II, 5) rarely employed in Latin, while I, 2), I, 3), and II, 2) would prevail. We cannot here take up this question in detail, but it will appear from the following pages that type II, 1) is far from uncommon in Latin.

A second characteristic of the Latin language, in which it differs decidedly from the modern languages of Europe belonging to the Indo-European group, is found in its pronounced attributive character as opposed to the predominatingly predicative character of the latter. This results in a great increase in the importance and complexity of the minor elements of the sentence.2 In fact the distinctive character of a Latin sentence may be more clearly reflected in them than in the relations between the primary (major) elements. Sometimes an attributive element, either by virtue of its dominating character or complexity, attains an importance equal to or greater than that of a subordinate clause. Furthermore such an attributive member or some part of it is apt, like any other element of a sentence, to awaken a train of associative additions. The main framework of a sentence may thus almost entirely disappear in the network of subordinate elements.3 The student of Latin sentence structure is constrained by these conditions to examine all the minor elements with special care, with much greater care perhaps than would yield profitable returns to the student of the modern sentence. A con

1 On the general distinction between attributive and predicative structure, see Wundt, Völkerps. I, vol. ii, ch. 7, § V, 3 ff.

2 The major elements of the sentence are those taken account of in the above formulæ, i.e. 1) main and subordinate clauses, 2) subject and predicate. All others are minor.

8 Whether this is a stylistic merit or defect, will, of course, depend upon the clearness or muddiness of an author's thought.

In this respect the Sanskrit stands on the same basis as the Latin. An admirable example of attributive structure is afforded by the following sentence (Hitopadeçaḥ 2, 4): ityālocya (gerund) tena (sc. sihena) grāmą gatvā (gerund) dadhikarṇanāmā bidālo māsādyāhārena satoṣya (gerund) prayatnād ānīya (gerund) svakandare dhṛtaḥ (participle). Translated ad litteram: "By the lion thus having reflected, having gone to a village, having won the confidence (sc. of a cat), having carefully led it (sc. to his den) the cat kept (sc. was).” Contrast with this version the following idiomatic English translation with predominating predicative form: "(The lion) thus reflecting went to a village, won the confidence of (literally satisfied') the cat Curd-ear by meat and other food, then led him carefully to his den and kept him there."

crete example will make this clear. Post emensos insuperabilis expeditionis eventus languentibus partium animis, quas periculorum varietas (a) fregerat (b) et laborum, nondum tubarum cessante clangore vel milite locato per stationes hibernas, fortunae saevientis procellae (4) tempestates alias rebus infudere (B) communibus per multa illa et dira facinora Caesaris Galli, qui (C) ex squalore imo miseriarum in aetatis adultae primitiis ad principale culmen insperato saltu provectus ultra terminos potestatis delatae procurrens asperitate nimia cuncta foedebat (D).1 How inadequate would be the conception of this sentence given by the following formula, which takes account of the major elements only!

A third fundamental question is whether in general the works of Roman literature are sufficiently spontaneous to admit the application of tests which make assumptions as to the form an author's thought assumed before it was written down in the words in which we have it now. Without entering upon a detailed examination of this question, we may justly make the following statements. Many lines of evidence converge to prove that, while such a test may not profitably be applied to a rhetorical writer like Cicero at least not to all of his works yet there are authors who do exhibit a high degree of spontaneity. Witness Varro, Petronius, Tacitus, Apuleius' Metamorphoses, not to mention many of the Christian writers. Again the decidedly rhetorical character of a composition does not necessarily affect (i.e. modify) the backbone or groundwork of the author's sentences, but may be concerned chiefly with minor details of phraseology, leaving the relations between the larger or even smaller elements quite undisturbed, and so not obscuring the original organization of the thought. For example, the clausulae rhetoricae so extensively used ordinarily involve nothing more than the choice of a different word or an alteration of the order of

1 Ammianus Marcel. 14, I, I.

words at the close of a period or colon. Furthermore, writers who, like Seneca of the school of modernists under the early empire, developed a tendency to the use of short, pithy sentences may, as will appear in the course of this paper, show marked individual traits in the inner structure of such sentences and in their interrelations. The same is true conversely of writers like Velleius Paterculus (a notorious slave of rhetoric) and Pliny the Elder, who sometimes attempt elaborate periods. The student who approaches the study of sentence structure from the point of view of the mental processes involved in it ought therefore to find in the Roman literature a fruitful field.

The number of types described by Dr. Boucke is, of course, not final, and we may at the outset point out another distinctively marked type common in Roman authors which has not been recorded as current among the moderns. A good example is found in the third chapter of Tacitus' Dialog on Orators: adeo te tragoediae istae non satiant (ab), quominus omissis orationum et causarum studiis omne tempus modo circa Medeam, ecce nunc circa Thyestem consumas (cd), cum te tot amicorum causae, tot coloniarum et municipiorum clientelae in forum vocent (ef), quibus vix suffeceris (g), etiam si non novum tibi ipse negotium importasses (ij), ut Domitium et Catonem, id est nostras quoque historias et Romana nomina (k) Graeculorum fabulis adgregares (mn).

ab
Ed
ef

gh

ij
kl

m n

In general character it resembles Dr. Boucke's analytic type (II, 4), and might perhaps be described as an analytic form with descending construction. Such a sentence may involve any number of successive bisections (analyses) of the unit of thought and its elements, from two or three up to the point at

which the unit passes beyond the range of consciousness possible to the writer. Sentences involving three or four such analyses are not uncommon. The type differs from II, 4) in not involving a primary bisection of the unit of thought into two large sub-units. This fact carries with it two consequences of sufficient importance to justify us in setting up the form as a distinct type. On the emotional side sentences of this type lack the element of tension that necessarily accompanies the analysis of the first sub-unit of type II, 4), as also the feeling of relaxation that sets in at the beginning of the analysis of the second. This alternation of tension and relaxation is, of course, an important factor in style. On the ideational side the smooth and continuous forward movement that attends the organization of the sentence, together with the absence of a fixed terminus up to which the analysis must proceed before the relation of the elements to each other becomes clear, make it possible for the author to break off at various points without causing a violent or even noticeable anacoluthon. This gives to the type a certain looseness or freedom which lays it particularly open to the intrusion of associative additions. Such additions yield two sub-varieties of this type. (a) If after the completion of the associative addition or additions the analysis of the complete idea is again taken up, we obtain a type bearing the same relation to the interlocked (II, 3) that the type just described bears to the analytic (II, 4):

(b) If there is no return to the prior idea we obtain a frustum of a sentence plus a closely related associative addition. Three specific forms which such associative additions assume are the correcting quamquam and si clauses, and in many cases those loosely attached clauses introduced by a relative. pronoun that may be rendered into English by a copulative conjunction plus a personal or demonstrative pronoun.

A number of such sentences occur in Tacitus, — chiefly, however, in the Dialog, -notwithstanding that author's strong tendency toward the isolating and synthetic types. Similar associative additions occur occasionally in the second half of ascending sentences of the analytic type (II, 4). The second chapter of the Dialog offers an illustration: nam postero die, etc., to the end of the chapter.1 The passage begins as an ascending sentence of the analytic type, but no sooner does the organization of the second sub-unit begin with the words venerunt, etc., than at the mention of Aper and Secundus the author passes off into a statement of his own friendly feelings toward them, which in turn suggests the hostile attitude of others, their criticism of the two orators, and the refutation of that criticism. Only after all this has been disposed of does Tacitus come back to the prior idea, and then in an entirely new sentence beginning with an igitur (" to resume") ut intravimus, etc., a repetition of the thought of venerunt, etc., which clearly betrays the associative character of the intervening statements. One should not be surprised to find associative additions thus bearing the outward garb of apperceptive elements, because the apperceptive tendencies in the Indo-European languages as a class are so strong that their mechanical devices and schematic forms are forced even on associative additions.

Among the Roman authors Tacitus is distinguished as being one of the most spontaneous, most indifferent to the rules of rhetoric, and most individual. His uniquely organized mind, his strong individuality, his enthusiastic devotion to his work, and his serious purpose in writing, all combine to give his works a special value to the student of sentence

structure.

Tacitus possessed an intensely emotional nature. This is clearly brought out in Nipperdey's masterly characterization of the style of Tacitus, to be found in the introduction to his edition of the Annals. Such natures naturally tend to ex

1 What is ordinarily printed as a separate sentence beginning nam et Secundus is really an associative addition to the preceding clause. One must not lay too much stress on punctuation.