p 
 
 658 
 F68 
 1896 
 MAIN 
 
 UC-NRLF 
 
 B 4 02D MSb 
 
 IV 
 
 
 
LIBRARY 
 
 OF THE 
 
 University of California. 
 
 GIKT OF^ 
 
 Class 
 
THE NEGATIVES 
 
 OF THE 
 
 INDO-EUROPEAN LANGUAGES 
 
 A DISSERTATION PRESENTED TO THE FACULTY OF 
 
 ARTS, LITERATURE, AND SCIENCE, OF THE 
 
 UNIVERSITY OF CHICAGO, IN CANDIDACY 
 
 FOR THE DEGREE OF DOCTOR 
 
 OF PHILOSOPHY 
 
 BY 
 FRANK HAMILTON FOWLER 
 
 CHICAGO 
 
 €iit UniUxMtii of ari)trago ^uw 
 
 T896 
 
THE NEGATIVES OF THE INDO-EUKOPEAN 
 LANGUAGES. 
 
 
 PRIMITIVE INDO-EUROPEAN. 
 
 I. I.E. me: Skr. md, Av. md, O.P. md, Or. fxrj (Doric ix-rj), Arm. 
 mi, Alb. nio- (in mo-s). 
 II. I.E. «(?.• Skr. na, Goth, ni, Lith. //<?, O.B. ne, Lat. «<?, Osc. 
 ne, Umbr. ne. I.E. «^.- Skr. nd, Lat. //^, Osc. ;//, Goth, ne, 
 O.Ir. «/. I.E. hD: Lat. «^«, Umbr. no- (in no-sve), O.Ir. «« 
 «a. The Skr. words given under ne and «^ and various com- 
 pounds in the Iranian languages may, of course, be connected 
 with either nc or no, or with both. 
 
 III. I.E. nei: Av. //ae- (in fiae-cis, nae-da), Lith. net ne-, Lat. nei 
 nl, Osc. nei, Umbr. nei- (in nei-p), probably also in O.P. naiy. 
 I.E. noi: Lat. «^v- (in noe-num and in Duenos noi-si and 
 noi-ne {})), Lith. «^?/- (in nai-kinii). As in the case of I.E. 
 «^ and «^, forms in Aryan from I.E. nei and noi cannot be 
 distinguished. 
 
 IV. I.E. n: Skr. a- an-, Av. a- an-, O.P. a. Arm. a«-, Gr. d- 
 av-, Lat. /«-, O.Ir. ««-, Germanic un-. I.E. //.• Gr. vtf (Dor. 
 va-), Osc. a;/-, Umbr. an-, O.H.G- a-, O.E. ^-. 
 
 Of these I.E. negatives, without doubt fiei and «f/ are ne and 
 «^ strengthened by the particle t, seen in various places elsewhere.' 
 I.E. net and noi, then, are examples of a phenomenon — the 
 strengthening of a negative — which was often repeated in the 
 separate developments. 
 
 The relation of ne ne no no is plainly one of ablaut ; and it 
 would seem most natural to consider ;/ n as the weak ablaut forms 
 of the same particle.* The prevailing use of n n in the separate 
 languages as negative prefixes indicates that those forms were in 
 the proethnic speech the negatives of nouns. Now it is easy to 
 
 ^ E. g., in pronouns (Lat. quoi qui), in the Locative suffixes singular and 
 plural {-i, s-i beside s-u); probably also the same particle differentiates the 
 primary from the secondary verb endings. Cf. Brugmann, II, 414, 424, 256, 
 356, 909; O. Brugmann, Ni., 32; Per Persson, I.F., 2, 247 ; Thurneysen, K.Z., 
 27. 177- 
 
 ''So Schulze, K.Z., 27, 606, and otliers. 
 
2 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 understand that there should be differences between the prevail- 
 ing accentual relations in combinations of negative and verb 
 and of negative and noun which should give rise to different 
 ablaut forms of the negative particle. According to Delbriick' 
 this difference is to be connected with the enclisis of the verb 
 in independent sentences. But Zimmer' has argued conclu- 
 sively, I think, that the condition of verbal enclisis in I.E. was 
 not position in an independent sentence, but was simply a matter 
 of sentence accent. It would still be the case, however, that in a 
 collocation of nep'ative and verb the verb would generally be 
 enclitic, and the result of this is clearly shown in the conditions 
 of verbal enclisis in O.Ir. The only objection to considering 
 the various negative prefixes mentioned above as representatives 
 of I.E. // and // and these as weak ablaut forms of )ie is the pres- 
 ence of such dissyllabic forms as Av. ana-, Gr. dva-. But these, I 
 think, are to be disposed of otherwise (see p. 8). 
 
 The forms // // were the only ones used with nouns in the pro- 
 ethnic speech. With the exception of the Balto-Slavic, none of 
 the languages has ne, ne, no, or no used as a negative prefix.^ The 
 common occurrence of compounds of a negative with an indefinite 
 adverb or pronoun (<?. g., Skr. iid-kis, ma-kis; Av. nae-cis ; Gr. 
 /AT/'Tts, ix-q-ifixy, ixrj-8€t<;, ovScls ; Lat. iicjno, nun-qiiani, ne-cubi; Eng. 
 none, never, etc.) is no evidence for the use as a prefix of the neg- 
 ative so compounded.'' A sharp distinction is to be made between 
 such compounds and compounds like avtWos, intactiis, unhappy, 
 etc. With these indefinite compounds the negative prefixes of a 
 given language never occur. This is seen most clearly in Balto- 
 Slavic.^ In other languages representatives of I.E. ne may appear 
 compounded with indefinites, but here where alone I.E. ne becomes 
 a negative prefix it is not so used. The indefinite compound arises 
 from collocation in the sentence. In a sentence containing a 
 
 'Syn. Forsch., IV, 146. 
 
 'Festgruss an Roth, 173; cf. Hirt, Akzent, 304 f. 
 
 3 The very few words in other languages which have sometimes been men- 
 tioned as showing tie- used as negative prefix I shall speak of under the several 
 languages. 
 
 ■» Hirt (Akzent, 312 f.) has most recently fallen into this error of citing 
 indefinite compounds in support of the supposition that the form ve was used as 
 a prefix. 
 
 ^ Cf. Lith. nevens, "non unus sed plures," and nei vi'tts. "nullus; " and see 
 Gehauer, Archiv f. Slav. Thil., 8, 177 ff. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 3 
 
 sentence-negative and in which the verb has for its subject an 
 indefinite pronoun or is modified by an indefinite adverb the 
 effect is the same whether the negative is felt with the verb or 
 with the indefinite subject or adverb. The same is not true if 
 the subject, for example, is other than indefinite.' 
 
 So far as I know, the only alleged example of a proethnic 
 noun showing ne- as a negative prefix is nepot- (Skr. napdt, Gr. 
 v€'7roSes, Lat. ncpos, etc.""), regarded by many as a compound of He 
 and pot- (Skr. pati-, Gr. Trocrts, Lat. potis, Goth. fadi). But the iso- 
 lation of the word in this respect should make us skeptical of an 
 etymology, which, moreover, on its semasiological side is none 
 too convincing.^ 
 
 I prefer to start with a root nep-^, with meaning "to bring 
 forth " or something of the sort, to which was added a suffix 
 consisting of a dental preceded by a vowel. With this root is 
 to be connected the root )ieb-, in Gr. ve(3p6'5, which I shall have 
 occasion to discuss below. 
 
 As for the suffix -od or -ot, which we have to assume if we take 
 the root to be nep-, we may suppose that the suffix was -od, and 
 that the -d- stem became a -/- stem owing to the scarcity of the 
 -d- stems, except in Greek where -d- stems were more numerous.^ 
 
 ' Three other classes of compounds in which negatives occur might be 
 mentioned, viz., strengthened negative adverbs, e. g., I.E. net not, Skr. na-ni'i. 
 Gr. oi}-/ct, etc.; compounds in which for the most part the original force of the 
 two elements is retained, e. g., Lat. 7ieque, Goth. 7ti-h ; compounds which are 
 in origin elliptical expressions, e. g., Lat. nlmJrum, Skr. ndsti-ka. 
 
 '^ Cf. Prellwitz, sub v. vdirodes ; Osthoff, Perfect., 599 ; J. Baunack, Studien, 
 272 ; Leumann, Festgruss an Bothlingk, 77 ; Streitberg, LF., Ill, 334 ; de Saus- 
 sure, Mem. Soc. Ling., Ill, 196 ; Brugmann, II, 123 ; and see Panini, V, 3, 75, 
 for the origin of the derivation. 
 
 3 Skr. napdt tiaptr Grassmann defines as i ) Abkommling, 2) Sohn, 3) Enkel ; 
 the Av. napaf means "descendant," Av. napti "posterity;" O.B. netij'i " Ge 
 schwistersohn ; " Bohemian tieti " niece ; " O.H.G. nefo "grandson," " relative ;" 
 Lat. «i'/Jj "descendant," "nephew," "grandson;" Lat. «ir//w " granddaugh- 
 ter;" Gr. I'eVoSes "children;" Gr. a.ve\f/i6s "nephew" (see Prellwitz 5«^ v.). 
 The meanings of the words in the various languages ("nephew," "grandson," 
 "descendant") indicate that at the end of the I.E. period the word had the 
 meaning "descendant." The usual etymology leads to the meaning "weak- 
 ling." The supposition that the idea of " descendant " was derived from that 
 of "weakling" is, of course, possible, but hardly satisfactory. 
 
 * Cf. Spiegel, K.Z., 14, 392. 
 
 5 6/. Brugmann, I, 128; Osthoff, Perfect., 159. The presence of iruis in 
 Greek might exert an especial influence in keeping the -d-. 
 
4 NEGATIVES OF TIJE INDO-EUROPEAN LANGUAGES 
 
 The suffixal -</-, though never very common — except in Greek — , 
 is found after various vowels': -ad, Gr. <^vy-a8-; -id- (probably 
 I.E. in some cases), Gr. 6.<jit-&-, Lat. capid; -ud, V,-aX. pal-ud- ; iid, 
 Lat. pec-ud-; ed or ed, Lat. her-ed-. A suffix -od- would not be 
 strange, though unsupported by more than this one example. It 
 is to be noted, however, that three or four words in Sanskrit — 
 drsad "rock," hhasdd "back parts," (arad "autumn," and 
 vafidd{}) "desire" — may have I.E. -f^^/, and that the Germanic 
 
 words with -at e. g., Goth, laiih-at-jaii, O.H.G. lohazzen — 
 
 speak as much for I.E. od as -ad-, unless indeed -at-ja/i is to be 
 taken as equivalent to Gr. -a^civ. To be mentioned in the same 
 connection is the common Goth, and West Germanic sufifix -assiis.'' 
 Or, on the other hand, the suffix may be an example of the 
 class, limited in number, of suffixes consisting of -/- preceded by 
 a vowel, e. g., Skr. sravdt-, mariit; Cir. KiXii'i, KeX-riT-os \ hat. /rg^s, 
 scges, caput {^cap-ot); Osc. liimitu[m ; O.Ir. cing-, gen. ciiiged; 
 Goth, mitaps; O.H.G. liclid? In this case we must suppose that 
 the -/- was changed to -8- in Greek under the influence of the -8- 
 stems, as Osthoff has claimed. 
 
 THE NEGATIVES IN THE SEPARATE LANGUAGES. 
 
 SANSKRIT. 
 
 Skr. md, I.E. 7ne. 
 
 The simple negative was strengthened by various particles — 
 liim, n, sma — , the md keeping its distinctive meaning. With 
 -kis it formed md-kis, "nequis." 
 
 Skr. na, I.E. ne. 
 
 Skr. nd,'' I.E. )ie. 
 
 Skr. ned is a compound of na and id (as ced from ca -\- id), 
 the latter undoubtedly a particle from the pronominal stem i- 
 seen in Cyprian iV (Hesychius), O.Lat. Im, and in the deictive 
 
 ' Biugmann, II, 123. 
 
 "^ Cf. Brugmann, II, loS; Bahder, N'erbalabstracta, 119. 
 
 ^Cf. Brugmann, II, 123, for other examples, and Walter, K.Z., X, 194 ff., in 
 regard to Latin eqties, miles, etc. It can hardly be that these words are com- 
 pounds of -/-/-, "going," and even in case of coin-es and anti-stes the supposition 
 is dout)tful. Cf. Wharton, s. v. comes. See also de Saussure, Mem. Soc. Ling., 
 Ill, 197. 
 
 •»A. ^r., K.V., 10, 348. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 5 
 
 particle / mentioned above under I.E. nei. The compound was 
 Aryan (jiaid, Av. noit). Skr. na-klm, na-hi, nanu show iia 
 strengthened by various particles. 
 
 Skr. na was not used as a prefix. Of Panini's lisf napdt has 
 been considered above. The rest are either altogether fanciful, 
 as nakula, "ichneumon," or manifestly impossible, as navedas, 
 "knowing," with the exception of napuhsaka, "hermaphrodite." 
 But I hardly think the word can stand alone as an example of 
 the use of na- as a prefix. The na- is probably the same as that 
 seen in na-vedas, whatever that may be, and the quasi-negative 
 force comes from the diminutive. 
 
 In regard to the relation of na comparative and na negative 
 in Sanskrit, taking into consideration the existence of similar 
 pairs in Balto-Slavic and O.Ir., we should sav, I think, that, either 
 in proethnic times or in the separate languages, the comparative 
 was developed from the negative.^ Cf. dialectic English "better 
 nor that." 
 
 Skr. cand also has generally been taken as a compound of na 
 negative.3 But Per Persson is surely right here in supposing that 
 the negative meaning is a derived one. Whatever the order of 
 development of meaning of cand, the uses may be classified as 
 follows : 
 
 {a) Intensive: i) In positive sentences, "indeed," 2) with a 
 negative, "(not) indeed." 
 
 {b) Negative: i) Strong negative, "indeed not," 2) conjunc- 
 tional negative, "also not," "and not." 
 
 Manifestly three explanations are possible : 1) That cand, a 
 compound of ca and na, negative, was used, first as a negative, 
 then as a second negative intensifying the first, and lastly from 
 being a negative intensifier cand came to be an intensifier that 
 could be used in a positive sentence. 2) An explanation, the 
 direct opposite of this, that cand, a compound of ca and na 
 intensive, was used first as an intensive particle and especiallv to 
 intensify a negative, and finally as an adverb with a negative force 
 
 "Panini's list (6, 3, 75) is 7ial>hraj, napat, nasatyd, namiici, nakula, nakha, 
 napunsaka, naksatra, nakra, navedas, and uaka. 
 
 ''Cf. B. and R., sub v.; Grassmann, sub v.; Delbriick, Syn. Forsch., 5, 543; 
 Pott, Ety. Forsch., i, 352; and Per Persson, I.F., II, 203. 
 
 3 Grassmann, sub v.; Eva Channing, J.A.O.S., 13, XCIX ; Delbriick, Syn. 
 Forsch., V, 544; Per Persson, I.F., 2, 204 ; and cf. Brugmann, II, 421. 
 
6 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 of its own. 3) That cand intensive and cand negative were of 
 different origin. 
 
 The very frequent use of cand to intensify a negative (in R.V. 
 57 out of 86, in A.V. all of the 39 cases) would indicate some 
 sort of a connection between the negative and intensive forces of 
 the word. Against the first explanation it is to be said that a 
 composition of ca and na negative in the order ca-ua is hardlv to 
 be expected; and again, it is difficult to see how cand used as a 
 quasi-intensive particle with a negative, with the meaning "(not) 
 . . . . indeed not" — supposing it to have had that meaning 
 originally — could acquire a purely intensive (positive) force, 
 without being used as a true negative expressing with the pre- 
 ceding negative a simple strong negation. But a collocation of 
 two negatives in the sense of one probably does not occur in 
 Sanskrit,' and in fact a "not — not," such as this would be, is 
 impossible anywhere.' 
 
 ' Delbriick, Syn. Forsch., V, 544; Eva Channing, J.A.O.S., 13, XCIX. 
 
 'A distinction is to be made between collocations of true negatives with 
 the meaning of one and those which historically are collocations of negative 
 and int^sive, the intensive having acquired a negative force. True double 
 negatives are separated in their application. In general one negatives the sen- 
 tence, while the other is felt with an important word. The most common case 
 is that in which a second negative is used to negative an indefinite pronoun or 
 adverb, e. g., A^o one shan't do it; iVon miseret neminis, Enn. Erecth., frag. 
 4; Jura te nocitnruni nan esse hoinini de hac re nemini. Plant , Mil , 1411 ; to. 
 ■y' OX) k4 Tis ovd^ tdoiTo, Od., 8, 280. Very often the negative and the indefinite 
 form a compound. But a negative may be felt to belong to any important 
 word and so two negatives be admitted, e. g., O.E., TAer nys no table; N^eqiie 
 nucleis ad oleam ne utatur, Cato, R.R., 66. The need felt for a second negative 
 may be increased if one of the negatives is a conjunction, if the negative is 
 compounded with a verb, or if a number of words intervenes between the first 
 negative and some important word. But these causes act indirectly, and the 
 second negative is still felt with some important word. Of a different sort are 
 double negatives arising from syntactical contamination, e. g.. Forbade the boy 
 he should not pass those grounds, Shaks., Passionate Pilgrim. For other exam- 
 ples of contamination cf. Strong, Logeman, and Wheeler, 155, and for other 
 examples of true double negatives cf. Zimmer, Streifzuge, 90 ; Eva Channing, 
 J.A.O.S., 12, XCIX; Gebauer, Archiv f. Slav. Phil., 8, 177; Kent, Pub. Mod. 
 Lang. Asso., 5, 190; Richardson, Harvard Studies, I, 154; Spurrel, Welsh 
 Grammar, 158; Habich, De neg. usu Plaut., 6; Lucilius, ed. Miiller, 241. But 
 the collocation neqtie . . . haud can neither be classed with the examples of 
 contamination, as Strong, Logeman, and Wheeler, nor with neque . . . num- 
 quam, as Habich would have it (see below p. 27). Habich's separation of 
 neque . . . nuniqiiain from cases like non . . . nemini seems to me to be an 
 arbitrary classification. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 7 
 
 In favor of the second explanation, that proposed by Per 
 Persson, are the numerous analogies furnished by negatives 
 which have become such from being used to intensify negatives. 
 The fact, too, that the use of cand as a positive intensive dies out 
 in the later Vedic period, except with derivatives of ka, indicates 
 that the development was from positive to negative. It may well 
 be, however, that the resemblance in form of cand, "indeed," and 
 ca . . . na, "and not," assisted in the development of meaning 
 of the former. The correspondences of ca/id elsewhere speak 
 strongly for the positive as the original force of the word ; com- 
 pare Av. cij/a, O.H.G. -gin, "irgend," O.N. -gt.^ Only in the last 
 does a negative meaning appear, and there the other Germanic 
 dialects show the negative meaning to be a developed one. 
 
 Skr. nil as a negative was mostly used with cid (iiu cid, 
 "never "),/. ^., y/// was a temporal negative adverb. The more 
 common and doubtless earlier meaning of nu was "now," with 
 which meaning, or with one easily connected therewith, its cog- 
 nates appear in various languages : Gr. vv, Lat. nu-(dins), O.Ir. 
 nu no, O.H.G. nii no, Lith. nu-gi, and in Umb. nurpener'^ (from 
 mc-arpener). In Sanskrit it is found used with a force easily 
 derivable from that of the temporal adverb, viz., as an intensive : 
 nd nu nami, "surely not."^ It was undoubtedly from this use as 
 an intensive that the negative force came to be attached to the 
 nu* in the same way in which Yxtnch. pas became negative. Other 
 analogies will be given under the Greek negatives. 
 
 Skr. a- an- (negative prefix), I.E. //- nn-. 
 
 AVESTAN. 
 
 Av. md, I.E. me. 
 
 Av. mott is formed by the addition to md of the particle //, as 
 in the case of noit (Skr. ned, Aryan naid). Probably, however, 
 
 ' Cf. Brugmann, II, 241 ; Hiibschmann, K.Z., 24, 328, n. 2 ; Bartholomae, 
 Arische Forsch., II, 126; Jackson, Av. Gram., §30; Per Persson as above; 
 Scherer, Z.G.D.S., 475. But Goth, -huti is probably from I.E. stem q7t-; J. 
 Schmidt, K.Z., 32, 402. 
 
 ^Brugmann, Osk. und Umbr., 225 f. In regard to Umbr. nosve see below 
 (P- 29). 
 
 ^ E. g., R.V., I, 165, 9, ndki?- nit na tvdtidtti asti. For the use of the tem- 
 poral adverb as an intensive cf. Eng. 7iever emphatic for no. 
 
 "^But cf. Kretschmer, K.Z., 31, 365. 
 
8 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 it was formed after the analogy of noit rather than by a direct 
 union of the elements ma and //. 
 
 In Av. i/uuta is to be seen md -j- a pronominal stem (ta (I.E. 
 da), as in iiae-cta (I.E. nei or noi A;- da). 
 
 I.E. //r probably does not occur in Av. except in compounds. 
 The single place in which it has been supposed to occur is Yasna, 
 44, 19. Tradition, however, has taken the word nd here as the 
 nominative of nar, "man," and it seems hardly probable that 
 this negative should occur once and only once in the entire lit- 
 erature. The common «<?// appears in the preceding line. The 
 absence of the uncompounded nd in O.P. also argues somewhat 
 against the supposition of its appearance here.' 
 
 The common negative both in the Gathas and later Avesta is 
 noit^, corresponding to Skr. ned. 
 
 I.E. ne further appears in the compound nava of the later 
 Avesta and navdt, "minime," "neither," of the Gathas. 
 
 A v. nae-, I.E. nei or noi. 
 
 Av. naetta (Gatha nai'da) is I.E. nei or noi strengthened by -da"" 
 (cf. madd). It could hardlv be the more common /loit with added 
 -a, as this is properly a postposition which becomes attached to 
 cases. ^ A remnant of the (probably Avestan) use of nae as an 
 independent negative adverb is seen in nae-cis, "no one." 
 
 Av. a- an- (negative prefix), I.E. n- nn-. 
 
 In a few cases also a prefix a/ia- occurs — in ana-lTareta, ana- 
 h'arepa, anazdpa, anamarezdika. If, as has been claimed above, 
 the I.E. negative prefixes ;/ n, indicated by the ordinarv negative 
 prefixes of the various languages, are simply the weak ablaut forms 
 of I.E. ne ne no no, then, it is plain, we must seek an explanation 
 for Av. ana- other than as a variant form'' of I.E. // n. The dis- 
 svllabic forms of the several languages must be considered together. 
 In Ossetan^ the common negative prefix is ana-, though a- also 
 occurs ; in Gr. we have a few cases of dva-, dva/reATTTos, dvd/reSvos, 
 etc.; in Prakrit Goldschmidt* has noted several cases of ana-; in 
 
 ' Cf. Jolly, Ver. Syn., 33 ; Spiegel, Conim. 2, 358. 
 » Cf. Per Persson, I.F., 249. 
 
 3 Jackson, Av. Gram., 222; Johansson, B.B., 20, 96. 
 
 ^Cf. Johansson, B.B., 15, 310; J. Schmidt, K.Z., 23, 273 ; Kretschmer, 
 K.Z., 31, 408. 
 
 sHiibschmann, Osset. Spr., 21. 
 'K.Z., 24, 426. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 9 
 
 O.H.G. we have the single case of una- in una-Jwlda; and Zim- 
 mer' has claimed that ana- appears in the Celtic an-. The 
 O.H.G. una- indicates that the first element was I.E. nn-, and the 
 simplest explanation of the presence of the -a-, in most cases at 
 least, is that it has been "clipped" from the nouns having a- 
 initial. The process may have taken place in proethnic times or, 
 more probably, in the separate developments. An example of 
 such a clipping may be seen in Mod.Ir. am-h-.^ For the Celtic 
 Zimnier supposes that ana- arose before sonants [cf. O.Ir. bunad, 
 Lat. fundus). 
 
 OLD PERSIAN. 
 
 O.P. md, I.E. 7ne. 
 
 O.P. mdtya, "that not," is md -\- -tya, the neuter accusative of 
 a relative stem.^ 
 
 O.P. naiy, I.E. net or noi. 
 
 O.P. naiy may be, however, from Aryan naid (Skr. ned, Av. 
 noih. The presence in Iranian of I.E. nei not, in Av. naeda and 
 nae-cis, favors the connection with I.E. nei or noi. 
 
 O.P. a- (negative prefix), I.E. //-. 
 
 ARMENIAN. 
 
 Arm. mi, I.E. me.'' 
 
 This is the only one of the I.E. negatives appearing in Arm. 
 
 The ordinary negative in Arm. is oc, and the modern language 
 has also ce. The connection^ of Arm. o( and Gr. ovk has been 
 disputed by Hiibschmann. Lagarde, however, connected the 
 two negatives, and Bugge has done the same, seeing no difficulty 
 in the Arm. -^ and deriving the Arm. o- from o- from au, which 
 he takes as an ablaut form* of the ou in Gr. ovk. 
 
 Inasmuch as Gr. ovk is made up of ov and an added element 
 -k{i), the consideration of the connection of the two negatives 
 
 'K.Z., 24, 523 f. 
 
 ^ Cf. Zimmer, K.Z., 24, 536. In Gr. avdivoLvos, v^voivos, diroii/a, and rrotv^ 
 are to be noted. 
 
 3 Spiegel, Keilinschriften, 1S2; ^Yhitnev, 499 <?. 
 
 ^Brugmann, I, 71 ; Hiibschmann, Arm. Studien, 46 and 61. 
 
 5 Cy. Hiibschmann, Arm. Stud.; Lagarde, Arm. Stud.; Bartholomae, Indg. 
 Studien, H, 20 f.; Bugge, K.Z., 32, 30. 
 
 ^ Cf. Brugmann, H, 114; Wackernagel, K.Z., 29, 141 ; Solmsen, 2'iid., 92; 
 and Prellwitz, s/t/>. v. ov ovk. 
 
 
 ( uNlVERSi 
 
10 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 really involves two questions, viz., as to the connection of Arm. 
 0- and Gr. ov, and of Arm. -( and (Ir. k(i). 
 
 Bartholoniae has shown that Arm. "(. may be from qi, but not 
 ki. So -"( in o"( might well be for -*qi, used before words begin- 
 ning with a vowel, and hence to be connected with the I.E. 
 interrogative indefinite stem qi and to be compared quite directly 
 with the Skr. indefinite particle cid, Av. cii, O.P. ciy. The ques- 
 tion, then, as to the connection of Arm. -f and Gr. -K(t) will 
 depend upon what we consider that Gr. -ki represents. 
 
 As for the connection between o- and Gr. ov-, Bugge would 
 suppose that Arm. o represents I.E. au-, and that the relation of 
 the two particles is one of ablaut. I shall try to show that Gr. 
 ov belonged to the e:o series. 
 
 Bugge's law of Arm. o from I.E. aij would not, of course, 
 exclude one of o from I.E. oii (or eij). Hiibschmann, however, 
 has given — with some hesitation — the representation of I.E. oij. 
 and eif in Arm. as oy, and Bartholoniae has argued for this view 
 at some length. But the question is not so thoroughly settled 
 that we can overlook the comparison of Arm. o(, "not," with Gr. 
 ovK(t), "not," as tending to establish a law of Arm. o from I.E. 
 oij. Further to be taken into consideration is Arm. jbuel, beside 
 Skr. hdvanam, Gr. xocfos. And again, if Arm. sork, Skr. (uskas, 
 Gr. awXeos, and Arm. ostin, Gr. aiio-raAcos, belong to the a:o 
 series, why may not Arm. sork and osiin represent the ci-grade ?' 
 
 GREEK. 
 
 Gr. /Ar/ (Elean ju.a, Boeotian /xei), I.E. me. 
 
 With -Se iirj formed a compound /at^Sc, meaning either "and 
 not" or "not even" {cf. orSe). A number of indefinite pronouns 
 and adverbs were formed from (i) /A17 and (2) /nr/Se. The second 
 element was «/xds, els,^ t«, ercpos, or an indefinite adverb. 
 
 ■ Cf. Arm. loganol, Gr. \o{p)iu, Lat. lavere; B.B., 17, 123. and Thurneysen, 
 K.Z., 28, 154. Bugge's examples are as follows (K.Z., 32, 29): (i) sork, Skr 
 t;uskas, Av. hu'ska, O.P. uska, Lith. sausas, O.B. sitchu, O.E. sear,Qx. ava\^os ; 
 (2) os/in, Gr. av<TTa\4os, a{>aTr)p6s; (3) doc, Gr. irKpavcKu; (4) oc, Gr. om, cf. Lat. 
 hand (:^aud?); (5) p'ok'r, Lat. paucus. Cf. further Bartholomae, B.B., XVII, 
 100 f.; Hiibschmann, Arm. Stud., 59, 62, 78; Osthoff, Perfect, 484 ff. 
 
 "Breal (Mem. Soc. Ling., I, 205) thinks that /urjSeis and oi)5e/s contain the 
 pronominal stem do-, a supposition quite possible if we consider /i7?5e/s and 
 ovdih by themselves, but considered in connection with /aTj5a/i(5s, etc., ixy)bb\m, 
 fj.T]diiroTe, etc., unlikely. It would seem that, whatever the origin of ovdels and 
 ^TjSe/s, odeTva is to l)e in some way connected (cf. Brugmann, Gr. Gr., § 94.) 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES II 
 
 I.E. >ie does not occur in Greek as a negative adverb or as a 
 negative prefix. The words so explained by J. Baunack' are to 
 be accounted for otherwise. The list is aveu, dve/x,wAtos, vUrap, 
 vefipos, and veVoSes. The last word has already been discussed ; 
 dvev, which, as Baunack supposes, contains two negatives — a 
 thing thoroughly improbable — , will be discussed later on (p. 33). 
 As for di/£/AwAtos it is impossible here to support the presence of 
 two negative prefixes in the sense of one. A comparison with 
 avd€8vo<; dvdeATTTos is useless, for here, too, we cannot suppose two 
 negatives, and the prefix dva- is to be explained otherwise.^ The 
 word seems to be connected with dvc/xos.^ As for veKxap, the word 
 is not I.E., but a Semitic loan word (Semitic niqtar).'' With 
 ve/3po<;, which Baunack connects with /3opa, must be considered 
 ve/3pa$, "fawn," "a young animal," ve^prj (with 8opd), "belonging 
 to a fawn," ve^pL<i, "a fawn skin." The absence of etymological 
 connection with any word for "deer" and the more general 
 meaning sometimes attached to ve^pai^ indicate that the original 
 meaning was a more general one than "fawn;" but that the 
 word was originally an adjective there is no indication. Probably 
 we have in ve/3po<i a -ro- suffix and the other words formed from 
 this by analogy — v€fipa$ after Troprai, ve'yS/ois after Trdprts, and the 
 meaning later changed from "fawn" to "fawn skin" (used as a 
 garment). Now if we see in ve/?pos ("*descendant," "*offspring") 
 a root ned-, we shall be inclined to connect this with the root uep- 
 which I have claimed appears in vcVoSes, etc. A root not difficult 
 to connect with these in meaning, but with a still different con- 
 sonant, is //efi/i-, Skr. /md/i-, "to rend asunder," nabhaiia, "a 
 spring."* 
 
 'Studien, 271 f., and cf. Brugmann, Gr. Gr., 223; Wheeler, CI. Review, 
 III, 130, and Olavsky, "Die nhd. Partikel nicht, etc.," reviewed by Michaelis, 
 K.Z., VI, 309. 
 
 "See p. 8. 
 
 3 Cf. Prellwitz, sub v. dve/xos. 
 
 ^ Cf. Keller, Lat. Volksetymologie, 226, and the index ; Muss-Arnolt, Trans. 
 Am. Ph. Assoc, 23, 143. 
 
 5 Cf. Hesych. vi^paKcs ' oi dppeves v€ttol tGiv dXeKTpvovwv. 
 
 * For variations in root determinatives between/ — /> cf. Skr. vepate.- Lat. 
 vibrare. Per Persson, Wrzlw., 49, but cf. Fay, A.J. P., 73, 481. Variation between 
 /, h, hh is to be seen, e. g., in std-b-, sta-bli-, sta-p-. Per Persson, 59. In regard 
 to nep- and nebh- cf. Spiegel, K.Z., 13, 370, and 19, 392. See also Brugmann, 
 I, 469, 7. Lat. tiefrens, if it means "young," may also belong here, in which 
 case we should have the roots nep-, neb-, nebh- in three words meaning "offspring." 
 
12 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 The coniinon negative in Greek taking the place of the lost 
 I.E. lie is ov, the origin and connection of which has never been 
 very satisfactorily explained.' It was suggested by Bopp ^ that 
 Gr. ov is connected with the pronominal stem seen in Av. ava. 
 He conceived that not only Gr. ov, but Skr. iia and other nega- 
 tives as well, had developed their negative meaning from a 
 demonstrative one of remoteness. 
 
 The most serious attempt to find a derivation for ov has been 
 made by Henry,^ who attempts to connect ov directly with the 
 Skr. preposition ava-. To make this connection Henry sets up 
 oua- as the I.E. form of the preposition, and by so doing is com- 
 pelled both to reject Brugmann's law of the Skr. treatment of 
 I.E. in open syllables,^ and, on account of Lat. au-, to accept 
 Thurneysen's^ law of oij to au in Lat. That a prefix with a 
 "sens inversif " may become practically a negative prefix mav be 
 admitted, and perhaps the best example^ is Lith. be-, O.B. bez{ji)-^ 
 used as a negative prefix, compared with Skr. bahis, " out," " out- 
 ward." But there is no example by an} means sure of a case 
 where such a prefix has become a true negative adverb.'' Henry 
 himself intimates that the strongest argument in favor of his 
 derivation is that no other seems to be at hand. 
 
 It seems certain that the negative meaning of Gr. ov is an 
 acquired one. This being so, it would be good method, if we 
 are to seek a derivation for ov, to note the ways in which such an 
 acquisition of negative force has been made in the case of other 
 words not originally negative. We may classify as follows : 
 
 I. Negatives formed by composition of I.E. me or iie with 
 
 'some other word do not concern us, except as they show that 
 
 with negatives intensives were often used which sometimes coa- 
 
 ' Grimm, Deutsche Gram., Ill, 759, and Pott, Ety. Forsch., I, 405-8, have 
 discussed the origin of the Gr. negative, but hardly in a way satisfactory to the 
 modern philologist, llartung's explanation (see Bopp) I have not seen. 
 
 ^Ver. Gram., Ill, §§3/1, 379, and Scherer, Z.G.D.S., 331. 
 
 3 Mem. Soc. Ling., 6, 378, and (/. Brugmann, Gr. Gr., § 164, and the Nach- 
 trage ; Planer, De Neg. Haud, and others. 
 
 ^See now Streitberg, I.F., III, 364. 
 5K.Z., 28, 154 f. 
 
 ^ C/. Lat. ex/ex, " lawless " (Lucilius, frag. 22, Miiller), Skr. vi-hasta, "with- 
 out hands," etc., Skr. tiir-bliara, " without measure," etc. 
 7 On Alb. -JT see p. 18. 
 
l/NIVERSITY 
 
 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 1 3 
 
 lesced with the negative. Examples are Skr. name, O.H.G. 
 iiiwiht, and in I.E itself nei. 
 
 2. Words of certain seniasiological categories may develop a 
 negative meaning from the original : {a) Words meaning "away 
 from," "other," or the like, e. g., M.H.G. anders, under, mere (in 
 sense of "weiter"), haz znd furdaz;"^ {b) comparatives or superla- 
 tives, or even positives, having a diminutive force, e. g., Lat. 
 minus and minime {minus especially in quo minus), O.E. med- in 
 med-wis, etc. It is to be noted here, however, that, possibly with 
 the exception of Alb. s (Lat. dis- ?), none of the examples shows 
 the development of a full-fledged negative adverb. 
 
 3. Words which from their meaning may be closely connected 
 with a negative may take on a negative force not by development 
 from their own proper meaning, but from association. These 
 words are more or less plainly indefinite pronouns or adverbs, 
 and at the same time more or less plainly intensives of the nega- 
 tive with which they are used." The following is a list of nega- 
 tive pronouns and adverbs which have become negative by asso- 
 ciation. Doubtless others could be added. ^ 
 
 In M.H.G.'' dekein, kein, and deiveder could be used either 
 with or without ;///// to give a negative force to the sentence ; 
 hence the negative force oi kcin and weder in N.H.G. In O.N. 
 en-ge, "Niemand" {cf. Goth, ni ains-liun), man{n)-ge, "Niemand," 
 hver-ge, "nirgends" {cf. O.S. /// hver-gin), and vaettr, "nichts," 
 but also "etwas" {cf. Goth. /// ivaihts). In O.Bohemian = srt;^;^)' 
 from meaning " desideratus " came to mean "nullus,"and kto 
 from meaning " aliquis" came to mean " nemo." \xv 0.\x. nach 
 na, "ullus" "aliquis," is also found with meaning "nullus." In 
 Welsh* nef, dim, and byih are either positive or negative indefi- 
 
 ' Paul, M.H.G. Gram., 125. In Albanian -s (-z) became a quite commonly- 
 used negative, and, if it were really from Lat. dis-, would be the best example 
 of this sort of development. Particles giving a bad signification may deserve 
 mention here. Cf. Skr. dtis-, Gr. dvs-, Arm. t-; O.Ir. 7?n-, Germ. Jtiiss-. 
 
 ^No line can be drawn between the two classes. It is to be noted, too, 
 that some nouns, e. g., res and homo, may be used practically as indefinite pro- 
 nouns. 
 
 3 Cf. Paul, M.H.D. Gram., 124. 
 
 * Cf. Grimm, Deutsche Gram., Ill, 720 ; Paul, M.H.D. Gram., 125; and 
 Hahn, M.H.D. Gram., § 435 f. 
 
 sGebauer, Archiv f. Slav. Phil., 7, i88. 
 
 *Spurrel, Welsh Gram., 159. 
 
14 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 nites. French rien, "nothing," (Lat. rem) got its negative mean- 
 ing from association. 
 
 Skr. ////, "not," >iu cid, "never," I have mentioned above as 
 takinir its resrular force from association with na. I have also 
 followed Per Persson in supposing a similar origin of the negative 
 force in Skr. cafid. Lat. hat/d, which I believe belongs here, will 
 be discussed below. French /^as'^ and point are well known as 
 examples of this phenomenon. Italian mica miga, Proventpal 
 Plica j/iiga mia, O.French mie'^ (Lat. mica^ show the same shift in 
 meaning as pas and point. Modern Gr. StdAoi; has its negative 
 meaning from use as intensive to a negative. 
 
 It is to be noted, then, that the use of intensives with nega- 
 tives is common; that sometimes this intensive united with the 
 negative to form a compound ; but that it quite frequently became 
 a negative itself; and that of negative adverbs not connected with 
 VIC or ne, and whose etymology we know, all, with the possible 
 exception of Alb. -s, came into use as intensives to negatives.^ We 
 should do well, then, to suppose that the same thing took place 
 in the case of Gr. ov which we know took place in the case of 
 French pas.'^ 
 
 In searching for a connection for Gr. ov among words that 
 could be used as intensives we will hardly find the object of our 
 search among the names of diminutive objects {passus, whit, etc.), 
 but rather among more generally used intensives, more or less 
 closely connected with pronominal stems like Skr. //// and I.E. 
 'i. Such a particle seems to be at hand in the Skr. u? 
 
 ' In the Creole of the Antilles French ne has entirely disappeared, and pa by 
 itself is the ordinary word for "not." Cf. de Poyen-Bellisle, " Les Sons et les 
 Formes du Creole dans les Antilles, 50. 
 
 2 O'Connor, A.I. P., 2, 210. 
 
 3 The development of the conjunctionally used negatives, Goth. //^<7/ and 
 O.^.jeda, is in reality the syntactical development of the clauses introduced by 
 those particles. Probably in both of these cases the development was from an 
 indirect question to a clause of fear to a final clause. 
 
 * On the general connection between intensives and negatives cf. Strong, 
 Logeman, and Wheeler, Hist, of Lang., 102. Other words may obtain a new 
 force in the same way — cf. the Italian cosa with interrogative force from use 
 with die. 
 
 5 On this particle cf. Fick, B.B., 7, 270 ; Osthoff, Perf., 328, M.U., 4, 253 ; 
 Delbriick, Syn. Forsch., 5, 504 f.; Brugmann, Gr. Gr., 224; Kretschmer, K.Z., 
 31, 364; etc. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 15 
 
 Whatever may have been the original meaning and function of 
 the I.E. u, the value indicated for it by the separate languages is 
 simply that of a particle with no more definite value than that of 
 a mere intensive or at most of a demonstrative. In general it is 
 very similar to /. In giving the occurrences I assume that the 
 word belongs in the e:o ablaut series.' For the positive side of 
 the argument in favor of this the examples given will themselves 
 be all the evidence obtainable. It may be noted here, however, 
 that the supposition that a given weak form belongs in the e.o 
 series is antecedently more probable than that it belongs to any 
 other.* 
 
 The particle in proethnic times could become attached to 
 words. The locative plural endings -si and -sii were evidently 
 formed from -i' by the additions of the particles -/ and -21? Bar- 
 tholomae"* supposes also that there was a locative singular suffix -u 
 beside that in -/. The examples, though few and only adverbs, 
 show an I.E. use of the particle whether the words affected were 
 full-fledged case forms or not. 
 
 The union of -// with the pronoun so- is doubtless I.E.^ This 
 use of -u is paralleled by the more frequent employment of -/ in 
 the formation of pronouns : Lat. qui, etc.* 
 
 The particle was attached to the third person singular and 
 plural secondary endings {-t-ii, -nf-u).' Here, too, there is a 
 parallelism with the particle -/, if it is the latter that differentiates 
 the "primary" from the "secondary" endings.* 
 
 'On ablaut of particles cf. Osthoff, Perfect., 328, and Per Persson, I.F., 200 f. 
 
 ^Osthoff (Perf. 328), however, sees the strong form of Skr. u in Gr. aJb, and 
 so others — a view which Brugmann (Gr. Gr., 221) questions and Kretschmer 
 (K.Z., 31, 364) argues against. Sonne (K.Z., 12, 278) suggests the comparison 
 of Gr. aS and Skr. o (a -\-ti); cf. Prellwitz, sub v. aS and avepiw. 
 
 3 Brugmann, II, 356. 
 
 ^B.B., 15, 23, and cf. Brugmann, II, 256, Rem. 
 
 5 Skr. so a-sdu (for both genders), Av. haii (for both genders), O.P. kauv 
 {for both genders), Gr. ot-ros. Cf. Brugmann, Gr. Gr., 130; Delbriick, Syn. 
 Forsch., IV, 139; Sonne, K.Z., 12, 270, and Windisch, Curtius Studien, 2, 263 
 and 366 f. It is possible that Gr. ovTO'i is for *so-ittos, cf. Skr. uta, but the other 
 view seems preferable'. The use of i< after pronouns and even between pro- 
 nouns is common in Skr. 
 
 * Cf. Per Persson, I.F., II, 247 f. 
 
 ^Brugmann, II, 992, 1017 ; Thurneysen, K.Z., 27, 174; Hirt, I.F., I, 206; 
 Osthoff, M.U., IV, 252, 257. 
 
 * Brugmann, II, 909, 973. 
 
1 6 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 An I.E. compound arising from the collocation of two j)ar- 
 ticles is to be seen in Skr. o (beside the simple a), C}r. av. Lat. 
 au-t{i) aii-tan, Osc. av-ti av-ti, Umbr. u-te o-tc, Goth, aii-k, O.N. 
 au-k, O.E. ed-c, O.S. o-k, O.H.G. ou-h. I shall attempt to show 
 (p. 2)'^) that the Gr., Goth., and O.H.G. words for "without," 
 av£v, ifiu, dno, contain the particle //-, in the case of the Gr. in the 
 strong form en. The compound would be of much the same 
 sort as the one just mentioned. If I am right in claiming that 
 avev contains the strong form of I.E. ii, it is plain the particle 
 belongs in the e.o ablaut series, and forms of the o- grade may be 
 expected. 
 
 I.E. // appears in the Skr. intensive and conjunctional par- 
 ticle //,' and in composition in san-ti-tar'' (beside san-i-tur), and 
 ii-ta — an Aryan formation (Av. itfa, O.P. iita). In Gr. the weak 
 form of the particle appears in itav-v^ and the o- strong grade 
 form probably in ovv. In Latin our particle is probably to be 
 seen in )ie-u, se-u, ce-ii} In Goth. -// appears as an interrogative 
 particle^ and in connection with -h i-uh) as a conjunctional and 
 intensive particle. O.B. u- in ii-lw "oJv" is perhaps to be com- 
 pared with Gr. ov of ov-v. 
 
 I hold it to be reasonably certain, (i) that Gr. ov received its 
 negative force from use as an intensive, (2) that the particle whose 
 weak form is it was capable of being used as an intensive, and (3) 
 that it could have the ablaut form oij. Positive evidence in Gr. 
 itself that ov did so get its negative force would consist of a use 
 corresponding to the common one of (nc) . . . pas in French ; 
 but one negative — possibl}' the only one — from which ov derived 
 its negative meaning died out before historical times, and ov 
 having become a full-fledged negative, we ought not to expect to 
 
 ' Delbriick, Syn. Forsch., 5, 504 f. 
 
 *Gr. fire/o, Goth, sundro, etc., show the word without the particle. Arm. ev 
 probably does not belong here ; cf. lliibschmann, .\rm. Stud., 75 ; Brugmann, 
 1,63. 
 
 3 0sthof£, M.U., IV, 252; Brugmann, Gr. Gr., 130. 
 
 ■• Of these neu may be a form of nL"ue, but sen is not from sl-ve, and cen is 
 unexplained. Ce-ti contains the pronominal deictive stem ko-ke- seen in ec-ce 
 and elsewhere, and sc- in sen is the pronominal stem seen in the preposition se 
 {sed), "without," in the conjunction 5^</, and in the reflexive pronoun .ft',- cf' 
 Stolz, Lat. Gr., 346; Per Persson, I.F., II, 223; v. Planta, 145. 
 
 ^ Cf., however, Liden quoted by Per Persson, I.F., II, 213. J.Schmidt 
 (Vocalismus, I, 152), Sonne (K.Z., XII, 289), and Scherer (Z.G.D.S., 374) iden- 
 tify Goth, -u and Skr. 7/. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 17 
 
 find it used as an intensive, except perchance in some construc- 
 tion or collocation of words which would in some way protect ox 
 from being taken as a negative.' 
 
 Strengthened forms'" of Gr. ov are ovkL and ov^i and their ante- 
 vocalic forms ovK and ov^- Of ovkL and oux' it is possible that 
 only ovKi is original, and that ov^i arose by a sort of proportional 
 analogy between ovk, oix (before vowel with rough breathing), 
 and ovKi. More probably, however, we have in the -xt of oix^ the 
 correspondent of Skr. ^/,3 Av. zl* "for," "certe," O.B. zi, I.E. 
 ^/il. In Skr., Av., and O.B. the word is used as an intensive; so 
 also in Gr. besides ovx^ in 17x1 and vatxt- 
 
 The -Ki of ov-KL is generally considered as the same particle 
 that appears in vat-Kt and 7roXAa-Kt and other multiplicatives. 
 Wackernagel and J. Schmidt have connected this with Skr. ci'd, 
 I.E. pronominal stem ^/-, to which Brugmann and Osthoff 
 objected that the particle would appear as -n, and they therefore 
 connected it with a stem h'. It can hardly be claimed that the 
 1/ element of the velar was lost in enclisis, as Bechtel has sug- 
 gested, and so that dentalization did not take place, for we should 
 have to dispose of re, Lat. gue, Skr. ca, and besides it is improb- 
 able, though not impossible, that an enclitic word should have a 
 different treatment from an unaccented syllable (cf. Trevre).^ 
 
 Gr. mt'/ct* is either a barbarism for vat'xt or a transformation of 
 
 ' It is possible that we are to see survivals of the original force of ov in the 
 collocation ovk oiv (positive) and in the ov ix-fj constructions, otherwise explained 
 by Goodwin (Moods and Tenses, Appendix II). 
 
 2 Their occurrences in Homer are: 01), 1336; ovk, 464; ovx, 27; ovkL, 15, 
 and ovx'i, 2 (0 716 = IT 762). Among other occurrences of ovxi- Herodas has 
 twelve cases ; cf. Smyth, Ionic Dialect, 295. 
 
 3 Osthoff, M.U., IV, 239 f.; Pott, Wurzelworterbuch, I, 1567. 
 
 * Av. yezi has been explained by Bartholomae (Ar. Forsch., II, 8, n. 3, and 
 cf. Osthoff, M.U., 4,240) as from yad + zi [i. e., yadg/ii), and he compares 
 uzfdl>idi. But the cases are not necessarily similar, since uziiii>{di may have 
 been formed after ud- became uz- in Av. A better explanation, it seems to me, 
 is that '^yedi, "when," " if," was changed to yezi under the influence of zi, " if," 
 but without the original form being driven entirely out of existence, so that we 
 still have in the younger Av. yeiti, " if," the direct descendant of an earlier *j/,frt'/. 
 
 5 Cf. Wackernagel, K.Z., 25, 286 f.; Osthoff, M.U., IV, 241 f.; J. Schmidt, 
 Pluralbildung, 252 ; Bechtel, Hauptprobleme, 354 ; Brugmann, II, 409, Gr. Gr., 
 131; Solmsen, K.Z., 33, 2981.; de Saussure, Mem. Soc. Ling., 6, 161 ; and 
 Buck, I.F., 4, 156. 
 
 * Aristophanes, Thesm., 1183, 1218. 
 
l8 NEGATIVES OF THE E\'DO- EUROPEAN LANGUAGES 
 
 vaLxi. under the influence of ovkL The -k- in ovkl, in the -kl of 
 TToWaKt and other multiplicatives, and that in the troublesome 
 indefinite and interrogative forms of various dialects, probably 
 have a common explanation. In the case of ovkl considered by 
 itself there is no reason for not supposing that the -k- is from an 
 original velar after the /// The connection of -k- in owi with a 
 velar is preferable for several reasons : i) It permits of a direct 
 connection of -kl with Skr. c/d, Av. ci, O.P. ci'j, which have a sim- 
 ilar use. 2) The I.E. pronominal stem ki is rare, and we have no 
 evidence of its use as a particle. 3) Gr. -kl in ovkl will then be 
 connected directly with Arm. ( of of, and, if Arm. may be from 
 I.E. ou, Gr. ovKi, "not," will be connected directly with Arm. of, 
 "not." 
 
 Considering the very great frequency with which negatives 
 are used with indefinites, there seems to be no possible objection 
 to supposing with Solmsen that the -k- of the troublesome indefi- 
 nite forms arose in the same way — in connection with ov. And 
 a similar explanation is at hand for the -kl of ttoXXolkl and other 
 multiplicatives, if we suppose that -kl arose in *7roXv-KL (I.E. />//;? 
 qi{d), Skr. puru cid) and was thence transferred to a stem TroAAa- 
 of other adverbs. Having become established as a multiplicative 
 suffix, it was added also to stems of numerals, possibly displacing 
 an earlier -Tt(s) {cf. Taren. dixdTL<;, "once")." 
 
 The compounds formed of oi and 8e are analogous to those 
 formed of ixrj and 8e. 
 
 Gr. av-3 a- (negative prefixes), I.E. //- iin-. 
 
 Gr. vt]-, Doric vd (negative prefix), I.E. n. 
 
 ALBANIAN. 
 
 Alb. mo- (in tuo-s), I.E. }>ie. 
 
 Alb. funge, tunk, nuk Meyer derives from the Lat. uuviquam. 
 
 Alb. .5- (before voiced consonants 2) Meyer derives from Lat. 
 
 ' Brugmann, I, 427 ; de Saussure, Mem. Soc. Ling., 6, 161. 
 
 * Brugmann now accepts the view that the -k- of TroXXd-Ki and of the indef- 
 inites is from the velar, the peculiar treatment being due to enclisis. He 
 explains the dentalization in the enclitic Tk a.s due to the retention of the « ele- 
 ment of the particle {ki,ie) when used before words beginning with an accented 
 vowel — e being elided and ku- really forming part of an accented syllable. 
 Berichten der Konigl. Sachs. Gesellschaft der Wissenschaften, 1895, 32 f. 
 
 'In regard to a.va.- cf. p. 8. 
 
NEGATIVES OF THE INDO-EUROPEAN lANGUAGES 19 
 
 dis-. But this -J" is the same that appears in !no-s,^ which would 
 seem to preclude the possibility of a connection with a word of 
 negative or quasi-negative meaning. It at the same time indi- 
 cates that the origin of the negative s was in an intensive to a 
 negative. Compare with Alb. mo-s:s Skr. nanu.nii ("not"). 
 
 LATIN . 
 
 Lat. ne, I.E. )ie. 
 
 Lat. ne formed various compounds: i) Loose compounds 
 with coxi]nwQ.\\ow%, >iedii7n, neve, neu ["^ lieu) {y)\ 2) more or less 
 close compounds with various indefinites, necubi^ necunde, neqiiis, 
 with conjunctional force of ne ; neqita/n, nequiter, nequitia, neqiid- 
 quatn, and ne . . . qiiidem, in which ne appears with no conjunc- 
 tional or prohibitive value. 
 
 Lat. ne, I.E. ne. 
 
 The simple ne written as a separate word is, of course, not 
 common.^ But the ne- in nescio'' and neqiieo is to be regarded as 
 the retention of an independent negative with these particular 
 verbs, and the conjunctional negatives neqiie and nisi^ show the 
 particle restricted in application, but hardly in meaning. I pre- 
 fer to follow O. Brugmann^ in deriving nisi from nesei, which 
 form appears on the inscription from Spoleto. It is hardly pos- 
 sible that the original form was ^neisi'^ and that the ei I has been 
 shortened before -si as an enclitic. Such a shortening, especially 
 in dissyllabic words, seems hardly well established. Si>ie'' and 
 qiioque^ cannot be adduced as examples. It would seem that the 
 law of shortening, if it existed, would apply to neve and nidum 
 as readily as to *neisi. Moreover, in Sen. Cons, de Bacch., which 
 is older than the change of ei to /, we have nisei. The change of 
 
 ' So Meyer, Alb. Wort., sub v. mos. 
 
 2 Lucan, 9, 1059, shows the e of necubi to be long. 
 
 3 For the occurrences of ne cf. Lorenz' note to Plautus, Most., 1 10 ; Draeger, 
 Hist. Syn., I, 133 ; and Habich, De neg. usu Plaut. 
 
 ^ Nesapius (Petronius, 50, 5) is formed in imitation of nescius (from nescio), 
 cf. also Terentius Scaurus, De orthogr. Gr. Lat., VII, 12, 4. 
 SO. Brugmann, Ni., 'i,-i„ and cf. Lindsav, Lat. Gr., 611. 
 
 * Cf. Wackernagel, Gr. Ak., 22 ; Skutsch, Forsch. Lat. Gr., 9, and the 
 Nachtrage. 
 
 7 Per Persson, I.F., 2, 223, n. 
 
 * Lindsay, Lat. Gr., 598. 
 
20 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 e to / in nisi and nihil may be due to assimilation, as O. Brug- 
 mann supposes, or to a lack of accent, as Lindsay supposes for 
 nisi and Per Persson' for sine (from '^se-ne), ?nihi, and tibi, or 
 possibly both influences may have operated to produce the 
 change. 
 
 Ne further appears compounded with another particle in nee, 
 **not." The form nee may be a sentence doublet of neqtie, and in 
 its use as conjunction such is doubtless the case ; but nee, "non," 
 has generally^ been explained as containing a particle -ee. Green- 
 ough,3 however, has attempted to explain this use of nee without 
 separating it from neque. Whatever may be said of the possibil- 
 ity of such a genesis as Greenough supposes — and I confess it 
 seems to me extremely improbable — , account must be taken of 
 the fact that neglego, negotit/m, neg ritu have neg, not nee. There 
 is no way of explaining a change of e to g. For these forms it 
 is necessary to set up a negative neg, and for this further support 
 is found in Latin in the denominative nego and in Jiegumate{?). 
 With this neg is to be compared directly Lith. negi. Further, 
 Havet" has pointed out that the form nee of XII Tab. may as 
 well be neg as nee, and the same is true of the passages in Cicero 
 which are imitations of old laws. It may not be difficult to 
 understand, too, how the formulas nee opinans, nee reete, nee pro- 
 cul, res nee maneipi, etc., remained in use after the simple neg had 
 nearly or completely died out. The -g may have been changed 
 to -e through influence of old orthography (legal and religious), 
 or through influence of the spoken nee conjunction (compare ^/^"^ 
 ritu and nee rite). The few cases = which cannot be considered as 
 stereotyped formulas may show either the retention of the simple 
 nee {neg) or the extension of use from the stereotyped formulas. 
 The very few cases of neque, "non," may be the result of a still 
 further confusion \iiee ("and not") : nee ("not") :: neque ("and 
 not") : neque ("not")], or more probably are copyist errors.* 
 
 Quln in all its uses has commonly been derived from qui and 
 
 ' I.F., 2, 223. 
 
 2 C/. liabich, De neg. Plant., 31 and references. 
 
 3 Harvard Studies, 2, 129 f. 
 *Mem. Soc. Ling., 6, 118. 
 
 5 For examples see Greenough and Haliich as above. 
 
 ^ Cf. Havet, Mem. Soc. Ling., 6, 118, n. Probably -que for -c is to be seen 
 in the form '^doncque ( > doniqiie) for donee. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 21 
 
 the negative >u\ but Wharton has derived the conjunction from 
 qi/i Ar fii/m, and Per Persson' has claimed that the qulfi of alio- 
 quifi, ceteroqinn, aiquiu, and hercle qiiln contains the positive par- 
 ticle ne. 
 
 Sin does not contain a negative.^ 
 
 Ne also forms negative indefinite adverbs and pronouns : 
 neutiquam, neutique (late), neuter, nunquam, nusquam, niillus. With 
 the exception of iiUllus the simplex of these compounds begins 
 with u- from I.E. ^/^-,^and the initial velar should appear in com- 
 pounds as in ali-cubi and ne-cubi. Neutiquam and neuter may be 
 regarded as transforrnates of older compounds with the -c- or as 
 being formed after I.E. qu- had become u-. The existence of the 
 form necuter'' renders the first supposition probable for neuter. 
 Moreover neuter in contrast to neutiquam was pronounced in early- 
 Latin with lieu- forming two syllables. ^ It was, then, a compound 
 transformed from the compound ne-cuter under the influence of 
 the simplex uter. The absence of contraction in this case was 
 due probably to the accent on the ne. Possibly also the side 
 form tiecuter exerted an influence. 
 
 In early Latin ?ieutiquam was regularly pronounced with the 
 first syllable short.*^ This can mean nothing else, it seems to me, 
 than that neutiquam was a collocation of two words ^ and so pro- 
 nounced with elision of the final -e or, more properly speaking, 
 with slurring^ of final -e and initial u-. This collocation, formed 
 simply by the juxtaposition of negative ne and the indefinite uti- 
 quam, remained after the use of the simplex utiquam had died 
 out. Later the collocation suffered contraction and became a 
 compound. 
 
 ' I.F., 2, 212. 
 
 2 Per Persson, I.F., 2, 222. 
 
 3 Cf. J. Schmidt, K.Z., 32, 394 f. 
 
 ^C.I.L., VI, 1527; Lucretius, 4, 1217 (where read 7iec, not 7ieque), 5, 839; 
 Mart., 5, 20, II. Cf. J. Schmidt, K.Z., 32, 403. 
 
 5 Lindsay, Lat. Lang., 143. 
 
 * Lindsay, Lat. Lang., 143 ; Lorenz to Mil., 631, and Brix to Capt., 586. 
 
 ^ Otherwise, Brugmann, LF., 6, 84. 
 
 ^ That the Romans so pronounced is indicated by analogies of modern 
 speech. Cf. also Probus (apud Gellius, XIII, 21, 6), who says that turrim had 
 a more melodious sound than tt<rrem in fitrrirn in pracipiti stantem. See Lind- 
 say, Lat. Lang., 144. 
 
2 2 NEGATIVES OF THE INDO-EUROPEAN lANGUAGES 
 
 In the case of iiuiiqiiam and fii/sqiui/n, while earlier forms with 
 -cu- may have existed, the compounds as thev appear have no 
 direct connection with those possible earlier forms. Nuiiqiiam 
 and misgi/am and nullits are generally considered as showing 
 composition with elision of -e of the negative. The other 
 examples given by Stolz' as showing elision may be explained 
 otherwise. Noetiian will be treated below, /nltiqimm, so far as I 
 know, does not occur, sorsus is from *se-vorsus,'' and sf/df/s may 
 have been formed in Italic times and so the Fi be from -on- and 
 that from -cu-. An explanation of iiitnqua7}i, niisquam, and nul- 
 lits which should not need to suppose an elision would seem 
 desirable. In the case of nulliis, since Ulliis never had an initial 
 consonant, there is no difficulty in supposing it to be from *noul- 
 lits from ^/le-iillits. Later, when the initial of ii?iqiiam and iisqitam 
 came to agree with that of Ulliis — save possibly in quantity — 
 /ii/nquam and iiiisquam would be natural analogical formations ^ 
 {ulliis : iiullus :; iniqiiain : ininquaiii). 
 
 It is to be noted that neither ne nor ne in any of the com- 
 pounds mentioned is a true negative prefix. Each is used to form 
 negative indefinite adverbs or pronouns. Similarly ne enters into 
 the compounds nemo and nihiluvi from *jie-hemo* and *iie-h'ilum? 
 But we have two or three "^ words which appear to have ne- as a 
 prefix. If /lefrens meant "a young animal just weaned," it may 
 have been sorr^e sort of a corruption of Crr. vey8/ods, "a fawn," or 
 
 ' Lat. Gr., 276, Anm. Brugmann (I.F., 6. 80) also derives nwnquaDt from 
 n nnquam. If imtWangulus, given by Brugmann, I, 604, and iiiiiltanimis were 
 true examples of elision, they would not really affect the argument here, for 
 these words are compounds of another sort — the first member is a stem, while 
 compounds of ne are the result of sentence combination. In indipiscoi-, indago, 
 and the like, the first member had lost its vowel before the compound was 
 formed. In magtiopere and tantopere we can hardlv suppose an elision of a long 
 vowel. 
 
 = Solmsen, Stud, zur lat. Lautg., and cf. below, p. 24. 
 
 ^ There can be no difficulty with the length of the vowels. That the -u- of 
 iifillus is long is shown by C.I.L., X, 4787. The (juantitv of -//- in nunqiiam 
 and misquam follows either that of nfillus or of unqiniiii and iisqiiaiii. 
 
 *■ Cf. Havet, Mem. Soc. Ling., 5, 447, who thinks that niino is from *ne/iei)io 
 from *nehotno, but that a simplex /lei/io (see Festus, no, Miiller) is a fiction of 
 Verrius Flaccus. 
 
 SBy assimilation of vowels as in nisi. The /, beside Julnm, is from the ? of 
 nihil, which was shortened after the dropping of -iiiii. 
 
 *Lat. nefrefis {!), "kidney," goes with Gr. vi<ppoi, etc. Brugmann, I, 423. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 23 
 
 possibly has a root, I.E. nebh-;'' but, at any rate, it can hardly 
 mean a "not eating animal," for it was applied to animals ai a 
 time when they began to eat." 
 
 It is only^ in nefas and its connected adjectives'* that we have 
 ne- used practically as a true negative prefix. But doubtless this 
 was originally an elliptical expression for ne fas est, used at first 
 with more or less exclamatory force and later almost as a true 
 noun. Its original character may be indicated by its frequent 
 use as a parenthetical exclamation. ^ For a similar development 
 compare >Tim~irum. Somewhat similar are Skr. ndstika and itihasa. 
 
 Lat. no- (in tid-n{e)), I.E. no. 
 
 Lat. non has commonly been derived from noemitn from ne -\- 
 oinom. The derivation is impossible*^ on phonetic grounds, if we 
 suppose that the -0- of the dipthong is short. But Solmsen,^ fol- 
 lowing Thurneysen, Kretschmer, and J. Schmidt, supposes that 
 the result of the contraction of 7ie and oi/iom was noinom, and that 
 this long dipthong later became 0. Neither of these changes is 
 well supported. 
 
 For the law of e-o to the single other example brought for- 
 ward is nolo from nevolo, and the validity of this example depends 
 upon the law which Solmsen tries to establish of the dropping 
 out of intervocalic v before in prehistoric Latin. For this last 
 law the examples given are as follows : 
 
 i) nolo, etc. 
 
 2) deorsum, seorsittn. 
 
 3) sol < *sdol < *sdz^ol < *sdul. 
 
 4) deus < *deivos, Gnaeus < Gnaivos. 
 
 5) prdris K prdvorJs = Gr. irpiopa < Trpw/ratpa, I.E. J>iz'rid(l). 
 
 I may be permitted to give reasons for thinking that, leaving 
 'Cy. p. iiw. 
 
 ''References in P'orcellini. 
 
 sHavet, Mem. Soc. Ling., 6, 108, sees ne- in necesse. The word is probably 
 to be connected with Gr. iveK-. 
 
 '^On confusion of derivatives olfas and/rtr« see Breal, Mem. Soc. Ling., 5, 
 339. The words seem not to have been used freely before Cicero, yet see Cato, 
 39, 12 ; 40, 7 ; 42, 7 (Jordan), '\astud, C.LL., I, 812, Conway, LF., 4, 213, takes 
 as castled and thinks the inscription not Latin. 
 
 5 Catullus, 68, 91 ; Vergil, 7, 73; 8, 688. 
 
 * But cf. Osthoff, Arch. Lat. Lex., 4, 459. 
 
 7 Solmsen, Stud, zur Lat. Lautge., 53 f.; Kretschmer, K.Z., 314, 62; J. 
 Schmidt, K.Z., 32, 407. 
 
24 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 nolo out of account, the examples given lack much of proving the 
 existence of the law, and, that so, another explanation for nolo is 
 desirable. 
 
 Ft oris, if it really occurred, is best explained, as Solmsen him- 
 self thinks probable, as for prora — itself borrowed from the Gr. 
 • — after the analogy oi puppis. 
 
 Notwithstanding the Gr. transcription of Gnaivos without a 
 sign for v, beside 'O/craomav and 'AouevTivw in the Monumentum 
 Ancyranum, the v in the word cannot have had a different treat- 
 ment from that of aevus, avus, fugitivus, octavus, etc., which 
 Solmsen necessarily supposes retained their -v- until after v-o 
 before s had become v-ti.^ And the same remark would apply to 
 the derivation of deiis from *detvos.^ 
 
 Sol has been derived from *sdvel by Schulze,^ which deriva- 
 tion would fall in directly with Solmsen's law" of ave to ao to o, 
 were it not for novem, pover, which show that ve in final syllables 
 following the accent was not changed. But this would not pre- 
 vent sdvelis, sdvel-em, etc., from becoming soils, solem, etc., after 
 which the nom. sol would be a natural analogical formation. 
 
 The forms deorsitm, seorsum, dorsum, sorsum are best taken as 
 formed after the analogy of introrsus, retrorsus,^ etc., and the 
 variation as due to the influence of the prepositions. Solmsen 
 himself explains the late extrorsum and iiltrorsum by analogy. 
 So also horsiim is an analogical formation. 
 
 The presence in Plautus of nevoid and the difificulty^ with the 
 eo verbs speak somewhat against the derivation of nolo and tiolini 
 from ne-volo, nevelhn. But if we once admit the presence in Lat. 
 of a negative no-, the easier derivation of nolo and ndlim is from 
 
 ' P. 45 f- 
 
 ^ Cf. Bronisch, Osk. i- und ^-Vocale, i8o, n.; Thurneysen, K.Z., 32, 558. 
 
 3K.Z., 27, 428. 
 
 ♦ P. 82 ff. On the contraction of -ao- cf. Bartholomae, Stud, zur idg. 
 Sprachgeschichte, II, 142, and Buck, Osc.-Umbr. Verb-system, 151. 
 
 5 These from intro-ve- by change of -ve- to o (Solmsen, 82 f.), or from 
 intro-vo-, v being dropped between like vowels {cf. Solmsen, 109 f.). I see no 
 reason why, if ovo became 0, eve became e, ivi became 1, and ava d, we should 
 not suppose that ovo became 0. Both forms, dorsum and deorsuvi, are found in 
 Dec. Min., C.I.L., i, 199. The contemporaneous use of the two forms indicates 
 that one was not the phonetic development of the other. While Plautus has 
 deorsttni, it is always dissyllabic. {Cf. Wagner to AuL, 365.) 
 
 * Solmsen explains this chronologically. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 25 
 
 *no-volo, *>io-vclim, or ^iiovolim, against which derivation no 
 phonetic objections can be raised. 
 
 The support for a law of eo to 0, then, seems very slight ; but 
 starting from a form iwiiiom we still should have to show that -oi- 
 would become elsewhere than where final. Kretschmer's' only 
 example in support of the law he^idt?, poti/s poailum [cf. '$ikx.pdy- 
 dna-m), which counts for nothing when compared with Skr. 
 pdtave pdtra-m, is this same supposed *//dinom to ndii{iim). x\nd 
 the onlv other support for the law is the corresponding supposed 
 change of ov to r',^ which Solmsen himself has pretty well dis- 
 posed of. Schmidt^ has added o\i\y prod-, which he supposes 
 from * pro-id. This is possibly an ablative case form, or has its 
 -d from words beginning with that letter, to which it was pre- 
 fixed. The probable explanation, however, is that prod- is after 
 the analogy of ind- and other prepositions ending in -d {postid, 
 antid, red), a transformation favored by the proportion ///- : ind- '.'. 
 pro-: prod-. Solmsen suggests promo as a further example. But 
 this is to be explained as a transformation oi*proemo {ov *prumo) 
 under the influence of pro. The meanings of the simple verb 
 and the compound had ceased to indicate the connection of the 
 two words, while in the compound the idea oi pro was very appar- 
 ent. In much the same way coemo {^cFimo) became cdtno, with 
 the meaning "to bring together," while, when the compound had 
 the meaning of the simplex, the form coemo was retained without 
 even contraction taking place. In the case of promo demo may 
 have assisted in the transformation. The same sort of transfor- 
 mation as in promo and como is to be seen in sumo and probably 
 in dego,'' beside deamdre, etc. 
 
 The fact is that this derivation of nofi from uoinum, supported 
 by that of nolo from *nevolo, has been the mainstay of all three 
 suppositions just discussed. But even if it were perfectly sure 
 
 ' K.Z., 31, 462. 
 
 ^And cf. Buck, Osk. Voc, 163; Brugmann, Die Ausdriicke fiir den Begriff 
 der Totalitat, 54 ff. 
 
 3K.Z., 32, 407. 
 
 "• Otherwise Stolz, Hist. Gr., 219, who considers coino as well as lic'go due to 
 a regular contraction. But for coind at least we cannot suppose that the result 
 of the contraction of o e was affected by the fact that the first vowel belonged 
 to the preposition co-. The length of the vowel in como may be due to promo 
 and demo, or to the form — *coemo — of which it was a transformate, or to both 
 causes. 
 
26 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 that e-oi could become oi and oi o, we should still have to explain 
 how, if noetntin became iion, it also remained. We have two cases 
 of noemnn in Lucretius (III, 198, and IV, 712), though Plautus, 
 a century and a half before, used iion far more often than noe- 
 fiiim.^ And while we find noenian in the early literature — not, 
 however, in inscriptions — beside the far more common fioii, we 
 have no trace whatever of an intermediate form other than the 
 dropping of final -m. I derive, then, Lat. noii from *no-n{eY as 
 qiiln from *qtii-/i[e), and nolo from *no-voio as prorsus from pro- 
 vorsiis. 
 
 The evidence for an I.E. form no may be summarized here. 
 Representatives of I.E. noi are, of course, evidence of the exist- 
 ence of an I.E. no. The words which, as 1 believe, must be 
 explained as containing I.E. no are Lat. non and nolo just dis- 
 cussed, Umbr. no-svc, O.Ir. nd. Either I.E. no or ne or both 
 may be seen in Aryan nd and its compounds. Forms to be 
 explained as containing I.E. noi are Lat. (Duenos) >ioi-si\ Lat. 
 nocniim, Lith. nai-kal( and nai-kinu. Either I.E. noi or nei may 
 be seen in Av. nae-cis nae-ita, O.P. naiy, O.B. ///. Leaving out of 
 account, then, Aryan, which furnishes as much evidence for the 
 no forms as for the ne, and Arm., Gr., and Alb., which furnish no 
 evidence for either, we have evidence for I.E. no in Italic, Celtic, 
 and Balto-Slavic, while for I.E. ne we have evidence in Italic, 
 Celtic, Germanic, and Balto-Slavic. 
 
 Lat. nei nl, I.E. nei. 
 
 Besides the independent nei ni the word appears in nlm'irum, 
 but not as a negative prefix. The word is an "elliptic paratactic 
 protasis ' non est nuriim.' " ^ 
 
 Lat. noi {\n noe-ninn and /loi-si {T>\\tno?,)), I.E. noi. 
 
 The common derivation of noenum noefiK* from ?ie-oinom can 
 only be made by supposing an elision in n{e) ; but if I am cor- 
 
 ' Plautus probably wrote noenum in a good many places in which it has 
 disappeared from the text. Cf. Habich, De neg. usu Plaut., 35 ; Wagner to 
 Aul., 67 ; Brix to Mil., 653. 
 
 - The above is in practical agreement with the derivation of non given by 
 Wharton (Ety. Lat.) and with the discussion of non and noenum by Thomas 
 (Class. Review, 5, 378, 434; 6, 194). C/. further Wackernagel, Gr. Akzent, 19, 
 n.; Stolz, Hist. Gr., 130, and Brugmann, I.F., 6, 80. My discussion was written 
 some two years ago and before Mr. Thomas' article had come under my notice. 
 
 3 So O. Brugmann, Ni, 19, and I have explained nefas similarly. 
 
 ■• The forms ninum and tienii are doubtless editor etymologies. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 27 
 
 rect above in regard to fiiiwtpiam and n usq 11 am, snch. an elision in 
 sentence combination is unsupported. If /loe/ium is from I.E. 
 noi-, the second element is to be connected with the pronominal 
 stem no- or «//-,and in either case with Lat. tiiim!^ 
 
 Unless Duenos noisi^ is simply a dialectic variation of tiei, we 
 have in the word another evidence of I.E. noi in Latin. Solm- 
 sen's'' objection to the view that noi is an ablaut form of nei, on 
 the ground that there is no support for the form in related lan- 
 guages, amounts to little, even if the fact were exactly so, if it can 
 be shown that representatives of an I.E, no existed. 
 
 The negative hand has been connected by Corssen and others 
 with Skr. ava and the negative force derived from the preposi- 
 tional. 
 
 The final -d after the diphthong is in itself peculiar. It can 
 hardly be an original d. But if we take haut^ as the more original 
 form, it is easy to suppose that hat/d dccost by assimilation, after 
 the law of the dropping of final -d had ceased to work, and that 
 then -//was retained in certain sentence combinations and dropped 
 in others (before consonants — cf. se-pono from sed-pono). The h- 
 may not be etymological, and, rejecting this, we arrive at a form 
 aiit identical with the conjunction aii-t,'' "or." The original 
 meaning of the particle aii-t, as indicated by Skr. 0, Gr. au, Lat. 
 fl/z-Zi^w, was "further," "again," and I see no great difficulty in 
 supposing that this aut was used with a negative as intensive ^ 
 {nan aut, "not again," "not at all"), and that thence the negative 
 idea became attached to the word.*^ The weak //- was retained 
 permanently in the negative and dropped permanently in the 
 conjunction. 
 
 ' Per Persson, I.F., 2, 206. 
 
 ^ Cf. Conway, A.J. P., 10, 455. If Conway's reading of the Duenos inscrip- 
 tion is the correct one, noine does not occur. 
 
 sSolmsen, Stud, zur Lat. Lautge., 87, and cf. Conway, A.J. P., 10, 455; 
 Planta, I, 152. 
 
 -•Habich (De neg. usu Plant., 13), following Ritschl, takes hand as the 
 more original form. Cf. Stolz, Lat. Gr., 317. 
 
 s For connections see p. 16. 
 
 ^The use of antem as an intensive, not, however, of a negative, occurs occa- 
 sionally in Plautus ; cf. Pseud., 305, Amph., 901. 
 
 7 Wharton, Ety. Lat., supposes the h- to be unorganic and compares antem. 
 I do not know how he would connect the meanings of the two words. >^ \ ^ 'l.^t 
 
2 8 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 The origin of the negative meaning of hand is indicated by 
 the collocation neque . . . hand, of which most of the examples 
 are in Plautus. The examples are : ' 
 
 Neque ego haud committam. Bacch., 1037. 
 
 Neque id hand imt)ierito tiio. Men., 371. 
 
 Neque illud hand obiciet niihi. Epid., 664. 
 
 Neque mihi Jiaud imperito eveniet. Persa, 535. 
 
 Neque id haud subditiva arbitror gloria esse. Bacch., 26. 
 
 Neque haut Ion ge post. Gellius, 17, 21, 34. 
 
 Ne temere facias : neque tu haud dices tihi //on praedictum. Cave. 
 Ter. Andr., 205.'' 
 
 It is to be noted that in the examples from Plautus and Ter- 
 ence neque and haud are separated by only one word, in that 
 from Gellius not at all. The examples of neque . . . numquatn 
 differ in this respect, and, besides, such double negatives as neque 
 . . . nunquam (negative and negative indefinite) are to be expected 
 almost everywhere in literature and do not serve to explain neque 
 . . . haud.^ The nearness of the two negatives to each other 
 would seem to preclude the supposition that the negatives were 
 separated in their application, which is a necessary condition for 
 the use of two negatives in the sense of one. If it were the case 
 that the force of the neque was expended on the following word, 
 haud would be left as the special negative of the verb in four of 
 the seven cases, which is not the common use of this negative. 
 It seems necessary, then, to separate neque . . . haud from such 
 cases as Plautus, Cure, 579,'' and Cato, R.R., 66.^ 
 
 Neque . . . haud, except for the conjunction -que, is just like 
 French ne . . . pas — the two are practically one word negativing 
 the sentence.^ We may conjecture that the few examples we have 
 of the collocation are the survivals of a common form of expres- 
 sion. The dying out was the natural result of //i^?//*^/ becoming an 
 independent negative. 
 
 ' Of. Ziemer, Jung. Streif., 141 ; Habich, 7. 
 
 ' Probably the true reading, although the MSS. have hoc in place of hand. 
 
 ■iCf. Gebauer, Archiv f. Slav. Phil., 8, 177. 
 
 ■• Ut ego tua magnifica verba, neque istas fiias iiiagiias ///iiias, fion plitris 
 facia quavi ancillani meani. 
 
 5 .See p. 6. 
 
 ''The explanation given by Strong, Logeman, and Wheeler (Hist, of Lang., 
 155) of double negatives caused by contamination seems inapplicable here ; nor 
 is it possible that neque . . . hatui is a translation of a Gr. double negative. 
 
NEGATIVES OF THE INDO-EUKOPEAN LANGUAGES 29 
 
 It may well be that hand was allowed to stand by itself first 
 in adjectival and adverbial expressions not closely connected with 
 a verb and forming, as it were, an accentual unit. It is in such 
 cases that French ne pas has become /ai' and ne point point. This 
 would explain the common use of Iiaitd with adjectives and 
 adverbs. 
 
 Lat. ///- (as negative prefix), I.E. //. 
 
 OSCAN. 
 
 Osc. ne, I.E. ne. 
 Osc. «/', I.E. ne. 
 Osc. nei, I.E. nei. 
 
 All these Oscan negatives form compounds with -/ (Lat. -que) 
 the conjunction retaining its proper force. 
 Osc. an- (negative prefix), I.E. //. 
 
 u M B R 1 A N . 
 
 Umbr. ne- (in nc-p), I.E. ne. 
 
 Umbr. nei- (in neip), I.E. nei. 
 
 Umbr. )io- (in no-sve), I.E. no. 
 
 Umbr. nosve has usually been regarded as equal to Lat. fiisi, 
 though it was phonetically impossible to connect no- with Lat. 
 ni-, n'l, ne, or ne, and non seemed specifically Lat. So Brugmann' 
 has proposed to see in the word not a negative, but a representa- 
 tive of I.E. nit, "now." But the supposition seems impossible 
 for phonetic reasons.^ I.E. u became in Umbr. only under 
 certain conditions, which will not include nosve. The usual rep- 
 resentative was //, as shown hy fust, tuva, etc.; while u became 
 only before ni. The result before other labials was a " Mittel- 
 laut," which was sometimes written //, sometimes (sopa:supa). 
 
 As for the meaning required for the word by the passage^ in 
 
 ' Osk. und Umbr., 225 ; cf. Bucheler, Umbrica, 96, and Breal, Eugubines, 
 XXII. 
 
 2 Von Planta, §§ 51-2. 
 
 3 Ig. Tab. VI, B, 54. The passage with Breal's translation is as follows : 
 . . . eetu ehesH popln. nosve ier die esu pophi, sopir habe esme pople, portatu ulo 
 pne mersest,fetH urii pirse mers est; "ito ex hoc populo. Si non iverit ex hoc 
 populo, siquis incola est, huic populo [vectigal] portatu illuc ubi lex est, sacri- 
 ficato id quod lex est." 
 
30 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 which it occurs, it seems to me the negative is the more natural. 
 It would seem that the object was to get rid of the "peregrini," 
 and what they did after leaving the territory would be a matter of 
 indifference to the Iguvini. The most natural thing after the 
 decree of expulsion would be a threat against those who failed to 
 obey that decree. Habe'^ is used absolutely, /. e., with some word 
 for "property" understood in thought, which word would then 
 be supplied as the object of portatu. The sense would be then, 
 " Let these people depart, but if they do not depart, whoever' has 
 property shall contribute it for this people to be used in the com- 
 ing sacrifice." If Brugmann's interpretation is the correct one, 
 we should expect in place of nosve, "if now, '' pone, "when," as he 
 himself admits. Brugmann's interpretation apparently agrees 
 better with the briefer direction in Table I;^ but it may easily 
 have been that "svepis h abe, etc.," was understood as the pen- 
 alty attached to the non-compliance with the sentence of banish- 
 ment not here expressed. 
 
 Umbr. an- (as negative prefix), I.E. //. 
 
 OLD IRISH. 
 
 I.E. 7fie does not appear in Celtic. O.Ir. mi- (negative pre- 
 fix) probably corresponds to Germanic miss-.'' 
 O.Ir. /il, I.E. ne. 
 
 The negative is used with con and in the compound ma-nt, 
 "if not." 
 
 O.Ir. nd, I.E. /id. 
 
 The word further appears in the compounds ?/dd, ndch, arnd, 
 arnach, arnad, conna, connaro. 
 
 Also in Scotch Gaelic and Welsh ni and nd appear as the 
 representatives of I.E. ne and no respectively. 
 
 O.Ir. an-^ am-, e- (negative prefix), I.E. //. 
 
 ' Cf. Harper's Lex., sitl) v. habeo, II, A. 
 
 ^Sopir=^ "quisquis;" Brugmann, Osk. unci Umbr., 214, andv. Plauta, 152. 
 
 3 Tab. \, b, 17, 18, eturstamu tuta . . . : "svepis habe, purtatulu, 
 pue mers est, feitu usu, pere mers est," which Biicheler translates, 
 "exterminato urbem . . . sicjuis habet, portato illo (juo jus est, facito illo quod 
 jus est." 
 
 ■* Cf. Kluge, sub V. miss-, and Feist, sub v. miss-. 
 
 5 In regard to an{a)- cf. Zimmer, K.Z., 24, 532. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 3 1 
 GOTHIC. 
 
 Goth, ni, I.E. tie. 
 
 Goth, ni appears further in the compounds niba nibai, nih, niu. 
 The -/;' of nih is the I.E. qe, Skr. ca, Gr. re, Lat. -que, Osc.-Umbr. 
 -/. Goth, niu- (interrogative) contains the enclitic particle u 
 (Vedic u, etc.) and so would correspond in form almost exactly 
 with Lat. neu {<ine-u), though the composition in the two lan- 
 guages was doubtless independent. 
 
 Goth, ne, I.E. ne. 
 
 Goth, nei, I.E. nei. 
 
 OLD HIGH GERMAN. 
 
 O.H.G. ni, I.E. ne. 
 
 O.H.G. niwiht, niowihi, nein, nio, nioman, etc., show ni com- 
 pounded with various pronominal words. O.H.G. noh, "neither," 
 can hardly be explained as phonetically from I.E. ne + qe — the 
 form should correspond with Goth. nih. Paul's ^ and Braune's* 
 supposition of a change of .? lo ^ in enclitics and proclitics seems 
 uncertain and perhaps would not apply here if established. One 
 might see here I.E. no. But more probably the similarity of form 
 of noh (Goth, naiih, I.E. nu-qe) and the use of that particle as an 
 intensive to a negative caused it to assume the force of nih, driv- 
 ing that particle out of use as an independent adverb and thence 
 influencing the vowel in the compound nih{h)ein {noh{h)ein : nih- 
 {h)ein.^) 
 
 ^ Cf. Sonne, K.Z., 12, 279; Dahlmann, J.P., 2, 257 ; Brugmann, II, 411 ; 
 Scherer, Z.G.D.S., 476. Scherer would connect Goth, -h with Lat. -c, Gr. -/c in 
 
 2 Cf. Scherer, Sonne, and Dahlmann as above, and J. Schmidt, Vocalismus, 
 I, 152. 
 
 3P.B.B., 6, 248. 
 
 '• Ahd. Gr., § 29. 
 
 5 The usual derivation oi nihhein ixora. nih tf/« seems hardly satisfactory. 
 Nihhein and dehhein must be considered together. If the latter is from a 
 deh + ein {cf. Kluge, sub v. kein), there is no explanation for *deh-. The 
 occurrence of -kh- (beside -h-) in both words might not be difficult of explana- 
 tion {cf. Braune, A.H.G. Gr., §154, A, 6), but nih with meaning "and not" 
 could hardly enter the compound, and we have no evidence of the uncom- 
 pounded adverb meaning simply " not." It is difficult to understand under 
 what circumstances dechein became kein in M.H.G. Would it not be better to 
 suppose that the first member of the compound dehein is the article as in de- 
 
32 NEGATIVES OF THE INDO-EUKOPEAN LANGUAGES 
 NORTH (GERMANIC. 
 
 In North Clernianic all I.E. negative adverbs disappeared. 
 Besides the negative verbal suffix -at, which Noreen' connects 
 with Goth, aiiiata, a negative appears in the various dialects, e. g., 
 O.Ic. ekke, compounded of eit, neuter of eiii, and a particle -gc, 
 -gi. This particle "" is the same one that a{)pears in man{n)-ge, 
 "Niemand," hverge, "nirgends," and so is plainly responsible for 
 the negative force of ck-ke, etc. 
 
 Germanic t///-' (negative prefix), I.E. >/. 
 
 O.H.G. a-, O.E. ce- (negative prefix), I.E. [f-{'?). 
 
 In O.H.G. some ten or fifteen'' words appear with this prefix, 
 among which are d-tcil, "non-participation," and d-mdlit, whence 
 the N.H.G. Ohiimacht {^xdiX^oXAQ. Oh-inacht), with -;/ from the com- 
 moner prefix iin-} The number of examples could be added to 
 from M.H.G. In O.E. some ten or twelve words have ce as nega- 
 tive prefix, e. g., ce-meii, "unmanned," ce-iiot, "useless." If now 
 these prefixes are to be connected, as it seems they should be, 
 they point to an urgermanic vowel sound differing but little, if 
 any, in degree of openness from urgermanic e. But cj" as a neg- 
 ative prefix would stand entirely by itself, unless it is to be con- 
 nected with I.E. //. 
 
 weder? (This can hardly be for *deh-wedej; as Braune would have it — §295, 
 A, 2 ; cf. Ill weder.) If the compound were formed early, -////- would be regu- 
 larly for Germanic -/•-, and -h- would be a simplification seen in other cases. 
 We would have, then, for the last member a Germanic pronoun kein, and in the 
 case of nihhein the first member would be ni-, as in ni weder. This kein, or at 
 least its initial, is to be connected in some way with the -k of Goth, mi-k, thu-k, 
 si-k, which may be identified with the Gr. 7e. Compare also Skr. ha, gha, and 
 -g of Lat. >ieg. (Cf. Curtius, Grundziige, 526 ; and Havet, Mem. Soc. Ling., 6, 
 118.) 
 
 ' Altic. und Aitnorw. Gr., § 57, 4 /^ 
 
 2 Noreen, 71, and (/. above, p. 7. 
 
 3 On O.H.G. una- in una-holda, see p. 8. 
 
 " Cf. Weinhold, M.II.G. Gr., § 291, who connects O.H.G. a-, with Skr. a-. 
 Grimm, 695, connects O.H.G. a- with the preposition, Goth, us, etc. 
 
 5 Andresen, Deutsche Volksetymologie, 275. 
 
 * (\f represents the sound just mentioned and which, as I shall try to show, 
 was a more or less nasalized vowel sound slightly more closed than urgermanic e. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 33 
 
 It is to be noted that in all the O.H.G. and O.E. words the 
 prefix appears before a consonant. It might well be that an ante- 
 vocalic form of the prefix was displaced by the common uu-. So 
 we are at liberty to suppose for the urgermanic sound some sort 
 of nasal affection (nasalization, glide sound, or both), which should 
 disappear in O.H.G. before consonants, but which might be found 
 to appear before vowels. 
 
 The O.H.G., Goth., and Gr. words for "without" — dnu, inu, 
 and avei> — are of value in the discussion. Kluge' and others have 
 connected these words, but I do not think the phonetics have 
 been made clear. If we suppose, now, that the urgermanic sound 
 mentioned above was slightly closer than urgermanic e, there is no 
 diflticulty in supposing as representatives of it Goth, in-, O.H.G. 
 an-, before vowels, and O.H.G. a-, O.E. ^-, before consonants. 
 This will permit us to connect directly Goth, inu and O.H.G. 
 dnu d)io ana, O.S. ano, O.Ic. on an. They are representatives of 
 I.E. nn-u. 
 
 Gr. aveu has been connected with Skr. sanutar.'' To say 
 nothing of the initial smooth breathing, the identity in meaning 
 and the apparent resemblance in form of Gr. avev and O.H.G. 
 dnu, Goth, inu,^ should lead us to suppose that avev was connected, 
 not with sanutar,'' but with the Germanic words. This can be 
 done by supposing beside I.E. nn-u a form nn-eu. The last ele- 
 ment in these forms is the particle u, Vedic //, etc., and the first, 
 as I think, the I.E. // //, which appears in the separate languages 
 as a negative prefix. ^ 
 
 'Kluge, s. V. ohne; Feist, sub v. imt; Prellwitz, sub v. iLvev; Bezzenberger, 
 Adverb., 84 ; Noreen, Urgermanische Lautlehre, 85. Noreen connects the 
 words with the various forms of the negative ne, but in a way hardly satisfac- 
 tory, it seems to me. 
 
 ^Bartholomae, B.B., 15, 16, and cf. Meringer, B.B., 16, 227, and Johansson, 
 B.B., 15,310. 
 
 3O.B. vHiin is doubtless for vtnu, — cf. vine Skr. vind ; cf. Jagic Archiv. f. 
 Slav. Phil. I, 17, Per Persson, I.F"., 2, 213 and references. Possibly also a Skr. 
 auo is to be added ; see B. and R., sub v. For the Ossetan iind see Hiibsch- 
 mann, Oss. Sprache, 21. 
 
 * With Skr. sauu/ar sanitur are connected Gr. firep and oaap, the aspiration 
 being lost through the influence of fivei/ and avrap (Brugmann, II, 75). Gr. 
 &VIS from dvev after X'^P'S (Brugmann, Gr. Gr., 218). 
 
 5 This derivation of Goth, inu, O.H.G. dnu, Gr. dvev, as well as that of the 
 O.H.G. d-, O.E. te-, is in conflict with the theory advanced by Plerlinger (K.Z., 
 
34 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 There may be difficulty in supposing that a compound adverb- 
 preposition nti-eu was formed from //, which regularly appears as 
 a negative prefix, and a particle u. However, remembering that 
 originally // was an independent word negativing a noun and, 
 that so, it could have a particle as // attached to it, and remem- 
 bering that /i-ehjos came to mean "without a horse," we may be 
 allowed to conjecture that n eu ekuos came to have the same 
 meaning, although the words did not form so close a compound 
 that nn-eu would be unable to be separated as an adverb-preposi- 
 tion meaning "without."' Something very near the reverse of 
 this process is to be seen, for example, in Skr. iiirmaksikam, 
 "with freedom from flies." 
 
 L ITHUANI AN. 
 
 Lith. ne, I.E. iie. 
 
 The use of the particle was extended and drove out entirely 
 the representatives of I.E. n- and fi- as negative prefixes. It did 
 not form negative indefinite pronouns and adverbs. It was 
 strengthened by particles -gi and -gu, and formed a conjunc- 
 tional compound neba {cf. the Av. particle bd). 
 
 27, 436, and cf. Kluge, sub v.. Art., and Wood, Red. Verbs in Germ., 34) that 
 the Germanic representation of I.E. n was an. But the only apparent support 
 for this theory is furnished by Germanic gaggan and blandan, which have con- 
 geners with an -e- form of root, and O.E. and (Kluge, P.B.B., 10, 444). The 
 analogy of the long sonant liquids will give no support here, since the repre- 
 sentations of sonant liquids and sonant nasals in the separate developments 
 differ as often as they agree. Cf. further Hubschmann, Vocalismus, 134 f.; 
 Brugmann, I, 306: Osthoff, Perfect., 178, 417; and de Saussure, System Prim., 
 274. De Saussure supposes that the Goth, representation of // was un, and so 
 also Streitberg, I.F., 6, 141. I am not unaware of the narrow basis upon which 
 my own theory rests, but, on the other hand, I believe nothing very definite can 
 be urged against it. To be sure, n is not assumed to have existed before vowels, 
 as I have supposed in the case of the I.E. nn-tt : Goth, hiu, O.H.G. anu. But 
 even if the rule were definitely established, it could not be applied with certainty 
 to a case of this kind. The -u may have become attached to the n in a late 
 period of the I.E., when the accentual laws which had caused the differentiation 
 of ;/ and n had ceased to work. It is possible that I.E. //- (negative prefix) 
 occurs in Goth, in-winds, " unjust," and inwidan, " to refuse to recognize," and 
 I.E. m in sinteins, sinteino, " irdfTOTe." (For the suffix, compare Lat. mati'itlnus, 
 dit'ttinus, perendinus; but see Brugmann, Die Ausdriicke fiir den Begriff der 
 Totalitat, 23.) 
 
 ' Delbriick, Ver. Syntax, 299, gives Gr. 6.vev, Goth, inuk, as an example of 
 prepositions which were proethnic and which were not praverbia. 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 35 
 
 Lith. nei (ffe-), I.E. fiei. 
 
 In contrast with f/c uci was not used as a true negative prefix 
 and did form indefinite pronouns and adverbs, e. g., nei venas, 
 "no one" (beside nev'ens, "not one," "many"). 
 
 Lith. 7iai-, I.E. noi. 
 
 The word only occurs in naikaJi and the more common tiai- 
 kinii, "tilgen." Beside naikinii we have nekinti, "verachten," the 
 latter evidently being a denominative of nekas. We must suppose 
 that there once existed in Lith. the indefinite pronouns nai-kas 
 and nei-kas, from each of which a denominative was formed.' 
 
 OLD BULGARIAN. 
 
 O.B. ne, I.E. ne. 
 
 O.B. ni, I.E. 71 ei or 710 i. 
 
 As in the case of the corresponding words in Lith., O.B. ne is 
 used as a negative prefix and ;// to form negative indefinites. 
 
 SYNTAX.^ 
 
 None of the negatives of separate origin was prohibitive,^ /. ^., 
 used with volitive forms. Negative prefixes sometimes arose, 
 practically equivalent to the representatives of I.E. // or ;7, e. g., 
 Lith. be-, O.B. bez-. 
 
 Of the representatives of the I.E. forms, >ie ue no no nei noi, 
 me, the syntactical uses as they appear in the various languages 
 may be classified as i) prohibitive, 2) convictional, 3) conjunc- 
 tional, 4) negative of dependent sentences, 5) conditional, 6) 
 interrogative. 
 
 ■ The relation of iiyksth to nekinu and naikinu I do not attempt to discuss. 
 It may be an afformate to these words, 01 may be entirely distinct and to be 
 connected with ninkii tiikau, Lett, niktts, O.B. niknq.ti. Cf. J. Schmidt, Plural- 
 bildung, 396, n., and Leskien, Der Ablaut der Wurzelsilben im Lith., 279. 
 
 2 It is proposed here to discuss briefly the syntactical uses of the negatives 
 of the various languages, mainly with the view of discovering the original value 
 of the I.E. negatives and the relations existing between that original value and 
 the values attached to the negatives of the separate languages. 
 
 3 Sacrificing exactness to convenience I use the term ''''prohibitive " as indi- 
 cated, ''''volitive forms''^ for forms expressing will or wish, "convictional nega- 
 tive " as opposed to " prohibitive," i. e., the negative of expressions of convic- 
 tion, expectation, possibility, etc. On Goth, ibai and O.^.jeda, see p. 14, n. 
 
36 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 The last two' can only be regarded as specializations in the 
 use of the convictional or prohibitive negatives. A special nega- 
 tive for dependent sentences is found, I believe, only in Celtic, 
 and there the distinction was not always kept up. The use of 
 Celtic >id in dependent sentences is probably connected with its 
 use with volitive forms. 
 
 The conjunctional use of a negative is only to be seen — with 
 certainty^ at least — in the case of Skr. ned, Gr. \xy], Lat. nc, and 
 O.H.G. ni. In all these cases, except that of Skr. ned, it is easy 
 to see that the conjunctional force arose through the development 
 of the clause introduced by it from an independent through a 
 paratactic to hypotactic clause of purpose, etc. But the case of 
 Skr. ned is somewhat different. In independent sentences there 
 was no distinction between na and ned,"^ and neither was to any 
 great extent the negative of the subjunctive with its volitive force. 
 The ordinary paratactic expression of purpose was with nid and 
 the injunctive.'' But we may suppose that beside the paratactic 
 md with injunctive or even, perhaps, earlier than that construction 
 for purpose, there was a paratactic expression of purpose with 
 )ied (and na) with the subjunctive, not descended directly from 
 an expression of negative will, but formed by adding a negative 
 to the positive expression legitimately using the subjunctive.^ 
 It might well be that ned should be introduced into and 
 become the generally used negative in these sentences without 
 being freely used in expressions of a more evident volitive 
 character. 
 
 The facts indicate that the conjunctional use of the negative 
 
 ' The last is altogether uncertain. The only case is Goth. >iei, and that 
 occurs with interrogative force without an interrogative word but once. Lat. 
 nei ni and Osc. 7ie are the only examples of a conditionally used negative. For 
 nl, see O. Brugmann, Nl. Osc. ne only once and that with pun, — tie pit u, "nisi 
 cum." 
 
 = I venture the opinion that the sentences introduced hv lua in the Av. 
 should not be considered as dependent. The fact that ina was not so used in 
 Skr., O.P., or in the modern Persian and the possibility oi considering the 
 examples paratactic should lead one to that view. 
 
 sDelbriick, Syn. Forsch., I, 112, 121 ; Whitney, J.A.O.S., 5, 385, 399, 
 
 *■ Delbriick, Syn. Forsch., V, 546. 
 
 5 Beside the subjunctive and equivalent to it in these clauses is, of course, 
 the injunctive, and the optative occurs once (A.H., 8,_'23, 11). 
 
NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 37 
 
 did not belong to the parent speech.' The use does not appear 
 in near all these languages ; the negatives so used are not the 
 same in any two of them, the negative so used in Skr. is a special 
 Aryan strengthened form of I.E. ne ; only the purpose clause 
 appears in Skr., while in the other languages also substantive 
 clauses are so introduced ; the conjunctional use of O.H.G.' ni is 
 quite certainly a special development, since the use does not 
 appear in Gothic. 
 
 We have left, then, two values possible for the I.E. negatives 
 — prohibitive and convictional. 
 
 The value of I.E. me was either that of a prohibitive negative 
 or one from which such a value could be derived. In all lan- 
 guages in which it occurs it has a more or less restricted prohib- 
 itive value. 
 
 On the other hand, for the representatives of I.E. ne and its 
 ablaut forms, with the exception of // //, the separate languages 
 show variations. The form ne never became a special prohibitive. 
 In Germanic and Balto-Slavic, where no distinction was made, 
 I.E. 7ie^ was used as convictional and prohibitive ; and in Skr., 
 while na was sometimes used with volitive forms, md was pecu- 
 liarly the prohibitive negative. The form ne became specially 
 prohibitive only in Latin, but the convictional value is retained 
 in the collocation ne . . . quidem. The form nei was also used 
 in Latin as a prohibitive, but it never became restricted to this 
 use. The form no became prohibitive only in Celtic, and there 
 I.E. ne was not altogether excluded from use with imperative and 
 subjunctive. The use of O.Ir. )id in dependent sentences is pos- 
 sibly connected with its use as a prohibitive. It is noteworthy 
 
 'This would furnish some evidence in support of the statement that the 
 parent speech did not possess dependent sentences. Cf. Hermann, K.Z., 23, 
 481 f.; Zimmer, Festgruss an Roth, 173. If negative purpose clauses are not 
 proethnic, probably positive purpose clauses are not. Probably also all sen- 
 tences introduced by a setitence relative (?'. e., relative adverb) are to be classed 
 as not proethnic. Further than this, however, the evidence furnished by the 
 absence of conjunctionally used negatives would not reach. 
 
 ^O.H.G. ni, "dass nicht," "quin," only after negative sentences. 
 
 3 In Latin probabivthe use of ne as a prohibitive survives in a few cases of 
 neqtte [nee) with volitive subjunctive, e. g., Plautus, Asin., 775 ff. {<r/- Elmer, A. 
 I.P., 15, 299 ff., especially 319; Loch, Imperativus bei Plautus; and A.J. P., 16, 
 No. 4). In Oscan nep (Lat. nee) was used with volitive forms, and in Umbrian 
 nep is used with an imperative {neip . . . nep, VI, A, 6). 
 
38 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 that, while in Latin ne was prohibitive, nD-ii {e) convictional, in 
 O.Ir., so far as the distinction was made, na (Lat. no-) was pro- 
 hibitive and ;// (Lat. ne) convictional. The form iioi nowhere 
 shows signs of being a prohibitive. 
 
 We shall be safe in saying that originally none of these forms 
 of ne had a special prohibitive force, and that the development or 
 specialization took place in each instance to meet a want for some 
 reason not felt in the LE. period. 
 
 What, now, was the relation of me to ne and its ablaut forms? 
 The persistent use of me as a more or less restricted prohibitive 
 wherever it occurs precludes the supposition that me and ne were 
 used indiscriminately. And it is about equally certain that me 
 and ne were not contrasted as the negatives of volitive and non- 
 volitive forms respectivelv. If such a distinction had been estab- 
 lished in LE. times, we should expect to find it more generally 
 kept up, and especially in all those languages in which the dis- 
 tinction of prohibitive and convictional negatives appear we 
 should expect to find tne retained. In Italic and Celtic, however, 
 we have the distinction of prohibitive and convictional negative, 
 but me has disappeared. Whether, now, we suppose that in Latin 
 me died out before ne became prohibitive, or that ne became 
 prohibitive and displaced me, we should be obliged to suppose 
 that the distinction died out and arose anew, and this is not 
 probable. But the strongest argument against supposing that 7ne 
 bore the same relation to ne in I.E. that /u.7; did to ov in Gr. is 
 furnished by the state of affairs we find in Skr., supplemented by 
 that in Av., O.P., and Gr. The most important fact is that in 
 Vedic Skr. md'' is used only with the injunctive. So far as we 
 can judge, the older Avestan^ was the same as the older Skr.; the 
 Gathas have md only with the injunctive. In O.P. md generally 
 has the injunctive, sometimes the optative. In Gr. the use of \i.t] 
 with the aorist subjunctive, contrasted with the use of the same 
 particle with present imperative, would seem to be a continuation' 
 of the use of the negative with the aorist injunctive, which Avery's* 
 
 ■Grassmann, sub v. ma; Delbiiick, Syn. Forsch., V, 358. The only excep- 
 tion of importance, if not absolutely, is the use of iiia with the optative hlntjenia, 
 which itself does not occur without ma. 
 
 'Spiegel, Alter. Gr., 520. 
 
 3C/.C.\N. E. Miller, A.J.I'., 418 f., and DelbrUck, Syn. Forsch., V, 358. 
 
 M-A.OS.. 13, 326 f. 
 
NEGATIVES OF THE INDO-EUROPEAN lANGUAGES 39 
 
 Statistics show to have been the most common tense, especially in 
 the earliest Sanskrit. Possibly, too, the more energetic character 
 of the perfect in prohibitions in Latin and the preference for the 
 perfect after the negative in Oscan is to be brought into this 
 connection.' 
 
 Now it can hardly be that ma (I.E. mc') was originally used 
 freely with volitive forms and later was restricted to use with the 
 injunctive, for, even in Skr., the injunctive was the least distinct- 
 ively volitive of all the volitive forms and probably at a period 
 not much antedating the Rig Veda was not modal at all." The 
 conclusion is, then, that me in I.E. was, in a way, prohibitive, but 
 used only with the injunctive or, in other words, that the pro- 
 ethnic method of expressing prohibition and negative wish was 
 by means of mi- and the injunctive. ^ This conclusion is sup- 
 ported by the fact that in Vedic Skr. the imperative has no nega- 
 tive and by the restricted use of the negative with the imperative 
 in Latin. The true original force of mc appears when we con- 
 sider the probable original value of the injunctive. If the injunc- 
 tive was non-modal,'' /. e., did not express will or ivish, then it is 
 plain that in a collocation of me and injunctive expressing nega- 
 tive will or wish the modal idea must have been expressed by 
 the me, and only the verbal idea was conveyed by the injunctive. 
 So do)i''t in English is strictly speaking not the negative of a 
 following verb, and there may be no conscious connection with 
 the nesfative not, but it is an expression of prohibition or nega- 
 tive will to which various verbal ideas may be added by means 
 of the colorless infinitive. Indeed, instead of believing that me 
 was originally a negative of a modal form, we may rather con- 
 jecture that the modal force of the injunctive proceeded from 
 this very use with me in prohibition, or, at least, that the develop- 
 ment of modal force was thus greatly assisted. In the collo- 
 cation of me and injunctive me was negative volitive, the injunc- 
 tive verbal, later me was felt as negative and the injunctive as 
 volitive verbal. 
 
 ' Cf. Buck, Verb-system, 140; Elmer, A.J. P., 15, 115 f. 
 
 * Note the increasing modality of the injunctive, as shown by Avery's statis- 
 tics. The ratio of non-modal to modal cases in R.V. is 1:1, in A.V. i : 10. Cf. 
 Thurneysen, K.Z., 27, 172 f.; Brugmann, M.U., 3, i, f., and Grundriss, II, 414. 
 
 3 Delbriick, Syn. Forsch., IV, 147. 
 ■♦ Cf. Thurneysen as above. 
 
40 NEGATIVES OF THE INDO-EUROPEAN LANGUAGES 
 
 So long as the original feeling was kept up, as in Vedic Skr., 
 md {me) could not be extended to other volitive forms. But we 
 are not surprised to note that those other volitive forms some- 
 times took the negative used with non-volitive forms. So na was 
 used with subjunctive, optative, and even with the injunctive, but 
 never with imperative. But later in Skr., when md was felt as a 
 true negative, its use could be and was extended to other volitive 
 forms;' and the same thing took place in Later Av., O.P., Arm., 
 Gr., and Alb. On the other hand, in Germanic and Balto-Slavic 
 the process, begun in Vedic Skr., of extending the use of fie to 
 volitive forms was consistently carried out, and the expression of 
 prohibition thus produced drove out the older me with injunctive. 
 In Italic and Celtic, in the same way, the convictional negatives 
 came to be used with volitive forms, and the old form of prohibi- 
 tion died out. Further, in these languages one or another ablaut 
 form was specialized and became with more or less consistency 
 the special negative of volitive forms. 
 
 'To imperative, optative, future, infinitive, and gerund. Cf. Speijer, § 353, 
 4. The use of md with augmented past forms was an extension of the modal 
 force of the injunctive rather than an extension of the use of ind. 
 
Frank Hamilton Fowler, born November i8, 1866, at Brad- 
 ford, Illinois; entered the Preparatory School of Lombard University 
 in September, 1883. and the College of the same institution in Sep- 
 tember, 1885; graduated from Lombard University in June, 1890; 
 student at the Johns Hopkins University during the year 1890-92; 
 teacher of Greek and Latin in Peaster Academy, Texas, during the 
 year 1891-92; Fellow in The University of Chicago 1892-95. 
 
RETURN TO the circulation desk of any 
 University of California Library 
 
 or to the 
 
 NORTHERN REGIONAL LIBRARY FACILITY 
 BIdg. 400, Richmond Field Station 
 University of California 
 Richmond, CA 94804-4698 
 
 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS 
 
 • 2-month loans may be renewed by calling 
 (510)642-6753 
 
 • 1-year loans may be recharged by bringing 
 books to NRLF 
 
 • Renewals and recharges may be made 
 4 days prior to due date 
 
 DUE AS STAMPED BELOW 
 
 JAN 1 8 2006 
 
 DD20 15M 4-02 
 
U.C.BERKELEY LIBRARIES 
 
 CD^b^fla7bfl 
 
jar'!!lK.-r: 
 
 ^« " '.■1: