LATIN SENTENCE CONNECTION Latin Sentence Connection By CLARENCE W. MENDELL, Ph. D. New Haven: Yale University Press London: Humphrey Milford Oxford University Press MDCCCCXVII Copyright, 1917 BY Yale University Press First published, January, 1917 TO E. M. 357399 PREFACE The study of the syntax of the verb in Latin has for the most part taken the form of a careful examination of coordination and subordination and of the develop- ment of the former into the latter by way of parataxis. The present investigation is an attempt to discover a more fundamental standpoint for the consideration of sentence relations, and to do away with the somewhat artificial distinction between coordinate and subordi- nate by means of a more thorough understanding of the nature and origin of each. It is based in part on my own complete collection from Tacitus of the in- stances of adjacent sentences not connected by con- junctions ; in part on the results of a special study of Cato and Sallust and the younger Pliny entire, of some three hundred pages of Cicero and about the same amount of Seneca, of three books of Caesar, three of Livy, four of Quintilian, and four lives of Suetonius ; in part on material drawn from casual reading. My collections include some fourteen thousand cases aside from many listed by citation but not actually taken off on cards. For Plautus and Terence I have made use of Bennett: Syntax of Early Latin, Morris: The Inde- pendent Subjunctive in Plautus, and the standard text books on Latin syntax. Citations are made from the latest editions of the Teubner texts. To Prof. E. P. Morris of Yale College I have the via PREFACE deepest sense of gratitude, which it gives me pleasure to express, for his unflagging interest and constant help in the pursuit of the present investigation. The study is the outcome of his suggestions, made in his syntax seminar and in his Principles and Methods in Syntax, and the discussion with him of the various phases of the subject has made possible the present treatment. I would acknowledge also the help derived from the thesis of Miss Irene Nye {Sentence Connec- tion, Yale University, 1912) and from the welcome criticism of my colleagues, Professor Hemingway and Doctor Nichols. C. W. M. New Haven, December, 1916. CONTENTS PAGE Chapter I. Introduction. The sentence : difficulty of definition; psychological processes behind the for- mation of sentences ; the objective result. Sentence groups : no sentence in consecutive discourse iso- lated; sentence content and sentence function; necessity for expression of sentence relations . . 1 Chapter II. Introduction (Continued). Fundamen- tal types of connection : Repetition, Change, Incom- pleteness; their general characteristics; order of investigation 13 Chapter III. Repetition. 1. Repetition of Content: its fundamental characteristics; its various types; the relations expressed by it ; the conjunctions used to supplement its force. 2. Repetition of Function : its characteristics, typical uses, and significance; conjunctions used to supplement its force ... 21 Chapter IV. Retrospective Incompleteness. Defini- tion of the term Incompleteness. Semantic and functional incompleteness 86 Chapter V. Change. Definition of the term. Seman- tic change : necessarily abrupt to be effective ; narrow range of relations expressed ; supplementary conjunctions. Functional change : wider range but less precision; relations expressed; conjunctions used to supplement force Ill Chapter VI. Anticipatory Incompleteness. Types corresponding to those used with retrospective force : relations suggested correspondingly vague. X CONTENTS PAGE Anticipatory force due to irrelevant emphasis : anti- cipates a contrast to follow. Anticipatory force due to semantic vagueness: anticipates an explana- tory clause to follow. Use of supplementary con- junctions. Functional incompleteness : anticipatory use of the imperative mode the most important type. 141 Chapter VII. Parenthetic Incompleteness. The use of injected words logically incomplete in meaning, to give tone to a statement. So-called Parataxis. Resultant constructions 191 Chapter VIII. Conclusion. Summary of the types of sentence connection disclosed by the investigation. Indication of possible applications of the principles discussed. Suggestion of important extensions of the investigation 207 LATIN SENTENCE CONNECTION CHAPTER I INTRODUCTION^ Some working definition of the term sentence is a primary requisite to a study of sentence connection. Not so much for the purpose of sharply dividing sen- tence from sentence, for this is of no great impor- tance; and not for the purpose of distinguishing between sentence and clause, a distinction which means nothing. But it is necessary, in order to get clearly in mind the mental processes lying behind the making of a sentence, to understand what the sentence is and what it represents. And so the difficulty in finding a satisfactory definition is not altogether unfortunate; the ultimate definition is of less value than an under- standing of the workings of the mind in producing a sentence, an understanding which should result from the attempt at definition. That there is a real difficulty in the satisfactory formulation of such a definition is obvious from the many attempts that have been made by scholars in widely differing fields. The dictionaries have as a rule settled upon some form of the statement that ''a 1 The discussion of the sentence in this chapter is based very largely on Morris: On Principles and Methods in Syntax (especially Chapter II); Wundt: Voelkerpsychologie (Volume II); Paul: Prinzipien der SprachgeschicMe, 4th edition, (Chapter VI) ; and, to a lesser degree, on Sheffield; Grammar and ThinTcing. 2 LATIN SENTENCE CONNECTION vS sentence is a group of words" with, of course, cer- tain properties which they describe. The New Eng- lish Dictionary, for instance, offers this: "A series of words in connected speech or writing, forming the grammatically complete expression of a single thought." The Century Dictionary and Cyclopedia is more concise: ''A form of words having grammati- cal completeness." Largely under the influence of psychological study, it has become customary for the students of language to discard this type of definition and to condemn it without qualification. This attitude is itself not alto- gether right. For, from the objective point of view the sentence is a group of words. So far as the reader or hearer is concerned the sentence is a series of words which he must relate to each other in such a way as to understand their meaning as a whole. The mental process for him is a synthetic one, so that the essential unit is the word and the ultimate end the combination of the words into a larger unit which shall convey to him a single idea. But, from the point of view of the speaker or writer, the reverse is the case, and it is for that reason that any definition of the sentence as a group of words is inadequate and mis- leading. It looks at the sentence only in an objective way — as a finished product — without reference to the process by which it came to exist. This process is something as follows. An idea first presents itself to the mind, not articu- lated into words, but as a unit. This first conception of the idea is very likely to be vague and ill-defined. INTRODUCTION Some external stimulus gives to the mind a general impression, or something within the mind itself, some thought already conceived, suggests a further idea. Instinctively the mind focuses on that undefined impression ; instinctively it begins to analyze the con- cept. And from the whole conceptual unit, the mind extracts the part which is really pertinent to the matter under discussion, to the train of thought which it is pursuing. This part it analyzes into words for presentation, and the result of this analysis for presen- tation is the sentence. For example, the discussion of colour in a landscape might suggest to the mind a certain large white house built of wood and standing among trees. The first impression is of the whole object in its setting. The mind proceeds to analyze this impression, and from its whole selects the relevant part, the colour of the house. This item it analyzes into words and presents in some such form as. The house is white, or. There is a white house. This is a simple example, but the speech form which finally results from the mental processes should be noted more carefully. Each word used in the presen- tation of the analyzed concept has in itself a kernel of meaning, that is, it represents an idea which is a com- ponent part of the larger idea, presented by the whole. If the expression of the concept is clear, the various component ideas, conveyed by the separate words employed, are so presented, and especially their rela- tions to each other are so presented, that the hearer or reader is enabled, by combining them again in his own mind, approximately to reproduce there the rele- LATIN SENTENCE CONNECTION vant part of the conceptional unit which first appeared in the mind of the speaker or writer. From this point of view, then, the sentence is the unit and the analysis into words serves only to convey the concept from one mind to another. A working definition, therefore, of the sentence, less misleading than the dictionary definitions and more useful in any study which seeks to understand the means employed by language to express ideas, would be the following: A sentence is the verbal expression of the relevant part of a single idea, conceived and analyzed in the mind of its author. With such a definition one further danger of mis- conception, and a grave one, remains ; namely, that of considering the concept from which the sentence arises not only as a unit but as an isolated unit. It is not an isolated unit. All thought is associative. Too much emphasis can hardly be given to the statement made above, that either some external stimulus, some- thing seen or heard, suggests to the mind each con- ceptual unit, or else such a unit is suggested by some- thing in the train of thought, by some detail of another concept already analyzed by the mind. In the train of thought represented by consecutive discourse the latter is, of course, regularly the case. And the pro- cess is as follows. When one concept has been ana- lyzed and its relevant phase presented, the component parts once more combine and again form a unit in the mind, while the next idea, suggested by the first, be- comes the focus of the mind's attention and is in turn analyzed in the same way, itself to return pres- INTRODUCTION ently to the form of a comparatively vague unit. Furthermore the mind is capable of embracing more than a single idea at a single time. So that while one concept is immediately before it for analysis, not only are the previous ideas present in less distinct form but in the same way it is, vaguely at least, con- scious of the further idea or ideas which the one immediately before it suggests. The individual con- cepts expressed in sentences may be and often are themselves component parts of a larger unit already conceived in a more or less vague way in the mind. In such cases the sentences in a group of sentences are analogous to the individual words in a single sentence. Two types of thought process are thus suggested: the first, that in which idea follows idea in succession as the result of immediate suggestion; the other, the type in which the thoughts are component parts resulting from the deliberate analysis of a larger con- cept. The important thing to be considered at present is this : that in neither type are the sentences isolated units. They are very essentially related to one another. Furthermore, whichever type of expression is used in any given instance, each sentence is related in thought to the adjacent ones and if the train of thought is properly sustained and adequately expressed, these relations will be conveyed to the mind of the hearer. Language is successful as a means of expressing thought only in proportion as it enables the hearer or reader not only to grasp indi- vidual ideas but to group these into a conceptual unit LATIN SENTENCE CONNECTION which is the duplicate of that in the mind of the speaker or writer. A little study of the sentence by itself will make the whole question more clear. The analysis of an idea separates it into a series of ideas each more or less complete in itself. But that same series of ideas might conceivably be arrived at as the result of the analysis of some quite different concept. Or, from the opposite point of view, these ideas, combined in one way may be the component parts of one concept; combined differently they may be the component parts of a concept quite different. Take, for example, the illustration already used. The mind having analyzed its impression of the house visible in the landscape, and having chosen the relevant part, presents that part in the sentence, The house is white. There are three distinct component ideas represented by words and forming together the expression of a conceptual unit. They are (1) house, (2) being, existence, (3) whiteness. Had the idea originally been. White houses exist. That whiteness is a house, the three compo- nent kernels of meaning used in expressing it would have been the same. The mere statement of the com- ponent elements would not define the idea that the author intended to express. In presentation therefore in language, inasmuch as the object is to reproduce the original concept in another mind, something more than the series of com- ponent ideas must be communicated. Each one must be so expressed as to convey not only its essential underlying concept but also its relation to the other INTRODUCTION parts and its importance with relation to them. The words then which are used to express the analyzed concepts, are not a simple series of bare names desig- nating the elements discovered by analysis; in some way, the relations between them are also expressed. This is equivalent to saying that each word has, besides its kernel of meaning, a function in the sen- tence. This function is very variously expressed. In Latin, inflectional endings do perhaps the bulk of the work ; but in all languages, order of arrangement, the use of prepositions, and, in conversation, tone, and emphasis, — these and other elements go to make up the expression of function. It is perfectly conceivable that there might be cir- cumstances under which such expression would be unnecessary and the mere statement of the compo- nent elements would successfully convey the thought unit. If speaker and hearer were so thoroughly inti- mate that their accumulated store of knowledge was practically the same and that their modes of thought were identical; and if besides this, both were at the time following the same line of thought and if the idea expressed was largely objective, then the naming of the elements would be sufficient. But the existence of such a situation is, of course, a rare occurrence. If the relation of the various elements to each other is not fully expressed, there is a real danger, almost a certainty, that the hearer will combine them in his own way, and that his way will be a different one from the speaker's. Transfer this reasoning to a larger field. A group 8 LATIN SENTENCE CONNECTION of sentences might be strung together, each embodying an idea. If nothing further were expressed in them, the hearer might perfectly grasp each of the ideas pre- sented, but the total effect made by the group would depend on the relations which he himself made between them, depending on his own mode of thought and atti- tude. To convey his own idea of relation, therefore, or, in other words, to convey his total thought, the speaker must express not only the individual ideas in sentences but also the relation between them as it exists in his own mind. In dealing with the individual sentence and the words which are used to express its meaning, this fact has been generally accepted and the means of express- ing relation between word and word have been much studied. This is by no means so true of the sentences as wholes. Their relations to each other have not been worked out with the same thoroughness. The reason is not far to seek. First, there has been a failure to recognize that all adjacent sentences are related (at least in the thought of their author) to each other, and that therefore the subject of investigation is really the means which define the kind of relation. Second, one type alone of sentence connection, and that the most mechanical, has been so overemphasized as to appear to be the only type. This is the conjunctional usage. Handbooks of grammar have loaded upon conjunctions practically all the work of sentence connection, with a grudging recognition of the use of repetition and kindred means. The result has been the dangerously systematic division of all sentence relations into co- INTRODUCTION ordinate and subordinate, with a free use of the ill- defined term parataxis to describe those instances which show a subordinate relation without a subordi- nating conjunction. Such a division implied what was generally accepted, that the remaining sentences — those without conjunctions — were unrelated. This general treatment is a survival of the period of syntactical study before the psychological side of lan- guage was taken into consideration. It is one-sided and begins with a wrong point of view. The funda- mental fact is that all contiguous sentences in con- nected discourse are related, or more exactly, the thought units behind them are related. The mere fact, then, of their juxtaposition indicates to the reader or hearer the fact of their relation. Just what is this relation which they bear to each other, he must dis- cover if he is to understand what he hears or reads. And it is for the speaker or writer to express this. If he does not specifically express it, one of two possible results must follow. First, in the event of there being only one conceivable relation between two adjacent sentences, the whole thought of the speaker or writer is adequately conveyed. But second, if more than one relation is conceivable — and this is usually the fact — then his meaning is not clear and his audience is apt to choose the wrong relation. In spoken conversation it is not at all unusual to hear sentences following one another without any expression of relation, any at least of a tangible nature, discernible in the words, which, however, ade- quately convey the thought of the speaker. This is 10 LATIN SENTENCE CONNECTION made possible primarily by the limited audience rather than by the means at the disposal of the speaker which the writer cannot use. It is perfectly true that tone and emphasis and other musical elements enter into the expression of thought relation and are of extremely great importance. But a little consideration will show that this is not the prime factor. It is necessary only to consider various types of spoken discourse. Ob- viously the type in which tangible connectives are most conspicuously absent without causing ambiguity is that of intimate conversation between friends. The audience is limited. The hearer's knowledge of the attendant circumstances and his processes of thought approximate those of the speaker. But if the audience is enlarged, there is a greater variety of hearers, each with his own background of experience and his own mode of thought. To convey his own meaning, his own idea of thought relation, to each, the speaker must be more explicit. In addressing a large and unfamiliar audience this becomes all the more imperative. Now the same degree of difference between types of discourse appears also in written matter. But inas- much as the musical elements — including gesture and expression — are no longer available, written matter requires always more exactness of connectives than spoken. The nearest approach to the intimate con- versation is, of course, a letter. Much can be left to the understanding of the friend or, in other words, to the fact that his background of experience and his thought processes will lead him to combine the sepa- rately expressed thoughts into a conceptual whole INTRODUCTION 11 corresponding to that from which the writer evolved them. No doubt the intimate reader of the letter imagines, from past experience, tone, tempo, emphasis and even gesture and expression. But with the audi- ence enlarged, all such quick appreciation becomes less probable, and the same increased need of precision is felt that was found in speaking to a larger audience. In the case of matter written for the general public, the audience is indefinitely enlarged, till it includes people of different race and age from those of the writer. Here, then, the expression of sentence relation is of prime importance. Such, of course, is the material at the disposal of the classical scholar, for this is just what constitutes the texts which he seeks to understand and interpret. He has before him exactly what the writer considered adequate for the reproduction of his own ideas. To be sure, he has a very different background from that of the writer. The experience which he brings to bear upon the interpretation of what he reads is altogether unlike that of the author, as is also his mode of thought. These are likely to be different too from those of the first readers of the literary remains of Rome. But this makes it all the more important for him to understand so far as possi- ble all the devices at the disposal of the writer for the expression of thought relation. Obviously conjunctions do not perform the entire task. The relations are largely expressed by other and less easily recognized means. But with the entire material available — for in written discourse nothing is lost to us by the lapse of time — tone, emphasis and so 12 LATIN SENTENCE CONNECTION on were never an aid in written matter — it should be possible to discover just what means were actually used by Roman writers. It is not equally probable nor even desirable that these means appear always distinct and clear cut. Language has never been a purely mechanical instrument that can be taken down into parts susceptible to hard and fast classification. And in the department under consideration, into which rhetorical influences, often extremely subtle, are con- stantly entering, this is unusually true. Nor should it be expected that the means discovered will be always simple; it is more reasonable, in view of the other phenomena of language, to look for complex usages, for several overlapping types of usage employed in a single instance. Finally, it should be borne in mind that no study of any language can be carried back to its most primitive forms. Especially is this true of Latin, in which all of the material shows the language in a comparatively late and highly developed stage. Changes of usage can be traced, but the study of the origins of types of sentence connection would be perilous. The practical necessity at present is descriptive work, the presenta- tion of the means (discovered by analysis) employed in the language to express the thought relations that lay in the mind of the writer. CHAPTER II INTRODUCTION (Continued) The purpose of the first chapter was to present in a general way the psychological processes which lie behind the making of sentences and the expression of their relations to one another. An understanding of these processes is necessary as a guide to any study of the actual concrete means which result from them. Such a presentation furnishes the general principle on which must be answered the first of the three great questions which arise with regard to the present prob- lem as they do with regard to every linguistic prob- lem:^ first, what are the psychological processes be- hind the linguistic phenomena; second, what are the means used to express these ; third, what is the result- ing linguistic form. The order in which these questions are put represents the natural order of progress in the development of linguistic forms, not the order in which such forms must necessarily be investigated. It is imperative that the general psychological principles should be clearly in mind during any investigation, but only as a guide. For while these principles answer in very general terms the first question, they do not answer it specifically for particular cases. Just as 1 Cf. Morris: On Prmciples and Methods in Syntax, p. 115. U LATIN SENTENCE CONNECTION they are the necessary background for one who is going to investigate, so in their application to each particular case they become the goal of the entire in- vestigation. For the object is constantly to determine in every individual instance just what was the thought in the writer's mind. The general principles on which all minds alike work in producing sentences, are then a guide in the study of the means actually found in use, toward the comprehension of the exact meaning of the writer in each specific case. The practical problem at present is, then, to answer the second question : What means are used in language to express the relations existing in the mind of speaker or writer between ideas embodied in sentences? Or, more narrowly for the moment. What means are so used by Latin writers? To do this, it is necessary to consider the sentences for the time being objectively as groups of words, themselves forming groups of word combinations. Safety, in such an investigation, lies in keeping rigidly to what is objective and concrete. It may be possible in conclusion to win to some generaliza- tions, but it is essential first to discover and present the actual means employed. The most natural place in which to look for the expression of sentence connection is, of course, in the second of two contiguous sentences. It is far easier to refer to what you have already said and to express your thought relations by such reference than to pre- pare in advance for what you are going to say and so to anticipate the relation. It is not at all surprising, therefore, to find the bulk of the work of sentence con- INTRODUCTION 15 nection done in the second sentence. The more naive and unpremeditated the matter, the more generally is this true. But no written matter is unpremeditated and no Latin that has come to us can be called naive. We are dealing with deliberate and consciously elabo- rated material. And the more carefully any piece of writing is elaborated, the greater tendency will there be to anticipate connections in order to facilitate the understanding on the part of the reader of what he is reading. It does not, however, follow that because connection is anticipated in the first sentence, there will be no further expression of it in the second. Such reinforcement is very common, and the more means used, the more precise will be the expression of the author's thought relation. The others reinforce or further define the first. An analysis of the means of sentence connection actually used by Latin writers makes possible a rough division into three groups according to the chief ele- ment which gives to each its power to express thought relations and so convey them to the reader. The first of these is repetition; the second, change; the third, incompleteness. Each requires some preliminary explanation. By their very nature the first two can occur only in the second sentence of any pair. The first consists in the repetition in the second sentence of any element of the first, the element repeated being the bond which unites the two and defines their relation. When a child says, ''The man has a dog. The dog is yellow," he is unconsciously making use of this element. The repeti- 16 LATIN SENTENCE CONNECTION tion of the word dog holds the concept behind the word before the mind of the hearer until a new idea is built around it. The same principle underlies some of the most conscious rhetoric. ''If I ascend up into heaven, thou art there; if I make my bed in hell, behold thou art there." Obviously the effects produced in these two illustrations are very different, but the essential element is the same. It must be constantly remem- bered that this repetition does not make a relation be- tween the two sentences. Experience has brought it about that the very fact of juxtaposition indicates a relation and the relation so indicated existed first in the mind of the speaker. It will be found that repeti- tion does distinctly draw attention to the fact of rela- tion as does also incompleteness, but its importance lies not there but in its definition of that relation. It can be shown to what extent and how it so defines the relation; at present it is necessary merely to indicate the nature of this particular element. It is this fact, that the means here considered are de- fining rather than relating means — that the sentences are always related and require only an indication of the kind of relation — that makes it possible for two such opposite elements as repetition and change to serve in the same capacity. They define different types of relation. If the search were for those ele- ments which made relation it would end with the fact of juxtaposition, and repetition and change might be from this point of view mutually exclusive.^ But our search is for all the means which define every possible 1 Cf . Miss Nye, p. 27. INTRODUCTION 17 relation, and change and repetition can thus work side by side. Just as a word in the second sentence may repeat a word from the first, so a word in the second may, by abrupt change, either of meaning or of func- tion, stand in contrast with a word in the first. ' 'John went sailing. William was afraid." Experience tells us that these sentences are related because of their juxtaposition. The change of name, together with the abrupt change of meaning in the predicates of the two, defines their relations as an adversative one. It is a question to be considered later whether such contrast can stand independently as a defining means ; whether there is not present always some element of repeti- tion, however slight, in contrasted sentences. But assuredly the determining factor in the definition is contrast or change. It is obvious that the use of this element serves less to point out the fact of relation than do the other two. Hence the fact that it will usually be found to be anticipated in some way, and hence also its value for abrupt rhetorical effects. It is not true, however, that it does not at all indicate the existence of relation, for this is actually done by the use of contrasted words, less obviously than in the case of repeated words, but none the less truly. The third element has a characteristic which dis- tinguishes it from the other two : it may occur in either of the two sentences. It is perhaps more patent when found in the second, just as it is probably true that the earliest expression of sentence relation was con- fined to the retrospective type. But a little familiarity with its usage makes absolutely clear its employment 18 LATIN SENTENCE CONNECTION in the first sentence with an anticipatory force. It is this that gives it its great importance. For one of the greatest advances in clear writing and one of the greatest aids to rapid understanding of what is written is the development of means to suggest in advance the relation of each sentence to the one following. By incompleteness of meaning nothing so general is intended as the perfectly true and obvious fact that no sentence in consecutive discourse is complete in meaning by itself without its context. Incompleteness in its present use is something far more concrete; it lies in a specific word or phrase whose logical meaning is not clear without reference to something outside of its own sentence. ' 'Afterwards John sat down. ' ' This sentence is syntactically complete, but, by its very meaning, ''afterwards" is incomplete without refer- ence to what has preceded. Again in the sentence, "He spoke the following words," there is syntactical completeness but not logical, until something is added to give meaning to "following." Such words as deictic pronouns, comparatives, verbs in a relative tense, these all have this connective function by virtue of their incompleteness of meaning. Such a threefold division of the types of sentence connections is not hard and fast. Instances occur and occur frequently in which more than one element is made use of and in which it is quite impossible to say which has the determining influence. And more than this, they are often quite inseparable in one and the same word employed as a means of sentence connec- tion, as in the demonstrative pronoun. Undoubtedly INTRODUCTION 19 the demonstrative most frequently repeats the content of a noun from a preceding sentence. For example, Tac. Hist. 1.57.1: Proxima legionis primae hiberna erant et promptissimus e legatis Fabius Valens. Is die proximo coloniam Agrippinensem . . . ingressus, imperatorem Vitellium consalutavit. Is clearly re- peats the Fabius Valens of the first sentence and therein lies an element of connection. But also the Is by itself is incomplete. Alone it has no content, but by its very meaning must refer to some noun. To such an extent is this true that when a demonstrative is used and there is no noun in a preceding sentence to which it can refer, its very incompleteness forces the reader to suspend his judgment until another sentence or clause furnishes the key to its meaning. It has then become an anticipatory means of expressing thought relation. It is altogether natural that the two elements should be thus united in a single type. The expression of thought relation was not a deliberate invention but a natural development in language. The three elements — repetition, change, and incompleteness — were not consciously made use of; at any rate, not at first. They are discovered only on analysis. There is no reason why they should not all appear in a single case. But this does not make it less practical to proceed to our investigation on the basis of this division, for the object in hand is simply to find, so far as possible, all of the tangible means of sentence connection employed, and to determine the relations which they express. It will be convenient therefore to begin with the large 20 LATIN SENTENCE CONNECTION group in which Repetition, in one form or another, is the diief factor. For repetition is, so far as it is pos- sible to judge, the simplest and most natural of all the means employed. Continuing with the elements whose influence is effective in the second sentence of a related pair, I shall take up next the principle of Incomplete- ness in so far as it is retrospective in character. Fol- lowing this, a study of the element of Change will com- plete the survey of retrospective means of connection. The important field of anticipation of relation, that is, the study of those instances in which the relation is expressed in the first of two related sentences, is treated in the next chapter, that on Anticipatory In- completeness. Finally, I shall discuss, in a chapter on Parenthetical Incompleteness, instances in which a syntactically independent phrase is injected into the middle of a sentence in such a way as distinctly to modify its tone or meaning, the relation between the two clauses being indicated by the incompleteness of the injected phrase. This order of consideration will necessarily divide the discussion of the principle of Incompleteness, but with the elements of the problem so interwoven as they are in the present question, some violence cannot but be done to logical order and distinct advantages will be found in making the dif- ferentiation between retrospective and anticipatory means, the factor to determine the order of investi- gation. CHAPTER III REPETITION Igitur in stagno Agrippae fahricatus est ratem, cui superpositum convivium navium aliarum tractu moveretur. Naves auro et ehore distinctae. These two sentences from Tacitus {Ann. XV.37.5) leave no doubt in the reader's mind that they were related to each other in the thought of their author. And this is true in spite of the fact that there is no conjunction to give warning of such relation. The opening word of the second sentence is the first mechanical sign of the relation. It repeats in its entirety the content of the navium of the first sentence and so keeps that item before the reader's mind until the new idea is built around it. The relations in which naves finds itself are therefore quite different from those which surround navium, but the objects which the two words represent are identical. If, instead of naves, Tacitus had used Jiae, exactly the same effect would have been produced so far as the mental process in the reader 's mind is concerned : hae would simply have selected the detail from the first sentence which was to form the starting point of the second. Had he used quae, the result would have been essentially unchanged in spite of the fact that we should then have had what we call a subordinate sen- 22 LATIN SENTENCE CONNECTION tence. If, for variety's sake, we substitute lintres for naves, still the connecting element remains the same: the content of navium is repeated and repeated in its entirety. Nam si, quae nunc temporis causa aut decrevit senatus aut populus iussit, in perpetuum servari oportet, cur pecunias reddimus privatisf cur publica praesenti pecunia locamusf cur servi, qui militent non emunturf cur privati non damus remiges, sicut tunc dedimusf (Livy XXXIV.6.17.) Here again is direct repetition of a word, but with an obvious difference. There is no concrete or defi- nite concept represented by the word which the repe- tition serves to hold before the reader's mind until the new sentence develops the writer's thought about this concrete object or concept. There is, in the pres- ent case, repetition rather of a mechanical part of the sentence; the chief thing repeated is really the func- tion which the cur serves rather than any image which it suggests. This is perhaps more obvious in a more extreme case : Livy XXIII.9.5 : sed sit nihil sancti, non fides, non religio, non pietas; audeantur infanda, si non perniciem nobis cum scelere ferunt. In this instance the function alone is repeated. In audeantur the word itself, the person, even the voice is changed. Sit and audeantur have but one thing in common, their subjunctive force expressing a hypothetical command ; yet this repetition is sufficient to impress on the read- er's mind the sense of relation between the sentences as it existed in the thought of the writer. It is evident that there are represented in these REPETITION 23 examples two distinctly different general types of repetition. In each example the repetition in one sentence of an element from another is the mechanical sign of relation. And in each example it is enough to suggest the relation. But the relations so sug- gested are different. In the one instance the second sentence represents a development of thought from the first; it expresses a new idea logically proceeding from the thought expressed in the first sentence. In the other example the second sentence presents a new idea, but it does not proceed logically from the first. It is rather parallel with it, proceeding as did the first from something antecedent to both sentences. In the second example, therefore, either sentence could have been omitted by the writer without destroying the sequence of thought in the paragraph. Fulness of expression would have been sacrificed but not logical sequence. The two types are not always sharply defined. Rather they shade into each other and overlap. But, apart from the cases that lie on the borderline, the types are clearly distinct and the ground of distinc- tion is readily discovered. It lies in the general char- acteristics of words as they are used in language, pointed out in the first chapter (page 7). That is, all words used in connected discourse have both a kernel of meaning and a function in the sentence. Accord- ing as one or the other of these two properties is repeated, we have one or the other of the two types of repetition noted. Semantic and morphological repeti- tion they might be called, but the psychological pro- 24 LATIN SENTENCE CONNECTION cesses are perhaps more clearly kept in view if the terms '* repetition of content" and ^'repetition of function" are retained. These phrases will also prove less confusing in such instances as Livy XXIV.2.10, in which apparent repetition of case has no bearing on the sentence connection: acceptique a plebe primo impetu omnem praeter arcem cepere. Arcem opti- mates tenehant. The case of the word repeated, arcem, is obviously the same in both sentences but its function is not repeated ; clearly this is an instance of repetition of content. On the other hand there may be repetition of content witli repetition of function. The substantive used in the two related sentences, representing an object or a concept, may be in the same relation in each sentence to the remaining parts of the sentence. An example from Cicero {Pro P. Sulla 5.14) will illustrate this: Multa, cum essem consul, de summis rei puhlicae periculis audivi, multa quaesivi, multa cognovi; etc. Such cases, however, belong most naturally with the instances of repetition of function, which proves to be the determining factor in expressing the sentence relations. It will be convenient to illustrate the use of repeti- tion by dividing it along the line of this fundamental distinction. After each type is amply illustrated, it will be easy to find other distinguishing marks which belong to each, and not difficult, I think, to discover the sentence relations which each indicates. Finally, it will be worth while to study the use of conjunctions in the expression of these same relations, to determine REPETITION 25 what is the fundamental element in the expression, and just what is the function of the conjunction. The repetition of content alone is taken first for illustra- tion because it has the appearance of being the more natural. It follows the line of simple straightforward associative thinking. An idea is selected from one sentence and made the starting point of the next. Repetition of Content The essential characteristics of this type of con- nection are, first, a word in the second sentence actually calling up before the mind either some con- crete object represented by some word or words in the first sentence, or else some distinct though abstract concept represented in the same way; and second, the fact that in the two sentences the words so used show different relations to the rest of their respective sen- tences. It is obvious that the clearest examples of the usage will be those in which substantives are used, especially such cases as show the same noun occurring in the two sentences. Bed nostra omnis vis in animo et corpore sita est: animi imperio, corporis servitio magis utimur (Sallust, Cat. 1.2). Like the illustra- tion already cited from Tacitus (page 21), this instance from Sallust shows a noun in the second sen- tence repeated from the first but serving a new func- tion in its new setting. This is a very simple method, perhaps the simplest, of expressing sentence relation, the sort of method that a child is apt to use when he begins to produce 26 LATIN SENTENCE CONNECTION what may with any justice be called consecutive narra- tive. It represents too a very simple thought proce- dure. It is therefore in the simpler authors that we expect to find the examples of it. It is, however, some- thing- of a surprise to find how largely they are con- fined to such writers. Although the principle involved is constantly made use of by all authors of all periods, it is only in the inscriptions and in such unadorned prose as that of Cato that this crudest form of repe- tition appears with frequency. Cato makes no attempt to avoid the monotony of literal repetition : Eos lapide consternito: silapis non erit, perticis . . . consternito: si pertica non erit, sarmentis conligatis. {R.R. XLIII. 1.) The lapide, lapis and the perticis, pertica illustrate the point ; the repetition of consternito and of si belong to a later discussion. Ita aedifices, ne villa fundum quaerat <.neve fundus villam';>. Patrem familiae villain rusticam bene aedificatam habere expedit, etc. {R.R. III.l.) Whether the words in brackets are cor- rectly restored or not, the villam of the second sentence repeats in its entirety the content of the villa of the first. Similar examples, though rare, are not lacking in the more rhetorical writers. A very few mil be suffi- cient to illustrate the type: Caesar, Bell. Gall. III.18.7: ad castra pergunt. Locus erat castrorum editus, etc. Cicero, Pro Quinct. 5.22: Itaque ex eo tempore res esse in vadimonium coepit. Cum vadimonia saepe dilata essent, et cum aliquantum temporis in ea re esset consumptum neque quicquam profectum esset, venit ad vadimonium REPETITION 27 Naevius. Livy XXIII.10.9: Ita in castra perducitur; extemploque impositus in navem et Carthaginem missus, ne . . ., etc. Navem Cyrenas detulit tempes- tas, etc. Seneca, De Ira II.18.1: ea in educationem et in sequentia tempora dividentur. Educatio maxi- mam diligentiam plurimumque profuturam desiderat. Quint. Inst. Orat. 1.3.1: Tradito sibi puero docendi peritus ingenium eius imprimis naturamque per- spiciat. Ingenii signum in parvis praecipuum memo- ria est. Tac. Germ. 10.1: Auspicia sortesque ut qui maxime observant: sortium consuetudo simplex. Pliny, Epist. IV.22.4: Idem apud imperatorem Nervam non minus fortiter. Cenabat Nerva cum paucis. Sallust, Cat, 55.2: Ipse praesidiis dispositis Lentulum in carcerem deducit; idem fit ceteris per praetores. Est in careers locus, quod Tullianum appellatur. This last example from Sallust illustrates a fur- ther point than the mere repetition of content, namely, that the word repeated may occur in a sentence or clause other than the one immediately preceding. In this case the separating clause is short and is evi- dently treated as though it were an integral part of the first sentence. But often a word is repeated from some distance back in the narrative, very frequently a proper name. The nature of the connection is not different; the intervening sentences are for the time being overlooked or else their relation to the new one is expressed by other means. Such use of repetition as has been illustrated is a very simple and elementary method of calling atten- 28 LATIN SENTENCE CONNECTION tion to the relation of one sentence to another. The desire for variety, the natural and constant effort, both conscious and unconscious, to give more ''style" to what was written or spoken, led to the development of a great many substitutes for this direct repetition, to many variations not always recognizable as such. But before illustrating these it will be worth while to see just what this type of repetition in its barest form indicates as to the kind of relation between sentences. If speech were confined to the representation of action in the most exact form possible, there would be but one essential relation which a sentence could bear to its immediate predecessor. For action follows a definite line of progression. Cause and effect are consecutive and cannot be reversed. The time element is the one important element in determining the rela- tion of event to event. So, when language simply and accurately represents action, each sentence is subse- quent to its predecessor both temporally and log- ically. The description of an action given in the order of its occurrence needs nothing to make clear the rela- tion between sentences. To be sure, the unity given to an action by the constant presence of a given actor is often reproduced by the repetition of a word repre- senting that actor, but this is not essential. Speech, however, is freer than action. It has license to reverse or entirely to rearrange the order of sequence. Two events, actually occurring in one order of sequence, may be told of in speech in the opposite order, for speech looks back upon action and may pre- REPETITION 29 sent the details of the action as it will. The effect may be told before the cause. When, however, the natural order is upset, the hearer requires warning of some sort or he will instinctively assume that the action is continuous and that it took place in the order given. Furthermore, there is another way in which speech is freer than action. Speech may stop to describe something or to give personal observations on events or things. In other words, reproduction of action is by no means its only function. Such inter- ruptions will not be reversals or disarrangements of the normal order. At the same time there will not be the force of sequence which is furnished by the bare narration of action to carry along the thought of the audience. There will be necessary, therefore, some specific warning of the nature of the sentence relation. Transfer this reasoning to the parallel field of speech, the reproduction of abstract thought. The same general facts are true, although the time ele- ment is no longer the criterion of relation. Progres- sion is logical not temporal. And inasmuch as the reproduction of thought by speech is far more subtle (the audience has ordinarily far less pertinent expe- rience to call to its assistance), and as there is less apt to be one normal line of progression, so there will be need of more specific expression of thought relation. There is still a further phase of the representation of action and of thought to be considered. As so far discussed, both types of speech are the representation of something in motion: in the case of action, the motion is physical, the motion of concrete bodies: in 30 LATIN SENTENCE CONNECTION the case of thouglit, it is mental and intangible. But concrete objects and mental conceptions can be sta- tionary : tliey are not always in motion ; and language is employed to represent them in both aspects. De- scriptions of objects, the exposition of a single con- cept, inasmuch as they come within the range of expression by means of language, must, like every portion of the material at the disposal of language, be expressed by sentences, and the relation of each sen- tence to the adjacent ones must be expressed within those sentences. But in these cases there will be no order of progression at all, either temporal or logical, to guide the reader in his interpretation except in so far as the writer may adopt some arbitrary order to make clearer his meaning, which order then becomes of itself a guide to the understanding of the sentence relations. Whether, therefore, speech represents concrete ob- jects or abstract thought, or both together, and whether they are represented as stationary or in motion, there will be in any sort of discourse three relations which a sentence may have to the one imme- diately preceding it. It may be and usually is logically subsequent ; it may be, by a reversal of the usual order, logically antecedent ; or third, it may be logically coin- cident. The time element may or may not enter into consideration; the question is always one of the log- ical order of sequence. In narration of action this is determined by the temporal sequence. The third type is more common than might at first appear, for it includes all those cases in which a number of sentences REPETITION 31 have all a common relation to another sentence, and it contains also all instances of contrasted sentences. Such are the most general relations possible between sentences, and these are all that we can expect to find indicated by the broadest types of connection. More exact relations will be expressed by the more exact and specific subdivisions of the broad types of connec- tion. The examples of connection by means of direct repetition of meaning fall clearly under the first class of relations. The sentence in which the repeated word occurs is logically subsequent to the one from which the word is repeated. Seneca, De Ira II.18.1: Ea in educationem et in sequentia tempora dividentur. Edu- catio maximam diligentiam plurimumque profuturam desiderat. By the repetition of educatio from educa- tionem, Seneca selects one item from the first sentence around which to build a logically subsequent thought. In such an instance as the one cited (page 27) from Pliny {Epist. IV.22.4), the presence of the time ele- ment at first obscures somewhat the logical relation. Cenabat Nerva cum paucis, presents a fact contempo- rary with the fact of the first sentence: Idem apud imperatorem Nervam non minus fortiter. This is indi- cated by means of the imperfect tense of the indica- tive. But logically the second sentence carries the thought forward and is subsequent, not coincident. The commonest device employed by Roman writers for avoiding the actual repetition of a word and yet retaining the force of such repetition, in other words, for repeating the content of a word by means of a new 32 LATIN SENTENCE CONNECTION one, is the use of the demonstrative and relative pro- nouns. When Tacitus writes has rather than iniurias in Agr. 13.1, it is only the form of expression, not the nature of the connection that he changes : Ipsi Britanni dilectum ac tributa et iniuncta imperii munera impigre obeunt, si iniuriae absint: has aegre tolerant, iam domiti ut pareant, nondum ut serviant. The demon- strative is an empty word by itself and has meaning only as it brings up before the mind the content of the word to which it points. So true is this, that if there is nothing already in the context to which it can point, the mind of the reader is forced to look forward, wait- ing for some word or words to give it content. The relative used with an antecedent is in no way different in its connective force from the demonstra- tive. Sallust, Bell. Jug. 25.4: Legantur in Africam maiores natu nobiles, amplis honoribus usi. In quis fuit M. Scaurus. Very often the two are used in close juxtaposition, the change serving merely to avoid monotony. E.g., Cicero, In Verrem 1.10.31: cum his plebeios esse coniunctos ; secundum quos aut nulli aut perpauci dies ad agendum futuri sunt. One more pair of examples will be enough to illustrate the obvious parallelism of usage: Caesar, Bell. Gall. III.1.4: Galba . . . constituit . . . in vico Varagrorum, . . . hiemare; qui vicus po situs in valle . . . montibus undique continetur. Cicero, De Domo Sua 50.130: Q. Marcius censor signum Concordiae fecerat idque in publico conlocarat. Hoc signum C. Cassius censor, etc. Innumerable instances are familiar in which the rela- tive and the demonstrative are interchangeable. In REPETITION 33 the defective text of Tac. Hist. 1.58.12, it is impossible to say whether the one or the other should be restored or whether instead of either, some more direct form of repetition was what Tacitus really used: Interim ut piaculum ohicitur centurio Crispinus; < — > sanguine Capitonis cruentaverat. It makes no real differ- ence whether is is supplied with Haase, or qui with some much older editor, or whether vir is preferred for variety's sake, or whether, as is perhaps most probable, the supposed lacuna is really not a lacuna and the subject of <5e> cruentaverat was unex- pressed.^ More important to illustrate than the interchange- ability of demonstrative and relative, which is famil- iar, is the limit to which the two are really interchange- able. They are interchangeable so long as they refer back to an expressed antecedent outside of their own clause. Not necessarily to a definite word : the relative may follow a speech, referring back to the whole body of it just as well as the demonstrative may : quae cum dixisset is not different from haec cum dixisset. But the antecedent must be present in what has preceded. For if there has been nothing for the pronoun to ''re- peat," the two have distinctly different effects. The demonstrative, as noted, simply holds the mind wait- ing for something to follow, that something being naturally in apposition with the demonstrative. This is not true of the relative. The relative does, to be sure, anticipate something to follow, but in this case 1 For further illustration ef . Sentence Connection in Tacitus, pp. 62 ff. 34 LATIN SENTENCE CONNECTION the something is always a demonstrative either ex- pressed or understood. How this came to be so is not hard to conjecture. It is only necessary to glance at the earlier uses of the relative as they appear in the inscriptions and even in Cato and as they are preserved in later archaisms, conscious and unconscious. C. I. L. 1.197. §3: Z non iouraverit, is magistratum imperiumve nei petito. C. I. L. 1.205. II. line 20 : quod- que ita factum actum iussum erit, id ius ratumque esto. Cato, R.R. 1.4: qui in his praedia vendiderint, eos pigeat vendidisse. Cato, R.R. CXLVIII.l: Quod neque aceat neque muceat, id dahitur. It is clear that the usage qui transtulit, is sustinet, is older than the construction, is, qui transtulit, sustinet. It will appear later that the formal repetition in such phrases indicates that the phrases are logically coincident. Such they undoubtedly were at first. It was only when the first clause came to be looked on as conditioning the second that the idea of subordination came in and the second clause became logically subse- quent to the first. This was the process through which many such correlative pairs went. It is the explana- tion of the origin of the subordinating conjunction si. In other cases subordination never developed. Only the second member had an independent history as in the case of e^ . . . et. In several instances survivals of the older correlative uses persisted, as in cum . . . tum, meaning both . . . and, and in many instances the relative and demonstrative were always used REPETITION 35 largely in this correlative fashion: e.g., quantus . . . tantus, tarn . . . quam, ut . . . ita, and the like. In passing from its indefinite to its relative use in the correlative construction, qui acquired its semi- deictic character, never so clear cut as that of the demonstrative which was originally the second mem- ber of the correlative pair and pointed back to the first, but still fairly strong as it looked forward regu- larly to a demonstrative to follow. When it came then to be used in an anaphoric sense, its force was primarily that of repetition, as was that of the demon- strative. It is in this usage that the relative is impor- tant in the present study. The underlying element which gives to the retrospective demonstrative and to the anaphoric relative their power to express sentence relation, is that of repetition. Furthermore, the fact that they are to the extent indicated, interchangeable, illustrates clearly the fact already emphasized, that we are here dealing with repetition of content only. The form matters not at all. So long as the object or con- cept behind the repeated word is brought up again be- fore the mind of the reader while a new idea is devel- oped around it, the relation is expressed by means of the principle of repetition of content. Demonstratives and relatives are only two of the many means that appear with the greatest frequency in all writers, instinctively to avoid the actual verbal repetition of a word from sentence to sentence. For example, if two of a man's names are presumably familiar to the reader, it is very common to find one of them used in the first sentence, the other in the 36 LATIN SENTENCE CONNECTION second. The use of a title for the name produces the same effect, and a descriptive noun of any sort is really no different. Videt ad ipsum fornicem Fa- bianum in turha Verrem; appellat hominem et ei voce maxima gratulatur, etc. (Cicero, In Verrem 1.7.19.) The hominem in this instance amounts to nothing more than a demonstrative repeating Verrem. Had Cicero chosen to use a title or another name to *' repeat" the Verrem the result would have been the same. Tacitus uses the last method with great frequency; e.g.. Hist. IV.6.12: Hinc inter Helvidium et Eprium acre iur- gium: Priscus eligi nominatim a magistratihus iuratis, Marcellus urnam postulahat, quae consults designati sententia fuerat. The context has made familiar the names of Helvidius Priscus and Eprius Marcellus. Compare Livy 1.27.7: inde eques . . . nuntiat regi abire Albanos. Tullus in re trepida duodecim vovit Salios. Livy XXXIV.29.13 : tradidit Quinctio urbem. Priusquam Gytheum traderetur, etc. A much wider scope is given to this principle when it is applied to common nouns. The use of a synonym to repeat a word from the preceding sentence is quite the same thing as the use of a title or second name just illustrated. Sallust, Cat. 51.18 furnishes a good instance: Nam profecto aut metus aut iniuria te sube- git, Silane, consulem designatum, genus poenae novom decernere. Be timore supervacuaneum est disserere, etc. There is no distinction here between metus and timor: they are synonyms; de metu would have done as well in the second sentence aside from rhetorical considerations. In fact the very next sentence begins REPETITION 37 with an instance of direct repetition : De poena possum equidem dicere, etc. Further examples are: Livy XXXI.29.8: sic Sicu- lorum civitatihus Syracusas aut Messanam aut Lily- baeum indicitur concilium: praetor Romanus conven- tus agit, etc. Tac. Ann. II.69.8: Turn Seleuciam degreditur, opperiens aegritudinem. Saevam vim morbi augehat persuasio veneni a Pisone accepti. Cicero, Pro P. Sulla 3.9: Neque ego hoc partiendae invidiae, sed communicandae laudis causa loquor; oneris mei partem nemini impertio, gloriae bonis omni- bus. In this last example the synonyms do not cover identically the same ground; it is an open question whether they were intended to, but there is sufficient repetition to illustrate the principle. In Tac. Eist. V.26.6, two words ordinarily dijffierent in content are used as synonyms to avoid repetition of the same word: Cuncta inter nos inimica: hostilia ab illo coepta, a me aucta erant. All of the instances of repetition examined so far have been alike in showing two substantives, one in each of two related sentences and each representing the same concept in the mind of the writer, and there- fore calling up the same concept in the mind of the reader. But the examples of demonstrative usage and of various types of synonym have shown that repeti- tion of the exact word, any repetition of externals, is quite unnecessary: content alone is important. So it is not difficult to understand the very numerous cases in which no longer one substantive repeats another, but in which the essential content of one part of speech 38 LATIN SENTENCE CONNECTION repeats that of another. The repetition is less com- plete in that the functions of the two words are now more widely different than they were when both mem- bers of the repeating pair were of the same part of speech. But the effect is unchanged: the kernel of meaning, stripped of its functional garb, is still the fixed landmark in the second sentence to guide the reader to the thought relation. A few examples of the repetition of a word in which that word is not a noun but a verb will serve to make even more clear the fact that it is the repetition of content, that is, of the concept behind the word, that is the important thing in the present type of repetition. Also they will make the instances to follow more clear. Cato makes constant use of this type in his unrhetori- cal style: e.g., R.R. LXXIV.l: suhigitoque pulchre. Ubi bene suhegeris, defingito coquitoque sub testu. In this example, as in practically all of this group, the change of verb form has its very important influence in defining the relation of sentence to sentence. But the repetition of the verb content is the first indicator : the functional change corresponds roughly to the change of case in the examples with nouns. Compare Cicero, Pro Quinctio 13.43: Pecunia mea tot annos utitur P. Quinctius. TJtatuur sane, non peto. Seneca, De Vita Beata 24.1: numquam magis nomina facio quam cum dono. ''Quid? tu" inquis ''recepturus donas?" Tac. Ann. XV.67.6: "Oderam te" inquit, ''nee quisquam tibi fidelior militum fuit, dum amari meruisti. Odisse coepi postquam parracida matris et uxoris, auriga et histrio et incendiarius extitisti/' REPETITION 39 If the second sentence shows the gerund of a verb used in the first, the transition to the type indicated above in which the kernel of meaning is repeated in a new part of speech, has already been made. Such a case is Caesar, Bell. Gall. II.l.l: certior fiebat, omnes Belgas . . . coniurare, obsidesque inter se dare. Con- iurandi has esse causas: etc. In III.l.l, misit is simi- larly taken up by causa mittendi. Tacitus, with his passion for variety, tends to use a synonym in the gerund rather than the same verb; e.g., Ann. XVI.15. 2: Eo missus centurio, qui caedem eius maturaret. Causa festinandi ex eo oriebatur quod Ostorius, etc. But this usage will be illustrated more fully later. And even Tacitus does not always use a synonym: Germ. 26.3 : quos mox inter se secundum dignitationem partiuntur; facilitatem partiendi camporum spatia praestant. With such instances as those just cited in mind, a few examples of nouns repeating the kernel of mean- ing of verbs in preceding sentences will be sufficient. C. I. L. 1.603, line 9, furnishes a good example: TJbi venum datum erit, id profanum esto. Venditio locatio aedilis esto, etc. Compare Caesar, Bell. Civ. 1.2.7: Intercedit M. Antonius, Q. Cassius, tribuni plebis. Refertur confestim de intercessione tribunorum. Sen- eca, De Vita Beata 24.2: "Quid? tu" inquis " recep- turus donas?" Immo non perditurus : eo loco sit dona- tio unde repeti non debeat, reddi possit. Cicero, In Verrem 1.6.16: Mansit in condicione atque pacto usque ad eum finem, dum iudices reiecti sunt. Post- eaquam reiectio iudicum facta est . . . renuntiata est 40 LATIN SENTENCE CONNECTION tota condicio. Cicero, Pro Rose. Am. 34.96: "Non possum" inquit '' divinare." Eo rem iam adducam, ut nihil divinatione opus sit. Tac. Germ. 12.7 : Equo- rum pecorumque numero convicti multantur. Pars muUae regi vel civitati, pars ipsi, qui vindicatur, vel propinquis eius exsolvitur. These instances are almost identical with those in which the gerund repeated the verb ; the last one sug- gests a wider application of the principle. For while multae repeats the kernel of meaning in multantur, it does it in a different way, for example, from that in which reiectio in the Verres passage repeats the kernel of meaning in reiecti sunt. Multae is not the name of the action represented by the verb multantur. Both nouns have behind them the same original concept but the noun represents the thing extracted from the con- victed, not the act of extraction. Similar to this is such an instance as that in Tac. Hist. II.64.4, in which the noun represents the doer of the act repre- sented in the verb: atque ibi interfici iussit. Longum interfectori visum, etc. Provided the original kernel of meaning is repeated, the effect is always the same : it may be repeated in the reverse direction, noun to verb, or an adjective may take up the concept. Livy XXL49.11 : usque ad lucem portu se ahstinuerunt, demendis armamentis eo tem- pore aptandaque ad pugnam classe ahsumpto. Ubi inluxit, recepere classem in altum, etc. Cato, R.R. II.7 : servum morbosum, et siquid aliut supersit, ven- dat. Patreni familias vendacem, non emacem esse oportet. Cato, R.R. XCVI.l: Deinde lavito in mari: REPETITION 41 si aquam marinam non hahehis, facito aquam salsam, ea lavito. Cicero, De Orat. 1.26.120, offers an inter- esting illustration of the possible variation of the repe- tition, always using the same stem: Qui vero nihil potest dignum re . . . efficere . . . is mihi . . . impu- dens videtur. Non enim faciendo solum quod decet sed non faciendo id quod non decet impudentiae nomen effugere debemus. Quern vero non pudet . . . hunc ego . . . poena dignum puto. The conjunctions are supplementary, used because of the position of the repeating verbs late in the sentences. Furthermore, and quite naturally, corresponding to the simple use of a synonym, there are numerous instances of repetition of the kernel of meaning simi- lar to these last, but with the repeating word no longer showing any similarity of root but only of meaning. These are especially interesting as indicating the ex- tent to which this type of repetition was available for use. It will be worth while to quote several examples : Livy XXXIV.46.5: Consul ubi, quanta copiae, quanta fiducia esset hosti, sensit, nuntium ad collegam mittit, ut, si videretur ei, maturaret venire: se tergi- versando in adventum eius rem extracturum. Quae causa consuli cunctandi, eadem Gallis . . . rei matu- randae erat, etc. This example is cited for the use of cunctandi which takes up the kernel of meaning in rem extracturum. But it is an interesting example of the way in which various types of repetition are used in a single instance. Aside from the cunctandi, con- suli is direct repetition of the consul, and Gallis is here equivalent to hostibus, repeating the hosti of the first 42 LATIN SENTENCE CONNECTION sentence. Cicero, Pro Rose. Am. 22.61, furnishes an instance in which the repetition is delayed but without any change in the nature of the connection. De par- ricidio causa dicitur; ratio ah accusatore reddita non est, quam oh eausam patrem filius oeciderit. Livy XXIIL8.4: vietusque patris preeihus laerimisque etiam ad eenam eum eum patre vocari iussit, eui eon- vivio neminem Campanum praeterquam hospites . . . adhihiturus erat. Epulari eoeperunt de die, etc. Tac. Ann. 1.52.1 : Nuntiata ea Tiherium laetitia curaque adfecere: gaudehat oppressam seditionem, sed quod largiendis pecuniis et missione festinata favorem militum quaesivisset, hellica quoque Germanici gloria angehatur. Tac. Germ. 27.6: Lamenta ac lacrimas cito, dolorem et tristitiam tarde ponunt. Feminis lugere honestum est, viris meminisse. One more extension of the usage along this line was possible and in many ways it is the most important. For upon it depends the understanding of many of those sentences which are apparently independent but are in reality explanatory in each case of the preceding sentence. In such instances there often seems at first to be no visible sign of sentence relation and if the progress of thought is not familiar or obvious, the rela- tion is left obscure. But a further examination usu- ally shows that in the absence of actual repetition of a word or its equivalent or of a root or its equivalent, there is used in the second sentence a word which falls naturally into some fairly obvious common category with a word in the first. The effect is that of repetition of an element of meaning. But it is probably arrived REPETITION 43 at by a different mental process from that which leads to the understanding of repetition in a synonym. Be- tween the concept behind the first word or phrase and that behind the second, or repeating, word or phrase there is an intermediate concept embracing both. This tertium quid serves as a link between the two, so that the association is not immediate. For example, the opening sentences of Caesar's Bellum Alexandrinum run as follows: Bello Alexan- drino conflato, Caesar Rhodo atque ex Syria Ciliciaque omnem classem arcessit; Greta sagittarios, equites ah rege Nahataeorum evocat; tormenta undique conquiri et frumentum mitti, auxilia adduci iubet. The name Rhodo in connection with Syria and Cilicia suggests more than a mere place. The most prominent char- acteristic of Rhodes which it possesses in common with Syria and Cilicia is that of being a Roman dependency. In this larger concept Greta is also included so that there is an element of repetition in its use at the begin- ning of the second sentence. Again classem in the first sentence suggests the larger concept of military forces in its widest application and in this larger con- cept are embraced sagittarios and equites as well as tormenta, frumentum, and auxilia. Slightly different is such a case as Livy XXXIV. 52.10: 7^56 deinde Quinctius in urhem est invectus. Secuti Gurrum milites frequentes. The phrase est invectus suggests a concrete picture of Quinctius riding into the city in triumphal array. A part of the picture is the chariot in which he rides, the currum of the second sentence. Tac. Agr. 12.1: quaedam na- 44 LATIN SENTENCE CONNECTION tiones et curru proeliantur. Honestior auriga, clientes propugnant. In this example curru proeliantur sug- gests the concept of the warrior in his chariot with a driver and horses, the picture familiar to the Roman from his Homeric reading. This general concept includes the auriga of the second sentence. The difference between the two types is that in the first, Rhodo and Creta are equivalent elements of the larger concept, while in the second, currus is, in one example, a part of a concept really suggested in toto by est invectus, and, in the other example, with the addition of proeliantur, it presents a whole concept of which only a small part is considered in the second sentence. Further examples of the first type are : Caesar, Bell. Gall. IIL9.8: His initis consiliis oppida muniunt, fru- menta ex agris in oppida comportant, naves in Vene- tiam . . . cogunt. Socios sibi ad id helium Ossimos . . . adsciscunt ; auxilia ex Britannia . . . arcessunt. Tac. Germ. 23.1 : Potui humor ex hordeo aut frumento, in quandam similitudinem vini corruptus: proximi ripae et vinum mercantur. Cibi simplices, agrestia poma, etc. In this instance potui and cihi are equiva- lent elements of a common concept. So, in the follow- ing illustration, Peripateticis and Academici are united by their common relationship to the more general category of rhetorical schools: Tac. Dial. 31.26: ad hos permovendos mutahimur a Peripateticis aptos et in omnem disputationem paratos iam locos. Dahunt Academici pugnacitatem, Plato altitudinem, etc. The following examples will illustrate the second REPETITION 45 type outlined : Caesar, Bell. Alex. III.l : TJrhs fertilis- sima et copiosissima omnium rerum apparatus sug- gerehat. Ipsi homines ingeniosi atque acutissimi. Urbs, a populated city, connotes homines populating it. Cato, R.R. XIV.4: Villa lapide calce. Fun- damenta supra terram pede, ceteros parietes ex latere . . . indito. Tac. Ann. XV.3.9 : Et quia egena aquarum regio est, castella fontihus inposita; quosdam rivos congestu harenae ahdidit. This last illustration would fall under the first type if fontihus and rivos stood alone. But the larger common concept is expressed in aquarum. It is noteworthy that in all of these instances there is not actual or virtual repetition of a concept in the first sentence but either association by means of a ter- tium quid, or repetition of a part, not named but im- plied, of some concept presented in the first sentence. There remain two prominent types of repetition of content. The one is that in which the repeating word sums up in itself the meaning of more than a word or phrase, the other, that in which it repeats a part only of another word. The former is of two sorts. First, the summary word may be a descriptive noun refer- ring to a definite piece of text in an objective way. Such is the word orationem in the familiar phrase, orationem excepere, used after the report of a speech. Second, the summary word may summarize in itself the significance of what has preceded, as, for exam- ple, in such a phrase as terrorem auxit, following the account of some panic. The first is too familiar and too obvious to require any illustration whatever. It is 46 LATIN SENTENCE CONNECTION very similar to the use of ho^no to repeat a man's name. Moreover, the oratio or whatever the word used may be, very frequently goes back in reference to some actual words used in introducing the quotation or description, the body of text referred to, so that the repetition belongs really to one of the types already described. The use of the neuter plural of the demon- strative or relative will suggest itself at once in this connection : haec dixit, for example, or quae cum audi- verat. Very often a descriptive noun is added to make the repetition a little more explicit: haec verba, quod responsum; and the general plural, omnia, has the same function as the pronoun in this usage. The second of these two classes of summary repe- tition will be made equally obvious by one or two illustrations. Whereas the first class corresponded roughly to the repetition of one substantive by another having the same content, this second class corresponds in the same general way to those cases which showed the repetition of the kernel of meaning in a new part of speech and with a new root. Livy devotes consid- erable space (XXXIV.27) to the outrages of Nabis. In section nine he ends this account with the following statement: Hoc terrore ohstipuerant multitudinis animi. The terrore summarizes the effect of the sec- tions preceding, without repeating any single element in them. So causam confusionis (Tac. Hist. III. 38. 14) follows an account of distracted actions. In the following examples the matter summarized is less extensive but the principle is the same. Caesar, Bell. Gall. VI.13.6: Si qui aut privatus aut populus eorum REPETITION 47 decreto non stetit, sacrificiis interdicunt. Haec poena apud eos est gravissima. Pliny, Epist. IIL7.1 : Modo nuntiatus est Silius Italicus in Neapolitano suo inedia finisse vitam. Causa mortis valetudo. Suet. Titus 4: Tribunus militum et in Germania et in Britannia meruit. . . . Post stipendia foro operam dedit. Tac. Hist. IV.40.12: Repeti inde cognitionem inter Muso- nium Rufum et Publium Celerem placuit, damna- tusque Publius et Sorani manibus satis factum. Insig- nis publica severitate dies ne privatim quidem laude caruit. Such a pair of sentences as the following from Cicero {Topica 19.73) will perhaps make clear what has been mentioned as repetition of partial content : sed auctori- tatem aut natura aut tempus adfert. Naturae auctori- tas in virtute est maxitna, etc. If we consider the phrase aut natura aut tempus as a single concept, the subject of adfert, then the naturae is repeating but a part of that concept; the other part is taken up by a repetition of tempus in what follows. In such a case there is actual literal repetition of an entire concept which, simply for illustration, I have chosen to look on as part of a larger concept. Compare with this Livy XXIII.32.1 : Consules exercitus inter sese divi- serunt. Fabio exercitus Teani . . . evenit; Sempronio volones, qui ibi erant, etc. Here is a difference but not a very great one. There is no longer literal repe- tition, but Fabio and Sempronio each repeat a part of the content of consules. Together they are the con- tent of consides, as surely as though the word had been displaced by the phrase Fabius et Sempronius. 48 LATIN SENTENCE CONNECTION This is sufficiently obvious to need no further illustra- tion. Its commonest application is in the use of the correlatives alius . . . alius, pars . . . alii, quidam . . . ceteri, and the like. It is used also when a part only of a concept behind a noun in the first sentence is referred to in the second by such means as the following: pars, quidam, optimus quisque, qui parati erant, acriores, etc., the rest of the original concept being ignored. A descriptive noun, an adjective, a clause, anything which differentiates a part of the con- tent of a noun in the first sentence and makes that part, in the second, the beginning of a new clause, is an example of this ** partial repetition." - Before leaving the illustration of repetition of con- tent, it will be well for the sake of completeness to cite a few instances in which such repetition is implicit. These do not, of course, differ in principle from the instances of actually expressed repetition. In the Germania (32.7), Tacitus says: Inter familiam et penates et iura successionum equi traduntur : excipit filius, non ut cetera, maximus natu, sed prout ferox hello et fnelior. Clearly equos is understood with the excipit. There may be some question as to whether it is the incompleteness of excipit alone or the implicit repetition which has the greatest influence in defining the sentence relation; but implicit repetition of the content of a substantive is surely present. Whether equos be used in the second sentence or hos or quos or animalia or whether, as is the case, the specific word be left to implication, the effect is the same, and the prin- ciple utilized is the same. In Seneca, Ad Helviam I.l, REPETITION 49 it is rather the kernel of meaning in the substantive form consolandi that is repeated by implication in verb form : Saepe iam, mater optima, impetum cepi conso- landi te, saepe continui. TJt auderem, multa me impel- lebant, etc. Two examples from Tacitus will show how clearly the repetition is implied. In Ann. XV.49.1, he writes : Initium coniurationi non a cupidine ipsius fuit. But in 38.4, of the same book after speaking of the great fire, he has merely: Initium in ea parte circi ortum, quae Palatino Caelioque montibus contigua est. Cicero very frequently repeats by implication the entire verbal idea of the first sentence, usually forming a question in the second. One of many examples is Pro P. Sulla 13.36 : Ah Allohrogihus nominatum Sullam esse dicis. Quis negatf In the following examples it is rather a summary noun that is implied. Quint. Inst. Orat. 1.11.18: Neque id veteribus Romanis dedecori fuit; argumentum est sacerdotum nomine ac religione durans ad hoc tempus saltatio, etc. Livy 1.58.7 : Ceterum corpus est tantum violatum, animus insons: mors testis erit. Tac. Ann. 1.7.17 : Literas ad exercitus tamquam adepto principatu misit, nusquam cuncta- bundus nisi cum in senatu loqueretur. Causa prae- cipua ex formidine, etc. Similar examples might be multiplied, but in no instances do they show any real difference from the types of actually expressed repeti- tion. There is, however, a type of repetition of content not expressed in separate words which is of some interest and importance. This comprises those instances in which the subject of a verb is understood from the pre- 50 LATIN SENTENCE CONNECTION ceding sentence and repeated only by means of the person of the verb. Such instances are not really dif- ferent from those in which a direct object is assumed in the second sentence, or a limiting genitive, or the like. But a caution is necessary. At a casual glance this method of repetition would seem much more common than it really is. For those instances in which the subject of the main verb in each sentence is the same and is therefore repeated by means of the person of the second verb, are practically never instances of content repetition. It is true that there is repetition of content. But this is not the determining force. That is regularly repetition of function. This will be more clear after the study of functional repetition in the next chapter, but it can be indicated now. Sallust, Cat. 52.22, has the following: Pro his nos habemus luxuriam atque avaritiam, publice egestatem, privatim opulentiam. Laudamus divitias, sequimur inertiam. The three verbs have the same subject as shown by the person. But that subject does not pick an item from the preceding sentence around which to build a new idea. It repeats an item exactly and in the same form and in the same relative position and in the same use in each sentence. It emphasizes parallelism rather than logical advance. All of these points will be found to be indications of the second large type of repetition, the functional. Furthermore such use of verb person is regularly supported by repetition of tense and mood. Veni, vidi, vici (Suet. Julius 37), is a good instance. Eepetition of subject there surely is, but it "is not the sort of repetition that has been studied. Further con- REPETITION 51 sideration of functional repetition will show that person, tense, and mood all contribute to indicate the coincidence of the clauses and that this is deliberately used here in rhetorical fashion to depict the rapidity of action, amounting to actual coincidence. This does not mean that there is not any use at all of the person of the verb to correspond to the common use of implied object. The implied object too might be used in this functional fashion but as a matter of fact it was not so used to any great extent. With the implied subject the proportions are reversed. How- ever, such cases as the following show that content repetition could be implied by the person of the verb : Cato, R.R. 1.4: Ad villam cum venies, videto, vasa tor- cula et dolia multane sint: ubi non erunt, scito pro ratione fructum esse. The subject of the erimt in the second sentence is the vasa and dolia of the first. It is an item taken from the first and used in a new con- struction in the second, which advances the logical progress of the thought. There is not a wide field for this usage. It was not precise enough and so the sub- ject of the second verb was usually expressed. The great field for the efficient use of the repetition of the person of the verb will be found discussed in the next chapter. An analysis of the instances in which some form of repetition of content appears as a connecting element shows something beyond what was noted after study- ing those cases which showed repetition of an actual word. With the exception of some of the cases in 52 LATIN SENTENCE CONNECTION which it is merely the common category of two words that furnishes the element of repetition, all of the instances studied show the second sentence subsequent (logically) to the first. But on more careful examina- tion there prove to be at least three types of relation- ship indicated. In the majority of cases the second sentence is merely subsequent in logical order. An item of fact or observation is added to the first sen- tence, the repeating word acting as the point of de- parture for this new thought. In a considerable num- ber of cases, however, the second sentence is distinctly explanatory of the first, and in a smaller number it expresses the result of the fact stated in the first. For example, in such an instance as Pliny, Epist. IV.9.3, there is nothing more indicated by the repeti- tion of the name Rufus than that here is offered a new item connected with the preceding by the fact that it makes its point of departure the Rufus of the first sentence: Egit contra eum Pomponius Rufus, vir paratus et vehemens; Rufo suecessit Theophanes. But in Cato, R.R. III.l, a further element is obvious: Ita aedificas, ne villa fundum quaerat r ita ius deicito iudicia dato iudicareque iuheto cogito, etc. Cicero, Philip. XIV.5.14: An ut ego, qui Catilinam haec molientem sustulerim, everterim, ad- flixerim, ipse existerem repente Catilinaf Livy XXIII. 9.5 : sed sit nihil sancti, non fides, non religio, non pie- tas; audeantur infanda, si non perniciem nobis cum scelere ferunt. While repetition of the imperative and of the sub- junctive moods can regularly be made effective, and that of the infinitive very frequently, it is practically impossible to find in the repeated use of an indicative any such suggestion of parallelism, merely because the indicative is the regular mood of narrative. Its use, therefore, in a series of sentences or clauses does not mark them off from others adjacent. 76 LATIN SENTENCE CONNECTION In the examples given of the imperative and sub- junctive moods in this usage, there are usually supple- mentary means for reinforcing the definition of rela- tion. Especially noteworthy is the implicit repetition of subject, perhaps more obvious in such an instance as Cicero, De Orat. 1.51.223: qui sagaciter pervesti- get quid sui cives . . . cogitent, sentiant, opinentur, expectent. There may also be the repetition of tense, as later in the De Orat. (1.57.245) : Tu vero . . . si causam ageres militis, patrem eius, ut soles, dicendo a mortuis excitasses; statuisses ante oculos; conplexus esset filium flensque eum centumviris commendasset ; lapides mehercule omnes fiere ac lamentari coegisses, etc. These two supplementary means may be used in the same way with the indicative, and inasmuch as the mood repetition is not, as a rule, significant in such cases, the repetition of person and that of tense be- come really the primary means of defining clause rela- tions. For example, such familiar phrases as quae teritur absumitur (Livy XXXIV.7.4), or trucidant, spoliant (Tac. Hist. III.25.21), depend largely on the repetition of person, although other means are pres- ent and prominent. In the case of tense repetition the same principle holds true that appeared in the use of modal repetition: to be effective the repetition must be that of a tense different from that prevailing through the narrative. For example, Tac. Ann. VL24. 8: ut quis egredientem cubiculo Drusum pulsaverat, exterruerat, etc. Sallust, Cat. 25.4 : Sed ea saepe ante- hac fidem prodiderat, creditum abiuraverat, caedis REPETITION 77 conscia fuerat. Seneca, De Vita Beata 16.2: nihil cogens, nullo indigebis, liber eris, tutus, indemnis; nihil frustra temptabis, nihil prohiberis ; omnia tibi ex sententia cedent, nihil adversum accidet, nihil con- tra opinionem ac voluntatem. Occasionally a series of present tenses are thrust into the course of a narra- tive, just as historical infinitives are thrust in, and have then the same effect as these less distinctly narra- tive tenses. (See Miss Nye, page 120.) In reality, the use of mood, tense, and person, re- peated to indicate sentence relation, is not so restricted as this discussion would seem to indicate. The diffi- culty of analysis, however, makes it hard to determine its force with exactness and to illustrate it. If the verbs all have a common subject expressed, especially if they are all in a recognized subordinate construc- tion and all subordinate to the same clause, there is no doubt of the effect of the functional repetition in marking them as coordinate. Seneca, De Vita Beata 20.5: qui haec facere proponet, volet, temptabit, ad deos iter faciei. The three verbs with qui for their subject are clearly coincident logically. The actual repetition of the qui could not make this any clearer. The following examples show the same usage. They represent a very extensive class. Caesar, Bell. Gall. VI.24.1: Ac fuit antea tempus cum Germanos Galli virtute superarent, ultro bella inferrent, propter homi- num multitudinem agrique inopiam trans Rhenum colonias mitterent. Cicero, De Orat. 1.37.169: ut amicorum controversias causasque tueatur, laboranti- 78 LATIN SENTENCE CONNECTION bus succurrat, aegris medeatur, adflictos excitet. Livy XXI.10.7: quo lenius agunt, segnius incipiunt, eo, etc. In the following quotation from Sallust {Cat. 37.3), the subject common to the six verbs is still expressed, but the verbs are not in a subordinate construction: Nam semper in civitate quibus opes nullae sunt bonis invident, malos extollunt, Vetera odere, nova exoptant, odio suarum rerum mutari omnia student, turba atque seditionibus sine cura aluntur, quoniam egestas facile habetur sine damno. The various clauses constitute, however, a logical unit just as much as those in the preceding illustration. This is shown by a considera- tion of the next sentence : sed urbana plebs ea vera praeceps erat de multis causis. The urbana plebs ea is contrasted with the semper in civitate quibus opes nullae sunt in the first sentence, the special instance with the general truth, and to make this contrast, not very obvious by itself, immediately clear to the reader, the sed is used at the beginning of the sentence. The two sentences then as wholes are contrasted, and the element which indicates, even before the second sen- tence is reached, that the first consisted of several coincident clauses to be considered as one logical unit, is the functional repetition, chiefly the repetition of person in the verbs. The force of such functional repetition is often sup- plemented by repetition of category. That is, there is some word in each sentence which falls into an obvious common category with a word in each of the other sentences. Caesar, Bell. Gall. VI.13.4: Illi rebus divi- nis intersunt, sacrificia publica ac privata procurant, REPETITION 79 religiones inter pretantur. The rebus divinis, sacri- ficia, religiones are all related through their common category. This is true also in the following instance : Tac. Ann. XV.3.7 : Atque interim reliquas legiones pro ripa Euphratis locat, tumultuariam provincialium maniim armat, hostiles ingressus praesidiis intercipit. In this example from Tacitus, the subject is ex- pressed only by means of the person of the verb. A further illustration will show the same: Cicero, In Verrem 1.15.43: Itaque a populo Romano contemni- mur, despicimur, gravi diuturnaque iam flag ramus infamia. The fact that a populo limits both of the first two verbs and in sense the third, is also an aid to the quick understanding of the sentence relation but not an essential part of it. A slight difference is to be noted in the following example: Pliny, Epist. IL3.3: Prooemiatur apte, nar- rat aperte, pugnat acriter, colligit fortiter, ornat ex- celse, postremo docet, delectat, adficit. The difference lies in the fact that the verbs in the first part of the sentence represent the familiar steps in the progress of an oration. They give, at first, the impression that there is logical progression, that the clauses are not coincident. But such a conclusion is forestalled by the summary verbs at the end, which abandon the ora- torical order. This is a good instance to show that logical progression is quite independent of temporal, and to make clear the force of purely functional repe- tition to group the clauses together into a logical unit. These two points must be clear to make possible an understanding of the type which follows. This tjrpe 80 LATIN SENTENCE CONNECTION may be represented by Tac. Ann. 11.28.12 : Statim cor- ripit reum, adit consules, cognitionem senatus poscit. There is not the slightest doubt that the actions repre- sented by the several verbs are successive. The legal procedure is familiar enough to make that obvious to any reader. Therefore Tacitus did not need to mark it by conjunctions or adverbs. But there is also a dis- tinct impression of haste about the passage. This does not come from the ''omission" of conjunctions. The addition of these would not seriously affect the impression. It comes from the fact that the func- tional repetition in the person of the verbs inevitably suggests the coincidence of their clauses. The extreme instance of this double influence is the Caesarian Veni, vidi, vici. It has already been emphatically stated that juxta- position alone makes connection between sentences, and that the most natural relation between two adja- cent sentences is that of temporal succession. And in the present instances familiarity with the legal or military procedure under discussion supplements the force of these facts. But that is not to say that the further force noted, that of functional repetition, does not also have its influence. It is the presence of the two forces at once that makes these cases hard to classify. But they are no less instructive on that account. The repetition of the person of the verb gives the impression of coincidence while the other forces give the impression of succession in time. In the following examples, one element is sometimes in the ascendency, sometimes the other. REPETITION 81 Caesar, Bell. Gall. VI.38.2: Hie diffisus suae atque omnium saluti inermis ex tabernaculo prodit; videt imminere Jiostes atque in summo esse rem discrimine ; capit arma a proximis atque in porta consistit. Caesar, Bell. Civ. 1.30.1 : Itaque in praesentia Pompei sequendi rationem omittit, in Hispaniam proficisci constituit, duumviris municipiorum omnium imperat, ut naves conquirant, etc. Cicero, In Cat. III.4.8: Introduxi Volturcium sine Gallis; fidem puhlicam iussu senatus dedi; hortatus sum, ut ea, quae sciret, sine timore indicaret. Pliny, Epist. IV.25.4: Poposcit tahellas, stilum accepit, demisit caput, neminem veretur, se contemnit. It is probably not worth while to trace further this repetition of function. There is possibly effective definition of sentence relation expressed by repetition of number, probably by repetition of case. But the principle should be clear from the citations already made, and it is of greater importance to study the nature of the relation so defined. The fundamental relation between sentences indi- cated by the repetition of function is always the same : the sentences are always logically coincident. They may be parallel with each other in thought or they may be the reverse, contrasted with each other. The cases of contrast, sharply marked by an abrupt change, either of meaning or of function, are the subject of a later chapter. Leaving out of consideration for the time being those instances in which the specific rela- tion is one of contrast, there is no such variety of rela- 82 LATIN SENTENCE CONNECTION tion between sentence and sentence as was expressed by repetition of content. The variations in the type do not represent variations in the precise definition of relation. The repetition of function is a much nar- rower usage than repetition of content. In every instance the second sentence expresses some idea parallel to that expressed in the first. The two ideas are not only coincident but equivalent steps in the logical development of the writer's thought. It is not surprising, therefore, to find a very narrow range of conjunctions available for use with such repe- tition. As a matter of fact, et and -que are the only ones, with aut for negative sentences. The use of these can be very briefly illustrated. In the Tusculan Disputations, III.34.82, the repeti- tion of si serves alone to mark the relation between clauses ; Si quis error alicunde extitit, si paupertas momordit, si ignominia pupugit, si quid tenehrarum ohfudit exsilium, etc. The parallelism of clauses is obvious. In V.1.2, Cicero employs an et to mark the relation more conspicuously: Quodsi ah iis inventa et perfecta virtus est, et si praesidii ad beate vivendum in virtute satis est, etc. And again, in IV.35.76, the si is no longer repeated; the functional repetition of the verb forms supplemented by the use of et ex- presses the relation: Etenim si naturalis amor esset, et amarent omnes, et semper amarent, et idem amarent, etc. The meaning is essentially the same in all three sentences, the sentence relation identical. The same variation may be seen in the following examples. Tac. Ann. II.33.3: decretumque ne vasa REPETITION 83 auro solido ministrandis cibis fierent, ne vestis serica viros foedaret. Cato, R.R. XXXII.2 : caveto ne vitem praecipites et ne nimium praestringas. Nepos, Alcih. 5.1 : pertimuerunt ne caritate patriae ductus aliquando ah ipsis descisceret et cum suis in gratiam rediret. Cato makes his sentence relation no clearer when he uses et in addition to repetition in R.R. XXV.l : Quom vinum coctum erit et quom legetur, than when he uses the repetition alone in II.l : Pater familias ubi ad vil- lam venit, ubi larem familiarem salutavit. The et adds nothing to our understanding of Sallust, Bell. Jug. 85.25: scilicet quia imagines non habeo et quia mihi nova nobilitas est. For variety's sake, a writer sometimes uses both methods in one sentence. E.g., Sallust, Bell. Jug. 40.1: Interim Romae C. Manlius Limetanus tribunus plebis rogationem ad populum promulgat, uti quaereretur in eos, quorum consilio Jugurtha senati decreta neglegisset, quique ab eo in legationibus aut imperiis pecunias accepissent, qui elephantos quique perfugas tradidissent, item qui de pace aut bello cum hostibus pactiones fecissent. The use of item in this last clause is altogether identical with the use of a conjunction, and furnishes a good example of the method by which the majority of con- junctions first came into use, as adverbs to supplement the more fundamental means of expressing relation. The conjunction is used as a supplementary sign of relation with the other types of functional repetition, too. For example, Cicero, De Orat. 1.8.30: prae- cipue semper floruit semperque dominata est. The repetition of the semper is quite adequate to express 84 LATIN SENTENCE CONNECTION the sentence relation; so is the repetition of velim and of the second person present subjunctive following it, in Cicero, Ad Att. 1.7.1. The et is really an unneces- sary addition. Tu velim ea, quae nobis emisse et parasse scrihis, des operant ut quam primum habea- mus et velim cogites . . . quem ad modum bibliothecam nobis conficere possis. In the following examples the conjunction is added to reinforce the repetition, which is confined to the verb form. Cicero, Pro Quinctio 2.10: orat atque ob- secrat. Livy XXXV.7.8: fudit fugavitque. There is no essential difference between these and the phrases already cited from Livy and Tacitus: quae teritur absumitur, and trucidant spoliant; or the phrases so common in the inscriptions, dedit adsignavit, or habue- runt possiderunt. (Cf. C. L L. 1.200.3; 1.204, line 16 et passim.) In the following there is just as little actual need of the conjunction: Cicero, De Orat. 1.55.235: Et enim sine controversia et magna est et late patet et ad multos pertinet et summo in honore semper fuit et clarissimi cives ei studio etiam hodie praesunt. Caesar, Bell. Gall. 1.3.4: cuius pater reg- num in Sequanis multos annos obtinuerat et ab senatu populi Romani amicus appelatus erat. Cicero, De Orat. 1.21.95: si quis pari fuerit ingenio pluraque quam hie et audierit et lectitarit et scripserit, etc. The principle outlined above, in the discussion of repetition of content, that the conjunctions are very often merely supplementary as a means of expressing sentence relation, is borne out by these instances of repetition of function. The conjunctions found in REPETITION 85 such instances are those originating from adverbs indicating something like **in addition to." The repe- tition of function unaided shows that the two sen- tences are parallel and coincident. The conjunction does no more than supplement this indication. CHAPTER IV RETROSPECTIVE INCOMPLETENESS After so long a study of repetition and its use to express certain relations between adjacent sentences, it is necessary to remind ourselves that repetition does not make relation, does not relate the sentences. The sentences, by virtue of the psychological processes which produced them, were of necessity related before ever they were put on paper or even analyzed into words. Repetition is the means employed to define to the reader the particular nature of the relation. This fundamental fact, that the sentences are in abso- lutely every instance related, is the explanation of the further fact that not alone repetition of an element of meaning or function, but also change of meaning or function may indicate the type of sentence relation. The unconscious problem in the reader's mind is always ''what is the relation," not ''is there a rela- tion." The use of repetition has been found to sug- gest certain sorts of relation, often more precisely defined by further means; the use of change will be found to suggest others. But a third element already met with in the study of repetition and even more prevalent throughout the instances of change, can be more conveniently studied next, because of the necessity of understanding it in RETROSPECTIVE INCOMPLETENESS 87 connection with the element of semantic and func- tional change. For want of a better name, I shall continue to call it the element of incompleteness.^ It has a wide range from the very simplest types to the most rhetorical adaptations. The principle is this: any word or phrase either essentially empty of meaning when taken by itself, or incomplete in meaning in its particular setting, forces the hearer or reader to look outside the word or phrase to find reason for its employment. Within the sentence, this is illustrated by the use of the inflec- tional endings. These endings do not add to the mean- ing of the word itself but give to it an element of relativity : they exist to show the relation of the word to other words and therefore render it incomplete until the mind grasps those other words. Between sentences, the same principle holds true. It is opera- tive in the use of such empty words as conjunctions or of demonstratives, which have meaning only by borrowing, or of words or phrases whose meaning is essentially incomplete except by reference to some- thing outside their own sentence. Syntactically, the sentences which exhibit this type of connection may be complete, but to make their sense complete, to give them logical finish, they need the help of something from without. Illustration will make this clear. Assume a clause beginning with et: et triumphavit Caesar. This is syntactically a complete clause, an idea expressed in 1 Cf . Sentence Connection in Tacitus, p. 112; Miss Nye: Sentence Connection, p. 1. 88 LATIN SENTENCE CONNECTION words. But there is left in the mind after reading it a feeling of incompleteness, due not alone to the fact that the circumstances of the triumph are unknown, that the title of Caesar is ambiguous. The source of vagueness is to be found in et. Et apparently has no meaning whatever; or if it has, that meaning is due to something outside the clause. Probably another clause once preceded this one and then the et had some force. Its very presence at the beginning of the clause forces the mind to relate the clause to some other in order to give meaning to et. The incompleteness in the meaning of et can be understood by reference to its origin. The conjunc- tion was first an adverb with the general sense of ''furthermore" or '*in addition." The adverb itself was apparently developed to mark a sentence relation not sufficiently obvious without it. And because of the fact that it was simply a marker, used not to express any element of thought within the sentence, but simply its relation to another sentence, its mean- ing was naturally incomplete without reference to that other sentence. It was probably used at first simply with a sentence subsequent logically to the one preced- ing it, to mark the thought progression. That other means of indicating the relation, aside from mere juxtaposition, might be present, has been seen in the chapter on repetition. But the et, once familiar, became the handiest of all means. Furthermore, this use in the second clause seems to have been always the fundamental one with et. What- ever force other means may give to a clause, the et, RETROSPECTIVE INCOMPLETENESS 89 suggesting simple succession, has no part in thera. Ehetoric may, for effect, use an et with a clause whose abrupt change from the preceding clause indicates contrast. But this is simply to heighten the effect of the contrast by the unexpectedness of it after et. The correlative use seems to have been a later develop- ment, the result of the influence of functional repeti- tion, but such an empty word lent itself readily to the usage. For, having no meaning by itself, if nothing had preceded its clause to complete its meaning, it must needs hold the mind in suspense until its incom- pleteness was somehow satisfied. There are adverbs in actual use which show the same force that lay in the adverb from which et developed. Item, etiam, rursus, are quite analogous. Similar to et in origin and force are sed, at, autem, and also enim and nam. The first group originate in adverbs meaning in general ''apart," ''away from." Such ideas are as incomplete as the notion of " in addi- tion to, ' ' in the adverbial ancestor of et. But the mean- ing of the adverbs in this group fitted them to mark a different type of relation, and, because of their mean- ing, they became attached to contrasted sentences in exactly the same way that et became attached to log- ically subsequent sentences. The origin of enim and nam is not so certain, but, whatever the adverb from which they came, it obviously had the same incom- pleteness and a meaning which fitted it to become the more or less mechanical sign of an explanatory sen- tence. At ego quasi ex aliqua peregrina delicataque merce lusus meos tibi prodo. Pliny, Epist. IV.14.1. 90 LATIN SENTENCE CONNECTION At performs in this sentence exactly the kind of func- tion that et performed in the sentence et triumphavit Caesar, a function that is not evident until the sen- tence is considered in its context. As the statement stands without context, at is an appendage, meaning almost nothing except by way of suggestion. So far as the clause itself is concerned et or enim or ergo might have been the opening word without affecting its meaning. But the preceding sentence considered with the present one makes clear the use of at. Tu fortasse orationem, ut soles, et flagites et expectas. As will appear in the following chapter, the abrupt semantic change between tu and ego (made obvious by position) and between orationem and lusus, indicate contrast, anticipated by fortasse in the first sentence. The contrast is indicated without at, which serves only to draw attention to it. In the following examples, the same principle is illustrated. Cicero, Ad Att. II.1.5 : ille autem non sim- ulat, sed plane tribunus pi. fieri cupit. Caesar, Bell. Civ. 1.51.6. Hoc pugnae tempus magnum attulit nos- tris ad salutem momentum; nacti enim spatium se in loca superiora receperunt. The essential mark of relation between these clauses is in the first and is dis- cussed in the chapter on Anticipatory Incompleteness (pp. 165 ff.). It lies in the deliberately vague state- ment implying an explanation to follow. In this instance the vagueness is not striking and the enim is therefore used to make sure that the second sentence is understood as explanatory. Cicero, Topica 24.91: Quarum fines ipsi declarant quihus utendum locis sit. RETROSPECTIVE INCOMPLETENESS 91 Nam iudicii finis est ius, ex quo etiam nomen. The primary signs of relation are two, anticipatory vague- ness in the first sentence and semantic repetition in the second, marking that sentence as subsequent log- ically to the first. Nam supplements the force of the vagueness in defining the explanatory nature of the second sentence. It is not intended at present to show the sentence relation brought out by the individual conjunctions, but to indicate that in themselves they are incomplete in meaning and that, therefore, placed at the begin- ning of a clause they inevitably direct the attention of the reader or hearer to a preceding sentence. They are themselves detached mechanical indicators of rela- tion, owing their power to indicate relation to their inherent incompleteness, their power of definition to their original meaning as adverbs. In Chapter VI (pp. 155 ff.) will be found similar mechanical indicators whose place is in the first sentence and whose func- tion is anticipatory rather than retrospective. Other conjunctions used in the second sentence will be dis- cussed presently. The next type of incompleteness is that found in demonstratives and relatives. This was touched on in Chapter III (pp. 31 if.) where, however, the element of repetition was primarily under discussion. That element was almost always prominent in these pro- nouns, until they became the favourite means employed to repeat the content of some preceding substantive without direct repetition of the actual word. But their ability to do this lies in their own emptiness of 92 LATIN SENTENCE CONNECTION meaning. The fact that they can repeat the most widely different sorts of content, shows that they are of themselves empty words. Haec statim dixit, or Quae vehementer dixit, are syntactically complete sentences. Logically, both of them are incomplete. It requires some reference to what lies outside the sentences them- selves to make their meaning complete, just as it re- quired some reference to what lay outside the et clause to make its meaning complete. Beyond this fact, how- ever, there is nothing in common between the demon- strative and the conjunction. The latter referred back to the preceding clause, but did not bring it or any part of it definitely before the mind as a part of its own clause. This the demonstrative and the relative always do: whenever they refer the mind to what has preceded, they always make all or part of that preceding thought a distinct and real part of their own clause. That there is a difference between the relative and the demonstrative in the range of their suggestive incompleteness, has been indicated in the last chapter (page 33). But this is chiefly of impor- tance when their force is anticipatory. For the essence of the incompleteness of the relative and the demonstrative lies in their deictic character. In their retrospective uses they are alike in this respect. But if nothing precedes to give content to their incom- pleteness, they are quite different. The relative, in developing its subordinating force, has lost to a large extent its general deictic character and can look for- ward only to a demonstrative, expressed or implied. RETROSPECTIVE INCOMPLETENESS 93 The demonstrative merely points. It has no further connotation. In so far as this demonstrative type of incomplete- ness is retrospective, its defining force is always that of repetition of content. The element of incompleteness indicates the existence of relation without defining it. The reason for emphasizing this element is that it lies behind the demonstrative even when retrospection is impossible, and the force of the demonstrative then becomes anticipatory. For the present, however, the former phase only is under discussion.^ It is the same element that explains the connective force of the de- monstrative adverbs : hie, hinc, hue, inde, turn, ita, tarn, and the like. It is in the same way behind the demon- strative adjectives like talis, only with them the ele- ment of repetition is rather more evident. It is be- hind the conjunctions itaque, quare, and tamen. And, finally, it is the key to an understanding of a great many of the subordinating conjunctions. These last points need a little further explanation. The non-subordinating conjunctions already dis- cussed are primarily adverbs modifying the whole clause in which they stand and developing each from an independent meaning of its own. Those that re- main to be discussed are, for the most part, developed from the demonstrative and relative pronouns and have therefore the same inherent incompleteness as the pronouns, resulting in a deictic function. Itaque, 1 Examples of the demonstrative usage are omitted because of the illustrations given in Chapter III. Many examples may be found in Sentence Connection in Tacitus, pp. 57 ff. 94 LATIN SENTENCE CONNECTION ideo and inde as well as quo and unde, quapropter and quare all point back to the preceding clause and effect a repetition of content, a summary repetition of the content of the whole preceding clause. The element of incompleteness indicates this retrospective relation, the element of repetition serves to define it. Unlike the conjunctions previously discussed those in the present category, by virtue of their function of repeat- ing the content of the previous clause, form an integral part of their own clause, instead of merely modifying its whole tone. Those previously discussed could be omitted without destroying the clause structure; the present group consists of conjunctions which cannot be thus omitted without destroying the sense of the clauses in which they stand. Quod also, in its use so common in Seneca and else- where in the phrase quod si, has the force of summary repetition, gathering into an accusative of specifica- tion the general idea of what has preceded. Such is its use, for example, in Horace, Odes LI: quod si me lyricis, vatihus inseres sublimi feriam sidera vertice. But quod had also another use in which it was more closely bound to the verb: Fecisti mihi pergratum, quod Serapionis librum ad me misisti. (Cicero, Ad Att. II.4.1.) Obviously this use is simply a further development of the same relative usage. Examples from Plautus make this even more clear: e.g., Capt. 586: Filium tuom quod redimere se ait, id ne utiquam mihi placet. Quam and cum (quom) have the same original force and have their demonstrative counter- parts in tarn and turn. Ut apparently has the same RETROSPECTIVE INCOMPLETENESS 95 underlying relative element and is similarly used both independently as a conjunction and in correlation with its demonstrative counterpart, ita. Whether the grad- ual development of subordinating force took place in each correlative pair independently or whether the adaptability to this twofold usage was already inher- ent in the relative from which each came is quite im- material to the present discussion. "With respect to the relative pronoun the case is clear. Originally in the usage of an indefinite pronoun and a demonstra- tive in adjacent clauses, the demonstrative repeated the content of the indefinite and this repetition indi- cated the logical subsequence of the clause in which it occurred. The other clauses came to be looked upon as not only antecedent but subordinate and the rela- tive became differentiated from the indefinite as a subordinating pronoun in the pair, is . . . qui. In the case of the other relative words the exact point at which their independent existence began is not so clear but the principle is the same fundamentally. On the other hand, the conjunction si almost surely attained its adverbial force as a form of the demon- strative pronoun. It was used in correlative clauses to which it thus imparted the force of functional repe- tition. But as one sentence came to be looked upon as the principal one and the other as the conditioning one, the subordinating use became established. Clearly the line between coordinate and subordinate was not always sharply drawn. Obviously not all conjunctions are alike in their ori- gin ; it is not strange therefore to find that they differ 96 LATIN SENTENCE CONNECTION in the origin of their element of incompleteness. But in their developed use in Latin this can be truly said of the conjunctions in general : they are empty words by themselves; and whatever the specific force which they exercise in expressing relation, they give mean- ing only as they are considered with reference to some clause outside their own. They are like the preposi- tions within the sentence. For prepositions have only an indeterminate content. Absolutely alone they mean almost nothing; taken merely with the word which they *' govern, " they gain some precision but very little. They must always be considered with reference to other words in the sentence. So it is with the con- junctions, except that clauses and sentences take the place of words. And, furthermore, it will prove true as already suggested, that conjunctions, like prepo- sitions, are not a primary means of expressing rela- tion. They were in all probability first used as a sup- plementary means and only when they had become familiar in this way, and had attained fairly perma- nent fields of usage, did they become regular carriers for all sorts of connections. The analogy with prepo- sitions strengthens this theory. Conjunctions developed a large field of their own, so large a field that it has often been looked on as the whole field of sentence connection. Their real func- tion cannot be altogether clear until we have consid- ered all types of connectives. At present it is simply the principle of incompleteness behind them all that is under consideration. How this came to be present has been shown in a few cases and will be shown in the RETROSPECTIVE INCOMPLETENESS 97 others from time to time. But that it is present in all of the conjunctions is probably obvious. None of the conjunctions has real meaning when consideration is given solely to the clause in which it stands. Even more clearly than in the demonstrative usage, it is repetition to which the element of incompleteness points in the next type, that of comparatives. When Tacitus writes {Ann. III.43.1) : Apud Aeduos maior moles exorta, the maior is logically incomplete without reference to the preceding description of revolt in two other German tribes. Moles repeats the general con- tent of what has gone before, but this fact does not prevent the comparative maior from having its own connective force due to its incompleteness. The same is true very generally in the use of comparatives: from their adjectival nature they are almost neces- sarily accompanied by actual or implicit repetition. One instance of implicit repetition may be cited as typical. Tac. Ann. XIII.39.5: et Corhulo, . . . excin- dere parat castella, sibique quod validissimum in ea praefectura, cognomento Volandum, sumit; minora Cornelio Flacco legato et Insteio Capitoni castrorum praefecto mandat. Undoubtedly castella is implied with minora and undoubtedly this is implicit repeti- tion of the clearest sort. But it is not the omission of castella which makes minora incomplete and this point is worth noting carefully. The implication of castella with minora is an instance of relation expressed by repetition: it would have been virtually the same no matter what particular method of expressing a part of the content of castella was employed. It is totally 98 LATIN SENTENCE CONNECTION independent of the incompleteness of minora, which is a relativity inherent in the nature of the meaning of a comparative. The incompleteness of the compara- tive points out the repetition, but it is not the same thing as the repetition any more than the incomplete- ness of the demonstrative and the repetition which it indicates are one and the same thing. The force of comparatives is not confined to adjec- tives in the comparative degree. Some adjectives are essentially comparative in their meaning. Alius, alter, ceterus, reliquus, par, and similis are such by nature. When used retrospectively they all imply repetition just as do the ordinary adjectives used in the same way, but the first four have a very strong element of incompleteness which becomes of importance in their anticipatory use. In the same category, of course, fall the corresponding adverbs, such as cetera, alias, pari- ter, and also contra. And finally, the conjunction ceterum is dependent on the same element for its force. Certain nouns belong to the category of words whose meaning is inherently relative or incomplete although it is not so easy to analyze their force as it is that of demonstratives and comparatives. They are such nouns as causa which by their meaning cannot be thought of without reference to another concept out- side of themselves. There can be, for the ordinary mind, no conception of cause apart from something caused. Such words have only an indeterminate con- tent, so that, while they are not quite so empty of mean- ing as conjunctions are, they still depend for any satis- RETROSPECTIVE INCOMPLETENESS 99 factory meaning on some word outside themselves. If the complementary concept is expressed in the sen- tence, as for example in Tac. Hist. II.54.5, Causa fin- gendi fuit ut, etc., the connection is regularly expressed by repetition of content. But if there is no summary repetition, as in this illustration, or specific repeti- tion, such as frequently follows causa, the incomplete- ness of the word, unsatisfied in its own clause, directs the mind to the one preceding and, like the compara- tives, implies the repetition which is not specifically expressed. It was probably as much the presence of this element of incompleteness as the force of the ablative case which singled out causa to develop a prepositional usage. The number of nouns like causa is more numerous than at first appears. Tac. Ann. IV.1.3, shows initium used together with causa: Cum repente turbare for- tuna coepit, saevire ipse aut saevientihus vires prae- here. Initium et causa penes Aelium Seianum. Argu- mentum, testis, testimonium, very often show the same incompleteness though, like all words of this group, they have a perfectly concrete and self-sufficient use beside. Pliny, Epist. IV.7.3: recta ingenia debilitat verecundia, perversa confirmat audacia. Exemplo est Regulus. Compare an instance in which the usage is supplemented by a conjunction but with the same force: et testimonio sunt clarissimi poetae. (Quint. hist. Orat. I.IO.IO.) In Cicero, Pro Rose. Com. 4.11, the repetition is made explicit: Ei rei ipsa verba for- mulae testimonio sunt, but the relation is just as clearly indicated by the incompleteness in the follow- 100 LATIN SENTENCE CONNECTION ing instances. Cicero, Brutus 6.25: Quare quinque artium concursus maxumarum quantam vim quantam- que difficult at em habeat existimari potest. Testis est Graecia, quae, etc. Tac. Hist, y.2.3: ludaeos Greta insula profugos novissima Lihyae insedisse memorant, qua tempestate Saturnus vi lovis pulsus cesserit reg- nis. Argumentum e nomine petitur. The list of words might be extended but extension would show no change in the principle. The verbs which, by their meaning, come under this category of the incomplete are few. Such a verb as malo undoubtedly contains the same element of incom- pleteness as the comparative adjectives and adverbs. But that is because of the adverb compounded with volo in the formation of malo. The verb, therefore, belongs with the comparative adverbs in the present study. Again, respondeo and similar verbs com- pounded with re-, have a distinct element of incom- pleteness imparted to them by the prefix. These verbs do not merely repeat by implication an element from the preceding sentence; they are rather like the ad- verbs item and rursus: their reference is to the whole idea preceding and does not select any part of it to be used as the starting point for a new idea. On the other hand, the tendency is strong to make the repeti- tion, such as there is, less vague. Often it becomes quite explicit. For example, Cicero, Pro Quinctio 18.57: Quaesivit a te . . . Quinctius, quo die vadimo- nium istuc factum esse dicer es. Respondisti statim: Nonis Febr. It is natural to assume something like ei from the Quinctius of the first sentence. Very often RETROSPECTIVE INCOMPLETENESS ' 101' ' "' there is actual repetition used in the rest of the sec- ond sentence. Such is the case in Seneca, De Const. Sap. 18.5: Antistheni mater harhara et Thraessa obi- ciebatur; respondit et deorum matrem Idaeam esse. But answering implies something said or asked and the incompleteness points to the repetition even though it can hardly be conceived of without it. More intimately involved with direct repetition is the incompleteness of transitive verbs used without an object. These have been discussed as examples of implied repetition and such they are, but the inherent incompleteness in their use should also be noted. Bel- lum secutum est; by its very meaning sequor must have an object, for following is inconceivable apart from the thing followed. Wlien Tacitus begins a para- graph {Ann. III.52) with the statement, C. Sulpicius D. Haterius consules sequuntur, the object of the sequuntur is necessarily assumed from the preceding paragraph. It is a case of implicit summary repeti- tion. However, the incompleteness of the verb alone is clear. The same is true of any transitive verb which cannot be used intransitively and whose object is not expressed in its own sentence. The same principle holds true with the class of verbs compounded with prepositions which retain their force in the combination. Accedo, subdo, cir- cumsto, are examples. The only difference between these and the simple transitives is in the nature of the repetition implied. The element repeated by implication is suggested by the preposition, not by the verb, and is therefore in a different relation to the 102 LATIN SENTENCE CONNECTION verb. But tliis is a minor distinction, as will be clear from two or three examples. Cato, R.R. LXXIV.l: Farinam in mortarium indito, aquae paulatim addito. Livy 1.57.11 : Et turn quidem ah nocturno iuvenali ludo in castra redeunt. Paucis interiectis diehus Sex. Tar- quinius . . . Collatiam venit. Tac. Ann. 1.57.12 : et ereptus Segestes magna cum propinquorum et clien- tium manu. Inerant feminae nobiles. The type of sentence relation with which retrospec- tive incompleteness of content is employed, is ob- viously not determined by the incompleteness itself, but instead, this element of sentence connection points to some other element which is the determining factor. Ordinarily that factor is repetition of content. This is especially obvious when the incompleteness appears in demonstratives or verbs, but it is also beyond doubt in the case of comparatives of all sorts. With the demonstrative adverbs the relation indicated is the same as that with the similar adjectives. The second sentence, therefore, in all such instances is subse- quent logically to the first. The incompleteness serves to call attention to the relation rather than to define it. If the repetition is not sufficiently obvious when thus indicated it is often expressed in words. Only with those conjunctions whose origin is not to be found in the demonstrative pronoun or in the relative does the relation indicated vary. Et, sed, and enim were found serving as mere indicators, modifying their en- tire sentences, and pointing to types of relation con- RETROSPECTIVE INCOMPLETENESS 103 siderably varied, each limited in range by its own adverbial meaning. To enumerate the conjunctions therefore that are found supplementing the various types of retrospec- tive incompleteness is largely fruitless for they are the same as those found with the types of repetition already studied to which incompleteness regularly points. The following are cited for special matters of interest attaching to them. For example, Quint. Inst. Orat. 1.6.22 : Recta est haec via, quis negat? Sed adiacet et mollior et magis trita. Adiacet and the comparative adjectives both indicate repetition of con- tent which would regularly indicate a sentence logic- ally subsequent to the preceding. But the rhetorical question, quis negat, injects an emphasis which will be found later to suggest a concessive tone.^ By itself this suggestion is not sufficiently strong to counteract the influence of the repetition, with the result that sed is added to reinforce it. Regularly et or que is the con- junction used with incomplete verbs if the need of any is felt. (Cf. Tac. Ann. XIV.32.11: et inerat, and Tac. Ann. XV.26.5: addiditque.) A conflict of tone is pres- ent in Sallust, Cat. 6.4 : Igitur reges populique finitumi hello temptare; pauci ex amicis auxilio esse; nam ceteri metu perculsi a periculis aherant. Although ceterus, as an incomplete adjective, exerts the force of repetition of content, it will be found, by virtue of its meaning which isolates one individual or group of individuals from a larger whole, to suggest contrast almost irresistibly (cf. p. 153). In the present 1 Cf. Chapter VI, pp. 151 ff. 104 LATIN SENTENCE CONNECTION instance, Sallust felt the second sentence to be explana- tory and made sure of conveying this tone by using nam. The opposite is the case in Livy XXI.29.7 : Mul- titudo timehat quidem liostem nondum oblitterata memoria superioris belli, sed magis iter immensum Alpesque . . . metuehat. Content repetition is not merely implied in magis: it is expressed in metuehat; but contrast is the chief relation and this is prepared for by quidem and plainly marked by sed. Compare Cicero, De Orat. 1.7.27 : ut dies inter eos curiae fuisse videretur, convivium Tusculani. Postero autem die, etc. In all of the groups illustrated so far in the present chapter, groups of conjunctions, pronouns, adjectives, adverbs, and verbs, the incompleteness inherent in the words has been semantic. There has been in each instance something essentially incomplete or relative in the meaning of the word itself. The principle, how- ever, like that of repetition, finds demonstration in the field of function as well. There are modes and tenses which are essentially relative and which cannot occur in consecutive discourse without the implication of other sentences outside themselves, in which their relativity can be satisfied. It is not now a question of the origin of modal uses in Latin, but of their actual status as they appear in Latin literature. The original use of the subjunctive was no doubt an independent one. Fortunately there is enough of this independent usage left in historical Latin to make it clear and to illustrate the develop- RETROSPECTIVE INCOMPLETENESS 105 ment of the subordinate uses.^ But in historical Latin the subjunctive had acquired a pretty strong notion of contingency, arising out of its very general function of expressing ^vill rather than fact. Even in the more precise imperative there will appear the same ten- dency when it is used in consecutive discourse. As used in Cicero, for example, the subjunctive mode gives to its verb an incompleteness or relativity such as their meaning gave to words in the preceding cate- gories. The difficulty in making this clear in illustra- tion lies in the fact that logical incompleteness in this modal usage was developed in Latin into syntactical incompleteness so thoroughly and so early that as a rule mechanical signs of subordination accompany the mode. But not always. Necesse est exhiheas, licet videan- tur, monehat rediret, exclamat irent, rogat quid sit, and the like are by no means unfamiliar. It can be shown later, I think, how these phrases originated and by what steps the subjunctive acquired its relativity. But for the present, the fact to note is that in classical Latin the mode itself suggests relativity: it has come to be felt as incomplete by itself. In all of these phrases the meaning of the verb which appears in the indicative does much to anticipate the subjunctive with its tone of subordination; the types of clauses which this construction can follow are few and familiar. Necesse est or exclamat act as signals to the mind to remain open for several possible sorts of expression 1 See especially, — Morris : The Subjunctive in Independent Sentences in Plautus. A. J. P. XVIII. (1897), Nos. 70, 71, 72. 106 LATIN SENTENCE CONNECTION that might follow and of these the subjunctive is one. But that the subjunctive plays its part is obvious when the order of the clauses is reversed: exhibeas necesse est, quid sit rogat. Two examples will be sufficient to illustrate this point. In reading Cicero, In V err em II.II.78.191 : Nam aut exhibeas nobis Verucium, the mind is left in sus- pense, not alone because of aut which by its own in- completeness implies another aut clause to follow, but by the incompleteness of the verb. And that again is due to the mode. The judgment is held in suspense until the following necesse est aut te Verucium esse fateare is read. Suppose the necesse est had been reserved to the end : the incompleteness due to the sub- junctive would still have continued after the incom- pleteness due to the aut was satisfied and until the indicative was reached. So we read in Tac, Hist. IV.60.15 : Simulata ea fuerint, and we have no logically complete sentence. To say that it is not syntactically complete either, is simply to accept the general recog- nition of the logical incompleteness which has led to this syntactical classification. The following clause adds a clear pointer to the nature of the relation to be expressed: an retinere saevientes nequiverint. There is no longer doubt as to the clause relations, but there is no logical completeness until the sentence closes with parum adfirmatur. These two illustrations of the incompleteness inher- ent in the subjunctive mode are really examples of anticipatory incompleteness and are given here only because they make the force of the subjunctive more RETROSPECTIVE INCOMPLETENESS 107 obvious. Compare Livy XXII.39.20 : malo, te sapiens hostis metuat, quam stulti cives laudent. The contrast between the two subjunctive clauses is not of impor- tance here, but rather the incompleteness imparted to each of them by the subjunctive mode, further illus- trated by the following instances. Tac. Ann. 11.17.6: exclamat irent, sequerentur Romanas aves propria legionum numina. Pliny, Epist. VI.16.15: In com- mune consultant, intra tecta suhsistant an in aperto vagentur. The verbs in the indicative mode used in connection with the subjunctives illustrated fall into certain rather narrow categories and play an essential part in making clear the sentence connection. But none of them really make necessary the subjunctive mode fol- lowing them so that an important part is played by the incompleteness of that mode. A discussion of the development of this type is reserved for Chapter VII. Even when the verb in the indicative does not fall into one of these categories of commanding, urging, asking, permitting, and the like, the subjunctive mode in ordinary use has still a relativity or incomplete- ness, usually marked by a conjunction but not depend- ing on any such external means for its primary con- nective force. That this is so, is indicated by the sub- junctive use with conjunctions so different in origin as si and ut, but in classical Latin the conjunction and the subjunctive are so closely united that it would be beyond the sphere of this present study to do more than indicate the essential incompleteness of the sub- junctive mode in actual practice. 108 LATIN SENTENCE CONNECTION The infinitive, too, has been looked upon as an incom- plete mode and in so far as it has any real verbal force it is incomplete by itself. But it is far more noun than verb in all its usages and hardly comes into a discus- sion of sentence relations. The phrase, in hostem ire exercitum, is surely incomplete as it stands alone, but there is more than mere logical incompleteness. The accusative indicates something different and the con- clusion of the sentence with dixit indicates that the phrase was really substantival, the virtual object of the verb. The relation expressed is no doubt analo- gous to that noted above with the subjunctive, but the actual parallel is with a noun within a sentence, its relation to the rest of the sentence marked by its case. The indicative mode, of course, has no notion of incompleteness inherent in it. In so far as it shows any influence of the principle of relativity it will be found to be due to the effect of tense and not mode or else to the meaning of the verb. The imperative, too, is primarily a "complete" mode, but one adapted to use in conversation rather than in consecutive dis- course. This latter fact will be found later to account for an anticipatory incompleteness of considerable importance.^ Finally, there are two tenses which are essentially incomplete. To say that anything had happened, im- plies that something else did happen to which it was prior, and to say that something will have happened implies an intermediate time between the present and some future time indicated. In other words, the plu- 1 Chapter VI, pp. 173 ff. RETROSPECTIVE INCOMPLETENESS 109 perfect and the future perfect are relative tenses and by the very meaning which they give to a word, render its total meaning incomplete. Tac. Hist. 1.67.4: Ini- tium hello fuit avaritia ac festinatio unaetvicensimae legionis; rapuerant pecuniam missam in stipendium castelli, etc. Sallust, Cat. 7.5 : Igitur talibus viris non labor insolitus, non locus ullus asper aut arduus erat, non armatus hostis formidulosus: virtus omnia domuerat. In all such cases the tense of the second verb implies something antecedent to the time of the first sentence. Regularly the sense is explanatory largely by virtue of the suggestion of antecedence lying in the tense, but partly also as a result of the anticipation inherent in the first sentence. In Caesar, Bell. Civ. 1.51.1, the demonstrative pronoun is used without changing the force of the pluperfect in any way: Nuntiatur Afranio magnos commeatus, qui iter habebant ad Caesarem, ad flumen constitisse. Vene- rant eo sagittarii ex Rutenis, etc. Very frequently a conjunction supports the pluper- fect tense and the conjunction is regularly enim or nam. For example, Tac. Ann. XV.2.1 : hunc ego eodem mecum patre genitum . . . in possessionem Armeniae deduxi, qui tertius potentiae gradus habetur: nam Medos Pacorus ante ceperat. Cicero, In Q. Caec. 19.63: Atque ille Cn. Pompeius ita cum C. lulio con- tendit, ut tu mecum; quaestor enim Albuci fuerat, ut tu Verris. It is clear that retrospective incompleteness of function is as inconclusive in defining relation as is that of content. It indicates rather than defines the 110 LATIN SENTENCE CONNECTION relation except in such a special use as that of the pluperfect tense. Its importance is not great but it must be understood, as must all retrospective incom- pleteness, primarily to make comprehensible the prin- ciple of incompleteness in its really important field, the anticipatory. CHAPTER V CHANGE If you say, ''The house is white; the house is old," the repetition of the word house and the repetition of the sentence structure combine to convey the notion of two statements starting from the same point and coordinate with each other. If you say, ' ' The house is white; white is my favourite colour," the repetition of the word white without the repetition of the sen- tence structure gives the notion of consecutive ideas, one beginning where the other left off. If you say, **The house is white; the barn is red," the first thing to strike your hearer's attention is no longer verbal repetition but verbal change. The repetition of struc- ture still conveys the notion that the ideas are con- temporaneous but the change indicates at once that they are in opposition. Here are three distinct sen- tence relations. The first two, by the very fact that they repeat in the second sentence an element from the first, call attention to the fact of sentence relation. The third can do this only indirectly, through the fact that every change which is sufficiently abrupt and clearly enough defined to mark a definite sentence rela- tion takes place within a given category, so that there is an unconscious sense of repetition latent in the 112 LATIN SENTENCE CONNECTION change. Otherwise change can only define a relation which is necessarily implied by juxtaposition. It is for this reason that change has sometimes been discarded as a means of expressing sentence relation.^ But the mere fact that it does not primarily draw attention to relation is no reason for discarding it as a means of defining relation. The principle of incom- pleteness does primarily indicate relation and in addi- tion frequently defines it, and is constantly used to indicate in advance a use of the principle of change to follow. In other words, the principles of repetition and of incompleteness occur many times to indicate or call attention to the relation defined by the principle of change. Obviously change is more limited in its effective range, and it is easy to see why the types of relation indicated by change are more sharply defined than the others, and why so much of the work of anticipation is directed toward making clear relations marked by change. The illustration used above shows semantic change. From what has been already discovered in the use of repetition and incompleteness, it is to be expected that there will be formal change as well as semantic; in other words, that the present principle will be found operating as well in the field of function as in the field of meaning. And this is actually the case. But in the instances of functional change the usage seems to be less narrowly confined to the expression of a single relation. Change of mode or tense is noticeable at once and is therefore effective in defining sentence 1 Miss Nye: Sentence Connection, p. 27. CHANGE 113 relation even when it is not forced into striking relief by contrast. Semantic change is effective only when it is so abrupt as to suggest at once some contrast; functional change indicates a greater variety of rela- tions. If you say ''John came home; the house had burned down," the change of tense, while obscured to a certain extent by the incompleteness inherent in the pluperfect, is still effective. If you say ''John came home; the house will burn down," the change of tense is effective in a different way and no longer obscured by any functional incompleteness. The change, in this case, indicates a fact subsequent to the first and not yet accomplished. In the use of Latin modes this be- comes more obvious and more easily illustrated. Necesse est eas: the change of mode is the distinguish- ing mark of the sentence relation. So in the very familiar use of the infinitive mode in indirect dis- course, the modal change is an indicator, even though that relation has been indicated already by a different means in the first sentence. It will be convenient to take up semantic change first. It is obvious that it is on the whole a verv exceptional sentence which does not show decided semantic change from the preceding sentence. One showing no such change is very nearly unique, a mere rhetorical curiosity used with some very special pur- pose. It is partly on this account that the most insig- nificant types of repetition are noticeable; and it is also on this account that, to serve as a tangible means of defining sentence relation, semantic change must be abrupt enough to compel attention, and that it must 114 LATIN SENTENCE CONNECTION also be confined within a sufficiently narrow range to be noticeable as efficient semantic change. In other words, there must be abrupt change of meaning occur- ring between words which are either essentially or at least temporarily in some common category. This fact is important for an understanding of the principle of change; it is also convenient as furnish- ing a means of classification. An illustration of ex- tremes will make it clear: *'It is a glorious day. Phalaris was a tyrant. ' ' There is semantic change here without a doubt. But it is not significant. It is not safe to say that there is no relation between the sen- tences, but it is reasonably certain that the relation was conceived in an erratic if not insane mind. ' ^ The day is heavy. My heart is light." Here again is semantic change and this time significant. And for this reason : heavy and light are adjectives in the same category, that is, adjectives of the same sort and applicable to the same sort of nouns. That fact and the further fact of verbal repetition in the is and for- mal repetition in the arrangement of the sentence, calls attention to the sentence relation. But the contrast, the abrupt semantic change, finally defines that rela- tion. Had the sentences been ' ' The day is heavy. My heart is sad, ' ' all the facts of the case would have been the same except for the absence of abrupt semantic change. And yet just that difference makes the sen- tence relations in the two .examples totally unlike. This is clear from a consideration of what conjunc- tions might be used to further emphasize the relation in each example. CHANGE 115 To repeat, the use of semantic change is possible only within common categories and only when the change is sufficiently abrupt to be significant. The illustrations may then be conveniently grouped according to the categories of the words which exhibit the change. The association between the words may be effected by their own essential meaning, in which case it is a real and permanent association ; or it may be effected by the special meaning given the words by the context, in which case it is a fictitious and tempo- rary association. Often it is not easy to distinguish the two, each element reinforcing the other. Probably the only words actually in absolute, nat- ural contrast, that is, words in which, by their essen- tial meaning, the semantic content of the one is the exact opposite of the semantic content of the other, are the positive and negative adjectives built on the same stem, such as aequus and iniquus. But, prac- tically speaking, the range is much wider: magnus, parvus; lente, celeriter; nunc, tunc; these and many more can scarcely be used in adjacent sentences with- out necessarily suggesting contrast. At the other extreme are such words as proper names which sug- gest no contrast whatever unless their particular con- text makes the change between them so abrupt as to be efficient. Between these two extremes fall many cases not belonging clearly to either group. Senatus and populus, or pedes and eques have attained their element of contrast by usage but it has almost become a permanent characteristic. 116 LATIN SENTENCE CONNECTION The fact that change, and in particular, semantic change, has so limited a range of effective usage, makes it unnecessary to go extensively into illustration. But though it is narrowly confined, perhaps because it is so confined, the usage is most important : it is employed to express a clear-cut and familiar type of sentence relation and underlies a very common conjunctional usage. This will be evident after a study of the exam- ples gathered here and rather arbitrarily grouped for the sake of convenience. As stated above there are many words whose mean- ings make them either the exact or the practical oppo- sites of other words. Such are primarily words like nocens and innocens, piger and impiger; but there is no sharp line between these and such other words as celer and lentus, albus and niger, olim and nunc. Each of these words has a wide range of usage by itself, but when they are used in close connection with their opposites, the fact of semantic change is so evident, the change itself so abrupt, as to make an unmistak- able indication of sentence relation. The sentences in which they stand are thrown into the same direct opposition that the words themselves have from the semantic point of view. Seneca, De Ira II.18.2 : facile est enim teneros adhuc animos componere, difficulter reciduntur vitia quae nobiscum creverunt. Though taken apart from their context, these two sentences stand clearly in opposi- tion to each other. The first one need not have been followed by a sentence in opposition to it. It is per- fectly conceivable that any one of several types of CHANGE 117 sentence might have followed: for example, a conclu- sion drawn from the fact stated, possibly introduced by ergo. But the first word of the second sentence at once fixes the relation, and this is because of the abrupt semantic change. The same is clear in all of the follow- ing examples. Caesar, Bell. Civ. III.87.2: Perexigua pars illius exercitus superest; magna pars deperiit. Cicero, Pro Rose. Am. 20.56: quod innocens, si accu- satus sit, ahsolvi potest, nocens, nisi accusatus fuerit, condemnari non potest. Sallust, Cat. 58.17 : Semper in proelio eis maxumum est periculum qui maxume timent: audacia pro muro habetur. Sallust, Bell. Jug. 10.6: Nam concordia parvae res crescunt, discordia maxumae dilabuntur. Livy XXXIV.13.5 : Adhuc prae- donum magis quam bellantium militastis more; nunc iusta pugna Jiostes cum hostibus conferetis manus. Quint. Inst. Orat. II.4.6: Facile remedium est uber- tatis; sterilia nullo labore vincuntur. Seneca, De Ira 11.28. 8: Aliena vitia in oculis habemus, a tergo nostra sunt. Tac. Hist. 1.15.19 : Fortunam adhuc tantum adversam tidisti: secundae res acrioribus stimulis animos explorant, quia miseriae tolerantur, felicitate corrumpimur. In these illustrations a number of facts stand out distinctly. First of all, it is the abrupt semantic change which determines the sentence relations. The ideas brought into sharp opposition may be contained in very different words as readily as in those that are in the same grammatical category. The range is all the way from such similar words as concordia and dis- cordia, or adhuc and nunc, to such unlike words as 118 LATIN SENTENCE CONNECTION timent and audacia. Ubertatis and sferilia, fortunam adversam and secundae res are midway between these extremes. Furthermore, the position of the contrasted words is only an aid to the understanding of the rela- tions. As a general rule it is true that some method is employed to emphasize the abruptness of the change. The two words may be each at the head of its clause, as in the example from Cicero, or the first at the begin- ning, the other at the extreme end, as in the case from the De Ira; or vice versa, as in the instance from Quintilian. The last is the most striking, the first is probably the most frequent. Again, it is very noticeable in all of these examples that although the abrupt change, the chief factor in defining relation, comes usually in the first word of the second sentence, making it sufiQciently marked to assure its effect, there is almost always much supple- mentary change to reinforce this effect. For example, in the instance from Livy: nunc contains the first sign of relation, an abrupt change from adhuc; but this is followed up by further noticeable change in iusta pugna from praedonum more. So in the example from Cicero : tiocens and innocens furnish the first and most important change ; but this is supplemented and rein- forced by the further contrast between ahsolvi and condemnari, and between potest and non potest. The same is just as noticeable in the Senecan case: the primary contrast is between in oculis and a tergo, but equally ^-ivid is the supplementary contrast between aliena and nostra. In Pliny, Epist. TV. 7. 3, there is a typical instance of the piling up of the contrast: ita CHANGE 119 recta ingenia debilitat verecimdia, perversa confirmat audacia. Finally, there is frequent, almost regular, occurrence of some sort of repetition in these examples. It serves to draw attention to their sentence relation and to mark the sentences as coincident. It ranges through all types of functional repetition and even repetition of meaning appears. It is more fully discussed below (pp. 124 and 125). A favourite rhetorical device of Cicero is based on this principle of abrupt change and falls under the head of changes, essentially inherent in the word meaning, a device by no means confined to Cicero : the use of a negative statement for the sake of emphasizing by contrast the following positive sentence. At present it is with the second sentence that we have to do. Anticipation there undoubtedly is, but the determina- tion of the sentence relation lies in the contrast in the second sentence. It was Cicero who chiefly used this in its simple form with a pure contrast. Pro Rose. Am. 27.73: Non quaero ahs te, quare patrem Sex. Roscius occiderit, quaero, quo modo occiderit. The same effect is obtained in the Pro Quinctio, 27.85 : haec omnia mitto; illud dico, dominum expidsum esse praedio. This use of mitto has the effect of a negative. The order may be reversed so that the negative clause is the second. The only real change is that there is then no apparent anticipation. Cicero, Pro Rose. Am. 21.58: Ego quid acceperim, scio, quid dicem, nescio. The importance of the repetition in this example, indicating the relation which the semantic 120 LATIN SENTENCE CONNECTION change defines, will be discussed after the other types of change have been illustrated. There are very many words without inherent con- trast into which persistent usage within a given cate- gory or field of meaning has injected an element of contrast, so that when they are used together within that category they are entirely like the words already discussed in the influence which they exert on the expression of sentence relations. Domi and foris furnish an extreme instance of such words, so extreme, however, that it is almost impossible to think of them without getting a sense of tacit contrast. Not so with populus and senatus or with verba and facta. There is, however, no new principle involved : a few illustra- tions will serve to make the case more obvious. Cicero, Pro Rose. Am. 5.13: Accusant ii qui in fortunas huius invaserunt, causam dicit is cui praeter calamitatem nihil reliquerunt. Sallust, Cat. 9.2: lurgia discordias simultates cum hostibus exercebant, cives cum civibus de virtute certabant. Sallust, Bell. Jug. 30.1 : Apud plebeni gravis invidia, patres solliciti erant.^ Tac. Ann. IL19.8: Hie pedes adstitit; equitem propinquis lucis texere. Tac. Ann. I.l.l: Urbem Romam a principio reges habuere; libertatem et consulatum L. Brutus instituit. Cicero, In Verrem 1.15.44: verbo illam poscere videbatur, re vera iudicia poscebat. The con- trast in Sallust, Cat. 9.4, is between m bello and in pace. There are left for consideration the examples of abrupt semantic change which is essentially due not to any inherent contrast of meaning but to a contrast CHANGE 121 temporarily given to the words by the context in which they stand. The only essential difference between these and the cases already studied is that more de- pends on the context for an understanding of the sentence relation, since from the context arises the contrast between the words. If they stood alone in a list they wouH not be thought of as contrasts. Aequus and iniquus can be nothing else than contrasts ; not so maiestatis and de pecuniis repetundis. These two seem to have more of similarity than of abrupt change. But in their setting in Tac. Ann. 1.74.22, the situation makes the contrast perfectly obvious: patiens tulit ahsolvi reum criminibus maiestatis: de pecuniis repe- tundis ad reciperatores itum est. The position of the words and the change in the meaning of the verbs supplements the effect of the context. A better illus- tration is Sallust, Bell. Jug. 102.7: quia parentis ahunde habemus, amicorum neque nobis neque cuiquam omnium satis fuit. Without the context, parens and amicus would scarcely be thought of as contrasted con- cepts; with the context, aided by the usual supple- mentary contrast, and the emphatic juxtaposition of the words, there is no doubt of their contrasted use. Other examples are the following: Cicero, Tusc. Disp. IV.5.9 : Peripatetici autem ad placandos animos multa adferunt, spinas partiendi et definiendi praetermittunt. Sallust, Cat. 52.6 : Non agitur de vectigalibus neque de sociorum iniuriis: libertas et anima nostra in dubio est. Tac. Hist. IV.17.24: Libertatem natura etiam mutis animalibus datam, virtutem proprium hominum honum. 122 LATIN SENTENCE CONNECTION A few characteristic uses of the principle of semantic change remain to be illustrated. They are not essen- tially different from those already shown, but form sufficiently distinct groups under the general divisions to be worth indicating. One is the use of the demon- strative pronouns, hie and ille. Cicero, De Fin. 11.26.82: Sed haec nihil sane ad rem: ilia videamus, quae a te de amicitia dicta sunt. The sane gives warning here of the abrupt change to follow. It does not occur in the following: Cicero, Pro P. Sulla 3.8: Ilia enim ad breve tempus severitatem postulavit, haec in omni vita misericordiam lenitatemque desiderat. Sallust, Cat. 54.2: Ille mansuetudine et misericordia clarus factus, huic severitas dignitatem addiderat. The usage is too familiar to need further illustration. The last instance above shows how easily the usage might pass over to proper names. The pronouns stand merely for Cato and Caesar who are so brought into contrast by the context that the names would have suggested it as well as the pronouns. The effect is frequently so obtained. For example, Caesar, Bell. Civ. III.30.4: Sed Caesari circuito maiore iter erat longius, adverso fiumine, ut vado transire posset; Pompeius, quia expedito itinere flumen ei transeundum non erat, magnis itinerihus ad Antonium contendit. Sallust, Bell. Jug. 52.2: Nam Metello virtus militum erat, locus advorsus; lugurthae alia omnia praeter milites opportuna. Quint. Inst. Oral. II.4.33: Apud Graecos enim lator earum ad iudicem vocabatur, Romanis pro contione suadere ac dissuadere moris fuit. CHANGE 123 There is no real difference when personal pronouns are used. Cicero, Pro Quinctio 10.35: Tu id semper facis, quia semper potes, ego in hac causa faciam, propterea quod in hac videor posse facere. Caesar, Bell. Civ. 11.32.11 : At, credo, Caesarem prohatis, in me offenditis. Sallust, Cat. 58.11: nos pro patria, pro lihertate, pro vita certamus, illis supervacuaneum est pro potentia paucorum pugnare. There is a noticeable tendency in such cases to add a conjunction to make the relation more obvious, as in Cicero, Pro Rose. Am. 30.84: Causam tu nullam reperiebas in Sex. Ros- cio; at ego in T. Roscio reperio. Adverbial clauses frequently take the place of simple adverbs, with exactly the same effect as single words. For example, Cicero, Pro Rose. Am. 43.126: Dum praesidia ulla fuerunt, in Sullae praesidiis fuit; posteaquam ah armis recessimus, in summo otio rediens a cena Romae occisus est. Sallust, Cat. 51.3: Ubi intenderis ingenium, valet; si luhido possidet, ea dominatur, animus nihil valet. Sallust, Cat. 3.2 : quae sibi quisque facilia factu putat, aequo animo accipit, supra ea veluti ficta pro falsis ducit. It has been suggested several times that, unlike the element of incompleteness, change of meaning is pri- marily a means to define relation rather than to draw attention to it. (Cf. p. 112.) This is very nearly axio- matic, for the very nature of this element makes it unsuitable for pointing out relation. Change of mean- ing is always present in a greater or less degree and only as the element of repetition (repetition of cate- gory, ordinarily) gives it significance, does it figure as 124 LATIN SENTENCE CONNECTION even a defining element. The extent to whicli more fundamental elements of connection, repetition and incompleteness, underlie the element of contrast varies with every instance. Most frequently some form of functional repetition is present. The contrasted clauses may be both subordinate to a third clause and even introduced by a common particle: Cicero, Pro Rose. Am. 31.86: cum viderent illos amplissimam pecuniam posidere, hunc in summa mendicitate esse. Quint. Inst. Orat, I. Proem. 4: ut operum fastigia spectantur, latent fundamenta. This is perhaps the most widely used type of contrast. The form, however, which the functional repetition takes is exceedingly varied as the following examples will show. Pliny, Epist. TV. 17.1 : Quod admones gratias ago, quod rogas queror. Cicero, Ad Att. II.20.2: si erit pugnandum, arcessam ad societatem lahoris; si quies dahitur, ah Amalthea te non commovebo. Cato, R.R. CXII.2: et ponito in sole hiduum aut triduum suh dio, si pluviae non erunt. Si pluvia erit, in tecto in cratihus con- ponito. Tac. Ann. XIV.9.2 : sunt qui tradiderint, sunt qui ahnuant. Seneca, Ad Polyh. 4.1 : Diutius accusare fata possumus, mutare non possumus. These illustrations are all fundamentally alike. Whether the repetition is that of modal construction or of meaning, whether slight or extensive, it is in each instance primarily repetition of function, and there- fore fixes the sentences as coincident. Incompleteness of meaning helps to draw attention to the relation in the following instances but does not change the funda- mental force of the repetition or the secondary force CHANGE 125 of the contrast. Cato, R.R. XCIII.l: Ad arhorem maxumam urnam conmixti sat est: ad minores arbores pro ratione indito. Seneca, Medea 159: For- tuna fortes metuit, ignavos premit. Seneca, De Vita Beata 10.3 : Tu voluptatem complecteris, ego conpesco; tu voluptate frueris, ego utor; etc. A different type and a comparatively rare one is illustrated by Livy I.lO.l: lam admodum mitigati animi raptis erant, at raptarum parentes turn maxime sordida veste lac7'imisque et querellis civitates con- citabant. The repetition is primarily of content; the type of sentence relation indicated is the subsequent. There is suflScient vagueness about the relation to lead Livy to use a conjunction, but that is a matter of style ; Tacitus would probably not have used the at. For the mitigati erant and the concitabant are in contrast and this, with the rest of the context, is sufficient to suggest the temporary contrast between raptis and raptarum parentes, suggested also by their positions in their respective clauses. The example is interesting as showing the power of the secondary element to modify very decidedly the force of the primary. The use of the conjunction is probably significant of the resulting vagueness. The contrasted element is sometimes emphasized by adverbs like certe, quidem, or sane. These are of comparatively slight importance in the present con- nection but of considerable importance later on, in the study of the means by which contrast is anticipated. (Cf. Chap. VI, p. 155.) They should therefore be noted in the following examples. Cicero, Ad Att. II.1.6 : Hoc 126 LATIN SENTENCE CONNECTION facer e ilium mihi quam prosit, nescio; rei puhlicae certe prodest. Cicero, In Q. Caec. 15.48: qui quid in dicendo posset, numquam satis attendi, in clamando quidem video eum esse bene rohustum. Cicero, Pro Rose. Am. 19.54: Vere nihil potes dicere; finge aliquid saltem commode. The adverbs are a help to the understanding of the relation because of the em- phasis which they bring to the contrasted element. There are a number of words whose meaning espe- cially adapts them for use in instances of contrast. They are of two types : first, words used as an integral part of the second sentence but with an incompleteness of content and a meaning suggestive of contrast, such as ceterus, alius, reliquus; second, adverbs, not an integral part of the second sentence or clause, but simply inserted to mark it as contrasted with the first, adverbs like rursus, contra, ex diverso. The first group owe their power to suggest contrast to that element in their meaning which selects one unit or group of units and definitely distinguishes it from others. Primarily these words are incomplete in mean- ing and this phase of their influence has been already discussed. (Cf. Chap. IV, p. 98.) But their meaning makes them adaptable to this particular type of in- complete usage. The type is too familiar to require more than a passing illustration. Cicero, Ad Att. III.lO.l : Acta quae essent usque ad VIII Kal. lunias, cognovi ex tuis litteris; reliqua exspectabam, ut tibi placebat, Thessalonicae. This is not the only use of these words. They are comparative in sense and it is only their particular CHANGE 127 meaning which makes their use in contrasts common. The development of the conjunction ceterum to mark a sentence or clause as bearing an adversative rela- tion to the one preceding it, is the result of this element of meaning in the adjective ceterus. The second group of words noted above is like the first in that the use with instances of abrupt semantic change is not their only one, but otherwise they are different. Contra, ex diverse, rursus, and similar ex- pressions, when used as adverbs introducing a clause or sentence and modifying it as a whole, are nothing more than indicators of relation of a mechanical sort. For example, Cicero, Pro Rose. Am. 29.79: Qiios neque ut convenire potuerit neque qua ratione . . . potes ostendere. Ego contra ostendo non modo nihil eorum fecisse Sex. Roscium, etc. The marks of con- trast, the functional repetition defined by semantic change, do not really need the contra to call attention to it. This adverbial use indicates clearly the origin of the adversative conjunctions which were first ad- verbs with the notion of ''apart" or "away from." A few instances of semantic change in which adver- sative conjunctions are used will show that these, like the adverbs, are primarily supplementary. Sallust, Cat. 51.26: Illis merito accidet quicquid evenerit: ceterum vos, patres conscripti, quid in alios statuatis, considerate. Pliny, Epist. IV.29.2 : Egit ille in senatu causam suam, egit autem sic ut deprecaretur. This is a nice case of functional repetition in which semantic change is present but inconspicuous and appearing late in the sentence, so that the autem is a distinct help to 128 LATIN SENTENCE CONNECTION the quick understanding of the relation. Cicero, Pro Rose. Com. 10.28: Panurgum tu, Saturi, proprium Fanni diets fuisse. At ego totum Rosci fuisse eon- tendo. Cicero, In Q. Caee. 7.23: reus ut absolvatur, non peto, sed ut potius ah hoe quam ah illo aceuse- tur, id peto. Livy XXXIV.7.4: Sed in purpura, quae teritur ahsumitur, . . . aliquant . . . eausam tenaeitatis video; in auro vero, in quo praeter manupretium nihil intertrimenti fit, quae malignitas est? An interesting instance of the familiar contrast between adverbs, sup- plemented as it frequently is, by the emphasizing quidem in the first sentence and by the conjunction autem in the second, is Cicero, De Orat. 1.4.14; the first sentence begins Ac prime quidem, the second post autem. The contrast between demonstrative pronouns is supplemented by a conjunction in Cicero, Pro Rose. Com. 1.1 : Erit in illius tahulis hoc nomen, at in huius non erit. Finally, Cicero, Tusc. Disp. V.19.55, illus- trates the use of the conjunction to supplement con- trast between names : Laelius si digito quern attigisset, poenas dedisset; at Cinna collegae sui, consults Cn. Octavii, praecidi caput iussit. Two adverbs which, in this particular relation, become practically conjunctions, should be included here. The first is nunc, used in the sentence following a condition contrary to fact. Tac. Agr. 34.1 : Si novae gentes atque ignota acies constitisset, aliorum exerci- tuum exemplis vos hortarer: nunc vestra decora re- censete, vestros oculos interrogate. The aliorum and the vestra, vestros are in sharp contrast, but there is no clause balancing the si . . . constitisset to mark the CHANGE 129 coincidence and the change. This balance is supplied by the nunc. Cicero, Pro Quinctio 14.47, shows that neither is necessary: nihil hoc tanto negotio, nihil tarn invidioso iudicio, nihil tarn copiosa advocatione uterer, si petendum esset; extorquendum est invito atque ingratiis. After other conditional sentences aliter serves the same function as the nunc does after conditions con- trary to fact. Tac. Hist. IV.59.16: Classicus corrup- tissimum quemque e deditis pergere ad obsessos iubet, veniam ostentantes, si praesentia sequerentur ; aliter nihil spei, famen ferrumque et extrema passuros. The incomplete adverb merely suggests a condition in con- trast to the one actually expressed in the first part of the sentence. The type of relation defined by all the examples of semantic change is the same. The two clauses are in contrast with each other and, except in the rarest instances, coincident. But the coincidence is deter- mined by other means. Only the opposition or con- trast is indicated by the abrupt change of meaning and that is always the same whether supported by the use of conjunctions or not. An unusual emphasis on one of the contrasted sentences may lead to the subordina- tion of one to the other but this can be more fully appreciated after a study of the means by which con- trast is anticipated. Early in the present chapter (page 112) it was stated that the range of relations expressed by functional change was less limited than the range of those ex- 130 LATIN SENTENCE CONNECTION pressed by semantic change. At first glance this seems not to be true. It is even doubted whether functional change can express sentence relations at all.^ This is partly the result of a confusion of two different ideas,^ one the shift of mode, tense, or person noted by Her- mann^ as a distinguishing mark of subordination and generally so accepted, the other the change of mode, tense, or person from the verb of one sentence to that of another contiguous to it. The former is the assumed change through which a verb form passes in its tran- sition from an independent to a dependent use, a change in the individual form itself. The latter is analogous in kind to the repetition of function already noted ; it is a phenomenon embracing two verbs in two different clauses. Another source of confusion lies in the fact that, like so many of the means uncon- sciously employed by language, this element of func- tional change is almost inextricably united with other elements contributing to the same end. Finally, it should be added that, although change of function expresses more different sorts of relation, it expresses none with the same precision that characterizes se- mantic change. Change of person can scarcely be regarded as a functional change. Eather, it is one of subject only and as such is essentially semantic. It is impossible to conceive of person in a verb apart from a concrete 1 Miss Nye: Sentence Connection, pp. 26 ff. 2 This confusion is my own. See : Sentence Connection in Tacitus, pp. 136 f. 3 See: Hermann: Gab es im Indogermanischen Nehensdtse; K. Z. 33, pp. 481 ff. CHANGE 131 subject, whether specific or general. So, whether there be repetition or change of person, it is inevitably the subject of the verb which the mind conceives as either repeated or changed. The result is that the principles already noted in the study of semantic change apply absolutely to change of person : ordinarily it will have no appreciable effect because it is the normally expected thing; but when the change is deliberately made obvious, in other words, when it is so striking that the two subjects are thrown into sharp opposition, the result is a contrast between them, suggesting at once an adversative relation between the sentences. The development in Romance languages of pronouns regularly to make clear the person and to take over a part of the function of the verb, is an indication of the semantic character of this element of function. In Latin it was usually necessary to employ the pronoun to make contrast obvious and efficient. This is illus- trated in the following examples. Cicero, De Imp. Cn. Pomp. 5.11: Illi libertatem inminutam civium Roma- norum non tulerunt; vos ereptam vitam neglegetisf Seneca, Troades fil^ : unum quaeris, ego quaero omnia. Pliny, Epist. IV.14.9 : Proinde, sive epigrammata . . . seu quod aliud vocare malueris licehit voces, ego tan- tum hendecasyllahos praesto. Compare an instance supported by the supplementary force of a conjunction : Pliny, Epist. 1.5.7 : Quaeris . . . quid sentiam; at ego ne interrogare qiiidem fas puto, de quo pronuntiatum est. While change of person is essentially a semantic change and therefore confined in range to the one rela- 132 LATIN SENTENCE CONNECTION tion expressed by semantic change, change of tense is different in character and has a wider range. It may be, and often is, deliberately made obvious and empha- sized in such a way as to produce the same result as that attained by emphatic semantic change, but it is also effective in its less striking forms to express less clearly defined relations. The first of these two uses is distinctly rare and usually, if not always, supported by further means. Illustrations are the following: Pliny, Epist. IV.12.1: Amas Egnatium Marcellinum atqiie etiam mihi saepe commendas: amahis magis commendabisque, si cognoveris eius recens factum. Martial V.9.4 : Non hahui fehrem, Symmache, nunc haheo. Seneca, De Ira, 1.5.1 : Quid esset ira quaesitum est . . . ; nunc quaeramus an ira secundum naturam sit. A further step is taken when a conjunction is used with the first clause : Cicero, De Imp. Cn. Pomp. 8.20 : Quoniam de genere belli dixi, nunc de magnitudine pauca dicam. Although rare in occurrence this type is easily understood after a study of the other changes which produce contrast. The chief field of influence for change of tense lies in the direction of less specific and clearly defined usage. The most obvious and frequently recurring change is from a narrative past tense to a pluperfect or the reverse. This change has a distinct influence in determining the relation between the sentences, origi- nating in the incomplete nature of the pluperfect already noted. (Cf. Chap. IV, p. 108.) That incom- pleteness marks the clause in which the pluperfect stands as antecedent to the other. Tac. Hist. III.84.27 : CHANGE 133 Laniata veste, foedum spectaculum, ducebatur, multis increpantihus, nullo inlacrimante: deformitas exitus mis eric ordiam ahstulerat. The summary repetition in deformitas exitus at once suggests a sentence logically subsequent to the preceding and only the tense change in ahstulerat corrects this impression. The resultant double impression is characteristic of almost all ex- planatory sentences. They are very often somewhat in the nature of afterthoughts; or else they are deliber- ately and for rhetorical effect kept back until after the clause of which they are the explanation. But logically they are antecedent to the other clause and this fact is brought out by the tense change. In the present instance, had there been a narrative tense in the place of the pluperfect, its clause would have been without ambiguity subsequent logically to the first clause and the sense would have been totally different. Had the first verb, on the other hand, been a pluperfect, there would have been a strong tendency to take the two clauses as coincident, which would hardly have been overshadowed by the force of the incompleteness of the tense in ahstulerat^ Tac. Ann. XV.16.16: Decesserat certamen virtutis et amhitio gloriae, felicium hominum adfectus: sola misericordia valehat, et apud minores magis. Caesar, Bell. Civ. II.4.4: Adventus enim L. Nasidii summa spe 1 Miss Nye (Sentence Connection, p. 26) expresses a different view. Such an instance as the following seems to me to show that it is the change of tense and not merely the incompleteness of the pluperfect which defines relation: Pliny, Epist. VI. 20.11: Nee multo post ilia nubes descender e in terras, operire maria; cinxerat Capreas et abscon- derat ; Miseni quod procurrit, ahstulerat. 134 LATIN SENTENCE CONNECTION et voluntate civitatem compleverat. Nacti idoneum ventum ex portu exeunt. The fact that the change of tense is, in these examples, accomplished in the reverse order, does not alter the significance: the clause with the pluperfect tense is antecedent to the other. This is the relation even when the clause with the narrative tense is made syntactically subordinate, as it frequently is. For example, Vixdum finierat Maternus concitatus et velut instinctus, cum Vipstanus Messalla cubiculum ingressus est. (Tac. Z>iaZ. 14.1.) The difference is that, in this last type, the emphasis is on temporal rather than causal relation. The temporal idea is often so clear in the meaning of the words of the first clause that the use of the conjunction et instead of cum does not change the effect: e.g., Vergil, Aen. 3.9: vix prima inceperat aestas, Et pater Anchises dare fatis vela iubebat. If the element of contrast enters in, usually marked by the use of a conjunction, since otherwise it is not obvious, the emphasis is again changed, but without changing the fundamental relation : Tac. Ann. 1.56.18: Fuerat animus Cheruscis iuvare Chattos, sed exterruit Caecina hue illuc ferens arma. That the causal relation, first noted, is often marked by a con- junction too, has been already illustrated (Chap. V, p. 108) in discussing the incompleteness of the plu- perfect. When the pluperfect clause comes first the conjunction igitur in the second clause performs this function: Tac. Hist. III.69.1: Praevenerat rumor eiurari ah eo imperium, scripseratque Flavius Sahinus cohortium trihunis, ut militem cohiberent. Igitur . . . primores senatus . . . domum Flavii Sabini complevere. CHANGE 135 When the clauses come in the reverse order, nam, enim and quippe are the usual conjunctions with the plu- perfect. Suet. Galba 9 : Nee diu cunctatus, conditionem partim metu, partim spe, recepit; nam . . . mandata Neronis de nece sua ad procuratores clam missa deprehenderat. In a rather general and undefined way, this explana- tory force seems to lie behind all changes from one tense to another indicating priority in time. If the tense changes in consecutive discourse from a future to a present or past or from a present to a past, it indi- cates a purpose on the part of the writer to guide the mind of the reader back to a point logically antecedent to the one already stated. For example, Sallust, Cat. 52 A : hoc, nisi provideris ne accidat, uhi evenit, frustra iudicia implores: capta urhe nihil fit reliqui victis. The repetition of content in the second sentence is undeni- able, but the change in tense defines the relation as explanatory. Again and again in Cato practical in- structions are given in the imperative, which has necessarily a future sense, followed by a statement in the present tense containing an explanation of the reason for the rule laid down. R.R. LXIII. is typical of these instances; Per aestatem boves aquam honam et liquidam hihant semper curato: ut valeant refert. The following examples show the change from a pres- ent or a future to a past tense with the same resultant effect. Sallust, Cat. 58.15: Si haec relinquere voltis, audacia opus est: nemo nisi victor pace helium mutavit. Seneca, Troades 886: Hie forsitan te casus excelso magis Solio reponet. Profuit multis capi. This usage 136 LATIN SENTENCE CONNECTION was never sufficiently common to develop any clearly defined function as a means of expressing sentence relation. But it was obvious enough to be made use of effectively by the rhetorical writers of Silver Latin, while its prosaic use goes back to the unadorned style of Cato. Not infrequently it is supplemented by con- junctions, as in the following examples. Cicero, De Orat. 1.1.4: Tibi vero, f rater, neque hortanti deero neque roganti. Nam neque auctoritate quisquam apud me plus valere te potest neque voluntate. Cato, R.R. CXLl : Si habebit aquam, vinum effluet, aqua manebit. Nam non continet vinum vas hederaceum. In the instances which fall under the present cate- gory just as in those which illustrated the change from a narrative tense to the pluperfect, it proves true that a reversal of the order of clauses does not change the relation of the clauses to each other. The clause whose verb shows priority of tense is still the explanatory or causal clause, the other the resultant. There is a dif- ference in the placing of the emphasis but not in the essential relation between clauses. There is not neces- sarily any anticipation of relation as there always is with the pluperfect clauses when they precede the clauses which they explain, for that anticipation re- sulted from the inherent incompleteness of the par- ticular tense. Otherwise there is nothing new in the present type. Cicero, Pro Quinctio 18.58: Profectus est una L. Albius homo cum primis honestus; dicet testimonium. It is true that the tacit repetition of Albius as subject of dicet suggests the subsequence of the second clause, but it has already been shown that CHANGE 137 change of tense may overshadow that suggestion, especially as repetition of subject is one of the least forcible types of repetition. In the present type, inasmuch as the two elements are supplementary, there is no avoidance of the explicit repetition of content. The citation above continues : Prosecuti sunt fami- liares et Alhium et Quinctium; dicent hi quoque testi- monium. The tacit repetition of a noun as subject of the second verb is the rule. Cicero, In Verrem 1.10,30 : Tres hi homines veteres tribuni militares sunt desig- nati; ex Kal. lanuar. non iudicahunt. Tac. Hist. V.23.1 : Civilem cupido incessit navalem aciem osten- tandi: complet quod biremium quaeque simplice ordine agebantur. That such repetition is not essential is shown by such instances as Tac. Hist. 111.29.12 : ceteri trepidis iam Vitellianis seque e vallo praecipitantibus perrupere. Completur caede quantum inter castra murosque vacui fuit. The present tense of complet after the perfect incessit is not really needed to point the relation which is obvious from the course of events, so that it is ordinarily called a present of excited narra- tive. But its selection at this particular point was surely determined by a desire to point more vividly the resultant nature of the second sentence. The device is made very familiar by Tacitus' use of it, nor is its force called into question by the numerous instances like intravi, conticuerunt of Pliny, Epist. II.18.2, in which the natural sequence of events is left to convey unaided the same relation. If a conjunction is used to make the relation more obvious it is regularly of the 138 LATIN SENTENCE CONNECTION igitur type, as in Tac. Ann. II.68.7 : neque vado pene- trari poterat. Igitur in ripa fluminis . . . vincitur. The sum total of evidence for the use of this means of expressing sentence relation indicates a type not sufficiently distinctive to be used extensively without being supported by further means of connection. The case is somewhat different with change of mood. The material to examine is ample, and there are obvious subordinate constructions based on this particular type of change. It seems to lie at the foundation of many of the regular subordinate clauses, as, for ex- ample, the ut clauses and those in indirect discourse. It is the type of sentence connection usually referred to as parataxis: concedas necesse est. (Cicero, Pro Rose. Am. 31.87.) Such instances are too familiar to require illustration here. Because they have been con- sidered, on the one hand, as examples of parataxis in its narrowest sense, examples of subordination without any expressed sign, and on the other hand, as examples of clauses between which there is a conjunction omitted and understood, they have been the subject of extensive study.^ They must, however, be considered not as an isolated type, but in relation to other similar changes, for the change is not confined to that from a subjunc- tive to an indicative and vice versa: Ite, Obsecror is quite as significant. The infinitive is so essentially a noun that its rela- tions to an adjacent finite verb assume the relation of noun to verb and the Latin infinitive never developed the various forms of subordinate clause that are to be 1 For a partial list of the literature see Morris, p. 113. CHANGE 139 found in the case of the subjunctive. The change from indicative to infinitive is, as a matter of fact, a con- crete sign of indirect quotation, and in this one type, I should say, is an efficient means of expressing the relation. But this is distinctly a relation of noun to verb rather than of verb to verb. The same is true of the other infinitive uses. Whether an infinitive is com- plementary or expresses indirect discourse is entirely a question of the meaning of the finite verb. The infinitive simply follows this verb in the relation that a noun would bear to it, the precise relation being defined not by the infinitive but by the finite verb. This last factor enters very largely into the deter- mination of the extent to which a simple change from indicative to subjunctive can be efficient to express relation without further means. Just so long as the finite verb has in it the notion of volition, expressed as wish, or request, or order, or even as the statement of necessity or need (which implies the volition of some compelling force), so long can the change be effective to produce the sense of relation. For the subjunctive by itself expresses volition, ranging from the mildest request to a real command. It is essen- tially imperative in sense. With the ordering type of verb, therefore, the relation is obvious, especially aided as it usually is by change of person ; volo adsis. When the finite verb was outside this category, that is, when the natural field of usage became extended, there was felt the need of supplementary mechanical means to make perfectly clear the relation of the subjunctive to the indicative. There developed the use of the 140 LATIN SENTENCE CONNECTION defining conjunction. In the chapter on Parenthetical Incompleteness the development of this usage will be discussed. At present all that need be noted is that the change from one mode to another is in itself not definite enough to express a precise sentence relation except under special circumstances. The same con- clusion is fair for functional change in general. It is not an effective element for the accurate expression of sentence relations. Its range was less circum- scribed than that of semantic change, but, without other more precise means of supporting it, it never developed general efficiency. CHAPTER VI ANTICIPATORY INCOMPLETENESS The long and rather obvious list of words and func- tions with an element of relativity, given in Chapter IV, is of importance almost solely for the sake of making clear a principle whose chief application lies in another direction. In the majority of instances cited, the element of incompleteness served hardly any further use than to draw attention to the fact that there was a relation existing between two sentences or clauses. It rarely defined that relation with any precision; that was done in most cases by further means employed. And when the relation is expressed in the second of two clauses the element of incomplete- ness is scarcely necessary, so instinctively do we assume relation between contiguous sentences. But there is a wider field for the principle of incom- pleteness. Repetition, by its very nature, could be of use only in the second clause : incompleteness may be satisfied either by what has preceded or by what fol- lows and can therefore exert its influence in either the first or the second sentence or clause. Hence it is the principle behind all forms of anticipatory connection. This is of great importance, for the practical value of an appreciation of the forms of sentence connection lies in the increased ability to grasp quickly and fully 142 LATIN SENTENCE CONNECTION the thought of any writer. This is somewhat furthered by an understanding of the retrospective means already studied. But if we can understand the various means developed for giving warning in the first sen- tence of the relation it is to bear to the second, our grasp of the writer's thought will be much quicker. To use once more the analogy of words and the sen- tence which they combine to form, each sentence will then suggest at once to the mind one or more possible relations which it may bear to the sentence following, and the means of connection expressed in this follow- ing sentence will determine for us more precisely which it is; just as each word that we read in a sen- tence not only expresses a relation to what has pre- ceded, but, by its meaning or function, suggests one or more possible relations with the words to come, the right one being determined by the words that follow. The principle is presumably clear already. If we read the sentence, haec verba dixit, there is a logical incompleteness due to the emptiness of meaning of haec. If any ''words" have been spoken, the mind instinctively refers to them the haec verba and the incompleteness is satisfied. But suppose there has been no quotation of anything said. The relativity still exists and the mind instinctively suspends judg- ment until a quotation following gives content to haec. In other words, the demonstrative pronoun, being in itself an empty word and so making its sentence or clause logically incomplete, forces the mind to look outside the sentence or clause to complete the sense. ANTICIPATORY INCOMPLETENESS 143 If this can be done by retrospection it is so done instinctively; if not, judgment must be suspended and an anticipatory means of expressing sentence relation has been established. It is apparent that this usage, anticipation of the connection, is largely rhetorical. Although the idea to follow the particular one that is being expressed is regularly present in the speaker's mind in a somewhat vague form, still the deliberate preparation for it in the framing of the sentence actually being spoken is in itself rhetorical : it is neither simple nor entirely nat- ural. But, as has been already noted, Latin prose as we have it is not simple and natural but complex and rhetorical, and it is therefore just as important to study the rhetorical development of a simple princi- ple as to note its more natural applications. In fact, such developments require more study and often lead to more definite results. Furthermore, the rhetorical nature of a usage is only a matter of degree. Every development in language as a means of expressing ideas is the result of intelligence, either conscious or unconscious. Most of the types that showed the element of incom- pleteness with retrospective force appear also with anticipatory force. In the use of conjunctions and demonstrative pronouns and also in that of the sub- junctive mode this has been already partially illus- trated (pp. 89, 33, 106), and a few instances will suffice to make it clear. With the so-called subordinat- ing conjunctions this anticipatory element is so obvious that there is danger of forgetting its character. Cum 144 LATIN SENTENCE CONNECTION esset Caesar in citeriore Gallia, does not make com- plete sense. To such an extent is this true that we call the clause subordinate. It is not the use of the sub- junctive that gives the incompleteness to the clause: Cicero begins his oration for Flaccus, Cum in maximis periculis huius urhis . . . caedem a vohis . . . depel- lebam. In neither of these cases can the cum clause refer to anything preceding, for there is nothing pre- ceding; the incompleteness in each case must be satis- fied by what follows, and so familiar is the usage that the mind is already prepared for the exact type of relation before the next clause is reached. In actual use, comparatively few of the non-subor- dinating conjunctions have anticipatory force. Enim, tamen, ergo, and their like have the force of incom- pleteness but it is retrospective ; the same is true to a considerable extent of et, neque, and aut. But these are not confined to the retrospective use. Livy XXIII. 41.3 : nam et filius Hampsicorae Hostus in acie cecidit; without the et this would be a logically complete sen- tence with no notion of reference to another. With the et this is changed : et is incomplete and renders the sentence incomplete. Furthermore, it can refer to nothing preceding and therefore attains more than a mere connective force: it anticipates the relation with the following clause. This is a good illustration be- cause the retrospective element in the second sentence is strong and clear : et Hampsicora cum paucis equiti- hus fugiens . . . mortem sihi conscivit. The incom- plete word et, at the beginning might be satisfied by repetition of either content or function as already ANTICIPATORY INCOMPLETENESS 145 indicated, but repeating, as it does, the et of the first sentence, it establishes functional repetition and at once indicates a parallelism of sentences. The repeti- tion of content in Hampsicora might at first be mis- leading, but the repetition of order throughout and of function in conscivit, adequately confirm the parallel- ism with the preceding clause ; but the et in that clause had already suggested such parallelism to follow. For the incompleteness of et in anticipatory usage is the same that it had in the retrospective. It had there the force of an adverb meaning ''moreover" or "besides." Now this had two possible uses when employed in the second sentence : it might mark either a subsequent or a coincident sentence. Its transfer- ence to the first sentence is rhetorical and in such use it is confined to the second type ; it gives warning that the sentence in which it stands is coincident with the one to follow. By its meaning it further defines the relation as one of parallelism rather than of contrast. This example will be sufficient to illustrate the force of the correlative pairs. In some cases, such as cum . . . turn, the differentiation of relative from demon- strative led to the development of a subordinating conjunction, and to a distinct type of sentence rela- tion. In others, such as alius . . . alius, the source of the incompleteness will be found to be rather differ- ent and the resulting type of relation different, too. But these differences are minor and result from the meanings of the words themselves. The fundamental principle is the same. The anticipatory use of the demonstrative has been 146 LATIN SENTENCE CONNECTION rather fully discussed already (cf. pp. 19 and 33). This is the only type of anticipation which Cato uses at all extensively and this fact alone distinguishes it as the simplest form. Cato, R.R. LXXXVIII.2: Id signi erit: menam aridam vel ovum demittito: si nata- hit, ea muries erit. Cato, R.R. XXIV.l : Vinum Grae- cum hoc modo fieri oportet. Uvas apicias percoctas bene legito. Cato, R.R. V.7: Nam res rustica sic est, si unam rem sero feceris, omnia opera sero fades. Caesar makes some use of the anticipatory demon- strative, Cicero more and Tacitus uses it very freely indeed. One more illustration will suffice. Tac. Ann. VI.50.6 : Illic eum, adpropinquare supremis tali modo compertum. Erat medicus arte insignis, nomine Charicles, etc. There is a sharp distinction to be noted here between the nature of the incompleteness to be found in the demonstrative and that in the conjunction. It is the same difference that marked their retrospective uses. The demonstrative is an empty word representing some concept, either simple, a single person or thing, or complex, a whole clause or several clauses. As regularly used, therefore, it simply stands for that con- cept and when employed in an anticipatory usage, it can imply only the verbal statement of that concept to give it content. The statement of that concept will therefore be, syntactically speaking, in apposition with the demonstrative. The conjunction, on the other hand, modifies the whole phrase in which it stands and renders it incomplete, but it does not stand for any concept outside its clause to which it points. It is ANTICIPATORY INCOMPLETENESS 147 therefore able to suggest a greater variety of relations between its clause and the one preceding or following as the case may be. It is especially interesting that some of the conjunc- tions have developed such specialized uses that they can no longer stand in a syntactically independent clause ; in other words, that in the development of the language, their incompleteness became so emphasized that their clauses became syntactically dependent as well as logically incomplete. Also that this same pro- cess was at work, as has been seen (Chap. Ill, p. 34), in the field of pronoun usage, differentiating the rela- tive from the demonstrative. In this field it was never completed. Quae statim fecerunt is essentially the same as Jiaec statim fecerunt, save as we read into the quae a certain notion of dependence by analogy with its more specialized uses. So long as its antecedent is outside itself, it is practically the equivalent of the demonstrative. But it developed a use in which it com- prised its antecedent within itself and became analo- gous to the conjunctions which we call subordinating. Qui moriuntur mortales sunt, shows the same intensi- fied anticipation that si moriuntur mortales sunt, would show. Now the same development is to be seen in the use of comparative adjectives when they appear in the anticipatory usage. Fortior erat Caesar, is syntacti- cally complete. Logically, it might be completed by the preceding sentence, or if there were none, it could conceivably anticipate some such contrast as sapientior erat Cicero. But the ordinary contrast implied is 148 LATIN SENTENCE CONNECTION minus fortis erat Cicero, and another construction more convenient and more common by far when once developed, grew up to express this relation, the con- struction with the adverb quam: Fortior erat Caesar quam Cicero (sc. erat). The principle is unchanged: the incompleteness of the comparative causes suspen- sion of judgment and in this case at least two possibili- ties are instinctively borne in mind until the following clause determines between them. It is the functional repetition which really defines the relation. The quam is the remnant of a correlative pair. The comparative adverbs and verbs are similar to comparative adjectives in their anticipatory uses. With the other verbs of incomplete meaning the case is a little different. The meaning of the verb plays the important part. Sequor could scarcely have an anticipatory force nor respondeo, in so far as the force of the re- is considered, and the same is at least par- tially true of the verbs compounded with prepositions. It is conceivable that, for the rhetorical effect of sus- pense, a writer might use such an expression as this: Circumsteterunt milites: imperator in medio erat. But, so far as any ordinary narrative is concerned, such verbs are essentially retrospective. On the other hand, there are verbs which apparently are essentially anticipatory in their incompleteness, again because of their meaning. But caution is neces- sary in these cases. Volo, dico, puto, and the like, seem at first glance to come under such a category. Examination, however, shows that they do not. As a matter of fact, the incompleteness of these verbs is ANTICIPATORY INCOMPLETENESS 149 not different from that of any verb which cannot be used absolutely. Frango, pello, or video have the same degree of incompleteness, but it is not such in any of these instances, whether of verbs of saying or of action, as to give incompleteness to the whole clause. It does anticipate an object, that is, a further word or words in the same clause in a certain relation to itself. But beyond that it does not go. In the place of the object noun may be an infinitive, but the infinitive is nothing more than the name of the verbal action and, in reality, a noun. Once more it is the meaning of the word which is the determining factor: the verbs of physical action must have a concrete object, those of mental action may be followed instead by the infinitive clause, an intangible object. Video, with both a literal, physical meaning and one figurative and mental, illustrates this distinc- tion well. And to this extent there is a difference in the incompleteness of the two sorts of verb. For the infinitive clause is subject to so much expansion and variation as to be scarcely recognizable as the object of the verb. Standing for direct narrative, it is often carried to great length and has within itself all the types of sentence connection studied. In all these verbs, then, of saying or thinking, the nature of the verb makes possible several construc- tions to follow, always with the rather remote possi- bility that the verb may be used absolutely. The re- maining possibilities are all alike in their fundamental characteristic: whether there follows direct quota- tion or indirect, or a noun in the accusative, the rela- 150 LATIN SENTENCE CONNECTION tion to the verb of saying or thinking is really the same. That their meaning gives these verbs a some- what greater relativity than that of ordinary transi- tive verbs is shown by the rhetorical development of this particular type of expression. In the place of a simple verb of saying there was used in Latin the widest range of phrases implying the notion of saying, which, on becoming familiar to the reader, suggest with the same force that dico shows, the object clause to follow. And when these phrases are employed, there must always follow a phrase or series of phrases, for they cannot, like the simple verb, take after them a simple noun object. They would all be taken as abso- lute statements without any relativity, the connection expressed solely by change of mode, if experience did not show them to belong to the dico class. Loquor and respondeo are like dico in being used both intransitively and transitively but loquor espe- cially is primarily intransitive and gains the antici- patory force by analogy. Caesar, Bell. Civ. 1.84.3: Audiente utroque exercitu loquitur Afranius: non esse aut ipsis, etc. It is the change of mode which finally makes sure the relation. Perhaps the simplest analo- gous phrase is verba facere: Sallust, Bell. Jug. 83.2: Ad ea rex satis placide verba facit: sese pacem cupere, etc. The folloAving examples will illustrate the exten- sion of the usage. Caesar, Bell. Civ. 1.24.5: Quem Caesar ad eum remittit cum mandatis: quoniam . . . facultas conloquendi non fuerit . . . interesse rei pub- licae, etc. Sallust, Bell. Jug., 88.5 : Nam Bocchus nun- tios ad eum saepe miserat: velle populi Romani amici- ANTICIPATORY INCOMPLETENESS 151 tiam, etc. Caesar, Bell. Civ. 1.20.5 : legatosque ex suo numero ad Caesar em mittunt: sese paratos esse portas aperire, etc. Tac. Ann. 1.48.3: praemittit litteras ad Caecinam, venire se valida manu. Tac. Rist. III.81.7 : Obviae fuere et virgines Vestales cum epistulis Vitellii ad Antonium scriptis: eximi supremo certamini unum dieyn postulahat. This last instance shows that the anticipation has a force which is felt even when a different construction follows it. Two classes of incomplete verbs should be noted in passing, both of them anticipatory in force but both of them quite obvious in view of the preceding dis- cussion. The first comprises the impersonal verbs like licet, oportet, necesse est, which are incomplete in meaning and which, regardless of what particular con- struction follows, clearly call attention to the fact that their clause is incomplete and related to what follows. The other class is much like this, namely, the personal verbs whose incompleteness of meaning requires some further verbal notion to make sense, the verbs regu- larly followed by a complementary infinitive, such as pergo, cesso, or desino. The difference between these verbs and the pello class is that these are intransitive ; their meaning prevents us from looking for a tangible object. The incompleteness is therefore different in the tone of its suggestion rather than in kind. Be- hind all of these classes of verbs is the same principle, that of semantic incompleteness. In turning to a consideration of the anticipatory use of the adjectives alter, alius, ceterus, and the like, 152 LATIN SENTENCE CONNECTION we reach a much more important field of anticipatory- force. It is primarily ceterus which exhibits this usage. Par and similis seem never to have acquired anticipatory force; alter and alius, used in the first sentence of a pair, are much like the correlative demon- stratives and will be considered briefly. But ceterus and the less frequently used reliquus have a meaning which singles them out for a rhetorical use of great importance. Alter and alius are essentially comparative in mean- ing. The notion of ''another" or ''the other" implies someone or something discussed in another clause. In simple narrative their use would therefore be con- fined to retrospective reference. But like the demon- stratives, perhaps by analogy with them, these words came to be used in a correlative way. If the items under discussion are clear to the reader, then the use of alter or alius for one of them or for one group of them, indicates a differentiation to be further made clear by a second alter or alius in the following sen- tence or clause. Like the anticipatory et, the anticipa- tory alter or alius becomes a warning of functional repetition to follow. Whether the exact relation of clauses will be one of parallelism or of contrast is not determined without further means. The usage is nar- row and does not lead to any very significant types of sentence connection. When, however, we speak, not of "others" but of "the others," or of "the rest," notions expressed by ceterus, we exclude a certain number of whatever ob- jects are under discussion. That number is limited ANTICIPATORY INCOMPLETENESS 153 either by what has already been said or by what is immediately to be said ; the very word ceterus implies always a definite part of some whole excluded from the ceterus clause. For example, Tac. Ann. XV.50.8: Natalis particeps ad omne secretum Pisoni erat, ceteris spes ex novis rebus petehatur. The ceteris refers to all of the supporters of the Pisonian conspir- acy except Natalis and Senecio (who was excluded by the previous sentence). So in Agricola 34.8, Tacitus makes Agricola say: sic acerrimi Britannorum iam pridem ceciderunt, reliquus est numerus ignavorum et metuentium. The dead Britons are excluded; all the rest are cowards. It is the force of the reliquus, which includes all the Britons except the specifically excluded, that gives to the appeal its rhetorical effect. Now turn this around. Suppose there occurs the sentence, Ceteri sine mora veniunt (Sallust, Cat. 46.4), in a context in which nothing has appeared to explain the ceteri. As yet there has been no exception defined from which ''the rest" are isolated. The effect is some- thing more than mere incompleteness. The fact that ceterus by its meaning isolates one exception or one group of exceptions, serves two purposes: it puts an emphasis on the exception and it suggests that the action or quality ascribed to the exception is distinctly different from the action or quality ascribed to the rest of the group from which it is taken. The conclu- sion, then, to the sentence quoted is not unexpected: Caeparius, paulo ante domo egressus, cognito indicia ex urbe profugerat. This is rhetorical, but it was a familiar and frequently employed usage and the prin- 154 LATIN SENTENCE CONNECTION ciple involved almost lost any rhetorical earmarks before classical times. A few typical cases of this usage with ceterus are the following. Seneca, De Brev. Vit. 7.9: De cetero fors fortuna, ut volet, ordinet: vita iam in tuto est. Here the cetero goes so far as to include everything existent with the exception of vita. The meaning of alius is sometimes extended and made the same as that of ceterus to serve in this usage : Tac. Ann. 11.38. 24: Egere alii grates: siluit Hortalus, etc. In Cicero, Epist. I.l.l, the principle of repetition is made use of in the second clause and plays its part in expressing the sentence relation, but the force of ceterus in antici- pating the relation is undiminished. Ego omni officio ac potius pietate erga te ceteris satis facio omnibus, mihi ipsi numquam satis facio. Cicero, Ad Att. 11.18. 2: Non dubitant iurare ceteri; Laterensis existimatur laute fecisse, quod trihunatum. pi. petere destitit, ne iuraret. Cicero, Ad Att. III.15.2: Nam ceteri dolores mitigantur vetustate, hie non potest non . . . cotidie augeri. In Tac. Hist. III.10.15, the exception is marked as such by the use of unus, "alone," but again this proves to be only a supplementary expression of relation ; fre- mitu et clamore ceteros aspernantur. Uni Antonio apertae militum aures. Much the same is the rein- forcement in Cicero, Pro P. Sulla 25.71 : Omitto ceteros, ne sit infinitum; tantum a vohis peto, ut taciti de omni- bus . . . cogitetis. Most frequently a conjunction introducing the second sentence is the means of mak- ing the relation absolutely clear. The conjunction is ANTICIPATORY INCOMPLETENESS 155 purely supplementary, however. Sallust, Cat. 52.1: Postquam Caesar dicundi finem fecit, ceteri verho aliis alii varie adsentiehantur. At M. Porcius Cato rogatus sententiam huiusce modi orationem habuit. Cicero, De Orat. 1.9.35 : Cetera, inquit, adsentior Crasso . . . ; sed ilia duo, Crasse, vereor ut tihi possum con- cedere. Cicero, De Orat. 1.3.12 : quia ceterarum artium studia fere reconditis atque ahditis e fontibus hauri- untur, dicendi autem omnis ratio in medio posita com- muni quodam in usu . . . versatur. Quite different, on the other hand, is the reinforce- ment in Tac. Ann. XIV.32.15, and it throws some addi- tional light on the principle underlying the usage: Et cetera quidem impetu direpta aut incensa sunt: tem- plum, in quo se miles conglohaverat, hiduo obsessum expurgatumque. In this instance the device used to supplement ceterus is in the same clause with it ; it is itself anticipatory. It lies in the word quidem. Qui- dem alone is practically empty of meaning. A study of its function throughout many instances indicates that it brings into relief some word in its clause, most often the word which it follows, giving it a marked emphasis. Sometimes it seems to emphasize the clause rather than a specific word, but regularly it emphasizes the word. Now when there is nothing preceding the clause in which the quidem stands that can account for this emphasis, the instinctive conclu- sion is that the word emphasized is being impressed on the mind to bring it into some clear-cut relation with something in the following clause. It is practi- cally the same effect which ceterus has by itself by 156 LATIN SENTENCE CONNECTION virtue of its meaning. And just as with ceterus, the suggestive emphasis produced by the use of quidem seems regularly to be the warning of a contrast to the word emphasized. This is borne out by the very frequent use of a con- junction in the sentence following the one with the quidem, always an adversative conjunction. Mark the contrast in Livy XXI.29.7, between timehat and magis metuehat: MuUitudo timehat quidem hostem . . . , sed magis iter immensum Alpesque . . . metue- hat. The contrast would not be obvious without qui- dem and the conjunction, for the repetition of function is not very marked and might be overshadowed by the semantic repetition. The two cases following are simi- lar, except for the fact that the contrast is more ob- vious without the mechanical indicators. Quintilian, Inst. Orat. I. Proem. 15 : Ac veterum quidem sapientiae professorum multos et honesta praecepisse et, ut prae- ceperint, etiam vixisse, facile concesserim; nostris vero temporihus suh hoc nomine maxima in plerisque vitia latuerunt. Cicero, In Q. Caec. 5.18: hanc Jiahent arcem minus aliquanto nunc quidem munitam quam antea, verum tamen, si qua reliqua spes est, quae socio- rum animos consolari possit, ea tota in Jiac lege posita est. Such examples are innumerable. An understanding of the force imparted by the quidem may be gained by a glance at two quotations from Cicero. In the Tusculan Disputations 1.33.81, he says : Vellem adesse posset Panaetius. This is an unfulfilled wish and it is conceivable that there should be nothing following. ANTICIPATORY INCOMPLETENESS 157 Cicero might be merely stating a wish with no ulterior thought. In other words, there is no anticipatory sign of relation in the clause ; if another clause is to follow, it will follow without prejudice. As a matter of fact, what does follow is an explanation of what Cicero would do if Panaetius were present: quaererem ex eo, cuius suorum similis fuisset Africani fratris nepos. In the retrospect the first clause comes to have the same effect as a conditional clause, but it did not have it by itself. Now compare another unfulfilled wish in the speech against Caecilius, 12.40: Utinam quidem essent! At once the quidem gives a different tone to this wish. The utinam, the sign of a wish, is empha- sized significantly; warning is given of a comparison, probably a contrast, to follow, and the first words of the next clause make this certain : verum tamen ut esse possent, magno studio mihi a pueritia est elahoratum. With such evidence of the force of quidem, the com- paratively few instances of its unsupported use are clear. For example, Livy XXXL36.3: Et equitatus quidem cessit, duces caetratae cohortis non satis ex- pectato signo ante tempus excitatis suis occasionem bene gerendae rei amisere. Tac. Germ. 6.5: Et eques quidem scuto frameaque contentus est; pedites et mis- silia spargunt. Livy 1.57.11: Et tum quidem ah noc- turno iuvenali ludo in castra redeunt. Faucis interiec- tis diehus Sex. Tarquinius . . . Collatiam venit. Tac. Ann. XV. 44.1: Et haec quidem humanis consiliis pro- videbantur. Mox petita dis piacula, etc. Tac. Agr. 17.7 : Et Cerialis quidem alterius successoris curam famamque obruisset; subiit sustinuitque molem lulius 158 LATIN SENTENCE CONNECTION Frontinus. In this last example, Tacitus shows his fondness for unusual arrangement by putting the qui- dem before rather than after the alterius. In a later chapter of the Agricola (38.1), his order is even more unusual, but even so, the force of the quidem is clear. Et nox quidem gaudio praedaque laeta victoribus: Britanni palantes mixtoque virorum mulierumque ploratu, trahere vulneratos, etc. Nox is almost the only word not contrasted with something! Sane, although not so empty a word, has much the same force as quidem. As an adverb it has an inde- pendent meaning and in this meaning is probably to be found the origin of the anticipatory force of this class of words. For it is an intensive adverb and therefore adds emphasis to the word or phrase with which it is used, and when that emphasis finds no ex- planation in the past context, it becomes suggestive of a relation to be developed, regularly that of con- trast, often contrast between what is true and what is not true, or vice versa. An instance from Plautus throws light on this development: Pseud. 662: Sane sapis et consilium placet. Sed vide sis ne in quaestione sis quando accersam mihi. The sane is on the line between its simple meaning of *'very" and its use as a mere indicator of relation. Most of the cases of sane used with anticipatory force are supplemented by an adversative conjunction in the following clause, but that this is not altogether necessary will be seen from the last two of the follow- ing examples. Cicero, In Verrem 11.53.132: Nihil sane vafre nee malitiose facere conatus est; sed ut ANTICIPATORY INCOMPLETENESS 159 studia cupiditatesque honorum . . . tollerentur, . . . ostendit sese in omnibus civitatihus censores esse facturum. Cicero, Acad. 1.7.25 : Bene sane facis; sed enitar, ut Latine loquar, etc. Sallust, Bell. Jug. 42.2: Et sane Gracchis cupidine victoriae hand satis modera- tus animus fuit: sed bono vinci satius est quam malo more iniuriam vincere. Tac. Ann. XIV.44.5 : Sane consilium occuUavit, telum inter ignaros paravit: num excubias transire . . . caedem patrare poterat omnibus nesciis? Cicero, De Fin. 11.26.82 : Sed haec nihil sane ad rem; ilia videamus, quae a te de amicitia dicta sunt. With these cases are to be grouped the following with other emphasizing expressions. Quint. Inst. Orat. 1.6.12 : Quaedam sine dubio conantur eruditi defendere . . . Illi autem iidem . . . etc. Quint. Inst. Orat. IL5.21: quae turn sine dubio erat optima, sed nostris temporibus aliena est. Sallust, Bell. Jug. 31.5: Certe ego libertatem . . . experiar: verum id frustra an ob rem faciam, in vostra manu situm est, Quirites. Finally, an instance of valde is a good borderline case to indicate the origin of the usage: Cicero, Ad Att. 1.13.5 : Quae laudas ex orationibus, mihi crede, valde mihi placebant, sed non audebam antea dicere. In Quintilian, 1.6.22, a phrase is used to give the desired emphasis: Recta est haec via, quis negatf Sed adiacet et mollior et magis trita. In such a sane example as that from the Verres speech, there is a further element of anticipation to be noted that will be given more study later on. The statement in which the sane stands is the statement which the speaker does not really believe, as he has 160 LATIN SENTENCE CONNECTION already shown. The statement has therefore a hypo- thetical tone. But this can be better estimated in con- nection with the imperative uses to be discussed later (p. 173). There are still left several groups of cases in which suggestive emphasis is effected in the first sentence in such a significant way as to indicate some contrast to follow. The most frequently used means to accomplish such anticipation is the use of a negative statement in the first sentence of a pair, a negative statement that does not deny or contradict any other statement already made or any idea naturally present in the mind of the speaker because of what has preceded. Its purpose is not immediately clear. By itself it is superfluous and usually irrelevant. Since nothing already stated can explain its meaning, the judgment is held in suspense until the following sentence clears up the doubt. This following statement is regularly positive, declaring what is true in contrast to what has been stated as not true. The purpose of the first sentence is then clear : it is a rhetorical device for emphasizing by contrast the positive statement of the second. Cicero, Pro Rose. Am. 27.73 : Non quaero ahs te, quare patrem Sex. Roseius oeeiderit, quaero, quo modo oeeiderit. The first sentence, the statement that Cicero does not ask a certain question, is inconclusive by itself. He has already stated that he will drop the question of motive and so, as he takes up a new sub- ject, this negative statement is quite superfluous, until the positive sentence gives it point. It is a rhetorical ANTICIPATORY INCOMPLETENESS 161 use of the principle of incompleteness. The exact repetition of the quaero serves to give a parallelism to the clauses that makes the contrast more clear, but the anticipation in the non quaero had already fore- shadowed it. Various supplementary means are used to make the contrast more obvious. Virtual repetition appears in Sallust, Bell. Jug. 31.6 : Nihil vi, nihil seces- sione opus est: necesse est suomet ipsi more praeci- pites emit. In Seneca, De Vita Beata 2.2, it is the force of a comparative: oculis de homine non credo, haheo melius et certius lumen quo a falsis vera diiudicem: animi honum animus inveniat. Most frequently of all an adversative conjunction supplements the force of the anticipatory device. For example. Quint. Inst. Orat. 1.2.14: Non enim vox ilia praeceptoris ut coena minus pluribus sufficit, sed ut sol universis idem lucis calorisque largitur. So in Caesar, Bell. Civ. 1.41.4, in which the verb veto with its negative notion takes the place of an actual negative: vallo muniri vetuit, quod eminere et procul videri necesse erat, sed a fronte contra hostem pedum XV fossam fieri iussit. These supplementary means of expression are not necessary, as can be seen from the following instances. Cicero, Brutus 81.280: industriam non sum expertus; studium certe fuit. Sallust, Cat. 52.6: Non agitur de vectigalihus neque de sociorum iniuriis: lihertas et anima nostra in duhio est. Seneca, De Provid. 6.4: non est ista solida et sincera felicitas: crusta est et quidem tenuis. Tac, Hist. 1.76.6: Nusquam fides aut amor: metu ac necessitate hue illuc mutahantur. There is one variation of this particular application 162 LATIN SENTENCE CONNECTION of the principle of anticipatory incompleteness which is apt to lead to confusion. It is illustrated by an instance from Tacitus {Ann. XIV.25.1) : At praesidium Legerda . . . non sine certamine expugnatum est: nam et proelium pro muris ausi erant et pulsi intra muni- menta aggeri demum et inrumpentium armis cesser e. The use of the conjunction nam suggests that the rela- tion here is not one of contrast but that the second sen- tence explains the first. And so it does. But there is also perfectly clear contrast between what was true and what was not true. The difference here is that an element of explanation is added and this is marked by the nam. The contrast is still present, prepared for by the suggestive use of the negative. The explana- tion too is prepared for by the suggestive generality of the first statement, but this will be more obvious after that type of connection has been studied (pp. 165 ff.). For the present, the element of contrast is the noteworthy thing and especially its anticipation in the first clause. In the following examples will be seen the same mix- ture of elements without any conjunction to serve as guide. Caesar, Bell. Civ. 1.75.2: Petreius vero non deserit sese. Armat familiam; etc. Sallust, Bell. Jug. 51.5: Sed ne Jugurtha quidem interea quietus erat: circumire, hortari, renovare proelium, et ipse cum delectis temptare omnia. Tac. Hist. III.47.1: Nee ceterae nationes silebant. Suhita per Pontum arma harharum mancipium, regiae quondam classis praefec- tus, moverat. The common expression non modo . . . sed etiam is ANTICIPATORY INCOMPLETENESS 163 based on this same principle, as is doubtless clear with- out illustration. There is an interesting extension of this use of an irrelevant negative to anticipate a contrast, which be- comes a very frequent rhetorical device. That is the use of a condition contrary to fact. By its very char- acter such a condition gives to the sentence in which it stands an incompleteness of tone. For, like the negative sentence which has no bearing on what has preceded, the condition contrary to fact is irrelevant or at least inconclusive without reference to something outside itself. If, then, such satisfactory fulfilment has not preceded, the anticipatory effect of the irrele- vant negative clause is reproduced. For example, Seneca, De Ira 11.33.6: Contempsisses Romanum pat- rem, si sibi timuisset; nunc iram compescuit pietas. Seneca, Ad Helviam 18.9: Numerarem inter magna solacia patrem quoque tuum nisi ahesset; nunc tamen ex adfectu tuo, qui illius in te sit cogita. Cicero, Pro Quinctio 1.2 : Neque hoc tanto opere querendum videre- tur, haec summa in illis esse, si in nobis essent saltern mediocria: verum ita se res habet ut ego, etc. In each of these three examples a word is used in the second clause to make the relation absolutely ob- vious, either nunc or an adversative conjunction or both. Nunc is probably the most frequently used; it is without temporal significance and is employed sim- ply to mark the contrast of a statement of fact with the unreal condition, like vvv Se in Greek. But neither of these obvious signs is absolutely necessary. The anticipation is always there and the contrast is very 164 LATIN SENTENCE CONNECTION often left to speak for itself. A few instances are enough for illustration. Cicero, Pro Quinctio 14.47: nihil hoc tanto negotio, nihil tarn invidioso iudicio, nihil tarn copiosa advocatione uterer, si petendum esset; extorquendum est invito atque ingratiis. Livy XXI.43.6: Si Sicilimn tantum ac Sardiniam parentihus nostris ereptas nostra virtute recuperaturi essemus, satis tamen ampla pretia essent: quidquid Romani tot triumphis partum congestumque possident, id omne vestrum cum ipsis dominis futurum est. Sallust, Cat. 18.8: Quodni Catilina maturasset pro curia signum sociis dare, eo die post conditam urhem Romam pessu- mum f acinus patratum foret. Quia nondum frequentes armati convenerant, ea res consilium diremit. There are other ways of anticipating contrast, but the general principle has been amply illustrated and may be stated as follows: Any method of markedly emphasizing a word or phrase or clause, when such emphasis has no bearing on the context already dis- closed, serves to indicate probable contrast to follow. It will be sufficient to call attention to two familiar forms through which the principle works. One con- sists of putting the word to be emphasized in an un- usual or prominent position, as in the case of the illis in Sallust, Cat. 51.26: Illis merito accidet quicquid evenerit: ceterum vos, patres conscripti, quid in alios statuatis, considerate. The other type is based on the singling out of some specific detail and so dramng attention to it, without apparent reason, until the sec- ond clause is reached. This is illustrated by Tac. Ann. XIV.57.10: cui caveri utcumque ah urhanis insidiis ANTICIPATORY INCOMPLETENESS 165 praesenti opera; longinquos motus quonam modo com- primi posset Wliy urbanis insidiis should be singled out from the general danger is not clear until the con- trast with longinquos motus is brought out. All of these methods of giving suggestive empha- sis to some word or phrase are methods of anticipating a contrast, an adversative relation, some of them natural, some of them highly rhetorical. Another type of rhetorical usage anticipates a different sentence relation. It is widely used and based like the other anticipatory uses on the element of incompleteness. In the present group that incompleteness lies in a word or phrase so general in meaning or so vague as to be practically valueless in the narrative without some explanatory statement or statements following it to give it point. Seneca, De Brev. Vit. 12.9, reads : Non est ergo hie otiosus, aliud illi nomen imponas. The first clause of course acquires its anticipatory force from the inconclusive negative sentence. But after the posi- tive statement has been made and the anticipated con- trast thus completed, there is still a sense of incom- pleteness about the sentence : aliud illi nomen imponas is quite inconclusive and the mind is left in suspense until the rest of the sentence gives a specific explana- tion which satisfies the anticipation: aeger est, immo mortuus est. An instance not complicated by the presence of the negative clause is Tac. Ann. III.31.7 : Ac forte parva res magnum ad certamen progressa praehuit iuveni materiam apiscendi favoris. Parva res, magnum cer- 166 LATIN SENTENCE CONNECTION tamen are intentionally vague and inconclusive, parva res in particular adding nothing specific to the narra- tive and serving simply to hold the attention until the explanation follows : Domitius Corbulo praetura func- tus de L. Sulla nohili iuvene questus est apud senatum. In Cicero, Ad Att. 1.1.3, the vague statement that there is something that Cicero wishes forgiven, has in its setting no conclusive or satisfactory meaning. A context can be imagined in which this statement, sed est quod abs te ignosci pervelim, might be logically complete. Suppose some action of his had been under discussion and he had said that he realized it to be something which it would be hard for Atticus to over- look, and had continued, sed est, quod abs te ignosci pervelim. There would have been no anticipatory vagueness at all. But as it stands, with nothing defi- nite behind it, the vague est quod requires an explana- tion which follows: Caecilius, avunculus tuus, a P. Vario cum magna pecunia frauderetur, agere coepit, etc. Not infrequently the vague word res is employed in this way. Seneca, Ad Polyb. 11.3: Deinde adiecit rem maioris et prudentiae et animi: ''Et huic rei sus- tuli." Or again, Seneca, De Provid. 1.1: faciam rem non difficilem, causam deorum agam. Frequently such a statement as Interea Romae multa simul moliri (Sallust, Cat. 27.2) is quite unsatisfac- tory in its context; the statement that many things were being planned merely leads the reader to look forward to a specification of what at least some of them were. And in this he is not disappointed : insi- dias tendere, parare incendia, opportuna loca armatis ANTICIPATORY INCOMPLETENESS 167 hominibus obsidere. Ipse cum telo esse, etc. Such an instance is Quint. Inst. Orat. 1.11.9: Nam frons pluri- hus generibus peccat. Vidi multos, quorum supercilia ad singulos vocis conatus allevarentur, aliorum con- stricta, aliorum etiam dissidentia, cum alterum in ver- ticem tenderent, altera paene oculus ipse premeretur. Again, the word which gives the tone of inconclu- siveness may be a word that specifically indicates an analysis to follow, as in the following examples. Sal- lust, Bell. Jug. 25.6: primo conmotus metu atque luhi- dine divorsus agitahatur: timebat iram senatus, ni paruisset legatis; porro animus cupidine caecus ad inceptum scelus rapiebatur. Tac. Ann. 1.80.4: Causae variae traduntur: alii taedio novae curae semel placita pro aeternis servavisse, quidam invidia, ne plures fruerentur ; sunt qui existiment, ut callidum eius inge- nium, ita anxium iudicium. Caesar, Bell. Civ. 1.39.4: Quo facto duas res consecutus est, quod pignore animos centurionum devinxit et largitione militum voluntates redemit. Cicero, Epist. 1.2.1 : postulatum est, ut Bibuli sententia divideretur. Quatenus de religione dicebat . . . Bibulo adsensum est; de tribus legatis frequentes ierunt in alia omnia. The vague statement is sometimes not especially vague or general in itself and contains no such half empty word as res or diversus, and may yet have the inconclusive effect of a vague statement because it is quite irrelevant until explained by a subsequent state- ment. Such is the statement in Tac. Ann. 1.29.12: Promptum ad asperiora ingenium Druso erat. In a description of Drusus' character this would not be so 168 LATIN SENTENCE CONNECTION general or vague as to suggest specifications to follow. But it does not occur in any such description; it is thrown into the narrative quite unexpectedly. The explanation is to be found in the rest of the sentence : vocatos Vibulenum et Percennium interfici iuhet. Com- pare Cicero, In Q. Caec. 17.55 : Cognosite ex me. There is nothing very strikingly unexpected in this injunction, but there is no apparent need of the me: cognoscite would seem to express the appeal sufficiently well. That the me was intentionally used is shown by the following clause explaining it: nam iste eam profecto, nisi plane nihil sapit, numquam proferet. And inas- much as the contrast between me and iste might sug- gest too significant a change, a supplementary nam is used. In general, this usage is found very frequently in the more rhetorical writers. Seneca especially abounds in it. It is not common in the more simple authors, although such examples as the following are not infre- quent in Caesar: Bell. Civ. 1.73.1: Postero die duces . . . de reliquis rebus consuUabant. Erat unum iter, Ilerdam si reverti vellent, alterum, si Tarraconem peterent. All of these types of vague or irrelevent sentences serve, by means of suspense, to make the sentence that follows emphatic. In this respect they are like the negative sentences already illustrated. But they antici- pate a different type of sentence to follow, an explan- atory sentence. This conclusion is borne out by a study of the conjunctions used to supplement the expression of relation. These are regularly nam and ANTICIPATORY INCOMPLETENESS 169 enim, which, from their original asseverative notion, came to be the regular indicators of an explanation. It has already been shown (Chap. Ill, p. 58) that, with repetition of content, they regularly mark an explana- tory sentence; also with such incomplete words as ceterus. Here, again, they serve the same function, but here, too, they are not primarily means of con- nection. Two instances from Tacitus will show the supple- mentary nature of the conjunction. In Hist. V.4.2, there is no conjunction : Prof ana illic omnia quae apud nos sacra, rursum concessa apud illos quae nobis incesta. The inconclusive vagueness of this sentence is merely to emphasize by suspense the details that follow through the paragraph, beginning, Effigiem ani- malis, etc. In paragraph 5, there is a similar vague- ness used with the same purpose : Hi ritus quoquo modo inducti antiquitate defenduntur: cetera instituta, sinis- tra foeda, pravitate valuere. But this time the ex- planatory details which fill the paragraph, begin. Nam pessimus quisque. The following instances with conjunctions will bear out the indications of this passage from Tacitus. Caesar, Bell. Civ. 1.51.6 : Hoc pugnae tempus magnum attulit nostris ad salutem momentum; nacti enim spa- tium se in loca superiora receperunt. Caesar, Bell. Civ. 1.48.1 : Accidit etiam repentinum incommodum biduo, quo haec gesta sunt. Tanta enim tempestas cooritur, ut numquam illis locis maiores aquas fuisse constaret. Sallust, Cat. 46.2: At ilium ingens cura atque laetitia simul occupavere: nam laetahatur intel- 170 LATIN SENTENCE CONNECTION legens coniuratione patefacta civitatem periculis erep- tam esse, etc. The repetition of laetitia in laetahatur shows the force of retrospective connection, but the anticipatory is obvious too and the nam supplements both. Suetonius, Caligula 19 : Novum praeterea atque inauditum genus spectaculi excogitavit. Nam Baia- rum medium intervallum ad Puteolanas moles . . . ponte coniunxit, etc. The conjunction has the same function of supplementing a relation adequately ex- pressed by other means, when it is used after a clause containing a word that implies analysis, as in Nepos, Pausanias 1.1 : Pausanias Lacedaemonius magnus homo, sed varius in omni genere vitae fuit: nam ut virtutihus eluxit, sic vitiis est obrutus. That the rela- tion between the clauses does not depend for expres- sion on the conjunction, is finally shown by the use of que in a similar instance, the que really performing no further function than to mark the subsequence of the second clause without defining its relation to the first : Tac. Hist. 11.57.11 : dein mohilitate ingenii, quod palam ahnuerat, inter secreta convivii largitur honoravitque Asiaticum anulis. No doubt, more variations of this principle of antici- pation by means of semantic incompleteness could be found, but those already discussed are the most com- monly used and serve sufficiently well to make the principle evident. There remain the forms of func- tional incompleteness that lead to anticipation. Most of these require little discussion. For in most types the principle is identical with that of the retrospective ANTICIPATORY INCOMPLETENESS 171 adaptations of functional incompleteness except that nothing already disclosed in the context satisfies the incompleteness, so that the judgment must wait for satisfaction until the following sentence or clause is developed. This fact was illustrated in passing in the section on retrospective incompleteness (e.g., Chap. IV, p. 106) and needs but brief illustration here. With the infinitive, the effect is obvious, for the infinitive is the name of an action and as a noun is normally de- pendent on some verb to complete its meaning in the sentence ; but, at the same time, this substantival char- acter of the infinitive somewhat removes it from the discussion. The numerous sentences, however, in Taci- tus ' Germania beginning with an infinitive surely illus- trate the principle. For example, 21.1: Suscipere tarn inimicitias seu patris seu propinqui quam amici- tias, is beyond question incomplete. The only alterna- tive possibility, that the verbs might be narrative, is eliminated by the nature of the context and by the lack of any natural subject for the verbs. The necesse est that follows comes as no surprise. The same is true of all syntactically subordinate clauses when they precede their **main" verb, cum clauses, ut clauses, and so on. But these are too ob- vious to require illustration. One thing, however, should be noted. Whether we have a conjunctional clause or one which the manuals call a substitute for a conjunctional clause, the logical incompleteness is the same and therein lies the fundamental expression of relation. The conjunctional uses were developed for 172 LATIN SENTENCE CONNECTION the more obvious definition of the relation and were so constantly used that they are looked upon as regu- lar, but when other means are sufficient, it is hardly accurate to say that their use is a substitute for the conjunctional use. The anticipatory effect of the simple subjunctive has been already discussed and illustrated (Chap. IV, p. 106). In actual practice it is due to the prevailingly dependent or contingent nature of the subjunctive. Bonum haberet animum (Tac. Hist. II.46.4), has no complete meaning by itself because of the function of haberet. The following iubebant completes it. The same principle is seen in Hist. 1.39.4. In spite of the prominent alii . . . alii, implying a plural verb to come, the incompleteness of rediret and of peteret is obvious : cum alii in Palatium rediret, alii Capitolium peteret, plerique rostra occupanda censerent, etc. Little need be said of the incomplete tenses used with anticipatory force. It is clear that if a statement that something had happened or will have happened implies that something did or will happen (cf. Chap. IV, p. 108), this implication will still be present even though nothing of the sort has been stated. In such a situation some statement of what did or what will happen is reasonably looked for to follow. The usage is not very common except that the pluperfect is some- times used with anticipatory force with the evident purpose of emphasizing the time for the purpose of establishing a contrast. In this use it is like those cases of suggestive emphasis already discussed. Tac. Ann. 1.56.18 : Fuerat animus Cheruscis iuvare Chattos, ANTICIPATORY INCOMPLETENESS 173 sed exterruit Caecina hue illuc ferens arma. Tac, Ann. XV.16.16: Decesserat certamen virtutis et ambitio gloriae, felicium hominum adfectus: sola misericordia valehat, et apud minor es magis. Further anticipatory uses of the pluperfect have been illustrated in Chapter V. The anticipatory use of the imperative mode cannot be passed over so casually. It is, to be sure, pretty generally recognized, but the explanation of it as standing for a conditional clause or a concessive, rather obscures the principle which gives to it its particular force. Unlike the infinitive and the sub- junctive, the imperative is primarily the mode of a syntactically complete sentence. The natural impulse, resulting from experience, is to look upon an impera- tive as complete in meaning also. And that is the way in which it commonly occurs in conversation and in the comedy. But it is obvious that this use of the imperative to express a direct command, is little called for in con- secutive discourse. It has its place — in drama, where actual conversation is reproduced, in the dialogue, in speeches, and finally, with verbs of mental rather than physical action, where its scope is wider. But the imperative actually appears much more frequently than such limitations would seem to permit. Again and again there occur imperatives which the context, or merely our general experience, tells us cannot be intended as actual commands to be carried out. It is 174 LATIN SENTENCE CONNECTION such imperatives which leave in the mind a sense of incompleteness, not syntactical but logical. When Cicero {Tusc. Disp. IV.24.54) says: Sic ira- cundus non semper iratus est; lacesse, it is perfectly clear that lacesse is not a command to the reader nor yet to Cicero's interlocutor to actually go and rouse up some irascible man. The imperative expresses syn- tactically a complete idea, at least it does when the object is assumed from the preceding sentence. But since it cannot be meant as a literal command, its sense is incomplete. And so it will remain so long as lacesse itself or the preceding context alone is studied. The following clause, however, completes it : iam vide- bis furentem. When Seneca, speaking of Eegulus {De Provid. 3.9) says: Refice ilium et mitte in senatum, there is obviously no conunand in the true sense, either literal or figurative. Once more the sense is incom- plete without the following clause : eandem sententiam dicet. Two things are clear from these examples: the im- perative used in consecutive discourse and not having literal imperative force, is logically incomplete, and the following clause may and usually does satisfy that incompleteness. The precise nature of the relation between the clauses can be clear only after a study of more instances and a further analysis of both clauses, the imperative clause and the clause that follows it. That there is a distinctly different type from that already illustrated can be seen from such an example as the following: Tac. Hist. IV.17.20: Servirent Suria Asiaque et suetus regihiis Oriens: multos adhuc in ANTICIPATORY INCOMPLETENESS 175 Gallia vivere ante tributa genitos. This is indirect discourse; the subjunctives represent imperatives in the direct. That Syria and the East are playing the slave is an admitted fact; Civilis is not urging it. Clearly the imperative is not literally meant ; its actual force is not clear until the second clause shows that it was permissive in a hypothetical sense. The contrasts between Gallos on the one hand and Suria, Asia, Oriens on the other, and between suetus regihus and servirent on the one hand and ante tributa genitos on the other, really define the relation. Such definition, however, belongs to the second clause and has been considered in the chapter on Change. At present the noteworthy point is the incompleteness which lies in the imperative used without literal jussive force. The result in each case is fundamentally the same : a sense of irrelevancy, of logical incompleteness, is forced upon the reader, turning his attention forward for satisfaction. There is a difference in detail between the two types but fundamentally they are the same. The development of this particular type of relation anticipated by incompleteness is not hard to indicate. In Plautus and Terence the literally jussive impera- tive is very frequently followed by a future tense indi- cating the result which will follow the performance of the command. Impera: imperium exequar. {Amph. 956.) Ausculta ergo, scies. (^5m. 350.) Oneris quid- vis impone: ecferet. {Phorm. 561.) Sequere me intus cetera audietis. (Phorm. 765.) These are familiar phrases. The result is often marked by an adverb, as in Bacchides 1023: Em specta! turn scies. 176 LATIN SENTENCE CONNECTION This turn of speech is extremely common and recurs wherever direct conversation is reproduced. For example, in the incident described in Livy III.2.9, in which the outposts of one army are taunting those of the other: Crastino die oriente sole redite in aciem; erit copia pugnandi; ne timete. Or again, Tac. Ann. XL2.5: "Interroga" inquit, ^'Suilli, filios tuos: virum esse me fatehuntur." In letters, which are one side of a written conversa- tion, this use is common. Cicero, Ad Att. II.25.2: Quare advola; aut expedies nos omni molestia aut eris particeps. Pliny, Epist. IV. 4.2 : Hunc rogo semestri trihunatu splendidiorem et sihi et avunculo suo facias. Obligahis me, obligahis Calvisium nostrum, etc. It has its place also in orations. For example, Cicero, hi Cat. 1.4.8: Recognosce tandem mecum noctem illam supe- riorem: iam intelleges multo me vigilare acrius ad salutem quam te ad perniciem rei puhlicae. All of these examples show actual commands in- tended to be taken literally even if not actually carried out. And furthermore the sense is complete with the close of the imperative clause : there is nothing to make the explanatory second clause felt as a necessity. So in English, ' ' Come unto me all ye that labour and are heavy laden," is logically a complete sentence. The addition of "and I will give you rest," adds to the meaning distinctly, but it was not required by any evident logical incompleteness in the imperative clause. When the command is one to mental rather than to physical action, the meaning of the verb usually gives to the clause a tone of lesser finality. It may be logic- ANTICIPATORY INCOMPLETENESS 177 ally complete and yet experience teaches us to find in an injunction to pay attention or to consider some- thing, the suggestion of some result to follow such consideration. This is seen in the example above from the first speech against Catiline. So also Cicero, Tusc. Disp. IV.24.53: Tracta definitiones fortitudinis : intel- leges earn stomacho non egere. Or, in poetry, Ovid, A. A. III.115 : Adspice quae nunc sunt Capitolia quae- que fuerunt: Alterius diceS' ilia fuisse Jovis. Or finally, with the fortifying adverb, Cicero, Pro Quinctio 25.79: Quod . . . a vobis . . . quaeso ut diligenter attendatis; profecto intelligetis illim ah initio cupi- ditatem pugnasse. From such examples it is no long step to instances which show distinct anticipation. The difference lies only in the meaning of the verb considered in con- nection with its context. The anticipation results from the fact that the meaning of the verb prevents us from understanding it in its particular setting as a literal command to be obeyed. The usage of an actual jussive imperative followed by a statement of the result of its assumed fulfilment was familiar and is readily adopted for these hypothetical imperatives. The imperative may be one which might, if the con- text were different, be meant as an actual command, but whose presence in consecutive discourse prevents its being so taken, so that the reader is forced to sus- pend his judgment as to its actual function in the sen- tence. Or it may, on the other hand, be a command quite impossible of carrying out. Or finally, it may be absolutely contrary to the will of the writer, so that its 178 LATIN SENTENCE CONNECTION intent is obviously not to secure compliance with what it orders. Any one of these three types produces sus- pension of judgment, anticipation of a clause of expla- nation to follow, due to the logical incompleteness of the imperative. A few examples of each will be suffi- cient. First Type. Cicero, In Verrem ILIV.25.55: Quern voles e conventu Syracusano virum bonum nominato ; producam. Seneca, De Tranq. An. 7.2: deme illis testes spectatoresque, non delectahit popina secreta. Juvenal, 1.155: Pone Tigellinum, taeda lucehis in ilia, Qua stantes ardent, qui fixo pectore fumant. Tac. Ann. 11.71.17: Ostendite populo Romano divi Augusti nep- tem eandemque coniugem meam, numerate sex liheros. Misericordia cum accusantibus erit. Second Type. Cicero, In Verrem II.I.23.61: Unum ostende in tabulis aut tuis aut patris tui emptum esse; vicisti. (Cicero's argument has already shown the impossibility of producing one such citation. The use of the perfect tense is a rhetorical device to show the immediateness of the hypothetical result.) Seneca, De Vita Beata 25.1: Pone in opulentissima me domo, pone aurmn argentumque ubi in promiscuo usu sit: non suspiciam me ob ista, quae etiam si apud me, extra me tamen sunt. In sublicium pontem me transfer et inter egentes abice: non ideo tamen me despiciam, quod in illorum numero consedero, qui manum ad stipem porrigunt. . . . Pone in instrumentis splendentibus et delicato apparatu: nihilo me feliciorem credam. . . . Muta stragula mea: nihilo miserius ero, etc. (This suggests also Horace, Odes 1.22.17 ff.) ANTICIPATORY INCOMPLETENESS 179 Thikd Type. Plautus, Rudens 1010: Tange! adfli- gam ad terram te. Seneca, De Vita Beata 20.6 : Gemite et infelicem linguam honorum exercete convicio, Mate, commordete: citius multo frangetis dentes quam im- primetis. Seneca, De Vita Beata 27.3 : adsilite, facite impetum: ferendo vos vincam. Examples might be multiplied, but the principle is evident: the use of an imperative which, from its meaning in its particular context, cannot be taken as a command to be actually fulfilled in either a literal or figurative sense, gives to its clause the logical incom- pleteness of an hypothesis, and the reader is pre- pared in advance for an explanatory completion of the thought in the following clause. This function of the imperative is made more obvious by the use of spe- cial words, modo and tantum, just as quidem and sane were found marking a different type of anticipation (cf. p. 155). It will be seen from a few examples that the fundamental principle is unchanged; only the ex- pression of it is made more clear and precise. Plautus, Amph. 286: Modo sis veni hue: invenies infortunium. Livy VI.18.7 : Ostendite modo helium; pacem hahebitis. Videant vos paratos ad vim; ius ipsi remittent. Cicero, Ad Att. III.23.4: modo res conficiatur, ero con- tentus. Tac. Ann. 11.15.10: Meminissent modo avari- tiae crudelitatis superhiae; aliud sihi reliquum quam tenere libertatem aut mori ante servitiumf Martial V.1.9: Tu tantum accipias: ego te legisse putaho. It is worth noting that, with modo to identify it, the imperative is used in the second clause as well as in the first in a hypothetical sense. Without the defining 180 LATIN SENTENCE CONNECTION word the relation would not be clear. Cicero, Ad Att. 1.6.1 : Non committam posthac, ut me accusare de epis- tularum neglegentia possis; tu modo videto, in tanto otio ut par mihi sis. Pliny, Epist. III.17.2 : Ego viati- cum, ego praemium dabo, nuntiat mihi modo quod opto. Seneca, De Ira III.42.1 : Poterimus autem, adnitamur modo. Seneca, De Vita Beata 1.4 : sanahimur, separe- mur modo a coetu. To consider the m^odo in these last three instances as standing for dummodo is to mis- understand the underlying force. The use of dum- modo originated in the reinforcement of the construc- tion with dwn by the addition of modo, in exactly the way that it is here added to this particular subjunctive usage and elsewhere to si clauses. It remains to consider briefly the other type of antici- patory imperative indicated above, the type illus- trated by the citation: Servirent Suria Asiaque et suetus regihus Oriens: multos adhuc in Gallia vivere ante tributa genitos. The key to its understanding is to be found in the fact that the imperative is incapable of construction as a command in any true sense. For the situation is one that even the speaker admits. Even hypothetically there is no order to do anything : it is simply an admission of an assumed contention. Lane (§ 1553) says: The subjunctive of desire may be used to denote wil];ingness, assumption, or conces- sion. And Kuehner (IIL47.10) also discusses the con- cessive use of the independent subjunctive. Blase (H. G. der lat. S. III.l. § 60.3) is more satisfactory: Der Imperativ dient auch dem Ausdruck des Zuge- ANTICIPATORY INCOMPLETENESS 181 staendnisses. And he goes on to point out the fact that concession is usually used with the third person which was lacking in the imperative formed on the present stem. Hence the frequency of the so-called future imperative and of the subjunctive in this usage. He calls attention also to the discovery of Woelflflin (Arch, f. 1. Lex. X.130) that licet does not occur in the Twelve Tables but that the positive use of the imperative often has a permissive rather than a mandatory tone. The key to the situation seems to lie in this last fact. There is latent in all the jussive forms a permissive tone. The situation determines whether or not it will become prominent. In the case of hypothetical im- peratives if there is no special sign, the natural impulse seems to be to expect a relation like that already studied, an hypothesis whose conclusion is to follow. But often the signs of irrelevant emphasis already examined appear with the imperative and the in- stinctive anticipation of contrast results, usually rein- forced by some element of the second clause, the result being what is called a concessive imperative. The imperative on the present stem was the predominant imperative in classical Latin and the absence of a third person necessitated the use of the subjunctive or of the less common imperative in -to. Hence the classi- fication in the grammars. The line of distinction is rarely clear between the two types of imperative when they are unsupported by special words. And this is natural, for there is no sharp cut line between the ideas which they represent, between condition and concession. For instance. Blase, 182 LATIN SENTENCE CONNECTION quoting Eberling, cites as a concessive imperative, Horace, Sat. 11.1.53: Scaevae vivacem crede nepoti Matrem: nil fiet sceleris pia dextera. I doubt very- much whether the imperative crede suggests more than a simple hypothesis. His further citation of Horace, Sat. II.7.73 is clearly wrong : Tolle periclum, lam vaga prosiliet frenis natura remotis. Obviously the impera- tive indicates an hypothesis, as is made certain by iam. His third example, Horace, Sat. II.3.69, is better but not good: adde Cicutae Nodosi tahulas centum, mille adde catenas: Effugiet tamen haec sceleratus vincula Proteus. In this instance the first clause leaves some doubt in the mind, a doubt dispelled only by the tamen in the second clause. I have found scarcely any good cases with the im- perative itself. Perhaps the best are the following in which the verb meaning suggests the concession of a point. Cicero, Tusc. Disp. 1.29.70 : Sed fac igneam, fac spirahilem; nihil ad id de quo agimus. Quint. Inst. Orat. 1.2.4: Da mentem ad peiora facilem, da negle- gentiam formandi custodiendique in aetate prima pudoris: non minor em flagitiis occasionem secreta praehuerint. It is the verb meaning which keeps similar phrases with esto from being ambiguous. For example, Cicero, Tusc. Disp. 1.43.102: Esto, fortes et duri Spartiatae; magnam hahet vim rei publicae disciplina. With the jussive subjunctive there are more in- stances, but the range is not wide; sit again plays a prominent part, and in such cases as Seneca, Ad Helviam 18.6, other means support the suggestion of ANTICIPATORY INCOMPLETENESS 183 the imperative : floreat reliqua in suo statu turha: nihil de orhitate, nihil de condicione mea querar, etc. But on the whole, the anticipation is not clear : the specific type of hypothesis is not sure until the second clause makes it so. This is illustrated by Petronius 61: viderint: narraho tamen. It is for this reason that, first, nearly all the hypothetical imperatives that antici- pate a contrast, in other words, express a concession, are accompanied by sane, and that, second, the licet usage had such a development in spite of starting so late. The instances with sane are precisely parallel to the other clauses with sane and to those with quidem already studied: the injected word defines the par- ticular type of anticipation. A very few illustrations will suffice. Cicero, De Nat. Deorum 1.24.68 : Sint sane ex atomis: non igitur aeterni. Cicero, Acad. 11.32.105 : Haec si vobis non prohamus, sint falsa sane, invidiosa certe non sunt. Sallust, Cat. 52.12 : Sint sane, quoniam ita se mores hahent, liherales ex sociorum fortunis, sint misericordes in furihus aerari: ne illi sanguinem nos- trum largiantur. Seneca, De Otio 7.2: Sit sane grande discrimen, tamen alterum sine altero non est. Tac. Hist. IV.58.22: Sane ego displiceam; sunt alii legati. The same effect is to be seen in the use of forsitan to point the coming contrast, as, for example. Quint. Inst. Orat. 1.12.8 : Mirum sit forsitan, sed experimentis de- prehendas. Or Cicero, Brutus 8.33: Quae forsitan laus sit; verum tamen natura magis tum casuque, numquam aut ratione aliqua aut idla observatione fiebat. 184 LATIN SENTENCE CONNECTION Some of the instances already cited of each type of anticipatory imperative have shown conjunctions in- troducing the second clause. As in all other sorts of sentence connection, the supplementary use of con- junctions bears out the results of the investigation of the cases without conjunctions. Et, enim, sed, at, tamen appear with the anticipatory imperatives. The first two are found with the merely hypothetical cases, the last three with the concessive. Without functional repetition, it was found (Chap. Ill, pp. 57 ff.), et regu- larly marks a clause subsequent logically to the one preceding, with no further connotation. It can there- fore be used after hypothetical imperatives with the clause expressing the result of the action urged. Seneca, De Provid. 6.7: Adtendite modo et videbitis quam brevis ad libertatem et quam expedita ducat via. No difference is apparent between this and the in- stances cited without the et. Horace, Epist. 1.18.107, uses the et and five lines below obtains the same results without it. Line 107: Sit mihi quod nunc est, etiam tninus, et mihi vivam Quod superest aevi si quid super- esse volunt di. Line 112 : Det vitam, det opes, aequum mi animum ipse parabo. When the second clause takes on more of an explan- atory tone, developing the purpose of the irrelevant imperative, enim is the more natural conjunction. It would seem that nam must have been used too, though I have found no instances. Cicero, Pro Rose. Com. 6.17: Quae cum ita sint, quis sit, qui socium fraudarit et fefellerit, consideremus ; dabit enim nobis iam tacite vita acta in alterutram partem firmum et ANTICIPATORY INCOMPLETENESS 185 grave testimonium. Cicero, De Orat. 1.24.112: Perge vero, inquit, Crasse, Mucins. Istam enim culpam, quam vereris, ego praestabo. It should be noted that these are instances which are scarcely distinguishable from ordinary imperatives without anticipatory force, almost exactly like the many instances in Cato of which the following will serve as the type. Cato, R.R. V.7 : Opera omnia mature conficias face. Nam res rustica sic est, si unam rem sero feceris, omnia opera sero fades. Passing to the cases of imperatives that express the admission or concession of some fact, it is of course the adversative conjunctions which are to be expected, as in all cases of contrast, whether anticipated or not. But once more they are purely supplementary. Cicero, In Q. Caec. 15.47: Esto; ipse nihil est, nihil potest; at venit paratus cum suhscriptorihus exercitatis et disertis. Horace, Sat. II.1.83 : Esto, si quis mala; sed bona si quis ludice condiderit laudatus Caesaref Martial V.15.6 : Non prosint sane, me tamen ista iuvant. Quint. Inst. Orat. 1.12.8: Mirum sit forsitan, sed experiments deprehendas. Such cases, in their anticipation of contrast, are quite like instances of fateor, the specific expression of concession. For example, Cicero, Ad Att. II.1.5 : Non consulare, inquies, dictum. Fateor: sed ego illam odi male consularem, Pliny, Epist. II.5.10: Fateor: in praesentia tamen et ista tibi familiariora fient et . . . etc. Two very similar types of usage ought to be con- sidered with the hypothetical imperative. One of them is the hypothetical use of an indicative, the other 186 LATIN SENTENCE CONNECTION the licet usage. In both types the underlying element which gives to the clause in which they occur an antici- patory force, is the irrelevance of the statement itself which the clause makes, or of the emphasis on the per- missive tone. They are therefore similar in the funda- mental element which gives them force and they prove to be similar in the effect produced, the one to the first type of hypothetical imperative, the other to the second. Take, for example, such a statement as fecit assem (Petronius, 61, line 16). Occurring in the description of a man, this is entirely irrelevant. It cannot be a mere statement of fact for it would be quite meaning- less in the context. The following clause, semissem habui, shows it to have been a hypothetical statement. A succession of such instances occurs in Seneca, De Tranq. An. 11.10 if. beginning with locuples es, a per- fectly irrelevant statement not understood until de- fined by what follows: numquid divitior Pompeiof And a little later: Honoribus summis functus es: num- quid aut tarn magnis aut tarn insperatis aut tarn uni- versis quam Seianus? And again: Rex es: non ad Croesum te niittam . . . etc. A good illustration is Cicero, De Finihus IL9.27 : Confuse loquitur; gerendus est mos, etc. Many more might be given ; one will do to represent the many occurrences in poetry. Juvenal, III.IOO if. : Rides, maiore cacchino Concutitur; fiet, si lacrimas conspexit amici, Nee dolet; igniculum hrumae si tempore poscas, Accipit endromidem; si dixeris "aestuo" sudat. One of the commonest forms in which this usage appears is with the verb in the future ANTICIPATORY INCOMPLETENESS 187 perfect, a future hypothesis: Cicero, In Verrem II.I.5.12 : Ex hoc quoque evaserit; profiscar eo, quo me iam pridem vocat populus Romanus. Martial III.78.7 : "Si nihil hinc veniet, pangentur carmina nobis; Audi- eris, dices esse Maronis opus." Et may be used in the same supplementary fashion as with the imperatives. Seneca, De Tranq. An. 11.3: Appelaverit natura quae prior nobis crediderit, et huic dicemus: etc. The clauses with licet are like the other type of imperative, that with permissive force, as the meaning of licet indicates. A permissive clause is rarely rele- vant in consecutive discourse except as anticipating a contrast. So true is this that licet comes to be looked on as merely a conjunction. But the origin of the usage must have been the use of licet to fill the want felt for a third person permissive imperative, which, by its meaning, it so readily fills. Instances are too common and too familiar to require more than passing notice. Cicero, Pro A. Caecina 14.41: "Queramur," inquit, "licet; tamen hoc interdicto Aebutius non tenetur." Seneca, Ad Helviam 10.6: Licet itaque augeatis census, promoveatis fines: numquam tamen corpora vestra laxabitis. The relation of the licet to the subjunctive is not under discussion at present, but the relation of the whole licet clause to the one that follows. The same characteristic element is behind the con- cessive use of quamvis. E.g., Cicero, Tusc. Disp. 11.25.61 : Nihil agis, dolor! quamvis sis molestus, num- quam te esse confitebor malum. It is the emphasis on 188 LATIN SENTENCE CONNECTION the imperative which anticipates the contrasted result of the hypothesis. A glance in retrospect at the instances of anticipa- tory incompleteness shows that this connective ele- ment, whether lying in the meaning or in the function of the word, marks the clause in which it stands as logically antecedent to the clause following. By thus conspicuously marking its own clause as logically ante- cedent, it emphatically draws attention to the follow- ing clause as the more important, logically, of the two. This accounts for the development of subordinate clauses to do much of the work which might have been carried by these incomplete clauses, resulting in the tendency to look upon the more fundamental and nor- mal types as an exceptional usage and as substitutes for the types with subordinating conjunctions. The truth is that the present usage does not differ, in the kind of relation expressed, from the retrospective incompleteness already studied, but by being anticipa- tory it puts a stronger emphasis on the logical sub- ordination of one sentence to the other. This is only a corollary to the general truth that expression of rela- tion in the first sentence is a conscious rhetorical de- velopment and therefore a natural step in the rhetori- cal development of subordination. Analysis shows three main divisions into which these clauses, logically incomplete, naturally fall. The lines of demarcation are not sharp but the prominent char- acteristics of each division are distinct. The first embraces those instances which are identi- ANTICIPATORY INCOMPLETENESS 189 cal with the cases of retrospective incompleteness except for the fact that they occur in the first instead of in the second of two adjacent clauses and therefore point in the opposite direction. They are quite as inconclusive, so far as definition of relation is con- cerned, as were the retrospective instances. They merely indicate the more important of the two clauses as being the one following: the precise definition of relation is determined by other and different means, ranging from the meaning of the clause to the mechani- cal indication by means of conjunctions. In this group are included the instances in which semantic incom- pleteness is evidenced in the use of demonstratives and conjunctions and incomplete verbs as well as the in- stances of functional incompleteness inherent in the infinitive, the subjunctive and the incomplete tenses. The second group comprises instances in which the incompleteness consists in an emphasis, marked but irrelevant, or at least without meaning until a further statement has followed. Such instances give to the clause following a distinctly adversative tone. So long as the incompleteness is semantic this is about all that it suggests, but when it is functional and lies therefore in some hortatory expression, it is sufiiciently promi- nent further to subordinate its own clause in tone and to suggest with more or less distinctness a concessive relation toward the clause following. This group in- cludes the instances of the use of comparatives, of selective words such as ceterus, of emphasizing words such as quidem, of irrelevant negative statements, of words unexpectedly emphasized by position, and, 190 LATIN SENTENCE CONNECTION finally, of anticipatory imperatives marked by empha- sis, with or without licet or quamvis. In the instances of the third group the incomplete- ness consists in a general vagueness or irrelevance of statement without however any marked emphasis. When the incompleteness thus effected is semantic it results in making more prominent the following clause, which takes the form of an explanation of the vague- ness or apparent irrelevance. When the incomplete- ness is functional the effect is somewhat more evident in the first clause, as was true also in the preceding group. In the present case however, it makes the first clause hypothetical, as expressing a contingency of which the statement of the second clause is the logical explanatory result. In this group fall the instances of words with a significantly vague meaning, or impera- tives without permissive tone, and of hypothetical indicatives. CHAPTER VII PARENTHETIC INCOMPLETENESS ( t 'Ride, si sapis, puella, ride" Paelignus, puto, dixerat poeta. (Martial 11.41.1.) Et quaeso con- siderate, quam convorsa rerum natura sit. (Sallust, Orat. Phil 13.) I licet. (Plautus, Most. 848.) Puto, quaeso and licet are none of them syntactically related to the sentences in which they stand. They represent one of the most interesting and most im- portant classes of sentence groups in which the element of incompleteness is effective. The group is composed of verbs, principally of saying or thinking, injected parenthetically into sentences quite complete syntac- tically (and often logically) without them. In a general way they are not unlike all parenthetical sen- tences and it will be perhaps the best approach to their study to notice a few familiar types of parenthesis. Cicero, Brutus 82.283 : Sed ad Calvum — is enim nobis erat propositus — revertamur. Livy XXXI.45. 10: Ad Prasias — continentis Atticae is locus est — Issaeorum viginti lemhi classi Romanorum adiuncti sunt. Tac. Hist. II.24.7: Ad duodecimum a Cremona {locus Castrorum vocatur) ferocissimos auxiliarium . . . componit. Tac. Agr. 22.2: vastatis usque ad Tanaum {aestuario nomen est) nationihus. Livy lXII.5.9 : auxilia . . . ad occupandos quae ad Anti- 192 LATIN SENTENCE CONNECTION goneam fauces sunt — Stena vocant Graeci — misit. Cicero, In Cat. 11.8.18: Horum hominum species est honestissima {sunt enim locupletes), voluntas vero et causa inpudentissima. Tac. Hist. III.21.9: dein sep- tima Claudiana agresti fossa {ita locus erat) prae- munita. These parenthetical statements all have an element in common : alone and for themselves, they are logically incomplete. Whether this is brought about by the force of a demonstrative, pointing to something outside the clause, or by that of an empty noun like locus, or by the assumption of object or subject from the main sentence, the results are the same. The parenthetical clause has no syntactical relation to the sentence into which it intrudes; logically it has close relation. Often, as in two of these instances, a con- junction adds precision to the expression of relation. The logical incompleteness may be due merely to the irrelevance of the parenthetical statement taken apart from the main sentence, as in Caesar, Bell. Civ. II.4.1 : Massilienses . . . naves refecerant summaque indus- tria armaverant — remigium, guhernatorum magna copia suppetehat — piscatoriasque adiecerant. Such parentheses are very often made more precise by the use of quippe or sane. Now, except for the kind of sentence in which they occur, and the kind of verb used, there is little dif- ference between these parentheses and the interjected verbs noted above. Puto, quaeso, licet are no more syntactically incomplete than the longer parentheses and logically they are quite as much so. They form a special group leading to distinct syntactical develop- PARENTHETIC INCOMPLETENESS 193 ments because of one point in which they are all alike : they are used to give tone to the statement or question or command into which they are injected. They are not explanations. They are expressions usually of the speaker's own personal attitude toward the statement or question or command; otherwise, of some im- personal factor which gives to the main sentence its tone. For example, notice the variations of tone in the command, ite, produced by the addition of iubeo, ohsecro, censeo, licet, or necesse est. To get a clear impression of the effect of these in- jected words, the first step is to note the determining force exercised upon them by the type of sentence into which they are injected. If it is a statement, they are restricted to one set of words, if a question, to another, if it is a command, or an exhortation, to a third. Quaeso, rogo, ohsecro, and the like, with their twofold meanings, appear both with questions and with im- perative sentences. Statements should be considered first. The injected words used with statements might be grouped in either of two ways : first, according to the mode in which they appear, indicative, subjunctive, or imperative; or, second, according to the type of verb. The latter is, I think, the more valuable grouping. There are, in the first place, numerous verbs of saying. Cicero, Pro Quinctio 11.37: In hanc rem te, te, inquam, testem, Naevi, citabo. Cicero, Ad Att. 11.11. 1: Narro tibi, plane relegatus mihi videor. The purpose of these inserted verbs seems to be to lay emphasis on the state- 194 LATIN SENTENCE CONNECTION ment made, to show the speaker's earnestness in mak- ing it. It is obvious without them that the statement is the speaker's own but the introduction of the first person of the verb of saying emphasizes this. Often however it has modifiers which give to it a different tone. Pliny, Epist. IV.17.6: Adulescentulus eram, et iam mihi ah illo honor atque etiam {audebo dicere) reverentia ut aequali hahehatur. The tone is almost apologetic, due to the meaning of the verb, audeo. Compare with this, Cicero, Pro Rose. Com. 6.17: qui medius fidius {audacter dico) plus fidei quam artis . . . possidet in se. Cicero, De Orat. 1.21.97 : quo in genere tu, Antoni, — vere loquar — nunquam mihi . . . defuisti. Or again, such extended phrases as Pliny, Epist. 1.2.3 : Nee materia ipsa huic {vereor, ne improbe dicam) aemulationi repugnavit. Cicero, In Cat. 1.1.3 : Nos, nos, dico aperte, consules desumus, and In Cat. 1.4.8: Bieo te priore nocte venisse inter falearios {non agam obseure) in M. Laeeae domum, are further varia- tions. Finally, the subjunctive mode gives a shading of tone to the injected verb, a tone of less finality usually reinforced by paene or prope. The origin of this tone in the subjunctive, sometimes miscalled "modesty," is probably to be found in the original "will" notion behind the mode. In a word of saying this cannot, in the first person, express much actual will, but, by analogy with other verbs, in which it shades from will into simple futurity, the subjunctive dieam has a less positive tone than dieo. Cicero, Pro Quinetio 13.44 : iam tu potes liheratus discedere moles- tia, prope dieam, non minore quam Quinctius. PARENTHETIC INCOMPLETENESS 195 With the instances of verbs of saying inserted in declarative sentences, there is no play for the impera- tive mode; at least, imperatives do not appear so far as I have noted. Another use of dico should, however, be noticed in passing : that in which it appears with an object which is also in apposition with a noun in the main sentence. Pliny, Epist. 1.16.1: Amaham Pom- peium Saturninum {hunc dico nostrum) laudebamque eius ingenium. Seneca, De Provid. 1.3 : Ne ilia quidem quae videntur confusa et incerta, pluvias dico nuhesque etc. Such instances are, however, analogous to the general type of parenthesis illustrated above and not to the tone words at present under discussion. They are explanatory parentheses pure and simple. The interjection of the second or third person as well as the first to mark simple quotation, as inquit, inquis, dixit, has been discussed before. (Cf. Chap. VI, pp. 148 ff.) The principle is the same as that behind these parenthetical verbs of saying interjected into non- quoted statements: the insertion marks the speaker's attitude toward what he is saying, in this case attri- butes it to someone else. If the quotation is hypo- thetical, inquies, dicet aliquis, or some similar phrase marks it as such. The importance of the meaning of the verb of saying is shown by the influence of such a word as fateor or its equivalent in a declarative sentence. By its very meaning it introduces the notion of concession and its sentence is regularly followed by one introduced by an adversative conjunction. But this influence is exerted on the sentence in w^hich it stands as a whole. 196 LATIN SENTENCE CONNECTION The relation of the fateor (as in Terence, Ad. 188: leno sum, pernicies communis fateor, adulescentium,) or non nego (as in Terence, Ad. 798: factum est, non nego,) to the clause in which it stands is the same as that of dico, inquis, and the like. Of the various verbs of thinking and other mental actions little need be added. They are fundamentally like the verbs of saying and they simply give shades of meaning to the sentences in which they stand. For example, Cicero, Pro Rose. Am. 28.76 : Litteras, credo, misit alicui sicario, qui Romae noverat neminem. Cicero, Ad Att. IV.2.7: Domus aedificatur, scis, quo sumptu qua molestia. Martial, Epist. ad Lib. II. line 11 : Puto me hercules, Deciane, verum dicis. Tac. Dial. 37.25 : Non, opinor, Demosthenem orationes inlustrant quas adversus tutores suos composuit. These words may be in the imperative ; they still mark the attitude of the speaker. Seneca, Ad Polyh. 9.9 : Est, mihi crede, magna felicitas in ipsa necessitate moriendi. Cicero, Ad Att. IX.6.4: Non sum, inquam, mihi crede, mentis compos. Cicero, Ad Att. IX.7.4: sed agetur, memento, foedissime. The verbs used parenthetically with questions are naturally, not verbs of saying or thinking, but verbs of asking, except when they are in the imperative, as, for example, Cicero, Ad Att. IL7.2: Narra mihi, reges Armenii patricios resalutare non solentf The more common type is illustrated by the following instances : Seneca, De Vita Beata 7.1: Quid est, oro vos, cur separari voluptas a virtute non possitf Cicero, Pro PARENTHETIC INCOMPLETENESS 197 Rose. Am. 40.118: Quid tandem, quaeso, iudicesf Cicero, Ad Att. IIL9.2; Obsecro, mi Pomponi, nondum perspicis quorum opera . . . perierimusf Suetonius, Claudius 40: rogo vos, quis potest sine offula viveref Plautus, Ps. 971 : ecquem in angiporto hoc hominem tu novistif te rogo. Cicero, Pro L. Murena 38.81 : Te, te appello, Cato; nonne prospicis tempestatem anni tuif The rather colourless age or agedum is not uncommon in somewhat urgent questions, especially in Seneca; for example. Be Vita Beata 11.1 : Age, non vides quam multa suasura sitf Terence, Andria 598: age igitur, uhi nunc est ipsusf This however is little more than an interjection and seems never to have developed into anything else. When it comes to the use of interjected verbs with the imperative mode, all the types of verbs which have been found with declarative and interrogative sen- tences reappear, and there is also a further group peculiar to the imperative sentences. For instance, Cicero, Pro Quinctio 25.79 : Die, inquam, diem. Seneca, Ad Marciam 12.4: Circumspice, inquam, omnis. Plau- tus, Men. 696 : heus tu, tihi dico, mane, redi. Cicero, In Cat. 1.3.6 : Muta iam istam mentem, mihi crede. These sentences all show inserted verbs which are used also with declarative sentences, and they are really used in the same way, to give a tone of earnestness to the sentence. On the other hand, in the use of quaeso and rogo with the imperative, there is only apparent simi- larity with their use in interrogative sentences. For, with an imperative sentence, quaeso and rogo have a 198 LATIN SENTENCE CONNECTION different meaning, urging rather than asking. For example, Cicero, In Cat. 1.11.27 : percipite, quaeso, dilig enter quae dicam. (This influence of the impera- tive to modify the meaning of the injected verb is seen in such sentences as Cicero, Ad Att. IX.6.2 : sed, opinor, quiescamus.) Finally the colourless age, agedum appears as an interjection to give emphasis to the command, as in Seneca, Ad Polyh. 4.2: Omnis agedum mortalis circumspice. Plautus, Cist. 544: Age perge, quaeso. Terence, Ad. 937 : age da veniam filio. Livy XXXVIII.47.11 : mittite, agedum, legates. Rogo and quaeso, in their meaning of ''urge" or "beg," belong to the third group of words used with the imperative, the most important group. Like the verbs of mental action with declarative sentences, the words of this group are used to give shades of tone to the imperative. They are either verbs of warning, ordering, asking, and the like in the first person, or else impersonals, giving tone in a more indirect way: opus est, licet, and the like. The most frequent are quaeso and obsecro; none are very common. For early Latin, they are collected in Bennett's Syntax. A few examples will suffice here. Cicero, Ad Att. IV.6.4: rogo, fac ut sumas. Plautus, Amph. 765 : mane, mane, obsecro te. (The doubling of the imperative has the same force as the injection of the obsecro.) Cicero, Pro Rose. Com. 7.20 : Oro atque obsecro vos, qui nostis, vitam inter se utriusque conferte. Seneca, Oed. 864: ignosce, quaeso. This is heightened in Terence, H. T. 1052, to age, quaeso, ne tarn offirma te. Plautus, Most. PARENTHETIC INCOMPLETENESS 199 848 : i licet. Pliny, Epist. IX.26.7 : Sed opus est exam- ine et libra. The presence or absence of the personal object with the oro or ohsecro has no importance, I think. The incompleteness lies in the meaning of the verb which requires some sort of object clause to give it satis- factory meaning. Somewhat different is amo te or amabo te. In this phrase it is the complete irrelevance of the statement except as taken with the sentence into which it is interjected which gives it a hypothetical sense. This is especially true of the future form which I should suspect was the earlier. Cicero, Ad Att. II.7.3 : Incende hominem, amabo te, quod potes. Origi- nally this was undoubtedly the equivalent of a condi- tional sentence although the amo te, amabo te, came to be no more than an interjection like age, agite, agedum. The great importance of the injected verbs with the imperative, lies in the light their use throws on the same usage with the subjunctive. The subjunctive was originally an independent mode with an indefinite notion of will which lent itself to the imperative uses and was often accompanied by the same interjected verbs, giving tone to the subjunctive. It seems certain that that tone was originally not at all precise in the subjunctive itself. It depended much on the person of the verb. In the first singular the idea of will appears as determination of various degrees, all the range from ''I shall" to '*I will." There can be no notion of imposed will, of command, when the speaker and the person addressed are one. In the first plural there is a change. The will notion applies not only to 200 LATIN SENTENCE CONNECTION the speaker but to others as well, so that there is not only the notion of determination but of command com- bined with it. (Hence volo and opinor both find place with the hortatory first plural.) In the second person the command is direct, in the third indirect; in both cases it is open to a wide range of interpretation, from humble request to arbitrary order. For Plautus these cases are thoroughly collected by Morris^ and conveniently arranged. There is little to add save instances from later Latin bearing out the deductions from Plautus. A few conclusions from the results of Morris's work are necessary however for an understanding of the usage. The second person instances show the same modifying verbs interjected into the sentences that appeared with the imperative, but distinctly more. The difference seems to be this. The injected verbs used with the imperative are either like inquam, dico (rare), emphasizing the command, or else they are words of request or beseeching, indicating that the imperative is not being used in its normal fashion as an abrupt command by one who has the right to order: oro, rogo, quaeso, ohsecro, opus est, amaho. With the subjunctive, on the other hand, the range is much wider: the injected verb may indicate wish {volo, malo, nolo), request {quaeso, oro, ohsecro), advice {censeo, moneo), command {iuheo, interdico), explanation {meliust, optimumst), and so on. The notion of will was precise and narrow in the imperative and modified by the injected verbs ; in the subjunctive 1 The Subjunctive in Independent Sentences in Plautus. A. J. P. XVIII. (1897), Nos. 70, 71, 72. PARENTHETIC INCOMPLETENESS 201 it is general and indefinite and is determined more precisely by the words injected. This usage never became much more precise in later Latin. Censeo desistas in Cicero (In Verrem II.V.68.174) may have been looked on as a subordinate subjunctive with a verb of advising. But it is hard to see how Maneat, quaeso, duretque gentibus (Tac. Germ. 33.7) can have been so regarded, nor does scias licet (Seneca, Be Tranq. An. 8.3) seem essentially different from i licet. The fact that, with the first singular, volo does not occur (it would be tautological), whereas it does appear with considerable frequency with the other persons (including the first plural), in which others than the speaker are addressed, indicates the defining nature of the injected verb. In treating the independent subjunctives without interjected words, Morris draws attention to the fact that any expression of will involves a speaker, a wilier, a hearer, an actor. In declarative sentences with the verb in the first person these are all four identical ; in the interrogative, the speaker and actor are one, the hearer and wilier another. In sentences with the verb in the second person, speaker and wilier are one, hearer and actor another. With the verb in the third person, speaker and wilier are usually one, but hearer and actor are distinct. Now these relations have an important bearing on the injected words. They explain why, in declarative sentences with the verb in the first person, not only is volo not to be found, but no verbs in the first person are injected. Also why such verbs are almost the only ones used in the case 202 LATIN SENTENCE CONNECTION of the direct relation established between two persons in the instances of subjunctives in the second person. And why the interjected words are almost evenly- divided between imperatives and first persons in the examples with the third person. Further discussion will be more clear with a number of examples given. They are divided first according to the person of the subjunctive and arranged within groups according to the form of the injected word. The cases from Plautus and Terence are taken from Morris and Durham.^ Subjunctive in First Person. Plautus, Ba. 707 : volo agamus. Plautus, Merc. 1015 : dicamus censeo. Cicero, Ad Att. II.5.1: Sed, opinor, excipiamus et exspectemus. Plautus, Trin. 681: dem suades. Plautus, Asin. 644: faciamus suades. Plautus, Ba. 24: sine te amem. Plautus, Men. 890: fac sciam. Plautus, Rud. 681 : afferam adigit. Cicero, Tusc. Disp. 1.49.119 : referamur necesse est. Cicero, Pro Rose. Am. 13.36 : ego crimen oportet diluam. Plautus, Asin. 448 : Nunc adeam optumumst. Subjunctive in Second Person. Sallust, Cat. 52.26 : miseriamini censeo. Cicero, Ad Att. II.lO.l : volo ames. Cicero, Ad Att. III.l.l: te ore des operam. Pliny, Epist. V.19.8 : rogo scribas. Pliny, Epist. I.lO.ll: te Jiortor permittas. (Malo, nolo, faxo, dico, interdico, quaeso, obsecro, obtestor also appear.) 1 Morris: The Subjunctive in Independent Sentences in Plautus; Dur- ham: The Subjunctive Substantive Clauses in Plautus not Including Indirect Questions, Cornell Studies in Classical Philology, No. XIII. PARENTHETIC INCOMPLETENESS 203 Sallust, Cat. 44.5 : fac cogites. Plautus, Asin. 902 : sine revenias. Pliny, Epist. VII.1.2: moneo resistas. Cicero, Pro Quinctio 22.73: doceas oportet. Plautus, Aul. 568 : optimumst loces. Subjunctive in Third Person. Plautus, Ps. 1123: volo accipiat. Pliny, Epist. V.14.9: cupio remittat. Seneca, De Vita Beata 25.5: malo iucundiora veniant. Pliny, Epist. III.12.1 : pacis- cor expedita sit. Tac. Germ. 33.7: Maneat, quaeso, duretque gentihus . . . odium sui. Cicero, In Verrem 1.17.51 : fac veniat in mentem. Martial III.25.3 : Roga lavetur. Seneca, Medea 189: iuhete sileat. Sallust, Cat. 32.2: mandat confirment. Cicero, Pro Rose. Am. 22.62: exstent oportet. Cicero, Pro Rose. Am. 40.116: videant necesse est. Sallust, Cat. 45.1 : agant permittit. Seneca, Ad Marciam 10.3: admonendus est animus amet. Cato, R.R. LXIX.2: Tepeat satis est. PUny, Epist. IX.33.11 : non est opus adfingas aliquid. The second person scarcely figures in the instances of inserted verbs except in the imperative because the speaker would rarely be likely to tell the hearer the latter 's will on any matter, but might readily urge him to a certain show of will. Even this is not natural when the subjunctive is in the first person so that the second person is found interjected there only in questions. Another thing to be noted in the examples cited is the group of cases of oro te and like expressions in which the direct object of the verb of ordering or 204 LATIN SENTENCE CONNECTION requesting or asking is stated. Pliny, Epist. VIII. 17.6: Teque rogo . . . quam maturissime solicitudini meae consulas. Cicero, Ad Att. III.l.l: Te oro, des operant. Pliny, Epist. I.lO.ll: Te hortor permittas. These are precisely like the corresponding type with the imperative. Cicero, Ad Att. IV.Sa.l: die, oro te, clarius. Cicero, Ad Att. III.18.2: fac, obsecro te, ut sciamus. With the personal object expressed there is even less feeling of any necessity of a subordinate clause to follow than there is without it. It is noteworthy that there is no shift of mode, tense or person indicating subordination. The subjunctives as they stand are quite capable of functioning alone. From such instances there undoubtedly developed a subordinate construction with ut. But the use of the construction without ut never became exceptional. The ut construction was never hard and fixed. Cato used facito ut amurcam cotidie suppleas (69.1), but much more frequently facito studeas (25.1) ; facito uti con- veniat (42.1) is less common by far than facito ehibat (71.1). And so throughout the literature, rogo scrihas and rogo ut advoles, animum advortas volo and volo ut facias, and so on, appear side by side. (Even in analogous cases in the imperfect to be noted presently, this is true. Compare Livy XXIX.18.16 : audita . . . vox est, ahstinerent manus, with Livy XXIV.31.1: conclamant , honum ut animum haherent.) Both con- structions were at hand to choose from. Neither was regular or irregular, but the use without the conjunc- tion was surely the more natural and the earlier. The conjunction, as usual, is originally supplementary. PARENTHETIC INCOMPLETENESS 205 The negative commands, prohibitions, are similar and equally interesting. For example, Cicero, Ad Att. IV.13.2 : Illud etiam atque etiam te rogo . . . ne istuc Jiospes veniam. In this case, to be sure, the notion of ''wish" that came into the first person of the sub- junctive through the use of velim (see Morris), is definitely present in the veniam and makes possible the use of rogo as a defining word. A more simple instance is Martial 1.116.5: si cupit hunc aliquis, moneo, ne speret agellum. In many instances the negative voli- tion came to be merged with the inserted verb. This is of course a secondary usage. Vetabo sit (Horace, Odes III.2.26), ames nolo, interdico aiant, and Cave tu mi iratus fuas (Plautus, Capt. 431), are examples. The secondary development of the construction by analogy is also to be noted, the use of other tenses of the subjunctive with injected verbs in the past. Pliny, Epist. IV.15.8: Te hortarer circumferres oculos. Tac. Ann. II.17.6: exclamat irent. Pliny, Epist. IV.11.11: praemonetur confugeret. Cicero, Pro Rose. Am. 32.90 : vellem viverent. Cicero, Ad Att. II.4.7: imperavi ne impediret. These are fully discussed by Morris. It hardly belongs to the province of this discussion to investigate the origins of grammatical forms which antedate the historical period of the language. But in the case of such constructions as the subjunctive after verbs like volo and iuheo, it has seemed worth while to go a little beyond mere description because of the clearness and significance of the evidence. The accepted origin of ut as an adverbial form of the rela- tive falls in with all other indications and suggests that 206 LATIN SENTENCE CONNECTION here was a verbal usage relying on incompleteness of meaning for the expression of its sentence relation- ship ; that this relationship was further defined by the use of an adverb which was in process of attaining a fixed conjunctional usage. The analogy of the impera- tive used in consecutive discourse with modo and sane^ strengthens this probability. It is quite a secondary consideration what point in the hardening process had been reached at any given time. With different interjected verbs and with different sorts of sentences, other constructions developed, as noted above: with declarative sentences and verbs of saying, the accusative and infinitive ; with interrogative sentences and verbs of asking, the subjunctive with an interrogative particle. In all these instances, the verb of saying or asking or urging or ordering was origi- nally incomplete from a logical standpoint and like most of the incomplete expressions studied indicates an idea logically antecedent to the adjacent clause (in this case to the clause in which it is inserted) . The fact that syntactically it developed into the main clause while the clause to which it lent tone, the one into which it was injected, became the subordinate clause, is only a further illustration of the essential difference be- tween logical and syntactical relation. It furnishes one of the best examples of this difference and of the more fundamental nature of such a connective element as incompleteness compared with the subordinating conjunction ut. CHAPTER VIII CONCLUSION The foregoing analysis of the fundamental means of expressing relation between contiguous sentences re- veals three general types which define relation only in a somewhat vague way except as special uses develop to express more precise definition. Such a result might at first seem meagre. In reality, it should be expected that any investigation going back of the precise and sharply defined means in general use in Classical Latin would lead toward means of less precision to be sure, but of more general application. For such a result is in accord with the universal law of progress in language. The original ablative ending expressed a wide range of possible relations, limited only by the meaning of the word itself and by the sense of the context in which the word was used. In some instances, as for example in the names of towns, this was always sufficient, and nothing further in the way of precision was felt neces- sary. But in others, many possible relations made it natural for adverbs to be used with the phrase in which the word in the ablative occurred, to indicate within a narrower range the significance of the ending. Another step was gradually taken by the development of the adverb into a preposition until finally the expression 208 LATIN SENTENCE CONNECTION of relation rests largely in this external element instead of in the case ending. Such a process is typical of much of the development in language. It is typical of the progress gradually made through long periods of time in the precise expression of the relation of sentence to sentence. Just as words spoken in succession are instinctively assumed to have relation and in fact must always bear some relation to each other if the utterance be that of a rational being, so when expression of thought in sentences had become an established fact, some rela- tion between sentences spoken in succession was beyond question. Inasmuch as sentences expressed more than single words, the range of possible relations was distinctly more limited than that between words. Each element in a sentence, by its bearing on the others, narrows down the range of possible relation which the sentence can bear to others. It is primarily for this reason and not because of any inherent diffi- culty in devising a modification of the sentence corre- sponding to case endings, that the relation between sentences was more generally and for a longer period left for mere juxtaposition to indicate. But even between sentences there were usually several different relations possible and there were at least three fundamental and natural signs that served to define the relation: repetition, change, incomplete- ness. It is not improbable that there were others not yet disclosed by investigation. Until the need of more precision was felt, the relations indicated in the most general way by such fundamental means were the only CONCLUSION 209 limitation to interpretation, beyond tlie meaning of the individual sentences and their order of succession, and even this last might often be misleading rather than helpful. Adverbs or phrases limiting the range of possible relation suggested by the meaning of a sen- tence were a decided step toward more precision and eventually these developed into conjunctions upon which devolved much of the work previously carried by the more fundamental means. But throughout the history of the language, whether with or without the supplementary force of the conjunction, repetition, change and incompleteness exercised their natural and universal power to suggest relation between sentence and sentence, and not infrequently defined the relation so adequately that none of the more artificial means was necessary. The extent to which they appear thus unsupported varied with their use by different writers and should prove a useful criterion of style, but their chief interest lies in the fact that they furnish another line of evidence for the natural development of lan- guage from the most general expression of ideas to the most precise. In the introductory chapters the general psycho- logical principles were outlined that underly all sen- tence connection. These led directly to the examina- tion of the fundamental relations which language seeks to express. The next question to arise, what are the means employed by language to express these relations, has been answered to a considerable extent by the results of the investigation. It remains to point out the indications of the resulting types of sentence not s» i^ny si:yYESCs —7* . • &I ^siee