλανσε'ιοντες γαρ λέγουσιν αντί τον κλανστικώς έχοντες. The old technical writers assumed two types of this most familiar and common kind of desideratives, viz. those in ειω (από ενεστώτων) and those in σείω (άπο μελλόντων). But as for the first type only the Homeric κείω and some few forms like θαλπείω, τελείω, ριγείω, όκνείω are quoted, the desiderative force of which is open to great doubt, and in some instances decidedly to be rejected, Lobeck (on Buttmann's Ausf. Gr. ii. 2 389) has rightly banished altogether this first type, pointing out at the same time that even some of the ancient grammarians e.g. in the Et. M. 750, 50 recognized only the second type. The desideratives proper occur, as these writers noticed, only in the present, so that the form ώ-φεον (for ώψειον) quoted from Sophron ' σεσημείωται ' ['is specially noted' J : on the other hand an imperfect may be formed from the verbs in -ε<ω. Of the desideratives in σείω I have succeeded in discovering the following 20 instances. άκονσείων Soph. fr. 820 according to Bekker Anecd. p. 369, 13. 534 DESIDEKATIVES, INTENSIVE?, AND FREQUENT ATI VES. ch. xxiiu άναγνωσείω Steph. Thes. ' OtL' άπαλλαζείοντες Thuc. i. 95. σνμβασείοιτα Thuc. viii. 56 (probably more correctly συμβησείυντα). βρωσείοντες Callim. fr. 345. γαμησείω Alciphron i. 13, iii. 37. γελασείοντα Plato Phaed. p. 64. γραψείω Steph. Thes. * Gl.' δειπνησείω do. Ιρασείων Soph. Aj. 326, Ιρασείετον Eur. Phoen. 1208, Ιρασείεις Ai\ Pax 62. ΰωσείειν Hesych. (M.S. ΰοσείειν), παραΖωσείοντα Thuc. iv. 28. ελασείυντι Lucian Charon c. 9. εργασεΐων Soph. Trach. 1232, έργασείεις Soph. Phil. 1001. κλανσείυιτες Apollon. Lex. 125, 23. vdv μάχη σείοντας Thuc. viii. 79. 386 όψείοντες Ξ 37. πυλεμησείοντες Thuc. i. 33. τνραννησείοντα Diog. Laert. i. 2, 18. ψενζείω Eur. Here. F. 628. χεσείω Ar. Nub. 296, χεσείη Equ. 888. The ancients, as we saw, derived these forms all from the future, a derivation to which in this case the meaning lends a certain probability, inasmuch as the wish is always directed to the future. And in Greek in particular the future does not reject the final usage, which is closely connected with the desiderative force. Still I do not see how we could get, in the face of the origin of the future examined on p. 466 if., either from the earlier form -σιω, or from the abbreviated -σ*ω, to -σείω. The desiderative shares the sibilant, which is essential for its form, not only with the Indo-Germanic future, but also with the Indian desiderative, which ends in -sct-mi. But we saw on p. 444, that most probably it is. not this termination, but rather the reduplication, which we can see in the stem of the verb e.g. in ρί -pa-sa-ti which properly carries the desi- derative force. Hence if the Greek desiderative is at all akin to the- Indian, it must have lost the reduplication, just as has been the case in a number of Sanskrit formations of the kind. But no one will readily maintain such a view in face of the entire absence of the slightest trace of such an expansion of the stem. And after all that would still leave the ει quite unexplained. Bopp has taken a different starting point for the analysis of the desiderative form, comparing a form like Ζω σείω with a *dasja-je (middle) which he deduces on the analogy of other forma- tions, and which, supported by the Skt. vrhd-je I become great, which he traces back to *vr-hant-je, he derives from *ddsjat-je t that is from the stem of the future participle. According to this ΐωσείω would be for οωσοντ-^ω. But all phonetic laws go against such an assumption, which Bopp was doubtless misled into making only by comparing the Latin desideratives in -turio, e.g. nupturio, esurio. But even in the case of these the derivation from the participle in -tilru-s is by no means certain. 387 For the difference of quantity (nupttirio, but nuptiira) [Roby § 976], which ' presents no difficulty' to Bopp, is a grievous hindrance. Now-a- days Bopp's view will hardly be defended by any one. Savelsberg Ztschr. xvi. 362 ff. brings the Latin forms arceesere (beside accersere 1 ), capessere,, 1 [Cp. Journal of Philology vi. 278 ff.] ch. χχιιι. DESIDERATiVES. 5Z5 lacessere into connexion with the Greek desideratives, which certainly come near in meaning. But even if we chose to start with the stems in I (e.g. capessi-vi) which occur elsewhere than in the present, we should still be far from getting an identity of stems. Hence no formation really corresponding to the desideratives has hitherto been discovered outside of Greek. It therefore remains probable that we have to do with specifically Greek recent formations, and we can only ask after what patterns they may have been produced. Derived verbs in ειω are to be found, as we saw on p. 240, almost exclusively in the language of Homer : νεικεϊω, τελείω etc. The desideratives in -σείω might therefore have been formed upon these at an early time, when such forms were in still more frequent use. Another circumstance is in favour of this. The diphthong ει is sometimes exposed to the same weakening into ε as in those presents. We have the isolated ώψεον preserved to us from Sophron. The Syracusan οψε'ω belonging to it, which we may assume, is related to οψε<ω just as -ελέω is to τελείω. The desideratives in -σείω are therefore, as far as their termination goes, verbs in -εω in their earlier form. It is more difficult to explain the preceding σ. We can hardly look for anything else in it but the remains of a stem-forming suffix. Now σ is indeed, apart from the widespread stem-forming suffix -as=Gr. sc, ο ς, which can hardly come into consideration here, a rare sound in nominal stem -formation. But there are a small number of stems with the suffix -σο, among which adjectives like (ppito -ς, κομφό-ς, τιθα-σό-ς, πνρ-σό-ς are found. Perhaps we may add also proper names like Αάμα-συ-ς, "Ελα-σο-ς, "Ερα-σο-ς, Σώ-σο -t. Now possibly έλασείω is to "Ελα σος just as Koipariwto κοίρανος or Ιιΰικεω to ciIlkoc. The desiderative force might have been produced by their likeness to the futures, which sound somewhat similarly. A second formation of essentially the same function, consisting of the 388 verbs in -αω and -ιαω, is seen at once to be of denominative origin. Per- haps the two forms, which have been very thoroughly discussed by Lobeck on Buttmann ii. 2 389, are better kept apart. The rarer verbs in -αω with- out a preceding ι evidently come from abstract substantives, e.g. from feminine abstract substantives in α, τομά-ω (Soph. Aj. 582 προς τυμώντι πήματι), μαχάν άντι τον μάχης ^εΤσέίαι Hesych., φονΰν (φοιφ νόος ηδη Soph. Phil. 1209), from masculines in ο, tiararav (Plato Phaed. p. 64), τοκώσα (Cratin. Comici ii. p. 208) =parturiens, λοπάν to incline to peel (of trees). Lobeck is certainly right in explaining the verbs of sickness [e.g. βραγχαΐ', λιθαν, ποΰαγράν, νέεράν : cp. Lobeck on Phryn. p. 80] as identical with those cited, appealing to the German termination -sucht in Gelbsucht [jaundice] beside Habsucht [greediness]. ψωρά> (Plato Gorg. 494) is related to ψώρα as τομαν to τομή. — The forms in -ίαω have certainly come originally from nominal stems which contained the <. Here Homer gives us an example: Μ 265 άμψοτέρω δ' Νια ντε κεΧεν- τιόωντ επί πύργων πάντοσε ψοιτητην, where the meaning is rather imitative than desiderative, ' playing the leader.' But in such formations the more delicate shades of meaning between the endeavour after a thing, the tendency or the inclination to a thing, and action in imitation of a man everywhere cross each other. Now as we find nouns in -ta-c f discussed by Lobeck Proleg. 487 ff., often formed with this sense, e.g. Τειρεσ-/α-ς, Χοζ-ία -c, ξανθ-ία-ς, κυματ-ία -c, I can well believe that κελεν- τιάω properly, or, as Pott is wont to say, ' in idea ' presupposes a 536 DESIDERATIVES, INTENSIVE^ AND FREQUENTATIVES. ch. xxiii. *κελευτ-ία-ς as a paragogeof κελεν-τη-ς, just as much as υνρηπάσ^ς in Ar. Vesp. 807 presupposes *ονρητίας and *ονρητής, άκουστιάν (άκονστικώς εχειν Hesych.) ^άκονστίας and ακουστής, ώνητιάν (Suid.) *ώνητίας,ώνητής. On the same principle γανριάν presupposes *γανρίας, μαλακιαν *μα\ακίας. It is certainly hard to decide whether abstract feminines may not also have contributed, as may be probable in the case of κλανσι,άν (Arist. Plut. 1099), όψιαν (Hesycb.). It is certain that in the course of the history of the Greek language the ι became constantly more firmly established, and -ιαω became a favourite termination of verbs of the kind, so that by the side of Ικτεράν λιθάν etc. Ικτεριάν, λιθίάν etc. established themselves. For these I may refer to Lobeck ad Phryn. p. 79 if. 389 The unmistakeably denominative character of the second formation evidently favours our conjecture that the case was the same with the first. As a rule these later offshoots of the verb do not seem to be formed directly from the stems of the verbs in question, but to pre- suppose usually the preliminary stage of a nominal stem, as we shall have to notice again in the case of the frequentatives. Π. INTENSIVES. A definitely formed intensive formation, characterised by particular terminations, does not exist in Greek, and therefore there is no Greek name for the intensives. ' I find in the ancient grammarians no mention of epitatic or paratatic verbs,' says Lobeck on Buttmann ii. 2 392. But there are remains of a primitive Indo- Germanic intensive form, to which attention has repeatedly been called. The sign of the intensity of the action is reduplication, which in this case is readily united with vowel- intensification in the syllable of reduplication. In Sanskrit, as may be seen from Delbriick's statement in his Old-Indian Verb p. 130 ff., the modes of treatment of the stem are diverse. A part of these forms have been mentioned already under the head of the perfect, e.g. p. 376 ff. One of the various methods of formation is that the reduplicated stem unites with the suffix -ja to form the present stem. According to Delbriick p. 1 31 this method is still ' rare ' in the Yeda, and becomes ' more common later on.' Its type is represented by ve-vixf-jd-te to let fly from the rt. vig, re-rih-ja-te from the rt. rih (lih) lick. But still the formation must be very ancient : otherwise Greek and Sanskrit could not agree in it. Such forms, so far as they are preserved in Greek, belong to the I-class of the present formation, and have accordingly been cited above, viz. p. 212 f. seven presents in -λλω, at -όλλω etc. p. 215 f. five in -ρω, γαρ-γαίοω etc. p. 217, eight in -ι»ω, αν-αίιομαι etc. p. 221, three in -σσω, ΰει-ΰίσσομαι etc., p. 226 f. sixteen in -ζω άρράζω etc. The whole subject has recently been treated so thoroughly by Fritzschein his 390 paper ' de reduplicatione Graeca ' Stud. vi. esp. p. 282 ff. that I need not follow it up further here, especially as I frankly confess that I cannot offer anything, which quite satisfies me, as to two phonetic difficulties which present themselves in these forms, viz. as to the ι, which appears e.g. in πα «-πάλλω, and as to the nasal which appears in ΰεν-ΰί\\ω. The very varied treatment which the palatal spirant,/ under- went in Greek has led the instinct of the language to part off these verbs, though fundamentally quite homogeneous. Hence we naturally could not talk of a definite and certain usage. CH. χχιπ. FREQUENTATIVES. 537 Intensives, which belong to any other class of presents, are extremely rare, perhaps do not exist at all, for even κω-κύ-ω beside Skt. ku howl (cp. Fritzsche p. 301) and νη-νε-ω beside j -c ω might easily have lost a j before the thematic vowel. III. FREQUENTATIVES. I have repeatedly, especially on pp. 236 f. and 243, called attention to the verbs in -τάω with the by-form -τάζω, and to those in -τέω with the by-form -τίζω, which have the plainest connexions with the Latin frequentatives in -tare or -itare. ευχετάομαι, όνοτάζω, ρνστάζω, αΐτέω, αιτίζω, ώστίζω may serve as examples, ριπτέω and jactare quite coincide in their meaning, fateri and φατίζειν do not essentially differ. The denominative origin of formations of this kind has been discussed already. Then' force is doubtless most correctly denoted by the name ' frequenta- tives.' But they often deviate into the category of the intensives. What Buttmann and Lobeck (Ausf. Gr. ii. 2 p. 392 f.) have further collected under the head of frequentatives consists of mere details. For instance, in γεμίζειν beside γέμειν, ερπύζειν beside ερπειν, σεβίζο> beside σέβομαι, it is merely a question of a further expansion through the medium of a nominal stem. We may also pass over here what is collected, as an appendix to this, with regard to diminutives. For the category of diminutives is developed only in nominal stems. If there are apparently some few diminutive 391 verbs, e.g. έξαπατύλλειν (Arist. Ach. 657), ηβνλλιάν (Ban. 515), these are rather ventures of the Comic poets than verbs actually in use, and even these have evidently arisen on the analogy of denominative verbs. The inchoatives, which exist only in the present, were discussed in Ohap. X. 538 ANOMALIES. CH. XXJV.. CHAPTER XXIV. ANOMALIES, We have but little information as to the arguments by which Crates of Mallus endeavoured to maintain the principle of anomaly, which he defended against Aristarchus. Still we cannot but suppose that the verb must have supplied him with especially abundant materials. In modern grammar the verb has always remained the special domain of anomaly. Buttmann in his Complete Grammar disposes of the regular verb in the first volume in 240 pages, the irregular in the second in 332 pages. When we read what he says at the commencement of the second volume as to the irregularities of the verb, we become aware how in spite of the extremely valuable investigations, of which I have everywhere thankfully availed myself, in the ' Survey of the Anomaly of the Verb' (ii. 1-89), the number of irregularities remains quite overpowering. When he has reached the end of this survey he himself prefixes to the list of verbs, which seemed to him indispensable as a last refuge, the remark : ' Here, with the exception of the verbs which are derived from other words according to a definite analogy, like the great majority of those in αζω, ιζω etc. [ati. p. 472 he adds those in αω, εω, οω, αινώ, ννω, ενω], we must properly place all the rest.' But even those excepted are not wholly lacking among the irregular verbs, for e.g. in the case of many verbs in αω and εω fluctuations of quantity, facts as to the moveable * χω " σχ€ ^Τ χ6? ' * σ χ ον ) aerr (eV) evveire „ σπ* (eVt-oW-s) σ«τ (βπ) Ζπομαι } , σπε (σπεσθαι). 5) Anomaly through transformation of σκ. Here belongs once more έρχομαι because of its χ, and also (cp. pp. 192, 197 f.) πάσχω and μίσγω. 6) Isolated application of reduplication to the formation of the present stems of thematic verbs. We must dwell for a little on this anomaly, because we have as yet said nothing of it. Six presents belong here, viz. 398 γί-γνο-μαι, from Homer (B 468 γίγνεται) onwards, in common use, with the by-form γίνομαι, which according to La Roche Textkritik p. 220 is the more usual in the M.SS. of Homer, but which is regarded with probability as post- Homeric because of its inferior originality. In Herodotus editors write yiVo/ucu (e.g. iii. 80 εγγίνεται), and so Tycho Mommeen in Pindar (e.g. Pyth. iii. 13), and after Aristotle, in whom Bekker approves of γίγνομαι, γίνομαι is commonly used. For the origin ch. xxiv. PHONETIC ANOMALIES. 543 of the ϊ cp. p. 193. The other present formation γείνομαι was discussed on p. 216. — The parallelism with the Latin gi-gno is very noteworthy. In Sanskrit too there are reduplicated forms : (ja-$an-ti (3 sing, in gram- marians according to the Pet. Diet.). i -ανω Homeric (e.g. Σ 259 ίαύων), then once in Soph. (Aj. 1204) and Eurip. (Phoen. 1538). The syllable of reduplication extends beyond the present stem in Ιαΰσαι X 261 and ίαύσεις in Lycophron (101), as in the Homeric ΰωώσω (cp. above p. 474), while άεσα for ά/εσα, discussed on p. 520, is formed from the non-reduplicated stem. 'ί-σχω, a stronger present form of the rt. σεχ, in extensive use from Homer (E 812 ϊσχει) onwards. μί-μνω, a poetical by-form of μένω, to which it stands exactly in the relation of 'ίσχω to εχω, occurring from Homer (N 747) onwards. πί-πτω common from Homer (A 69) onwards. The poetical by -forms πίτ-νω, πιτ-νέω have been quoted on p. 184. τι-τράω coming from τί-τρη-μι (quoted on p. 108) by a transition into the thematic conjugation : neither verb occurs before Appian and Galen (Lobeck on Buttmann Ausf. Gr. ii. 304). For τετραίνω or τιτραίνω cp. p. 217. Reduplication hence appears as a present strengthening within the thematic conjugation only under quite definite conditions, especially in the inchoative and in the I-class. Of the six verbs just quoted, two follow those forms of the I-class which have an intensive character, i -αύ-ω and τί-τρά-ω seem to be for *ί-αν-^ω, *η-τρα-βω, and are therefore related to the verbal stems nv and τρα just as *yap-yap-j(o (p. 115), *τί-τανγω, ^τε-τραρ^ω (p. 217), *βι-βαγω (p. 226) to their roots. To the precisely similar presents γίγνομαι, 'ίσχω, μίμνω, πίπτω the 399 question attaches itself, whether they, as was long assumed universally, have arisen by syncope from *γι-γενο-μαι, *μι-μενω, *πι-πετω, or not. The expulsion of a vowel like ε which is particularly common in the present stem, is, as no one can deny, very surprising. Now if we reflect that from all these stems there are also forms in which the first consonant is in immediate contact with the final consonant, and the vowel follows both, like -γρη~τος=Σι8ώ. gnd-tu-S, γνί]-σ-ω-ς, σχέ-σι-ς, σχη-σω, σχϊ}-μά, μναο-μαι, μι-μνη-σκω, μνη-μη (cp. Princ. i. 387), πε-πτη-ώς, πτώ-μα 7 πτώ-σι-ς, it seems to me more probable that metathesis took place here; and that therefore γί-γνε-ται differs from a Skt. ga-gan-ti only by this phenomenon, and by its middle form, and so far is the regular present to the aor. ε-γεν-το discussed on p. 130. For 'ίσχω the form σγί-ς men- tioned on p. 132 and discussed more fully on p. 279, carries great weight, for it can only be explained in this way. We must notice also the forms l -σχά-νω, Ι-σχα-νάω, which presuppose an *1-σχα (cp. pp. 182, 183). On ε-πλε-το we decided in favour of the same view. Just as in ε-σγο-ν beside σχέ-ς the thematic vowel suppressed the vowel of the root, so the same happened here and made it appear as though a thematic form were present here from the first. With this view the reduplication appears as a relic of the old unthematic method of formation, and thus still better suits τί-θη-μι, Ζί-ΰω-μι, 1-στημι. 544 ANOMALIES. cti. xxrw 7) Eoots fundamentally different, united to form one verb. This highest degree of anomaly appears only in five verbs, which we may therefore call in the strictest sense mixed verbs : i.e. 3 with two stems, and 2 with three stems. a) Rt. εδ with its three-fold, or if we include εΰ-μεναι (cp. p. 104), four-fold present εΰω (Homeric, e.g. εΰει Ο 636, Uoi Ν 322, and here and there in other poets), εσθω, and εσθίω, discussed on p. 501, its perfect again with various forms (cp. pp. 368, 415), and its future formed like a 400 present, supplemented by rt. ώαγ, with the fundamental meaning of participation, enjoyment, recognizable in the Skt. hhatf. (Princ. i. 370.) b) Rt. hp and Απ• united in the notion of saying, Ap in the present Αίρω and εφε'ω (p. 213), in the perfect ε'ίρηκα (p. 360), in the future έρέω, έρώ, in the aorist έρρήθην, with the Herodotean variant είρέβην and in the verbal adjective ρητός, Ρεπ only in εειπον, ειπον discussed on p. 291. The present, which afterwards becomes obsolete, is replaced by ψημί, λέγω, αγορεύω, so that really the form of the verbs of saying is still more diversified. c) Rt. Fop, rt. fiZ and rt. όπ. Fop has been preserved without further- expansion only in ορονται, οροντο (cp. p. 144). fopa must be considered denominative. It is superfluous to give references for the particular forms, for all are in constant use from Homer onwards. The present formation from the rt. AS according to the lengthening class Αγομαι (cp. p. 153) is also extant. In the perfect all three stems are represented : εωρακα, εώραμαι (Attic), Αηδα, olda in ordinary Greek only in the sense of a present, οπωπα (poetic). In the future too (ειίησω beside δψομαι) two roots compete. I have discussed the anomaly of this verb so fully in Princ. i. 124 that I need not return to it. d) Rt. τρεχ and rt. ΰραμ. Both occur since Homer (Ψ 520, Σ 30). Still the division here too is not at all smooth, for beside the ordinary conjunction τρέχω είραμον ΙέΙρομα (only in poetry) and ΰεΰράμηκα Ιραμονμαι a poetical aorist έβρεξα also occurs {επώρέζαντος Ν 409, περιθρέ'ζαι Ar. Thesm. 657) and in Ar. Nub. 1005 άποθρέζομαι. In this as in the preceding group we can see very plainly how at first the different stems were inflected throughout independently, and how it was only later that, owing to the prevalent use of particular stems in par- ticular tenses, certain branches of each stem died off. The definite union of the stems, which had become defective, in order to supplement each other, is only the last stage in these processes. e) Rt. φερ, stem ενεγκ, stem oi. Here the limitation of the first root to the present stem is common to Greeks and Romans, and is therefore doubtless to be regarded as very ancient. Just as certainly is ενεγκ beside 401 ενεικ reserved to the aorist (cp. pp. 291, 463) and perfect (cp. p. 407) and ol to the future ο'Ίσω, οίσομαι. The traces of an aorist from the latter stem are weak, and, with the exception of the Homeric οίσενε discussed on p. 461, late. Here therefore the individual stems, which are united into a whole, are the most defective. The comprehension of this last and highest degree of anomaly evidently leads us beyond what we can call the structure of the verb into the province of etymology and synonymic. 545 EXCUESUS ON SOME RECENT EXPLANATIONS OF THE VOCALISM IN THE THEMATIC AORIST. (Vol. II 2 . pp. 35-44.) The investigations of the Indo-Germanic vocalism, started by Brugman since the completion of the first edition of this work, and pursued zealously by many of our younger scholars, have been noticed once or twice above, e.g. pp. 49, 97 note. These have also led to the attempt to refer what seems the very capricious vocalisation of the thematic aorist to more definite principles. As a certain agreement seems to have been reached on this point among a number of scholars, in spite of by no means inconsiderable differences of opinion on various wide-reaching questions, it seems to me proper to say a few words about it. In doing so, I have set aside altogether the question as to the vocalism of the primitive Indo-Germanic language, and have kept in view in the first place the statement of the case which Fick has given in his essay, ' Zum Aorist- und Perfectablaut ' (Bezzenberger's Beitrage iv. 167), as this is the most systematic, and the most decidedly adapted to the Greek language : but I may add that I have read attentively the more im- portant of the other studies bearing on this question, especially those of de Saussure (Memoire sur le systeme primitif des voyelles, Leipzig 1879), Kluge (Beitrage zur Geschichte der germanischen Conjugation, Strassburg 1879) and Johannes Schmidt (Ztschr. vol. xxv.). The thematic aorist is fond, as a rule, of short stem-syllables. Hitherto this short stem-form, compared with that of the present with a long vowel or diphthong (ψυγεϊν ψενγειν), has been considered the more primitive, while the heavier form of the present has been held to be due to intensification. Fick and others invert this relation, starting from the heavier present form, and regarding the lighter form of the aorist as weakened from it. He thus really revives the view of the ancient grammarians, who always regarded the 1 sing. pres. act. as the πρώτη βέσις. This acute scholar attempts to defend his starting-point historically also by some general considerations. That the structure of the verb started with the present is my own view also, on which this book is based. But it does not at all follow from this that a particular, and in this case the heavier, present form must be older than the stem-form of the thematic aorist. As ΰίΰωμι was preceded by an obsolete *ΰώμι, so φεύγω may have been preceded by an older *ψΰγω, formed on the analogy of the sixth class of Ν Ν 546 EXCURSUS. Sanskrit verbs. It was only the opposition between the lighter and the heavier form which could call forth the difference of meaning between present-imperfect and aorist. In the same way we can understand the thematic vowel in the aorist in accordance with the explanation which I have attempted : for if the aorist stem was originally a kind of present stem, it was just as capable of receiving this vowel as the heavier kind. The chief advantage, according to Fick's view, seems to lie in the fact that he thinks he can point out one single definite motive for the weakening which he assumes, viz. the accentuation. In this he follows an explanation first suggested by Benfey in Orient und Occident iii. 65. Greek indeed shows the tendency to accentuate the final syllable in the aorists only to a very limited extent, i.e. in the participle of the active, in the infinitive of the active and middle, in the 2 sing, of the middle imperative of all verbs, and in that of the active of five especially com- mon ones : ιδέ, λαβε etc. But in Sanskrit, in the evidently cognate sixth class of presents (tudd ^=Lat. tundit), the chief accent always falls upon the thematic vowel of all non-augmented forms, and the same holds good of the corresponding aorist, e.g. vidd-t=Jri$e beside d-vida-t = εΐΰε. It is almost exclusively in the participle, e.g. nom. plur. viddntas =βιΰόντες that the two languages quite agree. It is certainly surprising here that precisely those aorist forms, which might have been regarded as the proper sphere for the establishment of the phonetic form, the augmented forms of the indicative, nowhere show the slightest trace of the accentuation of the final syllable. If therefore we refer the short vowels and the rejections of the vowel to be mentioned presently (σπΰν beside επειν) to the accentuation of the final syllable as the causa moven8 y we must assume either that these shortenings extended from what we cannot but regard (especially in the earliest time) as the much less common moods, participles or even infinitives, or that as in Sanskrit, so even in the primitive Indo-Germanic language there was a system of light forms, not without a present indicative also, in which this operation of the accentuation of the final syllable took shape. Fick's conjecture that the short vowel along with the accentuation of the final syllable was originally proper to the dual and plural forms only (*0vyo/xer, *φνγέτε beside 0ευγω, φεύγεις) is without any support from facts. We cannot show that the vowel of the stem varied under the influence of the personal endings except where they were attached immediately, as in the verbs in μι (Skt. emi, imds, veda, vidmd, Gr. ol<$a 'ίϊμεν). We see therefore that this principle of explanation does not settle the question so very simply and beyond all doubt. I do not wish to deny that there is any connexion between the aorist forms with a short vowel, and the accentuation of the final syllable : — the forms in which the two are united are too numerous for that : — but it seems to me by no means proved as yet that the accent was the prius and the short vowel the poaterius. But if even the post hoc is not established, much less can we regard the propter hoc as proved. We have also, as Misteli (Ztschr. f. Vblkerpsychologie xi. 234 f.) has recently pointed out, to take into careful consideration the manner in which we must conceive of the Indo- Gormanic accent, whether it was rather musical or in the nature of a stress. Cf. Verner Ztschr. xxiii. 115. It is only after scholars have attempted to state connectedly the conclusions to which they believe EXCURSUS. 547 they can arrive as to the accentuation of the primitive language, that we shall be able to decide more positively as to such ultimate questions. From the presupposed original accentuation, Fick, carrying out consistently his view, which always starts from the present stem, arrives at the following effects l : — 1 ) ά i ϋ of the present stem are shortened into ά Χ ϋ. In the present stem t and ϋ are so rare, that we really have only to consider the relation of ο (Ionic η) to α : κί-κάΒ-εϊ* beside κη^-ειν. 2) The ε of the root in the present is rejected in the aorist. The number of these instances, like κέλομαι έ-κέ-κλε-το is extremely small. We cannot admit that είπον belongs here. Fick traces it back to *ε•ϊε-Ρπο-ν. But the combination fn is exceedingly improbable. Hence with the exception of ίκέκλετο there remain only the forms dis- cussed on p. 279, which I explained otherwise, and ενε-/κε~ιν, if we derive this with Fick from εν-ενεκ-εΐν. The same loss of vowel however appears also in μίμνω, γίγνομαι. Are we here too to lay down a *μιμνώ, *γιγνόμαι 1 3) Under the same category 'rejection of the ε of the root' Fick brings the reduction of the diphthongs ει and ευ to ι and v, which neces- sarily results from his point of view : λείπειν Χιπεϊν, κενθειν κυθεϊν. Considered purely arithmetically this is correct; as in κέ-κ\-ε-το so in Χιπεϊ,ν as against the present λείπειν we have to note the subtraction of an ε; but for the ear, that is, for the living language, there is a very great difference between a vowel which drops out from between the surrounding consonants, that is to say, disappears altogether through the intermediate stage of an irrational vowel, and one which, originally united with a second vowel into the unity of the same syllable, then disappears before it. It would have been thought that this union, which was no loose one, must have preserved it. The young generation of philologists delights in warning us against constructions ' which can only cut a good figure on paper/ and points to the living languages as a principal source for the discovery of what is possible and real in language. Provided that we do not recklessly mix up the distinctive characters of particular languages and periods of language, I am entirely at one with them. 2 1 Fick includes in his discussion the vocalism of the strong passive aorist. But as these forms are specifically Greek it is better to omit them in questions as to the primitive Indo-Germanic vocalisation, of which relics present themselves in the thematic aorist. 2 I may remark in passing that it seems to me an unfounded charge against the older generation that they slighted the importance of the living languages. To say nothing of others, I will mention only Schleicher, who pos- sessed the most extensive knowledge of living languages, and made abundant use of them, especially in relation to Zetacism, [cp. also his comprehensive treatise 1 Die Sprachen Europas '], and Ascoli, who is probably surpassed by no living scholar in this respect. The doctrine of Bopp as to the < degeneration ' of sounds in the course of the history of language is based essentially on the views which resulted from a comparison of Gothic and Old High German full-sounding vowels with the half -mute εϊν. The two classes of the thematic present formation, which I give as the first and the second, — of the first, however, only those with ε in the stem-syllable — would be, so to speak, the only normal ones ; there would only have been presents like λέγω or φεύγω. This view suits very ill the actual stock of present forms in Sanskrit and Greek. Our survey proves that a much greater multiplicity prevailed. The variety of the present as contrasted with the uniformity of most other tenses is, I believe, a character indelebilis of the structure of the Indo-Germanic verb. We only get to the present forms of the I- and of the nasal class frcm stems with short vowels. ExcuKsua 549 We can understand λίσσημαι only from λΐ7, άλλομαι only from άλ, ΰάκνω only from duic. How these forms fit into Fick's system it is not easy to see. In any case there can have been no lack of the ' mutilations ' to which many investigators have now such a dislike. As Fick actually assumes disyllabic stems φύγε, φεύγε, he gets an almost infinite number of instances of syncope. Forms like πίστις, πέπιθμεν, εΰειζα must have lost not only the half of the diphthong proper to their stems, but also by syncope the final vowel of the * basis ' : and who can tell whether Ισ-τί will be left to us ? How does a man propose to make such hypotheses appear probable, and how can any one expect them to receive assent, so long as it is not shown in the remotest degree how the facts of the lan- guage are to be explained with such views Ί But let us return to the aorist. Among the non-reduplicated forms, of which we enumerated 118, the proposed rule as to the vowel suits 29, viz. άΰεΊν (ήΰομαι), βρυχειν, ΰάηται (3ήω), ΰρακείν, ΰραπών, ερικειν, εριττε~ιν, ερνγε'ιν, ι?ε7ν, ικέσθαι (if we start from the Doric εΐκω), κυθείν, λαθεϊν, λιπεϊν, παρΰεΤ,ν,ζ πιθέϊν, άμ-πνυε, πραθεΊν, πτέσθαι, πνθέσθαι, the two-fold σπεϊν, στιχεΊν, σχείν, ταρπώμεθα, τμαγεΊν, τραπεϊν, τραφέιν, τνχεϊν, φνγε'ιν, while 89 aorists cannot be explained without the aid of unheard-of present stems invented ad hoc of the kind noticed above. We recognised 41 redupli- cated aorists. Of these 12 fit the rule, e.g. κεκαΰεΊν, κεκυθωσι, λέλαθον, 29 do not. But further, the originator of this view has himself not failed to notice that a part of the aorists here coming into consideration stand quite outside of his rule, viz. those with an ε in the stem. We have indeed, noticed already (p. 279) that this vowel is the normal one for the present. Still there are 17 aorists with ε left, like γενέσθαι, ερέσθαι, τεκείν, τεμεϊν (beside ταμε'ιν), ελεΤν. These Fick regards as present forms, which were only later on fitted into the system of the aorist, and subjected to the aoristic rule of accentuation. A view like this attracts me much more than the attempt, so much favoured by other investigators, to explain the abnormal vowel by means of so-called ' in- clinations.' To what present are we to suppose that γενέσθαι and ελε ~ι ν inclined % If the Greeks retained the instinctive feeling that ε suited the present, and α the aorist, I do not understand how in particular instances without any recognizable reason by the mere play of chance the abnormal form could creep into the place of a normal form once extant. But I can understand very well that forms existing at an early date, like *γένεσθαι, *τέμειν, *ελειν assumed the force and accentuation of aorists in contrast to other forms. Evan the participles ιών, ε ων, εκών in which the force of a present still survives, have shifted their accent. The ο in θορεΊν, οΚέσθαι is certainly to be taken in the same way. The thematic aorist remains therefore, even according to Fick's way of representing it, a tense-group which did not spring up at once, but like so much else in the language, was formed by degrees from the union of elements of various kinds. The present is not distinguished from the aorist by any absolutely certain criterion. As there are ab- normal aorists with ε, so there are presents with a. For in forms like ayw=Skt. agami Lat. ago, αρΰω, άρχω, βλάβω, γράφω, μάχομαι, we 3 [i.e. πραδέΐν. Cp. l-7rpa5-e-s Hesych. iii. p. 164 Mor. Schmidt in the Doric verse : νυν δ 1 T/vOes es χορον (MS. -f)vdes Ζεχωρον) νυν 5' ewpades (Fiek, Bezzenb. .Beitr. iv. 173).] 550 EXCUESUS. cannot help regarding this vowel as a primitive one. And therefore I cannot see that as yet the view has been by any means refuted that an aorist like τραπεΊν differs no more from an Ionic present τράπειν than ϊλεϊν from the presupposed present *ελειν, and that therefore in every case the aorist sprang first from the contrast with a fuller present formation deviating in some way from it. But it is certainly noteworthy that the α is so extremely common in the aorist, so rare in the present. Out of 116 thematic aorists, 54 have a. According to the older view a was regarded as absolutely a heavier vowel than ε. How does this heavy vowel come so extensively into a system, which elsewhere loves light vowels. If we could succeed in re- ferring λσχεΊν to the same principle as λιπείν, φυγείν, as Fick attempts to do, the 54 forms with α added to the 32 with t and υ would give 86 forms which follow the same rule. To this extent I do not now reject the new view which has been expressed in so many quarters. I would formulise it thus : ά is in Greek not everywhere the same sound. By the side of the full sounding α of Άγω, άπό— Skt. άρα, which I continue to regard as primitive, there is an a, which I might call the weak a. This latter has partly sprung as an anaptyctic vowel, just as in αμείβω, άμέλγω, αλείφω, from the vocalic element in a liquid or a nasal, as in ταμεϊν beside τε-τμεϊν, partly arises from a minimal vowel of the same nature as that heard in the Indian r- vowel, as in ε-Ιρακο-ν, which now seems quite identical with ά -drea-m, just as βραϊΰ-ς answers to the Sans- krit mrdu-8. The peculiarity of Greek lies in the fact that this minimal vowel assumed the colour of the a, just as α is always a favourite sound in conjunction with those consonants. I do not see any decisive reason for assuming for Greek or for a preliminary stage to Greek syllable-forming nasals and liquids: I hold it rather the more reasonable course with Kluge and Joh. Schmidt to regard these deduced sounds as groups, consisting of a minimal vowel, which Schmidt denotes by a small a, and the consonant concerned ; thus ΰαρκεϊν, later ΰρακεϊ r. That creations like mntos (with the so-called η sonans), bharntns (= hharantas), t%rns (Lat. trans), tnnutai (Gk. τάνυται) ever existed any- where but on paper — which is in truth very long-suffering — I shall refuse to believe, until some one has discovered a living language, in which sounds of this kind accented as well as unaccented are pointed out to such an extent and in such complicated groups of consonants. At the end of the syllable the nasal disappeared altogether after this a, just as indeed we find the same disappearance after the full-sounding vowels e ? o, e.g. ες for εν-ς beside εις, in the Doric ace. pi. e.g. in καλός = *καλόν-ς Att. κάλους, and in ΰαίμοσι, ποιμεσι. Hence the fact which was established from the very beginning of comparative philology, that a Greek α commonly corresponds to the syllables am, an, Lat. em etc. in the cognate languages: cn-rci =septem, ε-κατόΊ>=<%ηίηηι etc. In verbal forms the a of the 1 sing, in ήα, ήια, of the 3 plur. act. in τιθεασι and of the 3 plur. mid. in αται, ατο, with the α of κατεκτα, κατακτάμεναι is to be taken in the same way. Certainly it might often be difficult to draw the line between the weak a, and the full-sounding a, which cannot be got rid of. There are questions enough remaining here. But it is possible to adopt the recently gained insight into the origination of the o, without therefore agreeing with all these hypotheses as to η sonans etc. and as to the EXCUKSUS. 551 accent as the impelling force for all abbreviations. Everywhere in the science of language it is better not to wish to finish off everything at once. I content myself with the knowledge that an aorist stem with α before the consonants mentioned e.g. Spatcely, πραθεϊν is the weak stem- form beside that with ε in the present, e.g. *5ή»Λ<ν, πέρθειν. In oppo- sition to the radical view that of different forms running parallel, one must be always the older, the other the later, I often prefer to assume from the beginning cognate duplicate forms, the use of which only by degrees became defined the one from the other, as they did here in such a way that the heavier foims took the durative function, while the lighter came to denote momentary action. The appeal to living lan- guages and dialects certainly favours this view. For there is probably no living language, nor even any popular dialect, in which all wavering between fuller and shorter, heavier and lighter forms is absolutely ex- cluded. But hereby too something is gained, in that the forms with α now take their place better in the analogies of the aorists with ι and v. Wherever the realm of caprice and chance in language is limited, we have to regard this as a gain. INDEX, άαττοτ, 82 f . We, 194 άά(Γκω, 194, 199 άβραχςν, 77 άβροτάζω, 281, 454 αβροτάξομςν (conj.), 446, 453 αβρΰνω, 254 ayaaaOc, 523 ayayas, 463 άγαγβΓί/, 291, 292 ayaypcri, 39 άγαγοίηι/, 336 °7"7°Χ α > 415 f. ayayvpTTjs, 365, 410 ay ay ω μι, 27, 39 ο7ά£«, 118, 208, 523 ay ados, 513 ^αίομαι, 118, 208, 235, 523 & 7 αμαι, 118, 208, 523 a7 415 f. A GREEK, ayfoxa, 367, 406, 415 f. ay-rivi, &yp, 38, 317 0711/-», -€ω, 177, 183 f. ayKapvtfaovrce, 306 ά7λαΐ€Ϊ(Γ0αί, 475, 481 &7?&, 177, 184 ayvos, 224 *7ΐα//«, 110, 218 οτγνώσασκε. 531 ατγνώσσω, 256 fayvwaros, 525 aytyipavri, 457 &yovn, 46 (bis) ayopa, 215 α7ορ-άομαί, -ά£ω, -ξύομαι, 235, 252 a7wpaeroi;i'Tes, 469 ^ορςυέμεν, 339 aypaeev, 502 α7ριαίΐΌΒ, 253 ay ρ6μ*νοι, 279 αγυρ-μός, -ttjs, 215 ά7ϋρτ(ί^ω, 236 S7*aCe, 223 ά 7 χιβ\ώε, 132 #7χω, 143, 145 α7χω/)ί|αι/τ6ϊ, 454 *7«, 143, 145, 183 -f., 406 θ7ωνίδατοι, 418 ψωνίζομαι, 250 άδάμαϊ, 353 άδά^ατοι, 514 αδδτ^, 372 άδ«ί/, 283 #δ6ίρ6»/, 77 άδη /fJTes, 372 Βήμα, 270 άδήσω, 270 αδικ-η-η, -βι 140, 190, 248 αδικοίη, 335 σ,Βικοίημζν, 335 αδικοσυλφ, 335 άδρακτον, 281 άδ /xis, 372 άδ /iiW, 254 £δι;τοί, 517 ae0Aeua>, 252 άβιδ^^αί, 339 άίίδ^σί, 38 άείδ«, 153 o.€ik4\ios, 513 άζίκιώ, 481 oeip», 110, 215, 236, 284 aetVeo, 461 aeVrey, etc., 120 ae|o, 144, 181, 265, 445 αίρσγ, 456 άερτάν, 236 άβσα, 194, 272, 520, 543 αέσκοντο, 194, 199 £ere, 208 ά£-άι>ω, -αινώ, αζω, 132, 224 &ξομαι, 205 SCa>, 226 αι'ιθεσσον, 256 άηναί, 341 άθνρω, 255 aWC», 237 αίδβΓο, 305 αϊδέυμαι, 258, 268, 522 α'ιδέσθητε, 510 αϊδεσυ, 268 αΐδήμων, 268 αϊδο /iot, 143, 258, 268 αίδώϊ, 268, 522 αΧθω, 58, 153 αικάλλω, 255 αίλβω, 261 αίμάσσω, 256 αΙματοΧοιχάς, 154 αΐνίω, 523 αϊνημι, 26 αίνησουσ ι, 242 αϊν^μα, 242 5 aiVt'Co/xou, 242, 523 αΐν'ισσομαι, 242 5 αϊνυμαι, 112 άίξασ /ce, 531 αϊολέω, 246 ο^λλω, 212, 255 536 554 INDEX. αϊρίτωσαν, 307 αιρίω, 112, 261, 262, 542 αίρω, 110, 215, 284 αϊσθανοίατο, 66 αισθάνομαι, 175, 182, 207 f. αΧσθεσθαι, 276 αισθίσθαι, 283 αϊσθήσομαι, 271 αίσθομαι, 182 άί<τ0α>, 503 f. αίσσω, 222, 227 άϊστίίω, 244 αϊσχννω, 255 «re», 112, 537 αϊτ-ημι, 26 αϊτίισων, 242 αίτίάομαι, 237 ■Mfo 242, 537 αΐτονμεσθα, 62 αίχμά£ω, 237 άί», 182, 208 α,κάμας, 353 ακάματος, 514 άκαχβ/ατο, 242 οκοχ-εΓ»/, •6σ•θαι, 291, 292 α,κάχ-ημαι, 367, -ή /ueiOS, 426 άκάχησΈ, 242 άκαχί£», 110, 242, and 6 άκαχμ4ρο5, 367 f., 420 άκέο, 305 άκε-, ακεί-ομαι, 240, 522 άκ€στ<ίί, 522 άκήδεσα, 266 άκήκοα, 368, 403 άκηχεδατα*, 65, 242, 417 f. άκ-ηχ*Μνπ, 242, 378 άκηχέμςνος, 417, 426 ακλαυτοϊ, 209 ακμή, 420 ακμ-ητ, άκαμαντ, 352 ακοντίζω, 250 ακουστό»', 310 ακουκα, 414 άκουσαν, 451 άκουσεί»;', 533 άκουστιαν, 536 ακούω, 253 άκωκή, 373 ολαί»Ό>, 185 αλαλάζω, 227, 237 άλάλημαι, 367 f. άλαλήμ^νος, 426 άλολησβαι, 378 άλάλτ/σο, 423 άλαλκάν, 291 f. &\αλκον, 265 άλαλύκτηααί, 236, 367 f. αλάΚυκτο, 368 αΚαΚύσθαι, 367 f., 378 αλάομαι, 185, 282 άλαττά^», 454 Λλα<ττο$, 514 ολγησ€Τ€ (conj.), 446 άλδαύ», 282 άλδήσασκε, 532 αλ,Ησκω, 189 f. (bis) άλεείνω, 254 aXtivai, 178 άλ(ΐφθ4ν, 500 αλείφω, 153 άλίξασθαι, 444 άλ6|€μ6ΐΌί, 339, 444 άλφυ, 304 ολεξω, 144, 156, 265, 445 ά\4ομαι, 194 OAeVflat, 283 oAerat (conj.), 313 άλεύατο, etc., 458 όλ6ί)μαί, 475 άλ€ύω, 194 άλε», 178 άλήλςσμαι, 368 αΚ-ηθην, 510 άλή0», 502, 504 άλήλίφα, 367 f. άλβ-αίκ», -ήσκω, &λθομαι, 185, 190, 194, 199, 264 άλθίσκω, 195 ct\teiS«, 252 ολΓι/αι, 178 αλίνονσιν, 178 a\iV», 170, 178 άλίο», 244 αλίσκομαι, 79, 133, 191, 195 α\ισμ4νοτ, 373 άλη-αί™, 117, 177, 185 aAiTei»/, 283 ολχτήμ-€»Όϊ, -»v, 373, 426 αλιφθερώκει, 394 αλί», 207, 249 αΚκάθω, 502 άλλαγηι/αι, 492 άλλάο-σ», 256 "άλλαχα, 365 άλλάχβτ?, 500 αΚλομαι, 203, 211 άλλόμαν, 58 ίλμ€ΙΌΪ, 90 άλοά», 244 αλοίην, αλφην, 329 άλο», 305 αλσο, αλτο, 90, 130 άλυί», 251 άλνκτάζω, 236 άλύ-£», -σ/ία;, -α», 195, 412 άλυσκ-ά\», -άνω, άλύσκω, 175, 182, 194, 196, 199, 237 α\φάνω, 175, 181, 283 άΚφίσίβυιαι, 272 άλφηστής, 272 αΚφοιν, 332 f . αΚώμεναι, 339, άλώναι, 341 a/xac 195 άμ£λυι/α>,.254 αμβ\ύσκ€ΐ, 195 αμβλυώσσω, 257 άμβλώσκω, 195 άμςίβω, 153 a/uetn-To, 131, 373 ά/*6ίψ6ται (conj.), 447 άμελγεϊ, 139, 383 άμ4\-/ω, 143 αμεύω, 153 αμμίνης, 372 . άμπάζονται, 224 8 άαπ67ταλώ»', 211, 295 αμ-κισχ-έΐν, -ίσθαι, 285 άμπισχνέομαι, 184, 283 άμπλακε»/, 283 άμ-πνυτο, -irvve, 129, 278, 286 άμυνάθςιν, 502 άμνσσω, 222 άμφαφάασκε, 530 άμφαφάω, 166 αμφιαχυια, 373 αμφιβαΚςνμαι, 475 αμφιβέβηκα, 380, 428 άμφιίννυμι, 113 άμφι4στισι, 39 άμφιμ4μυκςν, 377 άμφισβατήκειν, 394 αμφίσκω, 191, 197 άμφιώ, 480 αναβατό, 304 άίΌβιωσκοααί, 190, 199 f. άίΌ/8λυστάΐΌ>, 182 αναβροχίν, 491 άναβρώσκων, 194 άι/αγγ €λ ' 0,/τί > 46 άΐΌ7€7ράί>αταί, 66 αναΎ^γράφονται, 417 ανάγκη, 373 άι/αγ^ώοί/τι, 484 άι/αγ^ασεί», 534 άι/άδαστοϊ, 208, 521 άι/αδεδρο/χί, 373 ά» αδράαεταί, 468 άνα-θ4μ€ν, -θίμειν, 339 f. άί/αθί'σοί/τ«ϊ, 465 άναιλίθαι, 69, 350 άκαίνομα*, 217, 536 άναιραιρςμ4νο5, 367 άι>αίσ0ητοΓ, 515 ά^άκ€ίΚ6. 414 αναΧ-ίσκω, -όω, 195 INDEX. αναλώσω, 273 αναμεμίχαται, 418 άνανγελίοντι, 475 αναπαρΐ'ις, 493 αναπαΰεο, 304' άναπετώ etc., 480 άνασκολοπί€?σθαι, 481 άνασσείασκε, 531 ανάσσω, 256 ανασχο μην, 334 ά^άσχου, 305 άι/οτεθο, 491 ανατ(θ4καντι, 46, 385, 415 άνατίθεντι, 46 aitfdV», 156, 180 άνδίχά{α>ντί, 46 άι/δρί£ω, 250 afe'0€i, 148 ανζθίν, 127 οι/ίθε'τοί', 52 άν4θιαν, 49 αν4κτημαι, 358 άι/€κτθϊ, 515 άι/€λ(ίσ0ω, 308 άνελπίστο?, 515 άνςμόω, 238 άνίμορμυρεσκε, 529 ανίπταμ4νη, 358 άνεπτόμεσθ', 63 άι/€στακ€τα>, 422 ανΐστάκουσα, 394 &>^χ«ν, 304 avecpy-ov, -a etc., 81, 363, 393, 398 4 αν4ωνται, 417 άι/βώξεταί, 436 Οί/!ώσ0αί, 273 2 , 417 αν4ωχα, 407 άι/ήρ, 315 αφοθΐν, 399, 429 άντ)φθω, 366, 423 άνθελέσθων, 310 άι/ίά-ω, -£α>, 235 άι/ί -eer/ce, -ησκε, 529 f. άνκοθαρίοντι, 475 ανοιγί\σομαι, 493 ανοικοδομηθήμςιν, 340 ά• στα, 299 ανστήμεναι, 339 αντευπςποίηιαν, 374 άι/τιάσητον, 447 αιτιβολοίη, 335 αντιδωρησαίατο, 66 αντιλίγωντι, 46 ai'Ti-Jo), -άω, 235, 237 ά^τίώ» (fut.), 479 άΐ"Γίοα>ί'τωί', 306 άντιπςπονθ4μςν, 391 άντιπριάηται, 319 άί/τίΤ6τάχαταί, 66 &ντλην, 339 &ντομαι, 143 άίτ^'η, 335 άΐΊ>7τόδητο5, 517 άνό-«, -τω, 122, 144, 163, 168, 171, 283 (fut. 483) *w, 122, 144, 171, 178 αι/ωγα, etc., 373, 379, 387, 424, 429 άΐΌ>06θίΐ7, 335 άί>ωχ06, 384, 387 a£ere, 461 άξίάσβί, 246 aouros, 134 απα•γγ4λλζσκ*, 530 απαγγ4λλοισι, 47 απαιρ4ζσκον, 530 απαποΰμζσ& , 62 απαλλαγμένος, 372 a7raAAa|eioj/TfS, 534 ά7Γά^6ί7ΓΤΟ, 373 άπ-αξο'ι /Tt, 469 άπαστο*, 263 άποφβΤί/, 291 f. απαφίσκω, 191, 195, 199, 292 απφάφθη, 500 άπείθηι/οι, 339, 341 ά7Γ€ί0θί-775, -τ?, 335 ΟΤΓ€ΐλ€ίω, 241 απειλημένος, 372 απίκιξαν, 121 απ*κληϊσθησαν, 525 άπβσπάδαντο, 418 απεσσονα, 491 άπβσταλκαν etc., 384, 385, 414, 464 f. οτΓβστελλαν, 457 απεχθάνομαι, 182 άττεχβεσβαί, 285 απςχθησομαι, 271 απεψημ4νος, 372 απηλλά-γην, 500 απηλλάξεσθε, 436 απηλλαχα, 407 απηΰρα, 134 άπηχθετο, 276 απι4ωσι, 319 απίκατο, 418 άποαίνι /yuoi, 112 αποαίρεο, 305 απόβλητος, 515 αποβρίζω, 454 απάγ^με, 131, 144 απο'/ραφ4σθη, 70 f . άπογραψβν, 469 αποδαρ4ντα, 492 αποτάσσομαι, 208 ά-η-οδεδόαί'βί, 47, 49 άποδε'δρακα, 357 άττοδίδωτΐ, 38 αποδίνωντι, 178 απο-δόμεν, -δόμειν, 339 f. άποδύιτω, 306 άποδ^τω»', 306 άποδοσάι/τα);-', 465 άποδόσβα»', 310 οττοδρύφω, 144, 150, 166 άπο4ργαθ*, 504 άπΛερσΈ, 456 ά7το56ΐπά00ω, 69 οτΓ(ί0€στο5, 220 αποθρ4ξομαι, 544 αποθύσκειν, 197 άποκαταστασόι/τι, 469 αττοκίκλανται, 419 άτοκ€/«ί'δυΐ'€Οσ€ταί, 436 άποκεκύφαμες, 384 αποκινήσαοκ*, 532 αποκλάί, 127 A[FOK]TENEI, 452 απολ(λαμμ4νοι, 361 άπολελνγμ4νοι, 361 άπολελογηται, 373 άπολεσκετ», 531 άττολογίξάσβαχταϊ', 310 άπολοίατο, 65 απολονμεθον, 67 άπ^λωλα, 373 άπολύλη, 432 απομείρεται, 214 άπομνττω, 219 οττοί/αίοτο, 66 07ro|tWuTat, 114, 171 οποπ€<ρ€υγη, 428, 431 αποπλύνεσκε, 530 ΟΤΓ07Γί'ί7€Γ€1', 492 Ο7Γ07ΓΤ770ί, 298 άποροι, 336 αποσήθειν, 501 άποσκίδι/ασβαι, 117 άποσκλτϊί/αι, 132 αποσταλαμεν, 491 αποσταλθ4ντ€$, 499 αποστεροίη, 335 αποστηλάντων, 457' αποσφαγησοιντο, 493 άποτεβί/ασαι/, 428 ά7Γ07€ζσί? etc., 113, 141; 158, 469 αποτίνετον, 310 άτΓοτίι/οίαί/, 328, 451 απότμηται, 371 άττουρα?, 133 άπρακτος, 515 άπτόμεσθα, 62 όίπτω (apio), 82, 166 άττνδάαϊ, 465 άττυδ^εναι, 40, 339 oVu(ttos, 514 άττυτ€ί6τω, 158 απωλόμΐσθα, 62 αραιρηκώς, 367 f. αράβ-ησ€, -ιζς, 243 &ράρα, 368 αραρέΐν, 291 f. αραρίσκω, 191, 195, 199 α^άρ-, αρηρ-ύια, 402 556 INDEX. άράσσω, 219, 221, 227 apyea-T-fis, 242 αρΎμ4νο5, 372 άρδω, 143, 217 'Αρέθουσα, 502 ap-eiv, -4σθαι, 284 αρύων, 522 apeV-ets, -ορταί, 481 αρέσκω, 188, 191, 195, 514, 522 αρ4σται, 69, 279, 284 άρ6τάω, 237 αρετή, 514 άρήγω, 156 *άρ-ημ4νος, 372 &ρηρα, 367 άρήρεκα, 367 f. αρκ\ρεμ4νο5, 271, 417 άρηρομ4νη, 367 f. aptSei/ceros, 514 αριθμό*, 507 αριστάν, 237 άριστεύεσκε, 530 αριστεύω, 252 apt-, άρό-χ€τοί, 143 άρκ4ην, 342 άρκεω, 239 αρμενοε, 130 άρ/Αογή, 239, 242 5 αρμόζω, 239 αρμο'ιματα, 239 αρμοσμένος, 372 αρμό-σσω, -ττω, 239, 242 5 &ρνυ-μαι, -ται, 110, 168 άρνυσο, 304 αρόμμεναι, 340 αρπάζω, 167, 235, 453 f. αρπάμενοε, 133 f. αρκώμαι, 235, 479 άρρ-ά^ω, -ifa 226 f., 536 άρσαι, 456 άρτίαται, 372 άρτ4ομαι, 239 άρτύνθη, 499 άμτύ-ω, -ι/ω, 254 αρΰσσονται, 168 αρυτω, 163, 168 άρυω, 168 άρχ4μ*ναι, 339 άρχεύω, 252 αρχόμεσθ\ 62 άρχω, 143 "άσαμεν, 520 ασβίσθε, 77 aVj8e, 222 ούγά^ορ-αι, 237 ανδάω, 237 αύδήσασκε, 531 αύ4ρυσαν, 122 ανθίντηε, 283 ανλησευντι, 469 αύ£ άνω, -αν4ω, αΰξω, 175, 181, 265, 445 αυξήσω, 270 αυξούμενοί, 265 αυχ4ω. 240 αδω (kindle), 152, 155 αί/ω, αίίω (make dry), 155, ^ 226 άφαι\ησ4σθαι, 69, 542 άφάσσω, 166, 257 αφαύω, 155 άφάω, 166 a-6ίτ)Τ€, -βίτβ, 330 αφείλατο, 464 άφ4\αι, 463 αφεστήκη, 431 a(p€TT)s, 517 άφ4ω-κα, -μαι, -ντακ, 273, 417 άφήτ?, 315 άφ'ιητι, 38 a, 329 )8oiyco, 177, 185, 204, 216, 517 βαλ-4ειν, -εΊν, -4σθαι, 284, 349 )8a\eu, 304 βάλησθα, 34, 39 βάχτισι, 39 βαλ\4σκετο, 530 βάλλευ, 304 βάλλω, 202, 211 βάλοισθα, 35 βαμβαίνω, 217, 377 /θάπτω, 160, 162, 166 /Sape'co, 240, 243 βαρύθει, 502, 504 βαρύθεσκε, 530 βαρύνω, 254 βασενμαι, 469 βασιλεύω, 252 jSaVis, 517 βασκαίνω, 253 βάσ-κω, 189, 192, 200 βάσομεν (conj.), 446 βάσσων, 202 βατοϊ, 511 £οώ£ω, 237 i8a, 211 βεβάασι, 48, 386 )8e/3aios, 410 βφάμεν, 424 βεβάναι, 424 06&χώτα, 426 0<=j87j/ca, 379, 413 βεβίηκε, 413 β4βλα-φα, -μμαι, 358, 407 β4β\εφα, 400 β4β\ηαι, 417 βεβλήα-ται, -το, 65, 435 βεβ\-ηκειν, 432 βίβλ-ηκοι, 413, 423 0€/3λτ7χυΓα, 416 βεβολήατο, 271 β4βριθα, 378, 401 β4βροχα, 399 5 /Se'jSpwxa, 377, 401, 407 8 INDEX. 557 β^βρύχν, 376, 422 βεβρώθοιε, 379, 423 βεβρωκώς, 413 βεβρωτζς, 387 βςβωσα, 395 βςίομαι (βεομαι, βίομαί), 468 βαλτός, 514 βίρνώμ^θα, 118, 178 jSeppecu, 118 /SeTTOv, 464 β/ωμε»/, 320 0ί»0ι, 297 βηναι, 341 βήσεο, βήσΈτο, 461 βήσομζν (conj.), 446 )8ήσ•ο-ω, 218, 256 £ήα>, 315 βι-άω, -άζω, -αίω, 234, 235 0ι|8ά£ω, 226 βιβάκτ-ης, 226 0ij8<£* etc., 105 βιβάσθων, 503 f. βιβρώσκω, 193, 199, 218 0ι0δ, 479 βι-ησεαι (conj.), 447 βινεσκόμην, 527, 530 βιωατο, 65 βιωναι, 341 βιώσκομαι, 194 βλαβών, 489, 500 βλάβ€,αι, 144 βλαβ-ησομαι, 493 βλαβύσσειν, 167 0λά£«ν, 222 5 βλάπτω, 160, 167, 168 βλαστάνεσκε, 530 βλαστ-άιω, -4ω, 175, 182, 271 βλαστέΐν (aor.), 284 βλάστ-η-μα, -ais, 271 βλαστήσω, 271 βλαστέ, 281 βλείησθα, 35 βλέπω, 144 βλ^ται, 313, 315 βλίττω, 256 0λό$», 182, 222, 524 βλνστάνω, 182 0λύα>, 222, 524 βλώσκω, 190, 193, 199 βοαθησί~ω, -οντι, 468 f . βάλλομαι, 172 βόλομαι, 144, 172 βοσκίσ κοντό, 530 £(ί(Γκω, 189, 193, 265, 528 βουκολ&σκες, 530 βονκολιαξτ}, 469 βούλβταί (conj.), 322 βουλενησθα, 34 βουλώνω, 252 βούλη-μα, -σΐί, 265 βούλομαι, 172 1, 179, 265 322, 353 βονλόμίσθα, 62 βοώντων, 306 βραβεύω, 252 βραδύνω, 254 £ρά£ω, 220, 222, 225 βρασμός, 220 βράσσω, 220, 222 βραυκανασθαι, 183 0ραχ«>, 284 βραχεΐσα, 492 βρ4μω, 144 jSpe'xw, 143 0pif«, 224 0ρί0α>, 501, 504 βροντάς, 134 βροντάω, 237 0ρυ£ω, 226 0ρυκα>, 218, 411 βρύττω, 218 βρυχάν, 284 0ρύχω, 218 βρωσείοντςς, 534 0ό£•, 184, 222, 224 j8we«, j8iW, 178, 184, 222 0Λ», 184, 222 βωβύζίΐν, 226 βώλομαι, 172 βωτάζειν, 134 7ογ7 αι ' Ι ' €ί, '> 217 γάδβσθαί, 156 7atVeTat, 122 γαία», 112, 209, 501 Ύαμςτή, 514 7αμβ'ω, 261 f. (fut. 480) Ύαμησείω, 534 yavoco, 112 ηάνυμαι, 112, 194 -γανύσκομαι, 194, 199 7 αρ7αίρ«, 312 (bis), 376, 536 7αρ7αρ^ α '» 242 6 7ai»ptai', 536 (7) δοϋΐτβ'ω, 262 Ύβγάασι, 387 767άθ€ί, 391, 394 7e7cuceti', 394, 414, 424 7*7<"65, 282, 426 7e7f« / a j ueVos, 272 Ύ^έν-ημαι, 272 *γ4γ€υμαι, 525 7^βο, 378, 401 Ύί~/λυμμ4νο$, 358 7€7°» , α, 380, 399 7*7<ίί / «ίν, 394 Ύ^ραπτη (Boeot.), 61 yiypa -πτοι, 61 7€7Ρ«Ψ αται > 64, 67 7€7Ρ α Ψ" ται > 427 7«7ράψετοί, 436 yeypupas, 379 7€7ωί/€, 377, 401 (imper. 422) 767 424 7*7«ϊ / €τα>, 422 Ύ*Ίωνϊ\σω, 271 767^1'- ίσκω, -e'a>, 196, 262, 271 Ύ&γωσα, 395 y δίνομαι, 216, 543 γβλαι, 41, 300, 384 γ4\αιμι, 26, 39, 134, 247 Ύβλασείοντα, 534 7€λα(ΓΤ(^5, 515 76λάω, 235, 523 •γελώντων, 306 76/*€», 261, 268 γ*μίζ», 537 7«'μω, 144, 268 yeveia -σκω, -ζω, -ω, 194, 199 ytvwdai, 284 7eVe(Tts, 272 76ΐ'€σκ€το, 531 yeveTrj, 272, 514 7€J , e / Ti7s, 272 ytv -ησομαι, 272 yej /ητ οι, 61 yevoiaTO, 65 7*»'<>μ.€σ0α, 62 7€Vto (became), 130 7eVro (grasped), 131 7epatpo), 255 7€υμ*0α, 370 7et>, y -ηθομαι, 112, 240, 261 f., 501, 503 7ήλ€σ0αί, 179 γηράντεσσι, 134 γηράσκω, 134, 190, 194, 199 77?p€i's, 493, 496 y -ηρύω, 250 yiyv -ομαι, yiv-, 216, 258, 282, 542 f . 7*7ΐ/ώσκα>, 187, 189, 192 ff., 199 yίvυμaι, 112 yιvώσκω, 193 7λά^ω, 225 yλάπτω, 166 7λά(ρω, 144, 166 ΊΧίχομαι, 143, 150, 198 yλvκaίvω, 254 7λυφα>, 144, 150 yvάμ■πτω, 167 yvaώ0ι, 297 yv^tv, 320 γνώμεναι, 339 7»/ώνοί, 341 558 INDEX. -γνώομεν, 213 -γνωρίζω, 243 -γνώικω, 189, 193 -γνωσοίατο, 66, 486 -γρωσόμεσθα, 62 -γνωστός, 526 -γνωτόε, 511 γοάασκε, 190, 530 γοάω, 234, 273 •yoyyu^o), 226 -γογγυσμόϊ, 226 7οίδτ?μ», 383, 390 yoov, 284 Ύραμματεύω, 252 Ύραφεν, 491 7ραφί}ί'αι, 492 7ρά<ρω, 144, 265, 406 Ύραφώισι, 47 Ύραψείω, 534 Ύρίψω, 144 7pG, 7P"C«> 224, 237 Ύρυμπάνω, 180 Ύρυπ-αίνω, -ανίζω, 180 7p«^r, 175, 182 δαρθεϊν, 284 δαρθβίί, 499 δασάσκετο, 531 δασ^ί, 524 δασσάσθωσαν, 310 δασύνω, 254 δάσωντοι, 447 δατεασθαι, 459 (bis) δατεομαι, 208, 261, 521 δεα, δοο, 118 δέάτοι, 61, 318 δέγμενος, 104 δ€δο-ο, -ο»>, 267, 292 δεδάοκτβαι, 417 δεδάηκε, 413 δεδαίαται, 65 δεδακε, 292 δεδάρδαφε, 407 δεδαρμε', 424 δειδίσκομαι, 191, 197 δειδίσσεο, 304 δειδίσσομαι, 221, 411 δβίδοίκα, 378, 397, 413 δειδίσσομαι, 376, 536 δείδ«>, 393 δείκανάο/Αοι, 110, 176 f., 183 δείκνυε, 301 δείκνυμι, 110 δβίκι/ύι/αί, 341 δείλετο, 232 δειμαίνω, 253 δε?,/, 210, 265 δειττν-ίω, -Ίζω, 240, 243 δειιτι/ησεία», 534 δειπνησευντι, 46 δείπνε?!/, 481 δείρω, 203, 213 δείτητ*, 447 δε'κευ, 304 δέκτο, 131 δε'λλω, 211 δεμω, 144 δβί/δίλλω, 212, 377, 536 δεξαίατο, 66 δείοίατ, 66 δεξούνται, 469 δέομαι, 251 δερκομαι, 143 δερρω, 213 δε'ρω, 144, 213 δε'<πϊ, 517 δεσμοί, 519 δεσττ^α», 239 (bis) δέσποινα, 239 δεσπόσυΐΌϊ, 239 δεσττΛτηί, 239 δενκω, 155 δευμεϊ/ο,/, 370 δεύομαι, 251 δεύω, 265 δεφω, 144, 445 δεχαται, 104, 370 δίχνυμαι, 110, 183 δέχομαι, 110, 143 δεψω, 144, 445 δεα>, 517 δηθύνησθα, 34 δή«ω, 156 ',180 δηλήσ-εται, -ηται (conj.), 447 δηριαάσθων, 309 δηρινθήτην, 499 δτ?ρί-ο/χο«, 207, 499 δηρίω, 249, 254 δησάσκετο, 531 δΚ 467 διαβεβ-ηκει, 428 διαβεβλΐσθε, 423 δίοτ/^ντω, 306 δίαγ^ώι/οί, 341 INDEX. 559 διατγνωντι, 46 διαδ4λ\€ΐν, 211 διάδημα, 517 διάζεσθαι, 220 διακαθεξίομεν, 469 διακάρισται, 371 διακρινθ-^Ίτε, -ημεναι, 329, 499 διακω\ύ(Τ€ΐ (opt.), 452 διαλελαμμένοε, 419 διαλςχθηναι, 510 διαΧνσενντι, 469 διαλυτό*, διαλυτά, 515 δίάμςπττοί, 514 δίαπβττληχ^ (?), 407 δίαπ/πλίχβ, 402 δίαπβττολ^τϊσόμείΌΐ/, 436 διαπραθ4βιν, 349 δίαρ•η-ά|οί5, 455 διαρρίτττασκβν, 530 διάστημα, 518 δίοσκβδα?, 480 διατό/χ,νείΐ', 179 διατρυ<ρ4ν, 491 διαφ6ρ4των, 310 διαφθαρήσομαι, 493 διαφθ4ρσει, 457 διαφώσκειν, 193, 199 διδάσκω, 196, 200, 265 δίδει/τωΐ', 105 δίδη, 105, 517 δίδοί 41, 301, 384 διδο?-μ€ν, -Τ€, 329 δίδοισθο, δίδοΓσθα, 34, 37 διδόμεν, 339 διδόϊ/τωι/, 307 διδόσθω, 309 δίδοι/, 301 διδουναι, 341 διδράναι, 108 διδράσκω, 189, 193 διδύσκω, 199 δίδωθι, 298 δίδωμι, 105, 376, 517 διδώσω, 474, 543 δίδωτί, 38 (bis) δι*, 120 δίεδάσσαο, 450 δι*ί\(Ύμαι, 361 δΐίκδυη^αί, 494 διακριθώ, 499 διβλβγβι/, 490 δκ\4γησαν, 490 δί€λ€ Χ θ€1/, 490 διεξίμεναι, 339 διίρσαι, 214 δι4ρση$, 456 δίετβτάχατο, 66 δι4τμα-γ-€ν, -ον, 412, 489 δΐ€τρί'/3η, 500 δι4φθοραί, 399 δί&μαί, δί£ω, 106, 226 δικά^ητοι, 61 δινά^, 235 ff., 453 (bis) δικά>, 479 δικάσσω, 455 δικαωσι, 479 δί /cerf, 191, 284 δίΐ/€υ€σκ\ 530 διν4ω, δίνω, δ'ιννω, 178, 216 διόγνητο*, 511 διοικ4ν, 342 δίοίτο, 334 δίομαι, 148 δίττλά^ώ, 236 δίσδημαι, 106 δίχοίω, 234 δίψοίίΓί, 47 διψάω, 237 διωκάθω, 502, 504 διωκ4μςν, 339 διώκετον, 52 διψκηκα, 373 διώκω, 156, 411 δμηθ4ντα, 500 δνοπαλί£ω, 454 δο, Rt. 517 δοάσσατο, 118 δοΗναι, 346 δοθα, 491 δοίη-μεν, -re, δοΐμεν, 330 δοίηκ, 329 δοκά^ώ, 235 δο /ceoj, 258, 262 δοκιμα£οΊ/τωι/, 306 δοκιμάζω, 237 δοκιμαξόΊ/τι, 454 δοκίμωμι, 26, 247 δοκοί", 335 δοκοίη, 335 (bis) δοκο'ιησαν, 336 δοκοΊμι, 335 δό>€ΐ/αι, δό>€ί/, 339 (bis), 344 δόντω, 306 δθξ€?Τ6, 469 δό^, 298 δοσυ, 517 δύΌ-κον, 528, 531 δοτήρ, δότης, 517 δοτ(5ί, 512, 517 δου»/, 343 δούναι, 341, 346 δουπεω, 262 δραίνω, 216 δρακέιν, 284 δρακ4νπ$, 493 δραμ, Rt. 544 δραμέΐν, 284 δρα-πών, 284 δρασείων, 534 δράσσομαι, 218 δράω, 216 δρβ'ττ-ω, -τω, 144, 164, 168 δροκτά{>ί5, 236, 281 δούπτω, 166 οϊ/μβι/, 335 δρφην, 31 δρ'/ί?ί, 335 δρφμςν, 335 δρψμι, 335 δι/, Rt. 517 δυη, 329 δΟθί, 298 δνβμή, 579 δΓι/, 343 δνναι, 341 δύι/αμαι, 76, 117 δυνάμαι, 318, 524 δυνάμςσθα, 62 δύνα-νται, -ντο, 65 δι^άσθη, 524 δόνησι, 318 δύι>ω, δυι/ε'ω, 118, 177 f., 184, 198 δΰιττω, 167 δνρομαι, 216 δόσγω, 192, 197 (bis), 199 δύσβτο, 461 δνσις, 517 δύσκβ, 531 δυσκολαίνω, 253 δυσμή, 519 δυσχεραίνω, 254 δόω, 118, 148, 167, 517 δώ, 30 δώομεν, 313 δωρητός, 515 δωσείβι»', 534 ?a( = efy), 328 «άγην, eo|e, εαγα, 79, 362, 396 %άδα, 79, 362, 396 4ά\ην, 79, 491 Ιάλωκο, 362 4ά\ων, 79, 133, 273 ίάνασσε, 79 4αρδάλη, 79 « ϊάσομςν (conj.), 446 ea -ται, -το, 65 ϋ. έάψθη, 82, 498 4άω, 34, 363 Φάβη, 499, 518 ϊβασον, 84 4β€β\αστηκ€ΐ, 358 C07/J/, 126 εβητον, 54 e0iW, 134 1£λα0€, 275, 284 (aor. pass.). 491 4βλαστήκασ, 358 ϊβΚαστον, 281 4β\άφθησαν, 500 €/8λω, 132, 387 4βονΚηθην, 510 %ω, 133 fy»»'» 130 560 INDEX. 4γγ^ύημαι, 373 4γγ\νσσω, 256 4γγρα<ρ4μ*ν, 340 4yypae*, 171, 178 «(Το/και, 202 f., 222 4ζόμεσθ\ 62 ί£0γΐ|ν, 492, 500 βηκο, 80, 85, 106, 282, 464 4-f)u5av€, 79, 81 4θ(ίρω, 255 4θ4\-ησθα, -εισθα, 34 (bis), 37,39 4θ4\τ)σι, 38 €0βλω, 266 e'fleAayii, 27 1, 39 ϊθςμεν, 127 ?0τ?κα, 282, 411, 464 4θιονσι, 481 4θ\ασμ4νο5, 358 ?0os, 506 %6ξα, 544 tftfiev, 147 *0», 143 ^«/ca, 273 β'θώ/ίοτί, 385, 414 eT (fut.), 33 ςϊα-κα, -juoj, 363 βίασα, €Ϊων, 84 έίασκον, 190, 530 f'/a -ται, -το, 65 6?δα, 463 ζϊδ-4ω, -ΐίτ\ν, ~4ναι, ώί, 344, 391, 402, 422 «ίδήσω, 544 βϊδο/χαι, 153, 266 €Ϊΰομ(ν etc. (conj.), 313, 422 «Τδο*/, 84 βίδυΓα, ιδυΓα, 402 €ίημι (== cT/«), 121, 247 φν, 329 ct -ησθα, 35 (ϊτιτον, 54 cWiCov etc., 85 «ίβίκα, 364 «Γκα, «f^ai, 85, 364, 414 f. ϊϊκάθω, 502 et/ceVat ν. 4οικ4ναι, 504 f. βϊκτον, 387 €Ϊκυ7α, 4οικυια, 402 β&ν, 154, 394 βϊκω, 153 €, 110 €ίρ7ω, 143 «Jfpco, 304 clpeu), 544 βίρη-κα, -/αοι, 360, 544 ανήκων, 394 INDEX, 561 An, 364, 414 έίρομαι (ask), 213, 263 ςίρπομες, 43 ίΐρττορ, 86 etpuoTOt, 364 €Ϊ>α> ( = ffep-jw), 173, 214 €Ϊρω ( = p-ja>, s ay), 213 268, 360, 542 (ίσα, 86, 465, 521 e/se ορακα, 463 ε1<τθα, 34 emSeW, 342, 349 itffKu, 191, 197, 200 ύσόμςσθα, 62 *1$ττ\ςυσοίμ•ην, 486 uspirfiffeadat, 493 ela-rriKuv, 86, 428 (ιστία-κα, -μαι, 365 (Ιστίων etc., 86 εΐϊφρηίΌί, 128 «Tt€, 329 (ίχ4την, 51 f?Xoi/, 86 efo, 316 *ία>0α, 85, 305, 363, 400 *κάΐ7, 491 %καμ€, 278 4κβωντα$, 148 ίκτγαμί-σκω, -£α>, 196, 199 f. ίκγ€γάασ•0€, 416 (Kytyapev, 424 6κγ€γάοντοί, 416, 417, 468 e/cSajS»?, 494 4 4κδαρ4ντα, 492 4κδυμ*ν, 329 εκ€κλ€το, 291, 293 4κ4κρα•γον, 290 (κΐκρατηρίχημ^, 407, 434 4κ4$σαμ(ν, 456 4κ4ρδανα, 457 ϊκςρσεν, 456 e/c7ja etc., 459 4κίαθον, 504 (bis) £κίατο, 120, 504 enXayov, 285 €κλα7π)μαί, 492 4κλά<τθη, 525 1κ\ε\άθων, 293 «κλήϊίτται, 358 4•κ\ίνθη, ~κ\ίθ-η, -κλίνη, 499 f. iK\oyiC4(rea>, 309 4κμαγηναι, 492 4κμαίνω, 216 4κνησθ•ην, 521 4κοί\αναν, 457 4κοινωνησάτην } 54 ?κομει/, 346 4κόρεσθ€ν, 489 ^cirecpcuOVTi, 385, 397 4κΊΤ€φΐ\τγοί•ην, 31, 423 4κΊτ\αγΙ\σομαι, 493 4χΊτρα£4ω, 46S 4κραγί\σονται, 493 4κρ4μω, 60 ίκρινν*, 457 ϊκρνφθίν, 489 4κσωζοίατο, 66 Ικταν, 130 4κτάνθη, 499 4κτάσα. 4Γ0 4κτ?:σΑσ., 358 4κτήσω, 449 e/CT, 259, 268 Ο 4λκι>(ΓτάΧ<ϋ, 236 Ιλκόα», 523 ?λκ&>, 143, 251, 268 eA\aj8e, 78 ^λλα-ί?/, -re, 119, 298 €λλ*1Γ€, 78 €λλυσίν, 49 €λλα>, 179 6λΐΓ€0, 304 Ζλπομαι, 144 eAcrat, 456 «?λώ (fat.), 479 Ιμα0ο;/, 501 4μάνητ€, 492 έμαράνθη, 499 4μάρναο, 60 f/t-jSa, -)8τ7, 299 f. 4μβαίημ*ν, 330 €>)8αλ6?, 141, 457 ςμβαλουμςς, 43, 475 ίμβητον, 310 4μβιβάξαντ($, 226 4μβραμ4να, 361 ίμβραται, 361 e>e'0a>, 502, 504 ί^β'λλετον, 54 4μ4μηκον, 290 (bis), 429» 4μςρΙξαν, 454 4μ4σ•ω, 481 εμετοί, 514 έμημΐκα, 367, 369 4μιγ4σκορτο, 529 4μίγην, 491 e/ii X 06V, 489 ςμμαθε, 78 %μμ*ναι, ίμεναι, ίμμζν, ϊμςν? 338 ίϊ. 4μμ^ν4οισι, 47, 476 4μμόραντι, 361, 385 ?μμορ-€, -ον, 286, 361, 399> έμνησθην, 525 ε><>λοί>, 133, 190, 193 ϊμορτεν, 163, 281 4μττάζομαι > 223, 224 8 tyiraios, 223 4μπα.(ττηρα5, 223 4μπ(πηχ€σαν } 407' 4μπ€ποζισμ4νος, 373 4μπίπ\-ηθι, 297 f. %μ•κ\ΐ)σο, 304 εμποδίζω, 250 ϊμυξαν, 456 4μφανιξάντων, 455 eVtvycwievfiai, 481 eVoipco, 255, 283 Ινορα, 282 4vape7v, 285 tvavbv, 458 ίναυσμα, 520 fr8wei£&r0», 309 ivUtuaK&ra, 373 ίνδιδύσκω, 192, 193 eV5iK«ω, 175, 182 evrai/tfeij/ (Fut.), 483 ίντασσιν, 352 έντεθνμημαι, 373 «Vti, 48 4ντρ4ιτ€υ, 304 «ντυψίω, 468 4ρνβρίξΐ)ς, 455 ^|αλβί•7ΓΤ€θί/, 514 έ£αλιφ#, 492, 500 4ζαμβ\ουμ€Ρ, 195 ^ανγ€λίω, 475 €'ίαι/δραποδί6ϋΐ/ταί, 481 έξαπάτασκορ, 527, 530 €ξα7τοτύλλ«, 537 2ξαΐΓάφ•ησ€, 271 Ι&Ύλυμμίνφ, 358 €ξ6?, 141 €|6«, 300 ^€?/tei/, 339 ϊξασθα, 34 φ*Π, 96 ^κλάιτησαΐ', 497 ^6λά06(ΓΚ€, 531 ΐ^Καυνόμίσθα, 62 φλθωντι, 46 ^£€j/api£«, 454 ^Ci^/iCi/OS, 359 ^{επισταίατο, 66 φττλη, 131 4ξ€ρύσασκ«, 532 ^c'Tp W (e'0a>, 4ρεθίζω, 502, 504 4ρείκω, 154, 503 4pe7o, 305 4ρΐίομιν, 313, 322 4ρείπω, 154 4ρ4πτομαι (esse), 166 f. 4ρ4τττω, 166, 168 4ρ4ριπτο, 369 4ρ4σθαι, 285 «=pe07a>, 78, 155, 180 4ρ€υθω, 78, 155 4ρ4φω, 144, 166 ^Ερεχβεύϊ, 503 4ρ4χθων, 503 4ρηρ4όΌ.ται, 65, 367, 369, 418, 503 4ρηρημ4νύ5, 369 4ρ•{]ρισται, 369 4ρηρότηκα, 367 έρήτυθν, 490 €ρητυσασ«6, 532 4ρητυω, 250 4ριδαίνω, 185 4ριδτ]σασθαι, 271 4ρΊζ*σκον, 530 4ρικέΐν, 285 4pi£aPT*s, 454 4ριπέιν, 285 4ριπ4ντι, 493 ΈρίχθΛίΊΟϊ, 503 έρτπτάν, 514 «ρπύ^ω, 251, 537 έρπω, 144, 251, 537 4ρρά5α-ται, -το, 65, 217, 360, 481 4ρραίσθη, 525 Ζρραπται, 360 «ρρηγίΓα, 391 •epp7J7M at } 359 %>τ?χα, 407 %η 7 «> 360, 378, 401 4ppiyr)(ri, 38, 422 4ppiyovri, 394 4ρρίζωται, 359 ZppHpsv, 287 4ppi, tV0a>, 207, 501, 503 f ., 544 eVi5eo-«€, 531 4σκεα'α<τμ4νο$, 359 εσκευάδα-τα;, -το, 359, 418 4σκ4ψομαι, 436 εσκηφα, 401 4σκηφ4ι>αι, 401, 407 έσκλήκασι, 385 ίσκον, 528 4σμapάyιζe, 243 4σττάρθην, 499 ίσπαρται, 359, 419 4σπ4μπςσκον, 529 4σπ€ρ4θοντο, 502, 504 4σπόμην, 86, 294 ίσπον, 279 f. 4σπούδακα, 380 eWai (/es), 372 ίσσα,ν, 102 icrvei-Tcu, -σθαι, 469 fWewa, 459 iWt, eft, el, 33 &r (τ6ρττ-), 83 €Τ€06|/, 490 4τεθήπ€•α, -as, 430 ε'τερμάξαι/, 454 4τ4ρρατο, 457 4τ4ταλτο, 419 ε'τετάχατο, 66, 405 €Τ€τμον, 294 f. ετεύχςτον, 52 4τητόμακα, 367 βτλτ^, 132 βτ/Λογο^, 282 Ιτοί/λά^ω, 237 ετορε, 287 4τράπην, 492 ίτραφον, 21 ο, 490 εττακαι/, 464 4τύπαν, 491 4τύπη, 492 etfa, εΰασον, 84 εί>αδο>>, 79 εΰά£ω, 237 ευάλωκα, 79, 362, 414 ευδαιμονοΐτον, 54 εύδάνω, 181 ε8δτ?σ0α, 34, 39 ευδρσι, 39 εΰ'δω, 266 564 INDEX. €ύ4θωκα, 85, 273, 363, 414, cbepyer-nKOiffair, 894 εϋιδον, 84 εύκτάζου, 23G ίΰκτίμχνο*, 129, 353 ίίκτο, 131 «νι/ά^εσβαι, 235 κίνησα, 235 €tf£ecu (conj.), 447 efaeiaros, 514 (υράκοιμ(ν, 414 «υραν, 464 «ίγ>α<Γ0αί, 464 ««5/jet»', 283, 285 €νρ4τ-ηι>, 54 «ύρβ-τόί, -t4os, 514 €υρηκ€ (imperat.), 422 evp -ήμα, 271 «ύρτισω, 271 ευρίσκω, 191, 196 €νρομ€{, 43 ςυρύνω, 254 *5Ti (?), 48 εύτυχοί/Αβ^, 335 ευχαριστωμ^, 43 ευχετάασθαι, 198 βίχίτοί, 322 βύχετάομαι, 537 €i?X€U, 304 etfxo/«u, 143, 192 (bis), 198 €w«, 155 €<ράγαμ«ί/, 461 έφαναν, 461 €φασκ€, 527 IT. Ιψατον, 53 f . 4<ρ4ασθ(ν, 70, 498 ϊ<ρ*ίω, 316 €φ6λ€(Γ0α>ΐ', 310 €, 457 €φ€<Γσ•αι, 521 «ίττπ, 38 4φί\--ηθ*ν, -αθαν, 490 έφιορκίοιμι, 335 *>λαδοι/, 223, 287 *φ\4γην, 493 4φομαρτ(7τον, 310 (tpoptvti, 141 ϊφραίιν, 288 (φρηκα, 464 *, ί ν σι, 319 «ώθεα, 430 ϊώθουν, ίωσα etc., 80 eawct, ίωσμαι, 363 iipKiiv, 81 4ώ\ττξΐν, 81 '4ωμ€Ρ, 320 4ώνημα.ί, 363 4ωνονμτ\ν, 80 βρροχό'βί, 81 tWri, 46 ΐώρα-κα, -μαι, 363, 544 εώρακαν, 385 ^ώργβι, 81, 363 έώρταζον, 81 βώρων, 81 " Ζωατμαι, 363 /εογβ, 79 /efcrirj-pora, 270, 362 /€ΐδώ$, 426 /6I1T77V, 291 /ήλω, 179 /tVJa-vri, 427 βίσα-μι etc., ν. ίσ-αμχ /otJa, 379, 544• /οί5τ;Μ'» 26, 383, 390, 392 fparpo, 360 (α€λ({(£μαν, 58 ζαμιόντω, 306 Core's, 303 ζ4βυται, 357 £*&**«?, 114, 178 Ceioira, ^είουσα»', 520 Ce'AAa», 211 ζ4*ννμι, 114 CeWev, 146, 520 Cetrnk, 146, 520 Cei^yvO/ie»', 340 Cetryu/xej/cu, 339 Cevyyuiu, 110 ίευχθ€?σα, 500 ζ4ω, 114, 146 £)&, 298 CrjKos, 520 ζητ4ω, 105 Coe'y, 302 Coi /σθω, 114 Cuyo», 238 ζφην, 335 ζώννυμι, 114, 521 ζώννυιται, 319, 447 ζωννύσκ€το, 529 ζάωττι, 46 ^α, fa etc. (eV), 119 ηβά•σκω, -ω, 190 (bis) r 194, 199, 237 . 7]βυλ\&ν, 537 ί/ΤΤβλκα, ^14 τρ/~ϊραμμ4να.ν, 374 4 νγ€μορ€ύω, 252 ήγέΌμαι, 240 •JfyepeflojTai, 502 yyiveffKov, 529 ήγωνίσαντο, 455 #δ-εα, -et, 429 f. ?^δ€ί-/Α6|/, -Τ€, 433 ^δβί»/, 433 ^Se-juev, -τ€, 432 τ/δ-ησθ', -βίσΡα, 35 ί}δοιιαχ, 156, 180 ήδομεοΰα, 62 ^δύνα», 254 ηζυσμ4νθ5, 420 ήίίδηΓ, #δη?, ήίίδτ;, ήδη, ήΐδ^εν, 80, 431 f. tfetv, f}ia, $α, 433 f . ^ € φ€ (σ/«ρ), 81, 214 ήασθα, 35 η6ρ60οι/-τοί, -το, 502, 504 TjepiteVos, 419 $0«», 261 f., 503 ^ία, τ7α («Τ^Ο» 88 f. W,85, 411,464 ήκαζον, 88 ήκηκόη, 431 ηκροάσο, 60 4κω, 394 ήλα<Γ*«(>, ήλάσ•κ«, 189, 195, 199, 237 INDEX. 565 3λδαν€, 275, 282 f. ■ήλθαμε, 464 ή'λθετον, 54 (bis) ή\ιτον, 281 7?λκ7;σ€, 85 ήλλαξάτη»', 54 ήλσατο, 456 * ■^λσον, ελσών, 285 -f?\u0 oi >, 284, 502 f ., 542 ^Αφον, 283 ιΊΚωκα, 362 ήλωκβί/οί, 424 fyxcu, 103 ■/ίμαρτον, 163, 281 ήμβλακον, 195, 283 ήμβ\ωσ€, 133, 273 ήμβροτον, 91, 279, 281, 283 ^/w, 339 V, 103, 181 ημττισχόμην, 283 ημπλάκηται, 271 ϊμπΧακον, 282 1, 412 ημύναθον, 504 ■ημφίεσμαι, 374 ^οί, 339, 341 ■ηναντίωμαι, 373 ^αρο»/, 282 ^νδαί/€, 79, 180 ^vry/c-o, -ον, 291, 462 f. ^θο,ιιβί, 43 fytfey, 285 ηνιοχεύω, 252 -ηνίπαττον, 292 f. -ηνοίγην, 492 ήνοισται, 374 ή^τ', 65 ή^το<, 319 ■ηντςβόλησε, 95 ηί/ώ-/6α, 430 ήν^βυ/, 432 -ηνώχληκε, 374 •ηνώχΧουν, 95 7^ο/, 486 ■ητΓ€ροτΓ(ύω, 252 9*0r, 291 ■ηράμαν, 58 -ηράσσαο, 450 -ηρεμ-ίω, 4ζω, 243 $ριη-κα, ~μαι, 368 γρ-ημΐσθα, 63 ήρ-ηραν, 433 ■ηρ-ηρασθα, 35, 432 Ίψ-ηραοττο, 419, 434 ■ρρήσεται, 436 7]ρίσταμ(ν, 388 ^ρκα, 414 ■ηρκέσατον, 54 -tfpvye, 155 ^ΡΧ«, 401 . ήσ0α, 35 f. ίσββν, 310 7/σί, 38 3<™e, 92 ήσκηται, 366 ^σο, 304 ^στην, 54 ήσχυμμένοε, 366, 420 ήτί, 38 ητιμώσςται, 436 ί)χα, 406 fawci', 103, 181 r?xe'a>, 373 θαάσσω, 256 0άγω, 156 0άλ€, 285 θαΚίθτισι, 39 θαΚίθων, 502, 504 0άλλ», 211, 262 θάλιτησι, 40 0άλττω, θαλττύω, 144, 241, 269 θαμβαίνω, 254 θαναταν, 535 θαι/6?ί/, 285 θότττω, 166 0apcrea>, 229, 240 0αυμά£ω, 238 θαυμαίνω, 233, 253 0βφ, 315 θύημςν, 330 θβί-ϊϊ^, θίΓβν, 329 GeT/ic»', 329 0€?»/<«, 341 fcfrw, 216 (bis) θίίο-μαι, μεν, 313, 314 0βίω, 210 θίλ Ύ (σκ€, 529 0e\otj/, 332 0€λω, 144 θέλωντι, 46 0€>ej/ai, 339 0ei>e?>/, 285 0€j/Ta>i/, 307 0€'ο, 304 0eWo, 334 0epijj/at, 491, 496 θ4ρμ(το, 232 θβρμω, 144 θίρομαι, θερείομαι, 241, 269 θερσόμενος, 456 0epa>, 144 0€'y, 298 06σμ<ί?, 598 0€σ•7Π€€ί!/, 481 θίσσεσθαι, 220 0€T(is, 517 0eWt, 319 0^ot, 320 Orryavw, 175, 181 βήγω, 156, 181 0W, 0ή?7, 315 f. 07JK7J, 412, 517 0ηλεω, 262 θημένη, 353 0τ?ρά«, 233 θηρεύω, 252 07?σ6υ/*6σ0', 469 07ρ-€ΐ5ω, 252 θιγγάνω, 174, 180 0/γ€?»', 285 0170»/, 303, 450 0iV', 120, 247, 50S 0λάσσω, 220 0λάα>, 147, 220 θλιβηναι, 492 0λί0α>, 157 θναίσκω, 190, 194 θνάσκω, 194 θνγσκω, 190 f., 191 θνήσκω, 191 θολύνω, 254 θορΰν, 285 θόρνυμαι, 110, 194 0οΟ, 305 θράσκειν, 193 f ., 199 θράσσω, 218, 222 θρανσθίντα, 524 θρέξασκον, 532 θρησκεύω, 193 0ρήσκ«, 193 θρηχόω, 244 θρύπτω, 166 θρύσκω, 110, 191, 194 θρφσκω, 191 0υ€ί/, 342 θύςσκ*, 527 0ι>ία>, 203, 209 θυίωσιν, 147 0^evoy, 129 θύμμενοε, 370 0iW, 0iWa>, 171, 177, 178, 184, 216 0ύ«, 147 βαητεύού, 165 0ώπτω, 165 θωρήσσω, 256 0«aoj5 J u€0 > , 469 0α>τά£€ΐϊ/, 227 2 θωχθείς, 156 'icuVeTO, 89 «ίλλω, 136, 212, 365 iWOof, 365 ίάπτω, 167 ίάσκω, 191, 197, 199 ίαυω, 197, 520, 543 ίαχ€α>, 268 "Γαχο»/, 80 ίαχυΓα, 373, 377 ιαχω, 143, 268 ίδίΓί/, (δε'σ0α<, 285 Ϊδ6(τκ€, 191, 531 %y, ίο??, 316 f. ίδησώ, 270 ίδί», 202 f ., 208 ϊδμαι, 372, 419 566 INDEX. ίδμβν, -at, 340, 344, 387, 424 ίδρύνθησαν, 499 Ιδρίω, 111, 223, 254 ίδωμι, 27 fci, 301 iei 7 , 329 livai, 341 ίεράομαι, 237, 245 lepevro, 373 Upeva, 252 ίερίτευχε, 416 ίεσσα, 121 iCaw, 173, 175, 182 #ev, 304 ιζησομαι, 271 ί'^ω, -OM«i, 86, 111, 182, 202, 222, 242, 266, 271 lr\yopeiv, 368, 394 ?V/, 106, 517 ϊ-ησθα, 34, 39 ϊγσι, 38 f. ί£<π, 39 ιθαίνςσθαι, 153 Ϊ0ι (bis), 297 ί0ύα>, 250 "Ζκ -ave, -€TO, 89 χκαι /Ti, 154, 385, 394 ίκάνω, 173, 175, 177 (bis), 181 ικίσθαι, 285 'ίκεσθον, 54 (bis) 'ίκε'τει/σα, 89 ϊκμενος, 131 Ικνίομαι, 177, 181, 184 ίκόμζσθα, 62 «crep-mj/, -ay, 536 Ικτερώσσω, 257 Γκτο, 131 7κα>, 151, 158, 181 ίλα0*, 119, 298 'ίλαμαι, 195 ίλάσκομαι, 195 f., 411 ίλασόμεσθα (conj.), 447 1\4ομαι, 195 7λ,,δ«, 119, 298, 411 ίλήκ-ρσι, 411 Ιληγκρης, 336 ϊλλα>, 211 ίλλαττ-ε'ω, -ί£«, -tw, 165 ιλνμενον, 122 7μαό>, 458 Ιμάσσω, 256 ιμζίρω, 255 i>xe*/, 339 "i>erot, 121, 340 Ιμιτραδν, 458 ιμιτάταδι/, 458 ιμφαίν(ν, 342 *Ya7(Wa>, 306 ινδά\\ομαι, 256 ιρδικάζητοι, 61 Ινκαταττάταόν, 458 tvpvtr, 111 Ιννςσθαι, 111 fcw, 461 ϊο*, 334 tohj»/, 333 V € »/,311,312*,313f.,316, 317 ίππευα», 252 ίπίΓοτετρόφηκα, 374 "ττταμαι, 108 Ιντίσθαι, 292 ίπτομαι, 165 i, 191, 197 <σό«, 244 '/στα, '/στη, 300 Ίσταντο, 65 ίστάνω, 170, 175, 177, 179 ί'στασο, 304 ί'στασχ', 529 ϊστ^ι, 107, 518 '/στ^, 383 ί'στω (imperat), 305 Ιστώμεσθα, 62 ϊσχανάασκον, 530 , -ανάα>, 175 f., 177, 182, 183 1, 543 ίσχι/άνασ', 457 ισχυρίεΤται, 481 ?σχ«, 283, 543 (bis) Ιτίον, 517 ίτη?, 517 ιτητέον, 236 Γτθαι, 71, 350 ίτω-ν, -σαν, 307 £. π5£α>, 237, 454 ίχθυάασκον, 190 *χ0υάα>, 237 ϊψ-εται, -αο, 165, 44D W0i, 47 κάβασι, 298 Kayicalvu, 177, 185, 217 κά•γκανον, 107 καΎκομ4νηε, 177, 185 καγχάζω, 227 κάδω, 157 κατ}»/αί, 494 καθαίρω, 253, 255 κα0αυαίι/ω, 155 κά0ε, 299 καθ(5ονμαι, 223, 480 κα0ε?σε, 80 καθεξής 469 κα0εσάαε»Ό5, 465 καθ^στακυίαν, 425 κα0εστήξεί, 436 «άθευ, κάβου,. 304 £* κα0εύδα>, 266 κάθημαι, 334 καθίγνυσθαι, 110 Ka0i£a>, 266 κα0-ίξε?, -ίξα$, 456- καθίξτ;, 241 κάθιξοί/, 456 9 καθίστα, 301 καθίσταται, 318 καθίώ, 482 καθορψ, 335 καιίμςν, 339 καίνυμαι, 113 καίι/ω, 216 (bis) καίω, 209, 524 κακκά^είϊ/, 227 κακοτ€χνησίω, 468• κακόω, 244 κακΰνω, 254 κακχά£α>, 227 κα\4(σκ€, 530 καλε'σκετο, 530 καλε'χεί, 143 καλεω, 240, (fut, 480) καλή£ω, 248, 524 κάλημι, 26 καλήω, 248 καΚλννω, 255 καλύβ-η, -ιον, 167 καλύπτω, 167 καμέιν, 285 καμέιται, 475 κάμνω, 171, 179 κάμπτω, 165, 167 καμπΰΚΚω, 255 κανάχ-ησε, -ί£ε, 243> κα^Γί/, 282, 285 κάπετοϊ, 514 καπνίίων, 241 καπνω, κάτττω, 165 καρ??, 493 καρκαίραν, 215 καρπευσήται, 469» καρτερηαεί', 339 καρυξώ, 469 κασάί/είϊ, 122 κασκαλίζω, 357 * καστορνΰσα, 112 κατάβά, 299 καταβήμεναι, 339> καταβηομ^ν, 313 κατα7ίε7ν, 481 καταΎρίντων, 307 καταδίδαρ&ηκό'ί, 271 καταδο^λίττασθαι, 455 καταδυμεναι, 339 καταείνυσα»/, 113 καταζηνασκί, 531 INDEX. 567 καταθεΓτε, 330 κατάθοιτε, 334 κατακείαθεν, 504 κατακείαταχ, 65 κατακλιεΐ, 482 (bis) κατακλινηναι, 492 κατακλινήσομαι, 493 κατακοσμήσ-ησθε, 447 κατακτείνχισι, 39 καταλελάβηκε, 36 L καταλίπεσκε, 531 κατανενοχε, 399 καταπεφρόνηκα, 373 καταπιεί, 467 κατατΐτακών, 287 κατατττήτην, 127, 282, 412 καταρρώδηκαε, 372 κατασαττήσεται, 493 κατασβεσει, 521 κατασκεναν, 479 κατασκενάττη, 455 κατασκιωσι, 479 κατάσσω, 218 κάτασχε, 299 κατατεθτ)πειν, 394 κατεβλαφότεε, 358 κατεΎΚωττισμενον, 358 κατεγνωσθεν, 490 κατεδίτ^τα, 95 κατεδίκάσθε^, 490 κατείλοχε, 361 κατεκείαθε, 502 κατεκΧγσθ-ην, 525 κατεκλάζατο, 525 κατεκτάθεν, 499, 518 κατεΧηλνθοντοε, 394 κατενασθεν, 489 κατειτλ'η'γη, 491 κατερράχθτην, 219 κατεστράφθησαν, 499 κατεσωξαμεε, 454, 523 κατεσωσαμεε, 43 κατετάμομεε, 43 κατγκισται, 366 κατ-ήνοκα, 369 κατίσχεαι, 322 κατο<κτίεΐ, 481 κατορυχησόμεσθα, 492 κατυψρονηναί, 339, 341 καυάξαίϊ, 79 καυστόϊ, 524 καυχάσαι, 60 καχάΧω, 227j καχαξώ, 469 καχλάΧα>, 227 κάα>, 209 κεα-ται, -το, 66 κεδαταί, 116 κείατο, 65 κεΓ,ααί, 100 κείρω, 177, 214 κεΓσο, 304 κε?ται (conj.), 320 κείω, κε'α>, 467 κεκαδ-ε?»/, -ών, -ήσομαι, 157, 223, 270, 293, 435 κεκαομενοε, 131, 420 κεκαλυμμένος, 420 Κ€Κά>ω, 293 ■ κεκασμαι, 113, 131 κεκασμενοε, 420 κέκασσαι, 419 f. κεκαφηώε, 165, 270, 378 κεκελευστο, 524 κεκευθα, 397 κεκευται, 420 6 , 525 κεκηδα, 378, 396 κεκ-ηφε, 396 κεκίνδι^εύσεταί, 436 κεκΚα -yya, 377, 401 κεκλεβώε, 400, 408 κεκλεται, 276 κεκλήατο, 65 κεκληγα, 377 κεκλτ) 7 οντεε, 290, 393 κεκλη Ύ ώε, 290, 394, 396, 401, 426 κεκλήο, 423 κεκλήστ), 436 κεκλίαται, 65 κεκλοφα, 399, 407 κεκλυ-θι, -τε, 129, 290 κε'κλι/κε, 422 κεκμηκα, 413 κεκμηώε, 426 κεκοκεν, 414 κεκομικώε, 413 κεκόμιστη, 60 κεκονα, 400 /ce/coirc$s, 379 κεκορυθμενοε, 420 κεκοτηό-π, 378 κεκοφα, 407 κεκρά-γα, 290, 375, 377, 396 κεκρά^ετε, 387, 422 κίκρανται, 421 κεκραχθι, 290, 376, 387 κεκριγότεε, 377, 401 κεκρίκα, 385, 414 κεκρίκαισιν, 385 κεκτ-ημαι, 358, 375 κεκτγτο, 423 κεκύθωσι, 293 κεκυφα, 401 κελ-άδ«, -αδεω, 268 κελει/θείοντεί, 241 κελεύομεν, 322 (bis) κεΚεντιόωντ, 535 κελεύω, 251 κεΚεΰωνσι, 47 κελτ]τί£α>, 250 κε'λλω, 211 κεΚομαι, 131, 144, 266 KeWai, 263, 456 κενιέω, 263 KeVrrj, 300 κεντο, 131 κερ-άω, -αίω, Κεραμαι, 115, 120, 209, 235 κεράννυμι, 115 κερρω, 214 κερχνει, 179 κερχω, κερχνόω, 179 κερωνται, 318 κε'σκετ', 529 κεΰθω, κευθάνω, 151, 155, 175 (bis), 181 κε'χά>α, 396 κεχάναιτι, 385 κεχανδα, 181, 401 κεχαρήσω, 436 κεχαρηώε, 271, 378 κεχάροντο, 294 κεχείμα»τα<, 421 κεχήνετε, 422 κεχ^^τα, 378, 396 κεχλάδ-εί»', -orras, 394, 401 κεχληδενα*, 377, 401 κεχλοίδε, 196, 397 κεχοδα, 399 κεχολώατο, 65 κεχολώσεταί, 436 κεχολαχτο, 434 κεχρημενοε, 379 κεχυνται, 65 κεχωρίδαται, 242, 418 κεα>, 467 κήαιεν, 451 κηδαύ/εί, 185 κήδ-ω, -o^at, 157, 266 κηκία-, 249 κ-ήττιχάριτται, 455 κηρύσσω, 256 κίαβθϊ>, 113 κίασθαί, 100, 120 κίαταί, 64, 100 κί-γκραμι, 107 Ki7XaVa>, 175, 179 κιθαρίζω, 250 κίκλήσκ», 190 f„ 200, 377 Κίκλήσκω, 191 κΐί/αθίΧ», 503 κΐί'εω, 113, 177, 184- KiWyuat, 113, 177, 184 κίνυντο, 65 κιννρομαι, κινυρίζω, 255 κ/ξατο, 121 κιρνάω, κίρνημι, 115, 117, 176, 183 κίχά,/ω, 121, 175, 179 κίχείτ?, 121, 329 κίχεΤν, 121, 285 Κίχεί-ω, -ομεν, 313,316 κιχημεναι, 121, 339 κίχ'/Α", 121, 179 κίχλί^ω, 227 κίχρημι, 107 κίω, 113 568 INDEX. κιών, κιάν, 276, 285 Khayy -άνω, ~4ω, 180 κλάζω, 180, 223 6 , 224, 454 κλαίςσκ€, 530 κλαίβισθα, 35 κλαίω, 209, 266, 524 κλοξώ, 525 κΧαπίντςς, 492 κΧαυσ^ίοντςτ, 533 f. κλανσ-ίάω, 244 κ\αυσούμ(θα, 470 «λουστώ, 524 κλ<£α>, 147, 223 6 κλ6ί«, 249, 524 k\4xt€, 210, 501 κνίζω, 223 Όασσ-αι/, -ου»/, 246 κνώ<τσω, 221 κόβω, 240, 261 «οίκυλλω, 212 Koivaffavrcs, 246 κόλα, 479 καλάπ-τ», 166 κολοίω, 253 κομ4αν, 243 κομ((των, 310 κομιζόμ^νοε, 243 κο/ζίά>, 481 κονάβ~ησ€, -ιζ(, 243 *coj/-ia>, -ί£ω, 207, 249 κο7Γ€?<Γαι>, 492 kJttt», 165 K0/)€€i (fut.), 480 κορ4ννυμι, 115, 196 κορίσκω, 115, 196, 199 κορίσκονται, 190 κορυπτ-ί\θ5, 'όλης, 167 κορύπτω, 167 κορύσσω, 202, 256 κορυφουσθαι, 245 κοσκνλμάτια, 357 ■ κο<τμηθ(ΐμ€ν, 329 κοτίομαι, 522 /couriers, 481 Kpayuv, 285 κρα^ω, 224 κραίνω, 216 κράτ€σκ€, 530 κρατύνω, 255 κραυγάζω, 224 κρανγανάομαι, 183, 224 κρ4κην, 342 κρ4κω, 143 κ Ρ 4μαμαι, 115, 120, 123 κρςμάννυμι, 115, 120 κρ€/Λω (fut.), 480 κρημνάω, 183 κρήμνη, 300 κρί)μνημι, 115, 117 κρώΜμεν, 223, 340 κρί£ω, 223 f . Kpfoe, 224, 285 κριν4σθω, 309 Kpivevvn, 475 Kptve4vres, 499 κ/>ίΐΌ>, 170, 172, 185, 216, 518 κρίνωνσι, 47 κρίνωντι, 46 κροαίνω, 178, 186 κρούω, 178, 186 κρυβόν, 286 κρνβηναι, 494 κρύ/3α>, 166 κρύπτασκε (-βσκε), 530 f. κρύπτοισι, 47 κρύπτω, 166 κριφβί*, 122, 492, 500 κρνψθ4ντα, 500 κρώ£ω, 224 κτάμεναι, 130, 216 κτανάν, 282, 286 κταν4οντα, 475 4 KreivcffKf, 530 κτείννυμι, 114 mfe», 114, 216 (bis) κτςίνωμι, 27 /fTei/e?, 475 κτ4νναι, 457 /cTeW», 114, 216 κτ^ρΰζω, 454 κτίριουσι, 481 KTepiffaiev, 451 κτ4ωμςν, 319 κτησοιντο, 486 κτι£οίατ', 66 κτ/£ω, 129, 223, 353, 524 κτίμςνο*, 129, 223, 524 κτίννυμι, 114 ktuW», 261, 263 κυα»/ω, 186 κυΖ-άνω, -αίρω, ~ιόω, 181 f., 254 κϋ€ω, 268 κ«50€, 286 κυ.'σκα), 196, 200, 268 /ίϋκαι/(£«, 183 ««καώ, 183 kwccok, 183 κυχλ-€α>, -Jo?, 246 κυλ-ί», -ίνδω, -i»/5t'«, 207, 249 κυμαίνω, 253 κυν4ω, 184 κόπτω, 166 κνρ•4ω, κύρ-ω, •ομαι, 214, 260, 268 κνρκανάω, 176, 183 κυρονντων, 306 κύρσα$, 456 κυρτόω, 244 κυ'ρω, 214, 268 κι5α>, 196, 268 κα>/ίύα>, 376, 537 κωτίΚΑω, 255 λαβ -w, -4σθαι, 286 Κάβεσκον, 529, 531 \αβ4τωσαν, 307 λά£ο«/, 31 λά^οι/, 303, 450 λαγάο-σοΗ, 157 Kay χάνω, 180 λο^οίατο, 65 λάΧο/χα*, 122, 180, 226 λάγυσβαι, 122 f., 180, 226 λά£ω, 223 7 λαβ-εΓ»/, -co-flat, 280 λοθ«, 151, 157, 180 λακ*»/, 286 ΧακΊ]σομαι, 271 λαλά£ω, 227 λαμβάνω, 174 (bis), 180 λαρΛΈταν, 236 λαμπρύνω, 254 λά>™, 144, 236 λάμψομαι, 475 λαι/βάνω, 157, 180 λόξασθοί, 460 λάπτω, 167 λόσκω, 191, 271 λοκρυο-σ-ω, -r?w, 52, 167 λαχβΓ»/, 286 λαχ^ι,ν, 336 f. λαψρ, 469 λάω, 147 λβ'γί?, λ67«, 139 f. λ^γόμ&ν, 63 λ67ό'»τωΐ', 306 λβγω, 145 AetaW, 253 \ύβω, 154 λβΛω, 238 λ*ί™, 151 f., 154, 180 λ€ϊστ($ϊ, 515 \tiTovpy4v, 342 λ€ίχω, 154 λβ'κτο (counted), 131 λ€κτο (laid himself), 131 \(\αβ4σθαι, 293 λ€λά/37ϊκα, 270, 261 λίλοθο, 396 INDEX. 569 λ4λαθον, 293 λ€\άκοντο, 290, 293, 295 λ€λοκυΓσ, 402 λ4λαμμαι, 361 \4\αμτΓ€, 377, 401 Χ4\ασται, 419 ΧςΧάχασι, 361 χ4Χαχον, 293, 295 λίλεγο, 361, 400 λ€\ζίμμ€θον, 67 Χ(Χ€ΐμμ4νο*, 420 λβ'λβίττται, 419 λβλειχ-, \(\ιχ~μότ€ς, 378, 408 λελβίψΐτο», 436 λε'λβκτοί, 361 λ6λ€χο, 407 ΧίΧηκωε, 360, 377, 402 λζλ-ημμένη, 361 ΧςΧίημαι, 378 λϊλιημένος, 209 λ€\ιμμ4νος, 166, 378 x4xoya, 361 λβλογχα, 400 λ€λ($7χασ•ί, 360 f., 385 λ€λ(57Χϊ?, 391 XeXonra, 360, 397 Χ4λυνται, 65 λ€λυπτ)μ€σθ\ 62 ΧςΧνσομαι, 436 λβλΟτο, 423 λ€{€0, 461 λ€7ΓτΟίΌ>, 254 λ€πω, 144 λβσχτ?, 191 λβυκαίνω, 253 λβυσθηι/οί, 524 Χ^νσσω, 202 f., 218 \τγγ4μ^ναι, 339 Xrfya>, 157 ληθάν», 175, 180, 182 λή0β(ΓΚ6, 529 λή0ω, 157, 180 Χηκ4ω, 271 λτρττό'ϊ, 514 At^ei, 154 At'7|6, 224 λί£«, 224, 226, 454 λίθά^ω, 237 λίθ-ιαμ, -Sf, 536 ΧιΧαίομαι, 209 Χιμττάνω, 154, 173, 180 Χιμώσσω, 257 AiV, 153 λιπ-ββίρ, -€?ν, -4σθαι, 286, 343 Xfcrcr, 491 λίτττω, 166 λίσσ€ο, 304 Χισσ4σκ(το, 530 Χίσσομαι, 202, 220 Χιταίνω, 186, 220, 253 λίταν-ίυα), -6ίο,186, 220, 253 λιτ4σθαι, 286 Χίτομαι, 143, 150 Kupeppovvrcs, 166 λιχ /ta-f», -ω, 154 λθ€0) ν. λοιίω, λόιβάομαι, 231 λοξόΌ>, 238 λοπαν, 535 λο5, 303 λοΰμαί, 104 λούω, λ(5ω, λοε'ω, λούε», 240, 261, 269 Χ^άνομαι, 180, 224 λυ£ο, 224 λΟθι, 129 Χνμαίνητοι, 61 AutrcaiVo», 253 λύσσεταί, 221 λντο, 129 λυτο^, 518 λόττω, 224 λύω, 148 λωβάομαι, 245 Χωβασέίσβς, 469 λφτ;, 335 Χωοντο, 104 μαδ-άω, -ί£α>, 243 μα06?ί', 286 μαθ4τωσαν, 307 μαθενμαι, 480 μαθ-ησομαι, 270 μαιμάω, 104, 376 μαίνομαι, 203, 216 μαίομαι, 209 μακαρίζω, 250 μακώ>, 286 μαλάσσω, 161, 256 μανθάνω, 180 μανπνομαι, 252 μαπ&ιν, 286,291, 343 μαρ'.ΐΐν, 249 μαρμαίρω, 215, 377 μαρμαρύσσώ, 257, 454 μάρναμαι, 118 μάρναντο, 65 μάρναο, 304 μαρνώμςσθ 1 , 318 μάρπτω, 165 μαρτυρ4ω, μαρτνρομαι, 255, 260, 269 μάσσω, 219 μαστηοί-ην, 336 μαστίζω, 250, 454 μαστίω, 249 ματήσΐτον (conj.), 447 μαχαίταε, 191, 235, 241 μαχαν, 535 μαχείομαι, 240 μαχίοίατ, 65 μαχ4οιντο, 332 μαχ4ομται, 480 μάχ«σβοι/, 310 μαχ4σκω, 196, 4 μαχητό s, 514 μάχομαι, 143, 264, 269, 406 μαχόμςσθα, 62 μνγαίρω, 118, 255 /χ€7αλύ^ω, 254 μ&4ων, μ4δων, 261 μ45-ω, -ομαι, μίθ4ω, 143, 157, 269 juefletVflc», 423 /Α€β«ί«, 316 μ€θ4\*σκ€, 531 μΛ04μ*ν, 339 /*«β^ϊ7, 315 μ^θτηρμόσμεσθα, 62 μΐθι4μ^ν, 339 μΐθίχισι, 38 μ€0υίω, 190 μίθύσθην, 343 μςθύσκ-ω, -ομαι, 190, 195, 200 μβθέα, 250 μ^θώμΐν, 320 μεί-γνυμι, 111 μζιοιάασκ*, 530 μειΧίσσω, 256 μύρομαι, 214 μςΧαίνω, 253 μιλάει, 182 /Χ6λ€δαίΐ/&7, 253 /*€λ€ί, 211 μελ€τάα>, 236 μ, 76, 211, 266 μ4Χπω, 144 μ*λα>, 144, 266 μ4μαα, 104 μίμάασι, 360 μεμαθήκάσι, 385 μ(μαθηκ4τω, 422 J μςμακυΐα, 402 μςμάμςν, 387 j μ(μάποιβν, 291, 293 | μεμαρπώς, 401 ; μβμαχότβ*, 407 /ue/xocis, 378, 402, 426 μ4μβ\ς.ται, -το, 266, 378, 417 μ4μβ\ωκα, 380, 413 μεμβΧώντων, 387 μεμενάκουσα, 394 μςμ€τιμ4}Ό$, 374 μεμηκώς, 377, 396 /Α€/«ϊλ6, 212, 360, 378, 397 μϊμ-ηΚγ, 422 /i6>7jva, 378, 396 μεμίαμμαι, 421 μεμίασμαι, 420 μεμίξεται, 436 μεμισθώσονται, 427, 439 μ4μνζθ, μ4μνγ, 41 7 μςμνζώμζθα, 423 570 INDEX. μέμνημαι, 378 μεμίήμην, 423 μίμνηντο, 65 μ4μνη-σαι, -at, 417 μίμνησο, 423 μςμνησομαι, 436 μςμνρτο, 423 μςμνώο etc., 423 pl/uora, 360, 378, 399 μςμόρηκε, 271 μςμόρηται, 361 μςμορμίνον, 361 μ*μορυχμ4να, 420 μςμπτός, 515 μεμυφτε 408 μεμνκα, 360, 398, 413 μέμφομαι, 144, 145 μενετόί, 514 μζνίτωσαν, 307 μενείντ:, 46 μενοίνή^σι, 249 μενα>, 212, 258, 264 (bis), 266, 543 μβρθεΊο-α, 498 μέρμςρα, 365 μςρμηρίζω, 242 6 , 454 μπα?οικ4οι, 335 μεταχειριεΤταί, 481 μ^τ^γγραφ-ησ^ται, 493 μετείω, 316 μ€Τ€κΙαθ-€, -ον, 120, 502 μετεχόντων, 306 μετριάζω, 238 μήδομαι, 157 μηκ-άομαι, -άζω, 273 μηκύνω, 255 μηνίω, 207, 249 μη-τίομαι, 249 μιάν-θην, -θησαν, 489, 499 μίγήρϊ, 315 μιγηναί, 500 μιγίΐο~€σθαι, 5, 490 μίγνυμι, 111-, 198 μιηντ}, 457 μικτό, 131 μίμεο, 305 μιμναίσκω, 190, 235 μιμνησκω, 189, 190, 193, 194, 199 μιμνγσκω, 190, 543 μίμνω, 543 μιννθίσκον, 530 μινύβω, μινυθ4ω, 269, 503, 504 μινύρομαι, μινυρίζω, 255 μίσΎ^αι, 322 μι<ΤΎ4σκ€το, 530 μίσγω, 111, 189, 192, 198, 542 μιχθήμεναι, 500 μναίσκω, L98 μνάομαί, 11)3, 543 μπάσκετ*, 530 μνησαίαθ', 65 μνησάσκξτο, 532 μνησώμεθα, 447 μοιμύλλω, 202 μόλε?»/, 286 μο\-(ύειν, -οί•€ΐν, 253 μο\ΰνω, 183 μονόω, 244 μορθηναι, 498 μορμολύοΌΌμαι, 257 μορμύρεσκς, 191 μορμύρω, 215, 377 μόρναμαι, 118 μοχθ-ησΐΐν, -ίζοντα, 243 μοχθίζοντι, 46 μύ£ω, μυζ-άω, -4ω (suck), 226, 269, 273 μύ£ω (groan), 204, 226, 454 μυθ4σκοντο, 530 μυθήσομαι (conj.), 447 μνκάομαι, 204, 273 μνκ -e, -ον, 286 μύλλω, 204, 211 μύρομαι, 214 μΰσσω, 219 μύω, 147, 202 μωται etc., ΙΟΙ ναετί}ρε$, 210 ναΐξτάασκς, 190, 210 ναιετάω, 236 ναίω, 210, 220, 521 va/cros, 221 νόσσω, 221 ναυμαχησείοντατ, 534 ναυσιάω, 244 ναυτίλλεται (conj.), 322 ναυτίΧΚομαι, 255 νεά^ω, 231 νεά&>, νεόω, 231, 236, 238, 246 νίΐκααντΐρ, 458 νζικάαρ, 458 νείκείεσκε, 530 ί/εικείω, 240, 522 νείο, 305 νείψει 154, 158 νεμεθοντο, 502, 504 νεμεσίξομαί, 250 νεμζσσάω, 237 νεμεσσηθείομε»', 313 νεμεσσήβτ/τε, 510 νεμόμςσθα, 62 f . νέμω, 144, 266 i^ycKr/ueVai, 221 νίνασται, 221 νενςμήσθω, 423 νεί/«ύκασο>, ;585 ν(νησμ4νοι, 521 ΐ'ενοήκων, 394 νενόμίκα, 380 viVo ε, 399 νίομαι, 146 νεοω, 231 νενα-ουμενο*, 470 νευ(Γτά£ω, 236 νευω, 155 νεω, 148, 156 ν€ωο~σ€ΐ, 257 ντ?γάτεο5, 512 νήβω, 501, 504 ντ?νεα>, 376, 537 i/ijo-oiWt, 469 νηφαίνω, 186 νήψω, 144, 186 νήχω, 199 νί£ω, 167, 224 νίκαν, 339 νικάσκομ^ν, 527 νικά»;, 237 vt/c<£, 335 νικψ-η, -τε, 336 νίτττω, 162, 167, 168, 224 yiW(yiat, 146, 210, 220 νίψει, 158 νιφέμπ, 339 νιφ-ησομαι, 167 νοε'»/, 342 νομςνω, 252 νομιςυμςν, 481 νοσόΐ, 335 νοσαίΐμ αν, 335 νοσφίίϊς, 481 νοσφ'ιζω, 250 νοσφίξαιντο, 455 νιτχεί, 143, 150 νύσσω, 219 νυστάζω, 155, 236 £αίν«, 216 ξείνί^ω, 454 ξεω, 147, 216 ξηραίνω, 253 ίυγγίΌΤμεν, 330 ξυμ-βαίημςν, -βάΐμςν, 330 |υμ£λί}ταί, 320 ξυμβλ-ητην, ξνμβΚηντο, 65, 132 ξυνέβητον, 54 ξι/νείλεκται, 361 ξυνελεγημεν, 492, 497 |υΐ'ε£ερευθείτϊ>', 494 £ύνιε, 301 ξυντηχβεί?, 500 ξνρ-ω, -ομαι, ζυρ•4ω, -άω, 214, 246, 260, 269 ξύω, 147, 216 οΎκάομαι, 231 όδά£-ασ0α», -εσθα*, 445 οδά£ω, 144 οδοδυσταί, 365 οδοιπεπορήκαμεν, 374 οδύρομαι, 215 INDEX. 571 όδνσθηναι, 498 δδύσσομαι, 222\ 521 ίδωδα, 367, 369, 377 δδωδνσται, 369, 521 Ζζω, 203, 223, 264 υθομαι, 143 οί~/νυμι, 111 οΊδα, 153, 264, 371,397, 544 οίδ-άνω, -αινώ, -εω, -άω, 181, 270 οίδη$, -θα, 389 ι , 390 δϊζνω, 250 οϊζω, 237 ο7κα, 371 οϊκεντι, 47 οικήμαι, 372 οΐκήσοισι, 47 oi/cioDpTes, 481 οικοδομηται etc., 372 οίκοδομήται, 423 οίκτείρω, 255 οϊμώζω, 237, 454 οϊνοβαρείω, 240 oiVoiroToC 6 "''» 236 οίομαι, 267 οίόω, 244 dee, 450, 461 οίσευμες, 469 ο?(Γ0α, 35 1, 42, 383, 395 ο7(Γ06, 384 οϊσ-ω, -ομαι, 544 οϊχ^εσκε, 530 οίχνεω, 184 οϊχομαι, 184, 267, 417 otco, 207 δκελλω, 211 o/ij/eta>, 240 ολεκεσκεν, 530 ολβ'κο, 115, 143, 164, 411 δλεσθαι, 286 δλεσσω, 272 ολεσω, 480 9 όλιγοδραι/εωϊ', 216 ολισθάρω, 175 υλλυβ, 301 ολλυμι, 115, 1721,272 δλο'ιατο, GG ολολύζω, 227, 365, 454 δλόμενος, 172 δλοφύρομαι, 255 ολωλο, 367, 369 όλώλτ?, 422 όμαλίζω, 243 ίμαλόω, 238, 243 ομαρτον, 281, 286 δμ-η-γυρίζομαι, 250 δμιχεω, 261, 263 υμιώμεθα, 475 and note ομννε, δμνϋ, 301 δ'/^υβί, 292 ομνυμι, 111, 274 υμνυν, 339 ομνΰντων, 306 δμοιόω, 244 δμοκλ-ησασκε, 532 ομόρΎνυμι, 111 ομουμαι, 475 and note ομώμοκα, 367, 369 ονειρώσσω, 257 δνήμενος, 426 δνησο, 304 δνθεμεναι, 27, 339 ovfwj/ii, 108, 292 οί/ομ-ά^, -atVa>,238, 239, 253 υνομαι, 236 δνυμαίνω ν. ονομάζω δνόμασθεν, 489 ονομαστός, 515 ovorcifc, 121, 236, 537 o£iW, 254 07ri£a<, 250 ότπττεόω, 108, 292 όττλί^ω, 454 όπλισόμεσβα, 62 όπυίω, 251 οπ«η-α, 367, 369, 377, 544 οιτώπεοαν, 428 07Γ£ϋ7Γη, 391 οπωριεΟ^Τ65, 481 δρ•γα'ινω, 253 οργά^α*, 457 ορ-γάω, 240 δρεγνυμι, 111, 176 ορβγω, 78, 111, 143, 145 δρεΊται, 475 ορεοντο, 111, 261, 263 δρεχθεον, 503, 504 δρεχθεω, 111 δρτ?, 309 3 W , 26 ορθοω, 238 δριγνάομαι, 111, 176, 183 δρίννω, 172 fyiVo, 111, 172, 177, 183 δρκιξάτω, 455 ορκίζω, 468, 483 όρμαίνω, 253 ορμαον, 458 ορμάω, 237 δρμεατο, 372 δρμώμεθον, 67 δρνυμαι, 261 δρννμεν, 339 ορνυμι, 111, 172, 177 οροθύνω, 111, 503, 504 δρομαι, 144, 544 υρορεΊν, 291 δρούω, 111 δρράτω, 457 δρσασκε, 532 οοσεο, 462 ύρστιοΊ, 39 ορσο, 131 υρσομεν, 456 ορσωμεν, 447 οριτ^ηνοι, 494 δρύσσω, 222 όρυχ», 143, 150 δρχεομαι, 229, 240 δρωρα, 367, 369 δρώρεται, 417 δρωρεχα-ται, -το, 367, 369,. 378, 405, 418 δρωρεχότες, 400, 407 ορα>0ί7, 422 δρώρη-ται, -το, 417, 423 δρώρυχα, 367 δρφτον, 54 οσσΌμαι, 219 δσφραίνομαι, 186, 283 δσφρεσθαι, 286 δσφρήσομαι, 271 o^Ae, 120 ονλόμενος, 172, 179, 353 οϋνεσθε, 121 ούρ-ητιάστιε, 536 ο5τα, 134 ουτάμεν, 339 ούτασκε, 531 οΰτασται, 372 ούτ-ά-α,', -£α>, 134, 236 ούτήσασκε, 532 οφειλέτη, 61 δφειλόντων, 306 οφείλω (o), 181, 182, δφελλειεν, 457 δφελλω, 256 δφλάνω, οφλω, δφλ'ιο~κω Γ δφλισκάνω, 175, 181, 182, 183, 196, 276 οφλ4\σω, 271 δφρυωμενο$, 372 οχβ'ω, 240 όχλ -euifat, -Ίσσειαν, 243 οψ6ΐθϊ/τ€ϊ, 533, 534 υψοιι/το, δψοίαθ\ 66, 486 ϋψο/^βί, 219, 544 οψο»/, 359 ττά77?, 491 τταγί\σεται, 493 παθαίνω, 254 τταθ-εειν, -ε7ν, 286 πάθησθα, 34, 39 πάθ-ρσι, 39 πάθωντι, 46 παιδεύω, 252 τταί£α>, 250 ποι|ου^τοί, 456, 470 3 παιπάλλω, 213, 536 παιφάσσω, 221, 377 ιτα/ω, 210 πάλαιμι, 26 παλ-αίω (τ?α>), 191, 234 ιπίλλα>, 211, 213 7τάλτο, 131 παματοφαγεΊσται, 69 572 INDEX. παμφαίνω, 45, 183, 217 παμφανάω, 183 πάμφθαρτοε, 515 πανθάνω, 180 παπταίνω, 217, 377 παπταΚάομαι, 217 παραγγ^Κωνται, 457 παραδωσώοντα, 434 παραθίντων, 306 ■παραλαμβαν4τωσαν, 307 παραλύομαι (conj.), 447 παραμεινάντω, 306 παραμ^ινάτωσαν, 307 παραμβ/ψβταί (conj.), 447 παρανενόμηκα, 373 παραπ^πτωκ4τω, 422 ττοράστό, 299 παράσχω 132, 299 παραφθαίησι, 39, 316, 325 ΐΓοφδβΓ»/, 286 παρ4βασκ€, 528 παρέΐαν, 328 παρςίΚηφαν, 385 παρ-(ίσχηται, '^ισχησθαι, 340 παρςκίσκ€τ\ 529 παρεκοινάτο, 246 •7ταρ€ξ€λά<Γ?7σ0α, 34 ■τΓορεΐου^τί, 469 ■χαρίστηκαν, 385 •π•αρ€τά£α>ί/(Π, 47, 455 παρεχ4τωσαν, 307 παρ4χοιν, 332 7rape'xot(raf, 332, 452 τταρεχόιπ-ω, 306 ττάρην (παρίημι), 339, 343 παριστάμ^ναι, 339 παρίσταται, 318 παρμ4μβ\ωκ€, 379 παροίχωκΐν, 366 παρτςτύμβπ, 408 παρώξυμμαι, 420 παρψχηκ^ν, 366 7ταρψχωκ€, 413 -τάσσομαι, 221, 261, 521 ιτάσσω, 220 πασ-χΛι/τωι/, 306 πάσχω, 192, 198, 542 πατάσσω, 256 *ατ4ομαι, 261, 263, 521 ποΟ, 303 •καυσώμςσθα, 62 •τα<ρλά£«, 227 παφωι/, 291, 293 •>τάχ6Τ05, 514 π&άασκον, 530 ■ΠΈδίάλΟΗ, 457 ircifleu, 304 ιτβίθω, 154 irei««, 168, 217 ir€ipa{ofi€j/, 455 π€φασ€?σ0€, 469 ■παρασόμςσθα, 62 πειρασοΟμαί, 469 πςιρά-ω, 'ζω, 236 1Γ€/ρηθ€ί/Χ€ί/, 329 7r€ip7j0i)i>ai, 510 π€ίρί}τι£ω, 236 πείρω, 214 πείσβαι (conj.), 447 ττίΐστίον, 514 π4κτω, π€Κτ4ω, 162, 168 261, 269 αέκο>, 168 7Γ6λο^0^6σθα, 62 7Γ€λά-θω, -ω, 117, 502, 504 πελάσσ^τον, 462 7re\e/u£a>, 454 ireAeV/ceo, 191, 527, 529 7re\i7{ax, 455 8 π4\ομαι, 120, 280 ιτ€λα>, 144 πβλώ, 479 π4μπ(σκς, 529 πεμπόντων, 306 π€>π», 106, 144, 145 ττ^^ψαίοθ', 66 π4μφωμ^ν, 447 πενθείω, 240 πενιχρό*, 198 πίνομαι, 180, 192, 198 τ^πάγαΐίπΟ), 4 ?> 385, 396 π€πα 7 ο'ιην, 293, 336, 423 π€7ταθϋΓα, 402 7Γ€1ΓαΙχ€ί', 407 πςπάλασθε, 379 πβπαλών, 293 πεπίϊι/αί, 257 Trerrope?^, 293, 295 π^παρμ [vos, 214, 419 π^παρψνηκα, 374 πίπαυται, 380 π€π€ΐράσθω, 423 νίπνσθι (τΐΈττισΑι)» 290, 387, 402 π4πεισμαι, 419 π€π(ρασμ4νος, 420 πίπηγα, 396 π^π^ασιν, 385 7Γ€πηλ(5τί, 397 7Γ6πιασ^»/05, 421 τΓβτΓίθ-ίΓ^, -^σ0ο», 293 πίπΚΐχα, π4π\οχα, 399', 400, 407 πβπλήαταί, 66 7 π4π\ηΎα, 375 π4π\ηγον, 290, 429 πειτλήγορτ**, 290 7Γ€7τλ77γυ?ο, 379, 402 7Γ€7Γλϊ77«ϊ, 397 π4πλ-ηθα, 379, 401 7Γ6•7τλή{ομαι, 436 π^πληρώκοντα, 394 π4π\ησμαι, 357, 521 πί'πλοχο ν. ττ6πλ«χα ΐΓ67τοήκ[ω>>], 394 π4ποιθα, 379, 397 π4ποιθαν, 385 7τ«ποί0€θ, 430 πποίθ-ης, 391 π^ποιθοίη, 423 πίποίθομ^ν, 313, 423 7Γ€ποί6ω, 422 π*ποΚιορκΊ]μ4νος, 373 π4πομφα, 399, 405, 407 π^πόνημαι, 380 π4πονθα, 400 7re7ri{j/0ei/u,€j/, 428 7re7roi/0€/x€s, 432 πεπόνθίσαν, 428 π€7Γ(ί|/07}ί, 391 π^πορέιν, 294 π4πορ$α, 400 π4ποσθ€, 71, 384, 387 π4ποσχα, 198, 400, 408, 541 7Γ€7ΓΟτήθΤαί, 65, 378 7Γ67Γρ070, 401 π4πραται, 423 π4πραχα, 407 π€πρ7ΐσμ4νος, 521 π4πρησο, 423 π€πρ•ηχ4ναι, 404 7 πζπρωγγυςνκημ€ν, 390 π*πρωγ/ύηκα, 374 π«τται/ται, 65 π€πτ€ρύ'/ωμαι, 358 ππτηώϊ, 127, 387, 412, 542 7Γ€7ΓΤθ'ϊ, 512 π4πτνκται, 358 πβ'πτω, 165, 168, 219. π4πτωκα, 273 2 , 542 π^πτώκ€σαν, 428 π4πυθα, 401 ir6Tu0cV0oi, 294 π4πνκται, 358 ■7Γ€7Γυρ6χ(ίτ65, 400, 407 ιτ4πυσμαι, 419 π4πνσσαι, 420 πβράαι/ (fut.), 273, 480 πςράασκε, 530 π4ρόΌμαι, 143, 145, 280 πίρθαι, 131 π(ρβ(μ4νων, 27 π4ρθω, 143 πτριδώμϊθον, 67 1T€ptt€i€l', 329 ΠβρίΚλύ/ΛίίΌί, 129 π^ρικτίον*ς, 524 ιτβριμαξάτωσαι/, 307 π€ριστΊ}ωσ\ 314 f. ireptcpOvat, 341 π4ρνημι, 117, 128, 193 π4ρσοι*ν, 486 πβσ^ονται, 470 π4σσω, π4ττω, 165, 218, 219, 445 πίταμαι, 120, 267, 280 τπτάννυμι, 115, 123 INDEX. 57 a πετειν, 286 πετομαι, 120, 143, Ho 280 πεττω, v. πεσσω. πευθοίαθ', 65 πείθομαι, 156, 180 χευσεΐσθαί, 470 πεφα-γκα, 414 πεφασμενο*, 420, 524 «-έφευγα, 398 ιτεφεύγοι, 423 τπψη, 291 πεφτηνα, 397 πεφήνασι, 385 πεφήσεαι, 436 πεφ-ησομαι, 217 πεφθακα, 358 i/y?a, 402 ττεφυγγωΐ', ^94, ^01 πεφιτγμενος, 419 •π-εφι/^τεϊ, 378, 408 πέφνκα, 380, 385, 413 πεφνκασι, 385 πεφύκειν, 394 πεφνκη, 391 πεφνκη, 422 πεφυκώε, 358 χεφυλαξο, 423 ττεφυλοχο, 407 πεφντευκημεν, 390, 414, 424 ττεφυωταί, 426 τήγνυμι, 111, 219 τηδώ, 336 πηκτό, 131 πημαίνω, 253 πήσσω, 219 -ττήττω, 111 7πά£α>, 241 irtaiVa), 253 •jrteeij/, 342 f. «tcf«i (iri€^ea>), 344 f. iriei, 263 irteTj/, 286 πίησθα, 34 1Tl0-eiV, -cVflai, 28G «too, 304 πιθεσθων, 309 πιθ-ησω, 270 ιγΓΛ, 129, 263, 298 πιθώμεσθ', 63 πικραίνω, 253 ίτίλνα /io», 117 ■τπλνάω, 183 πίμ-πλά, -πλη, 301 πιμπλάνω, 170, 177 7Γί/Λ7Γλ€0>, 176 πίμπλημι, 45, 107 ττ/μπρη, 300 πίμπρτημι, 107 τη»/, 343 πίνεσκε, 529 πινόντων, 306 πιννμενη, 122 πινύσκω, 192, 197, 200 πιννσσω, 256 τπϊ/υτ^, 122, 192, 197 rim, 179, 216, 263, 518, 542 πίομαι, 467 πιπίσκω, 189, 193, 199, 200 πί7Γ7Γ-, τμγ-Ι^μ, 227 πιπράσκω, 117, 120, 189, 193, 199 πίπτησι, 39 ■πίπτω, 543 πίστευα, 252 πιτνεΐν (aor.), 286 πίτνημι, π'ιτνω, πιτν•άω, -εω, 114, 116, 117, 176, 183, 184, 282, 543 πιφαΰσκω, 193 ττίφρπμι, 107, 128 ττλάγχβη, 112 πλά&>, 224, 454 πλάναιμι, 26 πλε-γνυμι, 111 πλεία>, 156, 210, 524 •λΐη», 111, 143 πλευσοΰμεθα, 470 πλέω, 156 πλ^νυμι, 112, 224 πλήθος 107, 501, 504 πλημ-, πλημμ-ύρω, 255 πληρώνω (mod. gr.), 254 πλάσσω, 203, 219, 224 ττλητο, 127, 132 πλίσσο/χαί, 219 πλΰνω, 210 wei», 310, 524 πι/ευσΈΐται, 470 πνέω, πνεύω, 155, 156 irj /ίγω, 158 ποίντω, 306 ποθεω, 522 ττοθήω, 248 ποθόρ-ησθα, 34 ποίειμι, 26 ™eV, 342 ποιεεσκε, 530 ποιεοιν, 332 ποιεί, 139 ττοίηα/, 458 •η-οοίίτάτωσαϊ', 307 ιτοιησοΟντί, 469 ττοιήσα.'ί'Τί, 46 ποικίλλω, 161,212, 233, 255- ποιμαίνεσκε, 530 ποιμαίνω, 253 ποίοΓ, 335 ποιοίη, 335 ποιπννω, 377 ποιτρόπιοε, 142 ποίφνγμα, 221 ποιφΰσσω, 221 7Γ<ηώί/τί, 46 πολεμησείοντας, 534 πολεμιζεμεν, 339 πολεμίζω, 454 πολεμίξομεν, 453 Ή•ολί£α>, 250 πομπενω, 252 ποντοπορευεμεναι, 339 ΐ7•07πτύ£α>, 227 χορεΓί/, 287 πορενε-σθων, -σθωσαν, 310 λ 7TOpet θήίΌ£, 510 Tropevy, 252 πορεύω, 252 πορίζω, 243 πορνάμεν, 117 πορφΰρειν, 215 πορφυρεω, 215 iroTeoi/ieias, 392 TroTeos, 518 ποτητύν, 236 7roTtoe'7yU6i/oy, 370 7Γθτι-ίίλαίγ»(Τα, -κλαΐγοί',. 455 7τοτίπ€•7ΓΤ77υΤα{, 402 ■jrpa0-f'e:i/, -ε7ν, 287, 343 πραξίομεν, 469 7τρα£-ία>, -ets, 260 Trpa£oVTi, 447 πράσσω, 219 πρέπει, 144 πρεσβενσόντας, 469 πρί\σσεσκον, 530 πρίασθαι, 117, 120, 125 ττρίασ-ο, 60, 304 πρί£», 223, 251 irpt'a., 223, 251 ττρ<ί£ά, 299 προβάλεσκε, 531 προβεβηκν, 422 προβεβονλα, 265, 379, 401 προβωντες, 148, 299 προΎράφηντι, 318 προδιδ^μεν, 339 προειδεμεν, 391 προεληλυθοίης, 423 ττροθβΓτο, 334 προθεουσιν, 148 προϊδόμενος, 153 πραίϊσσομαι, 218 προκαλ-ίσσατο, -ίζετο, 243 (bis), 523 προλελε,γμενοι, 361 574 INDEX. προξεν4οι, 335 προοΊ-το, -ντο, 334 προςαρ-ηρεται (conj.), 423 irposZeKeo, 304 προσδοκάω, 235 προς-ηιξαι, 419 προςθεΐμεν, 330 πρόεθοιτο, 334 προςκυνειν, 184 προσταν, 148 προτίδεγμαι, 104 προτιθ4μειν, 340 προτίθηντι, 318 προψασωυνται, 481 προφύγοισθα, 35 προφύλαχθε, 384 προχωροίη, 335 «jptrvyvcueifpcr, 414 πταίρω, 112 πταίω, 211 ■7Γτορε«/, 287 πτάρνοισθε, 171, 179 πτάρνυμαι, 112, 171, 179 πτάρω, 144 πτερόω, 244 ■π•τ6ρύσ<τω, 256 πτ4σθαι, 280, 287 πτ-ησσω, 127, 219, 411 -π-τίσσω, 160, 203, 221, 241 πτολεμίζω, 242 7ττυί}ί/αί, 494 πτύρομαι, 214 πτύσσω, 219 τττιίω, 147, 161 πτωσκαζ4μεν, 219 τττωσ/ίά^ώ, 237 πτώσσω, 219, 256 πυ4ω, 501 πυ34σθαι, 287 πύθεσθαι, 504 πυθοίατο, 65 ■7τύθα>, 501 ττϋΐ/θάί/ο/ιαί, 156, 174, 180 ■jtiW, 501 πυρ4σσω, 256 πυστ4ον, 514 *■£, 300, 518 <η-ώ0<, 129, 298, 542 πω\4σκετο, 530 •π-ιέί/τ)*, 139 ™W, 179, 518, 542 7Γα>τά<τκ€-το», -το, 195, 199 ρ^ω, 225 0αί™, 217, 360, 521 £airif«, 360 ρ^τττω, 161, 166 0άσσατ€, 217, 521 £dV, 210 pWgat (dye), 225 9 pe'™, 78, 144, 145 ρ~4ρευκα, 360 Ρερίφθαι, 360 ρ"ενσεϊται, 470 £ea>, 78, 156 p -ηΎνυμι, 78, 112 ρ-ηγνυνται, 319 pVyvvvTO, 65 ρήγνυσκε, 529 ρηκτόε, 514 ρ-ησκομαι, 194, 199 p^jo-o-», 112, 219 £ήσ•σ•α> (stamp), 221, 222 Ρητός, 514 fiiy-4e>, -0», 78, 241, 246, 261, 263 piy&vTi, 249 ρΧοΌ>, 78 ίΦ», 226 ρίπιΧω, 165, 250 ριπτάζω, 236, 245 ρίπτασκ-ε, -ον, 236, 245, 530 ρίπτω, ριπτ4ω, 78, 154, 165, 236, 245, 269, 360, 537 pi^wu, 492, 500 ριφθεντες, 500 ροίζ-ασχ\ -εσχ, 531 ρο<ρεω, 78, 240 Ρ^ω, 201, 225 ρ^η, 491, 537 Ρυίσκομαι, 196, 199 ρνπόΌ>, 165, 360 ρύπτω, 165 ρύσκευ, 529 /$υσ>ιό>, 420 ρυστάζεσκεν, 530, 531 pWrdCw, 122, 236, 537 pvros, 156 ρώεσθαι, 114 ρώννυμι, 78, 114 ρώομαι, 78 traiVw, 217 σούρω, 214 σαλπίζω, 250, 454, 456 σαόΧ«, 524 σα<ίω, 244, 524 traiHj»;, 314 f., 491 σάττω, 219 σάω, 248 σοώσ6τοι/ (conj.)» 446 νβίνννμι, 114, 128, 621 σ«7ο, 459 (rcico, 78 σ€λαγ-€'α>, -ί(α>, 243 σ€λάσκ«, 195, 199 σεμνύνω, 254 σεσάχθω, 423 σ4σεισται, 360 σεσ-ημανται, 360 σεσημασμένος, 420 σ4σηπε, 360, 397 σεσηρ-4ναι, -ώς, 378, 397 σεσησμ4νος, 521 σεσ'ϊγηται, 360 σεσόανται, 417 σίσυρκα, 414 (ΓβΟτα*, 104 γ€ω, 263 στυγήσεταί, 492 crroeti/, 104 στικρελί^ω, 454 στύφω, 158 στωμυΧΧω, 256 σνγγεμος, 131 <τυγγ*Όίτο, 128 σΰΧασκε, 530 συλίύω, 252 συΧΧεγησόμενος, 493 συλώ, 335 συμβασείοντα, 534 συμμεμιχα, 407 συμπλακ-ρ, 492 σύμπωθι, 518 σνμφερην, 342 συνα,'γά'γοχα, 415 συίΌ7«7 Χ 6 " ι > ^91 σννοτγωνι\α.μενων, 455 συναρ-ηρακται, 367, 368 συναχθησοΰντι, 490 σννΰιαφυΧαξίομεν, 469 συνεαν, 328 συνεβάΧονθο, 47 συνεθεμαν, 58 συνεθοντο, 334 ! συνείκη, 394 σι/ϊ/βίλίχώ?, 400 συνειΧοχα, 361 συνεκεκΧειστο, 525 συνεμετρ-ησαμες, 43 συΐ'€ξ€δο0€ί', 490, συνερραφεν, 287 συΐφδβ'ατβ, 430 συ»/0ύξω, 475 σνίΌΚωχότβ, 369 συντριβησόμεθον, 67 συρίζω, 456 συρίσδει, 139 σύρω, 214 συσσημαινόσθων, 309 σύτο, 104, 130, 156, 518 σφαγεία, 492 σ<ρά£ω, 219, 225, 454 σφοληι/οι, 492 σφάΧΧω, 204 σ<ράττω ν. σ<ρά£ω. σφίγγω, 145, 150, 174 σφύζω, 225, ο-χά<Γω, σχάω, 223, 236, 524 575 σχε, 298, 299 σχεθεειν, 503, 504, 505 σχβίν, σχεσθαι, 287 σχεμεν, 339 σχ«, 132, 279, 287, 543 σχ6τό*, 279, 518, 543 σχήσ-εισθα, -εσθα, 34, 37 σχήσοί, 486 σχήσω, 279 σχ^ω, 223 σχοίατο, 334 σχο'ιην, 333 σώςώ, σ4>£ω, 114, 523 f. σωλίγξαι, 455 9 σωννΰω (σώννυμι), 114 σώοντο, 104 σωφρονίζω, 250 σώο, 523, (fut.) 484 τα/φ, 500 τα /fTjvat, 492 τάκω, 157 ταλαί-μοχθοί, -φρων, 235 ταμ-εΐν, -εσθαι, 287 ταμιεύεσκε, 530 τάμνρσι, 38 τάμι>ω, 179, 279 τάί/υταί, 113 τανύω, 113, 522 ταράσσω, 202, 218, 222 τάρβειμι, 26 τάρπημεν, 491, 498 ταρπώμεθα, 2S7 τάρφθη, 498 τάρχη, 222 τάσσω, 220, τατ<ίϊ. 217, 512, 518 ταφεΐν, 287 τα<ρί]ί/αί, 492 τέγγω, 143, 145 Τ60αλυ?α, 402 τεθάρσηκα, 380 τεθαρσήκασι, 413 τεθεικα, 414 τεθηΧα, 379 τεθηπα, 378, 397 τεθήσατο, 466 τεθηεν, 293 τεθΧασμένος, 358 τεθΧιφα, 407 τεθμός, 519 τεθναί-ην, 422 τεθνάκην, 391 τεθνάμεναι, 424 τεθνάσι, 387 τεθνεωσα, 395 τεβνεώτα, 426 τεθνηκα, 380, 413 τ€θ*ήξω, 436 τεθορείν, 293 τεθραμμαι, 419 T60a>7-, τίθατγ-μεναι, 156 TetW, 113, 202, 217, 506, 18 576 INDEX. relpu, 202, 214 τασαμενόϊ, 158 τ€ΐχ-4ω, -ίζω, 243 τεκει»/, 287 Ύ€κέϊσ•θαι, 481 τ^κμαίρ-ω, -ομαι, 255 τελε'βει, 502 τελε'θεσκε, 530 τελε'0», 504 Τελεία», 240 τελεσκο•/, 529 τελε'<τκω, 196, 199 τελεϋτάσοοιτί, 46 τελευτάα*, 345 τελεω, 522, (fat.) 480 τ€\ι<τκόμ€νο9, 196 τςμ•4ΐν, •4σθαι, 287 τέμ;/α>, 179, 279 τ4μω, 144, 179 Te'j /δω, 143 τεξείεσΑε, 470 3 τερεμι*ο$, 353 τε'ρττευ, 304 τέρπτισι, 38 τέρπω, 144 τε'ρρα•, 214 τερ<ταίι/α•, 253 τ4ρσαν, repcrcu, 456 τερο~-ήμε•/αι, -ηναι, 491, 497 τ4ρ<τομαι, 144 τερύσκ-ω, -ομαι, 189, 192, 195, 199 τ6το7«ίν, 294, 295 τεΓακα, 414 τ4τακτοι, 61 Τ6Ταρ7Γ€ΤΟ, 294 Τ€τάσθην, 435 τεταταζ, 419 τεταχα, 407 τετάχαταί, 418 τετελεντακούο -as, 394 τετεύξετα*, 436 τε'τευχα, 397 τετεύχαται, 65, 419 τετευχησβαι, 267, 270 τέτηκα, 378, 397 τετ<τ?<ίτ€ϊ, 378 τςτιμήμ*(Τθα, 62 τε'τλα, 422 τ4τΚαθι, 290, 378 τίτλοίη, 422 τίτλο/χεν, 388 τετλάμεναι, 424 τε'τλη /cas, 413 τε'τοκα, 400 τέτορε»/, 294, 295 τετραί™, 217, 269,543 τ4τραμμαι, 420 τετρά<ραται, 66, 405, 418 τςτράφθω, 423 τΐτραχυμ4νο$, 421 τ<τρ(μαίνω, 177, 186, 217 πτρηνα, 217, 466 τ€τρηχυΐα, 379, 397, 402 rirpiya, 357 T€rptyv7a, 377, 402 nrpiySnas, 426 τ4τριφα, 407 τετρίφαται, 405, 418 τ4τροφα, τ4τραψα (τρ4πω), 398, 400, 407 τ4τροφα, τ4τραφα (τρ4φω), 400 τετταρεϊ, 70 τετύγμη*', 434 τετυκεΓν etc., 156, 294 τέτυξο, 428 rertmovres, 294 τετυσκ-ων, -ετο, 197 τςτύφωμαι, 408 τετυχηκώϊ, 413 τετύχρσ*, 294 rerv -χβω, 423 τεύχω, 156, 180, 267 ττ}, ττ)τε, 303 τακτοί, 514 τήκω, 157 τιεσκόμενοι, 196 τίβεί, 301 τιθέΐμεν, 330 τιθ4μεν, 339 ηβ4μςσθα, 62 τί0εσκε, 529 τίθεσ-ο, 304 τίβτ;, 41 ηθήμςναι, 340 rffc?/u, 107, 517 τί0τ;ϊ, 383 τίθ-ησθα, 34 τίθητι, 38 Tt0ou, 305 τίκτε»', 342 τίκτω, 168 τίλλω, 212 τιμώσασα, 246 τίνυμαι, 113, 171 TtVa», 113, 170, 171 f., 179, 518 • ' τί<τετε (conj.), 446 τιταίν^τορ, 310 titcuW, 113, 217, 376 τίτί£ω, 227 τίτρημι, τιτράω, 108, 217, 543 τιτρώσκω, 115 τιτύσκομαι, 191, 197, 200 τία>, 113, 158, 179, 518 τλοΓει/, 329 τληθί, 298 τμάγεν, 491 τμ-ηγω, 157 το-κο-ι -je ( = δάκο*ε), 411 τοκώσΌ, 535 τολμά», 237 τομάω, 535 τονθορ•, τονβρ-νζω, 227 το£ά£ο/«α, 237 το|ει5», 252 τορενω, 252 τορεΌ>, 269 toVoois, 465 τραγεΐν, 287 τραπ-4ΐν, -4<τθαι, 287 τροττείομεί', 313 τραπε'ω, 240, 269 τ ράνω, 275 τραφΐίν, 287 τράψτ?, 491 τραφθηραι, 498 τράφω, 275 τρείω, 204, 210 τρ4μω, 144, 145 τρε'™, 144, 279 τρ4σσ€, 146 τρ4φοιν, 31 τρ4φω, 144 τρε'χω, 143, 544 τρ«ία>, 146, 204 τρί£ω, 158 τρί&>, 202, 225 τριψθέίσα, 500 τροττε'α», 240 τρυ£«, 225 τρόχω, 192, 199 τρώγω, 157 τρωννύω, τρώννυμι, 115 τρωπασκ4σθω, 195 τρωπάσκζτο, 530 τρωπάσκω, 190, 199 τυγχάνω, 156, 174, 180, 220 τκκτ^ί, 514 τώ<«, 143, 150 τνμβογ4ρων, 408 τυττά£ε<ι>, 270 . τι/ττεΓν, 287 Tuirrjffet, 270 τύτττω, 160 f., 165, 267 Typamjo^ioyTa, 534 τυραννιάω, 244 τυρ/8ά^«, 236 τνσσςι, 220 τυφ\ώσσω, 257 τυ<ρ<5ω, 408 τι'<ρ», 158,371 τι/χεΓ»/, 287, 465 τύχησε, 270 τύχρσι, 39 τύχ«μι, 27, 39 τα-θά^ω, 227 2 υ/8ρί£ω, 250 ι/γί-αίι/ω, -<£{,«, 89, 177 υγρωσσω, 257 ύδείο/χεν, 241 ύδρευα*, 252 8δ« (λ^γα•)» 143, 153 δετό'ί, 514 ύλάε», 162 INDEX. 577 ύλακτ4ω, 89, 162, 256 ύλάσσω, 256 ύμίναιοΊ, 336 ύμεναίουν, 89 υμνείουσαι, 241 ti£oj/, 45ό ύπαρξβΰι/τί, 469 νπαρχ4μζν, 340 ύπάρχεν, 342 ύπ€ΐκαθ4ων, 503 ύττ€λθετ4οι>, 514 ύπεμνί)μυκ€, 359, 413 ύττ4ρ€ΊΓΤ€, 166 ύττςρράγη, 491 ύττ4ρσχοι, 334 ύπισχν4ομαι, 177, 184 ύπι/όΌ*, 244 ύπνώεσκον, 530 ύττνώοντας, 190 ύπνώσσω, 257 ύπογραψόρταί, 469 ύττοδρήσσω, 221, 412 ύιτό'θευ, 304 ύ7Γ0λί^«Ι/, 202 υ7Γ07Γ6πτηώτ€5, 426 ύττοττιμττργσι, 39 ύποπτζύΐβσι, 39 υποταγεί, 492 ύποτίθοιτο, 334 9 ύποφαύσκαν, 193 ντττιάγσι, 39 ύστερ-4ω, -ίζω, 243 ύφά-γ^ο, 305 ίφαίνω, 177, 183, 205 ύφανάω, 183 ΰφανεν, 89 υφελοίατο, 66 νφ{]φασμαι, 292, 369 ύψίητί, 39 ύφόωσι, 178 φαάνθη, 499 φογείν, 287 φay4μev, 339, 340 φά•γομαι, 467 φάε, 147 φα4θουσα, 502 φα4θων, 502, 504 φαε^ω, 217, 254, 284 <ροθί, 298 φαί-ημεν, φαιμεν, 329 φαίτ}ί, 330 φαί™, 183,217 (bis), 221, 236 <ρα?(π, 47 φα»/, 148 φανΰσθαι, 475 φαν4σθαι, 287 φάνεσκ€, 488, 532 «paj /η, 492 φανηρ, 315 φάνηθι, φανήτω, 490 φανησαν, 490 φανήσομαι, 493 φαί/τά£α>, 236 φάντι, 319 φάο, 304 φάpypυμι, 168, 220 φάρκπ-σθαι, 161, 168 φαρμάσσω, 256 ψ άρω, 145 φασ^άνεται, 183 φάσκα>, 189, 193, 528 φάο-χ', 529 φασώ, 469 φατ€ΐόε, 512 (pcm'Cetj/, 243, 537 φατώς, 304 φάω, 524 ei/, 342 φερετωσαι>, 307 φέ>ρ, 317 φςρόμεσθα, 62 φερόντων, 307 φερτά^ί, 236 φέρτε, 104 (pepros, 512 «, 128, 144, 145 φεύγεσκε»/, 529 φςύγην, 342 φςνγόντων, 306 φε«ί 7 ω, 151, 152, 156, 181 φεύ£ω, 237, 251 φευκτός, 514 φευξείω, 534 φευξοίατ , 66, 486 φ€ϋ|οίτο, 486 φευξονμβθα, 470 φήί?$, φή?7, 315 φτ?μί (dor. φα^ί), 96, 99, 544 φημί^ω, 250 φτ}<τ0α, 34 φ£<π, 38 φθαιη, 329 φθαίητ€, 330 φ0άί>ω, φθαι>4ω, 144, 179, 269 φθ4ΎΎθμαι, 143 φθείρω, φθ4ρρω, 214 φθ4ραι, 457 φθ4ρσαντΐ$, 457 φθ4ω-μ€ν, -σι, 319 Φθ^ρ, 315 φθίεται (conj.), 313 φθινύθ^σκε, 530 φ0ίνύ0ω, 171, 172, 503, 504 φ0ύ/ω, φθ^εω, 171, 172, 179, 269, 518 φθιόμεσθα (conj.), 313 , 223 φραζώμςσθ', 62 φράσσω, 112, 168, 204, 220, 223 φρ^, 107, 128, 298 φριξός, 535 φρίσσω, 220 φρυντίσδην, 342 φρι^ω, 158, 220 φρύσσω, 220 φιτγ 7 άνω, 156, 174, 181 φύγε?*/, 288 φύγεσκε, 531 φυ&"«, 123, 173, 175, 182 φύ£ω, 225 φυτ?, 493 φυίω, 147, 209 φυλάσσοντι, 46 (bis) φυλασσόντων, 307 φυλάσσω, 233, 256 φΟί/, 343 φύί/ω, 179, 216 φύρσω, 457 φυρω, 214 φυσάω, 199, 221 φυσιάω, 237 φυσ -Tis (?), 519 φυτειχτεΓ, 141, 447 φυτεύω, 518 578 INDEX. φυτάν, 518 φύω, 147, 518 φώγ-ω, -νυμι, 112, 157, 225 φώζω, 157, 225 φωραν, 231 χαΰέίν, 288 χάζω, 223 χαίνω, 170, 197, 217 χαίρετον, 310 (bis) χαίρω, 215, 267, 271, 272 χαλ-άω, -ά£ω, -αίω, 234 χαλίπτω, 161, 165 χαλκεύω, 252 χανΖάνω, 174, 181 χαι^, 288 χανύαν, 217 χαράσσω, 256 χαριέισθαι, 481 χαρίζομαι, 250 χαριξιόμ^θα, 455, 469, 483 χαρίσσονται, 455 χάροντο, 288 χάσκω, 192 (bis), 197, 217 χοτιΧω, 251 Χ^ω, 223 χ6ίμά£α>, 236, 238, 253 χςιμαίνω, 253 (bis) χ*ίω, 210 χςρ-νίπτου, -νίψαντο, 167 χεσβίω, 534 χβσΌυμαί, 470 χίω, χίομαι, 115, 156, 210, 518 f., (fut.) 484 χηρόω, 244 χιμ-αν, -αζτιν, 236 χ\οι54σκω, 196, 199 χ<ίλαί<π, 47 χο\όω, 244 χορεύω, 252 χ<5«, 115 χραίνω, 217 χραισμίω, 39, 261, 26Β, 282 χράομαί, 196, 526 χρ^μίθω, 502, 504 χρςμετίζω, 168 χρςμίζω, 168 χρίμπτομαι, 168 χρή, 100 χρηίσκομαι, 196, 199 χρηι/, 92 χρηννυμαι (?), 116 χρησθα, 34 χρησται, 69 χρήστω, 69 χρΓ, 303 χρίμ-πτω, 168 χριστά, 147, 521 χρίω, 147, 521 χρώ£ω, 115 χρώννυμί, 115 χρώσθω (»/), 309 χώεο, 304 χώννυμι, χωνρύω, 115 χωραξάτω, 455 χώσεταί (conj.), 446 ψα?μα, 113 ψαίνυμαι, 113 ψαίνυσμα, 113 ψαίρω, 215 ψάλλω, 212 ψά 143 ψ<ίδω, 143 ψεΓσαι, 113 ψεόδω, 156 ψ^ρβί, 144 ψηχω, 199 ψί£ε<Γ0α<, 225 • ψινύθω»/, 113 ψύβοϊ,' 156 ψύττβί, 221 ψυχηι /at, 493 ψύχω, 158, 199 ψωραν, 535 ώατωθησω, 490 ώ£ρατο, 361 ψδήκαντι, 46 ωΰοιΤξΐτοπ)μ4νη, 374 ώδυσαο, 449 ώ^ω, 237 φήθηι/, 510 tffletr/ce, 190, 530 ώβ^ω, 263 ώΚισθον, 503 ώ/Αοσο etc., 87, 274 ώνασςίται, 240, 469 ώνίομαι, 120, 240 ωνητιάν, 536 ωρετο, ωρτο, 286 ώρίνθη, 499 ώριξαν, 454 ώρορε, 293 ώρυω, 78 δρτο v. topero ΰισασκς, 532 ωστιζ6μ^σ&, 62 ώ(ττί£ω, 537 ώστιοΰνται, 481 ώσφρ-αντο, -οντο, 464 ώσφρ-όμην, -άμην, 283 άχρβλείσθωοΌ^, 310 ώφβλή/ο?, 431 ώφ(λητ4ο$, 515 &φί\ον, ωφΧον, 279, 283, 286 ώχριάω, 244 <£χωκα, 267, 273 ώψεοι/, 533, 535 J5. ITALIC. (Latin not distinguished.) aamanaffed (Osc), 441 abdo, 506 abnueo, 259 acceptus, 515 actud (Osc), 306 acuo, 250 adagium, 103 aegrotus, 238 ago, 145 ahesnes (Umbr.), 230 aio, 181, 203, 230 alb-are, -ere, 244 albicare, 256 alumnus, 139, 353 angetuzet (Osc), 373 ang-it, -et (Osc), 373 ango, 110, 145 animare, 238 apio, 166 apiscor, 82, 191 apparere, 293 aptus, 82 arceo, 239 arcesso, 534 arrugia, 222 augeo, 181, 445 ausim, 438 auspicari, 240 balbutio, 257 bovare, 231 INDEX. 579 cacare, 231 cadaver, 425 caecutio, 257 caedo, 151 calare, 240 calator, 240 Calendae, 240 cano, 236 canto, 236 capesso, 534 capio, 203 careo, 497 carino, 177 caveo, 240, 261 -cello, 525 censazet (Osc.)> 441 censeo, 240 cepit, 389 cerno, 170 cio, cieo, 259 clar-are, -ere, 244 claresco, 190 claudico, 256 cluo, clueo, 148, 259 coctus, 512 coepi, 380, 390 columna, 353 comare, 231 comminiscor, 193 comparascuster(Osc)) 188 condo, 506 coquino, 177 coquo, 219 corrugus, 222 credo, 506 cresco, 190 cubito, 236 cubo, 166, 236 cumbo, 166 cupio, 203 cupire, 203 damnum, 353 danunt (dane), 171 dapinari, 240 datus, 512 deded (Osc), 389 dedeit, 389 dedet, 389 deicans (Osc), 317 deico, 151 deikum (Osc.)i 151 deivatud (Osc), 306 deleo, 259 depso, 145 depuvit, 211 desipire, 203 desivare, 364 destinare, 177 dicare, 236 dicere, 152 -dicus, 110 disco, 196 dixe, -m, 449 doceo, 196 domitus, 514 domo, 234, 273 douco (Old. Lat.), 151 duco, 152, 218 3 6do, 104, 145, 265 eiscu (Umbr.), 189, 192 eiscurent (Umbr.), 188 eituns (Osc), 307 erugit, 155 es (Imperat. tort.ed), 298 escit (Old Lat.), 528 estod (Old Lat.), 303, 306 estud (Osc), 306 esurio, 534 etu, -to (Umbr.), 306, 305 2 explenunt, 170 exstinxe, -m, 449 facia (Umbr.), 317 facio, 120, 203, 411 facitud (Old Lat,), 306 factud (Osc), 306 f alio, 204 farcio, 168, 204, 220 fari, 524 fateor, 243, 537 faxit, 438 fef acid (Osc), 412 feido, 151, 154 feliuf (Umbr.), 392 femina, 353 fendo, 216 ferascit, 200 ferinunt, 173 fero, 104, 145 fertu (Umbr.), 306 ferv -o, -eo, 259 fido, 151, 154 filius, 392 findo, 170 fingo, 174, 180 finio, 249 fio, 508 fleo, 259 fodio, 203 frango, 112, fremo, 145 frequens, 497 frico, 234 frigeo, 360 fruniscor, 176 fugio, 202, 203, 225 fuia, fuiest (Umbr.), 147 fuit, fuet, 389 f ulcio, 203 fulg-o, -eo, 259, 497 fumo, 231 furari, 231 furo, 151 futuo, 250 fuueit (Old Lat.), 389 gaudeo, 209, 240, 261, 503 genitur, 278 genitus (bis), 514 gero, 236 gesto, 236 gigno, 543 glocio, 203, 2L8 gnatus, 511, 543 gnosco, 179, 193 gnotus, 511 gradior, 203 grandire, 249 grava-re, -ri, 210, 243, grex, 215 gustus, 155, 520 habetutu (Umbr.), 308 2 habitare, 236 heriest (Umbr.), 215 heriiad (Osc), 215 hiemo, 236 hieto, 192 hisco, 192 (bis), 197 ignarus, 243 7 ignoro, 243 impedio, 250 inclino, 231 inclutus, 512 indago, 240 ingemisco, 188, 191 inretio, 249 insece, 132, 145 inserinuntur (Old Lat.), 173 interieisti, 389 inveterasco 110 invictus, 515 irascor, 190 ire, iri, 206 itare, 236 jac-ere, -ere, 167, 20.'*, 236, 335, 411, 497 jactare, 236, 537 jugare, 238 junctus, 514 jimgo, 108, 110 labare, 152 labi, 152 lacesso, 535 lacio, 203 lambo, 167 lascivus, 146 lav-are, -ere, 240, 261, 269 lggo, 145 lenio, 249 levo, 238 libo, 231 licet, 497 likitud (Osc), 303, 306 lingo, 154 lino, 170, 171, 178 linquo, 154, 174, 180 λιοκακβίτ (Osc), 412 locutus, 515 lubet, 166, 240 lupuce (Etrusc), 412 580 INDEX. luxare, 238 madeo, 243 maneo, 264 manus, 209 maturesco, 190 mederi, 270 meditari, 236 mejo, 203, 261 memento, 378 memini, 190, 375, 378, 390 memor, 365 memordi, 356 memoro, 230 mereo, 214, 361 metuo, 250 -miniscor, 192 misceo, 189, 192, 197 modulor, 256 molo, 211 moltas (Osc), 232 morior, 203 moriri, 203 ' moveo, 153 mugio, 204, 226 multo, 232 mungo, 219 nactus, 515 nanciscor, 176, 200, 291 necto, 163 neo, 148, 259, 501 novare, 231, 238 novi, 375 nubo, 152 nuo, nueo, 155, 236, 259 nupturio, 534 mito, 236 obdormisco, 190 obinunt (Old Lat*), 171 odi, 375, 378, 390 olfacio, 283 olo, oleo, 259 orior, 111, 172, 2 203, 261 ostendo, 236 ostento, 236 paciscor, 191, 200 pango, 111 papaver, 425 parentes, 278 pario, 203 parlre, 203 paro, 243 pasco, 189, 200 patior, 192, 198, 203 pavio, 210 pecto, 162 pedo, 145 pend-ere, -ere, 497 pensare, 236 peposci (Old Lat.), 73, 356 pepugi, 73, 356 perdo, 506 perfines (Old Lat.), 302 persni- (Umbr.), 170, 188 peto, 145, 259 ♦petulo, 256 pingo, 170 pinsio, 203, 221 pinso, 160, 221 piscor, 231 plango, 112, 224 plecto, 163 ploro, 209 posco, 192 potior, 198, 249 potitus, 515 poto, 236 praefica, 412 pravescere, 200 precor, 170 prehendo, 174, 181 proficiscor, 200 profiteor, 243 prospices (Old Lat.), 302 prufatted (Osc), 282 quaero, 259 quatio, 203 rapio, 167, 203 redinunt (Old Lat,), 171 rego, 145 reminiscor, 193 repens, 145 rideo, 223 ructo, 78 rugio, 204, 225 rumpo, 170 runcare, 222 runco (subst.), 222 salio, 203 (bis), 211, 236 salto, 236 salv-are, -ere, 244 salve, 173 sapio, 203 sarcio, 161, 167 scat-o, -eo, 259 scindo, 116, 151, 170 scisco, 189 sedeo, 152, 223 sedo, 231 sequor, 83, 145 sero (serui), 81, 214 serpo, 86 sido, 162 siem, 324, 328 sileo, 240 • simulo, 238, 243 sinciput, 392 sino, 84, 171,364 sisto, 107, 359 solin-o, -unt(01d Lat.), 173 sono, 231 sons, 352 sorbeo, 240 specio, 161, 203, 204 stahituto (Umbr.), 308 a statuo, 250 sterno, 112, 171 sternu-o, -to, 112 stinguo, 225 stipo, 231 strangulo, 231 struo, 251 studeo, 156 subahtu (Umbr.), 306 subdo, 506 subfio, 203 superstes, 352 taceo, 240 tagam, tangam, 278, 294 tango, 294, 295 tece (Etrusc.), 412 tego, 145 tendo, 506 teneo, 506 terg-o, -eo, 259 tero, 214 terreo, 240 tollo, 151, 235 torqueo, 240, 269 torreo, 497 tracto, 236 traho, 236 tremesco, 191 tremo, 145 tribarakavum (Osc), 233 trudo, 152 tulo (Old Lat.), 151 tuor, tueor, 259 turbo, 236 turce (Etrusc), 412, 413 2 tusetutu(umbr.), 308 2 tutudi (Old Lat.), 395 ulciscor, 200 ululo, 365 unco, 231 upupa, 365 urgeo, 240 uro, 152, 155, 520 ustulo, 256 ustus, 155 usus, 515 vado, 152 venio, 185, 204 venum (ire), 80, 206, 494 verro, 86, 123, 147, 520 vertumnus, 139, 353 vestio, 249, 520 visio (β84ω), 146, 520 viso, 444 volo (wish), 145 voluntas, 352 volvo, 85 vomo, 367 INDEX. 581 C. SANSKRIT. akar, 130 akshan, 279 akhjam, 280 again, 126 agasishus, 440 agrhnam, 169 aghasan, 279 aghajami, 236 akakaksham, 429 akukurat, 258 agananta, 282 afanishta, 438 a|ati, 317 adarcam, 278, 279 adam, 126 adrcam, 278, 279 admi, 104 ad-dhi, 298 adham, 126 adhukshat, 438 adhvarjant, 255 apaptam, 289 apipatat, 289 apipet, 376 apsanta, 439, 444 abudhanta, 278 abodhanta, 278 abhut, 126 abhaishma, 438, 446 ajansam, 438 ajasisham, 440 ar (go), 212 aratijati, 249 arikat, 278 arkase, 438 arthaje, 233 ardh, 163, 185 alambhanta, 174 av, 147 avidat, 278 avedisham, 265 avokam, 291 acamat, 278 acana, 116, 173 asati, 311 asaham, 279 astham, 126 asmi, 101 asvarshtam, 441 ah, 181 ahan, 130 akamami, 151 ada, 365 adajami, 264 adar, 212 adunvasva, 109 aninat, 291, 365 apipam, 108 amamat, 291, 365 £jam, 89 ara, 365 arta, 126 ardidat, 291, 365 arpipat, 291 as, 103 asate, 65 asitha, 35 astham, 281 aha, 103 ing, 154 ikkhami, 188, 189, 192 indh, 153 invami, 171 ijarmi, 212 ishananta, 173 ishnami, 116 ihi, 297 ukshanjati, 253 ukkhami, 188, 200 uvaka, 369 ush, 155, 520 urgajami, 240 rgh£jami, 229, 240 f^ujati, 250 rkkhami, 188, 192 (bis) rnge, 111 rnadhmi, 185 j-nomi, 109 j-taje, 239 rdMnt, 352 rdhnomi, 185 egami, 154 edhi, 297 emi 96 ogaja-mi, 240 Oshami, 151 (bis), 155 kamami, 151 ku (howl), 537 kupjami, 202, 203 krpa-, krpan-, krpa-jati, '253 kratujati, 250 kramami, 151 klamami, 151 kshan, 216 kshanonii, 114 kshinomi, 170 khang, 226 gakkhatat, 306 gakkhami, 188, 192 gat a, 511 gah, 166 gung, 226 guhami, 151, 155 grnishe, 438 grhana-s, 173 grdhnus, 108 grama-s, 215 gharsh, 147, 521 kakaksha, 429 kaje, 158 kinomi, 113, 170, 172, 179 korajishjami, 234 khad, 116 khid, 116 gaganti, 105, 543 gagana, 356 fanami, 272, 282 ganijati, 249 gagarajami, 215 gagarti, 368 gaje, 216 glgati, 105 glgfiasami, 187 givase, 350 gishe, 350, 448 gugosha, 356 gush, 155, 520 geshi, 298 geshjami, 474 goshami, 152, 155 gnata, 511 gnejas, 329 tata, 512 tatana, 395 tatre, 357 tan, 506 tanishiami, 477 tanute, 113 tarishjami, 265 tashthau, 289 tasthivan, 425 tishthami, 106, 107, 289 tutoda etc., 356, 389 tudami, 150 turvane, 346 tula j ami 235 typnomi, 170 trmpami, 170 tras, 521 trasjami, 146, 210 tra^ajami, 240 dacami, 170 datta, 519 dada, 382 dadarca etc., 357 dadami, 105 dadfcvan, 425 dadrus, 357 dadhami, 107 dadhidhve, 416 dadhima, 389 dadhlta, 331 582 INDEX. damajami, daman jami 117, 177, 230 daj, 208 dardarimi, 390 dardharti, 376 da?, dac, 109, 178 damane, 344 damjami, 117 davane, 346 dasjami, 260, 471 didrkshante, 444 die, 110 dijami, 120, 148 dirjami, 213 dunomi, 209 duhami, 150 dejam, 329 devajami, 233 djami, 203, 209 dvish, 520 dvishant, 352 dhajami (suckle), 202 dharsh-ajanii, -ami, 229, 240, 259 dhunOmi, 171 dhup, 158 dhumajami, 231 dhurvane, 346 dhrshnu-s, 108 dhejam, dhejas, 329 namasjami, 233 nas, 146, 210, 220, 521 nah, 163 nanadati, 376 nig, 167, 224 nid, 121 nind, 150 nesha, 450 neshati, 441 neshatha, 446 nonaviti, 390 pak, 219 pakata, 514 panate, 117 panajami, 176 pattraiami, 244 paprakkha, 400 papra, 357 pacjami, 161, 204 pacjes, 328, 332 pata-ja-mi, 235 pahi, 297 piparmi, 107 pipasati, 444 paplshati, 439 pipfhi, 297 pibadhjai, 85 1 pish, 160,221,241 pld, 241 prkkhimi, 192 pj-nakmi, 111 prnati, etc., 170 ppiadhjai, 351 paurusheja, 235 prakshe, 448 prakh, 188 pra-brava, 29 babhdva, 73, 356 babhuv&n, 425 bibharmi, 107 bubhugmahe, 416 bodhami, 152 bobhaviti, 390 bravlmi, 390 bha^, 544 bhafi^, 110 bhar, 128 ^ bharadhjai, 351 bharadhve, 63 bharasva, 304 bharami etc., 138, 317 bhare-ma, -jus, 324 bharti, 104 bhavatat, 304 bhami, 96 bharajami, 248 bhash, 188 bhugami, 151, 156 bhuranjati, 205 bhurati, 151 bhujama, 329 bhfta, 512 bhotsjami, 474 mantrajami, 233 manthami, mathnami, 170 > mandajadhjai, 351 man j ate, 203 marata, 514 masjami, 210 malati, 232 malajati, 232 masi, 298 miksh, mimiksh, 197 mlv, 153 muk. 150, 219 murkhami, 188, 200 murta, 188 mr^ajami, 233 mrnami, 118, 170 medjami, 202 mriiate, 204, 205 .iakkhami, 188 ja^, 224 jagadhva, 62 ja^ja, 204 jas, 146 ja, 106, 121, 206 jajam, 324, 329 jasi, 298 jukta, 512, 514 jukkhami, 188 junagmi,408, 110 rakshami, 445 ratharjati, 255 ramate, ramnati, 119 radhnomi, 114 rajati, 225 rasate, 446 rikh, 154 rinakmi, 154, 174, 180 ririkshe, 416 rireka, 356, 395 rishanjati, 178 rihami, 150 rugant, 352 rudh, 503 rurOga, 395 re^ami, 227 rerihjdte, 536 reshati, 178 lambh-, 174 lash, 520 lashjami, lasami, 146, 209 lubh, 166 lelih, 378 10k, 219 lohitati, 232 vakshajami, 265 vakshjami, 477 vad, 153 vadh, 80 vand, 153 vap, 205 vam, 367 var (defend), 122 vas (clothe), 113, 520 vas (dwell), 520 vasnajami, 240 vahatat, 304 vankkhami, 188 f., 192 (bis), 198 vami, 120 vahajami, 240 vig, 222 vida-si, -ti, 317 vidushi, 425 videt, 332 vidmane, 344 vidvdn, 425 vidhami, 150 vivasati, 534 vicami, 150 vlrajami, 237 vrnagmi, 110 vrnomi, 118,123, 172 vi-haje, 534 • vettha, 34, 36, 383, 395 vlda, 153, 371, 379 vedmi, 153 vedane, 344 vedajami, 233, 264 vevigjate, 536 vokatat, 304 vokati, 317 i vo^etam, 332 vjaprijate, 120 vrag^, 188 casajami, 242 INDE£. 583 QaknOmi, 175 9am, 171 $inut£ 113 gush, 155 Qri-na-mi, 170 grnthati, 170 9ete, 100 9Jami, 202 9rathnami, 170 91'addha, 506 9ruta, 512 9rudhi, 126, 129, 297 9vajami, 268 9vas, 521 sakadhjai, 351 sakasva. 304 sad, 223 sanoti, med. sanute, 122, 171, 283 saparjati, 255 samkirati, 115 sasftva, 73, 356 sahadhjai, 351 sahasva, 304 sadajami, 231 su (excite, send), 364 sedhami, 151 skhad, 116 stava, 29 star a j ami, 251 str-nomi (-nami), 112, 116 sthapajami, 231 snu, 156 (s)pa9jami, 204 smarajami, 230 sravami, 156 svanajanii, 231 svadha, 506 svadate, 156 svidjami, 203, 208 had, 223 hanati, 311 ha j an tat, 308 harjami, 215 hasate, 438 hasmahi, 438, 440 hinomi, 170 D. IRANIC. (Zend not distinguished.) aeiti, 96 aistata (Old Pers.), 79 atiyaisa (Old Pers.), 89 ada (Old Pers.), 126 adinam (Old Pers.), 170 apabarois, 332 amariyatd (Old Pers.), 204 avaretha, 122 avaiti, 39 idi, 297 iririthare, 365 i9aitg, 189, 192 ishasoit, 188, 192 ukhshyant, 204 uzbarajat, 240 kerenava, 176 kerenvo, 109 qabda, 506 qh-je-n, 329 gaidi, 297 gaozaiti, 151 ga9aiti, 188 gasaetem, 52 zahyaninam, 471 zinat, 170 zdi, 297 tatashat, 291 tanva, tanava, 113 thrazdum, 63 daidyata, 331 dan, 126 dadhami, 105, 107 daonha, 471 dato, 512 drazh, 218 nadent, 121 patiyaisa (Old Pers.), pathyaiti, 220 perena, 170 peregd, 192 (fra)mairyeite, 204 fradadathd, 36 fsanajaiti, 179 barayen, 324, 332 barat, 317 bun, 126, 278 buidjai, buzhdyai, 351 bushyantem, 471 bvat, 278 maidhayanha, 270 mainyetg, 204 yaonh, yah, 521 yaozda, 506 vaedd, 379 vanh, 114 vakhshya, 471 varatha, 122 vidushi, 425 venhat, 438 verezyitmi, 202, 204, 225 verez-jeidjai, -idyai, 351 voi9td, 34, 36, 379, 383 9aete, 100 9U9ruma, 356 9taomaine, 344 9pa9ya, 204 9rutO, 512 hi9taiti, 107 hush, 155 584 INDEX. Ε. TEUTONIC. (Gothic not distinguished.) afiifnan, 180 aiaik, 365 aialth, 365 aiauk, 356, 365 airthakunths, 511 anabiuda, 156 ananiujan, 231 anasilan, 240 aukan, 445 babim (Ο. H. G.), 157 baida, 154 baira, 138, 145 beo (A. S.), 467 biauknan, 181 bidjan, 203 bimunigon (Ο. H. G.), 256 biuga, 152, 156 chriuzigon (Ο. H. G.), 256 dagen (Ο. H. G.), 240 drahjan(O.H.G.),240,268 dreskan, 189, 192 eiscon (Ο. H. G.), 189,192 faifah, 356 faifahu -n, -th, 389 faiflok, 358 faifrais, 358 faltha, 163 fairveitjan, 264 firstan(O.H.G.), 100 fiskon, 231 flewiu (Ο. H. G.), 210 flihtu (Ο. H. G.) 163 fodjan, 261 forscon (Ο. H. G.), 189, 192 fraihna, 170 frathjan, 203 fraujinon, 254 fregna (Ο. N.), 170 frigne (A. S.), 170 gahalgjan (A. S.), 256 gaigrot, 358 ginem (Ο. H. G.), 170 giisnjan (A. $.), 254 hafjan, 202, 203 halon (0. S.), 240. hlahjan, 203 hlinen (0. S.), 231 hlinian (A. S.), 231 holen (Ο. H. G.), 240 ita, 145 jesan(O.H.G.), 146,520 kausjan, 240 kiusa, 152, 155 knau (Ο. H. G.), 526 kustus, 155 laikan, 227 leihva, 152, 154 leskan, 189 lubaith, 246 lustus, 146 man, 378 merjan, '230 mikiljan, 256 misciu(O.H.G.), 189,192 naan, n&jan, nawan (Ο. H. G.), 148 • namnjan, 239 uasi-da, -dedum, 507 niwon (Ο. H. G.), 231 og, 378 quillu,qual (O.H.G.), 211 raska (Ο. N.), 189 rasko (Ο. H. G.), 189 rathjan, 203 rinnan, 109, 172 saisost, 37 saisoum, 389 sandjan, 106 satjan, 321 scawon (Ο. H. G.), 240 sinnan (A. S.), 106 skaiskaid, 356, 357 skapjan, 203 skathjan, 203 sokidedum, 441 spehon (Ο. H. G.), 240 staiga, 155 staistagg, 357 staistald, 357 stan (Ο. H. G.), 100 steiga, 152, 155 straujan, 251 tagrjan, 250 teiha, 152 tiuha, 218 3 toumen(O.H.G.), 231 thagjan (O.S.), 240 thahan, 240 thaursnan, 253 thlasjan, 240 thriskan, 189 thulaith, 235 usskavjan, 342 usthriuta, 152 vagjan 240 vahsjan 204, 445 vaist, vait, 34, 153, 371, 379, 383 verstan (Μ. H. G.), 100 vigam, 138, 139 . vitu-m, -th, 389 wunscian (O.H. G.), 189, 192 (bis), 198 ziljan (Ο. H. G.), 212 a. Lithuanian. alkstu, 163 ariu, 202 auginti, 173, 181 augti, 173, 445 aukszt :ts, 445 baid^ti, 507 b6gu, 152 bezdu, 146 F. LETTO-SLAVONIC. bijoti, 507 bluz-ni-s, 175 budinu, 180 bugstu, 163 M-k, 450 bundu, 173, 180 bitsiu, busite, 471 dirin, 203, 213 dristu, 163 d&kit^, 404 &lmi f 104 eimi, 97, 171 eitu (Low Lith.), 163 esmi, 101 gasti, 163 gelbu (future, gelbesiu), 259 jeskoti, 189, 192 jodamas, 607 kanku, kakti, 175 INDEX. 585 kavoju, 240 keptas, 512 laidoju, 202 laukiu, 203 leku, likti, 154 liiitu, 163 lydinti, 176, 507 marinti, 173 mirsztu, 163 mirti, 173 pfdinti, 176 plaku, 219 plauju, 210 sedzu, 202, 203 sirpstu, 163 spiriu, spirti, 295 tenku, 180 nka-s, 174 5dzu, 203, 223 velku, 85 verdu. 507 vesdinti, 507 vesti. 507 vein, 29 b. Church Slavonic. bada, 467 bera, 145 bSga, 152 bichu, 438 dltatl, 151 ouvaja, 240 davu (δίδωκώί), 425 dSja, 247 d&aja, 238 druzati, 218 ida, 507 iskati, 189, 192 jada, 507 jam!, έιηϊ, 104 jasu, 438 kupuja, 253 melja, 203, 211 nesu, 278 otubSgnati, 174 plsa, 221 pleta, 163 plov-ja, a, 210 pola£a, 180 porja, prati, 214 rastetl, 163 saditi, 231 stana, 170 stati, 100 stlza, 155 subiraja, 240 suchft, 155 supatl, 151 tuknati, 180 veza, 29 νΜέ, 391 vMSti, 260, 264, 391 zin^, 170 z^dati, 181 ilutejeti, 244 znaja, 202 G. KELTIC. Old Irish. ad-chon-darc, 399 ailigim, 256 birt, 281 caraim, 238 carub, 507 cechan, 389 cechladatar, 358 cechnatar, 389 cechuin, 389, gab-si, 427 gegrannatar, 358 g^n-sam, 427 insadaim, 231 ne-naisc, 400 rodamatar, 389, sescaind, 358 sescaing, 358 10» DON : PRINTED BY 8POTTI8WOODE AND CO., NEW-STREET 6QCARF AND PARLIAMENT STREET QQ ο- i