renascentur (-~_I_~), temperemus (-~|~), toto removeamus (__~= the Ciceronian esse videatur), malae rei temperamentum est (_| _~_|_u_), adnitamur modo (—~—), etc. The recurrence of these familiar cadences again and again at the close of sentences can have had no other effect than to accentuate the close of the sentence and so direct attention to it. Thus the individual sentences are

marked off with greater distinctness and sharpness and the degree of their isolation enhanced. Of course, one could hardly suppose that the use of such cadences could in itself develop an isolating style. But where both the tendency to use clausulae and the isolating style are present, each enhances the effect of the other and develops in the writer an ever-increasing facility which may eventually become almost automatic. The skill which Seneca developed in the use of clausulae is shown by the infrequency with which he disturbs the natural order of words at the close of his sentences. In this respect he stands in marked contrast with Cicero and Nepos. It is furthermore worthy of note that both the clausulae and the isolating sentence are characteristic of the Asiatic school.

The sentence structure of Seneca is closely paralleled by that of Cato the Elder, judging from both the de Agri Cultura and the fragments of his orations. His narrow field of consciousness and his clear and distinct apperceptive power make each concept stand out sharply.1 Some few subordinate clauses are found (temporal, conditional, purpose), but they are always brief and concise. The successive units of thought are, however, seldom compared, and the relations between them less frequently grasped than in Seneca. See the fragment of his oration against Thermus preserved in ab Gellius, 10, 3, 17, | ab-b1| ab | ab|ao|ao| ao b—ab |

cd

1 Incidentally it may be said that to this was doubtless due in large measure the fact that later grammarians were led by the study of his vocabulary to collect lists of synonyms from his works. See Isidor, Differentiarum sive de Proprietate Sermonum, ii, praefatio 5, 10 (M). The same tendency of mind led Cato to make his collection of ἀποφθέγματα καὶ γνωμολογίαι, an anthology of pithy sayings gathered from other writers. Many of the acts of Cato's public life reflect this same quality of mind.

ab) (ubi societas?) | a(b), etc. In some of his fragments, where the emotional tone is calmer, associative elements are found. Compare, for example, with the passage just quoted the sentence, scio solere plerisque hominibus in rebus secundis atque prolixis atque prosperis animum excellere, atque superbiam atque ferociam augescere atque crescere.1 The strong individuality reflected in Cato's sentence structure forces us to conclude, that, however much Roman editors may have modernized his phraseology, they did not materially alter the groundwork of the style in the de Agri Cultura.

A comparison of the sentence structure of Cato and Varro shows clearly how much in style depends upon the author's type of mind and how little may depend upon the subjectmatter. To the strongly marked isolating style of Cato, the equally strong analytic tendency seen in Varro's Res Rusticae stands in sharp contrast. Varro's field of consciousness is wide, and a multiplicity of minor elements enter into his sentences. When most Varronian we have sentences represented by formulae like the following:

The type is in complete harmony with the careful and systematic analysis to which Varro subjected the subjectmatter of his works.2 Sentences like the above are, of

1 Apud Gellium, 6, 3, 14.

2 Compare the elaborate analyses descending sometimes almost to formalism and schematism which a table of contents of his Antiquitates shows. Similar analytic treatment of his theme is shown in numerous passages in the de Lingua Latina, especially at the beginnings of the books and larger subdivisions of the matter, where he is dealing with the general aspects of his subject rather than with the details.

course, not found on every page of Varro, but in general the organization of the sentences must be considered as quite elaborate, especially in view of the fact that the Res Rusticae are written in the form of the Platonic dialog, which is much less favorable to such a treatment than is the Aristotelian.

Quintilian is deserving of special attention as a notable example of a medium apperceptive type. The units of thought are of moderate size. The relations of their elements to each other as well as the relations of successive units are clearly grasped. Associative additions are frequent, but seldom overrun the thought or crowd out other elements. The sentences of Quintilian accordingly often bear a resemblance to the narrative type. In the following passage it happens that the associative processes are little in evidence: Institutio, 3, 6, 81-83: sed non statim, quod esse manifestum est, etiam quid sit apparet. hoc quoque constituto novissima qualitas superest neque his exploratis aliud est ultra. his infinitae quaestiones, his finitae continentur: horum aliqua in demonstrativa, deliberativa, iudiciali materia utique tractatur: haec rursus iudiciales causas et rationali parte et legali continent: neque enim ulla iuris disceptatio nisi finitione, qualitate, coniectura potest explicari. . . . discant igitur ante omnia, quadripertitam in omnibus causis esse rationem, quam primam intueri debeat qui acturus sit. Formula:

ab a b c d ab|ab|ab|ab ab cd ef

са

ef

The discourse moves on in a calm and steady flow, perspicuous and for the most part carefully articulated. The sentences seldom pass beyond the average normal range of consciousness. Here we have, to use Quintilian's own words, his nuda praeceptorum traditio. This is Quintilian the man. In the proemiums to his different books we see traces of Quintilian the rhetorician. Note only the following:

I, proem. 5, ego cum existimem, etc.

See also 3, proem. 2-3. Quintilian shows good taste in employing comparatively few of these overloaded sentences.

There appears to be no Roman writer, at least before the third century A.D., who was addicted to the use of the interlocked form (Boucke's type, II, 3). Those who do employ long sentences either follow the types described above (pp. 39-40) or the strictly analytic type (II, 4). This is plainly due to the same general characteristic of the Roman mind, which made it possible for them to create the two most remarkable products of that nation, the Roman legal system and the organization of the Roman Catholic Church. However, we do find sporadic examples of the interlocked type. Velleius Paterculus and Pliny the Elder, for example, occasionally become entangled in their own web of thought when they attempt elaborate periods. A single sentence from Velleius Paterculus (2, 18, 1-3) will be sufficient to illustrate the lengths to which he could go in extravaganza. Even more marvellously constructed than this sentence is the one which fills sections 143 and 144 of the seventh book of Pliny's Natural History.

The Roman historians of the classical period are representatives of a style approaching very closely to Dr. Boucke's narrative form (II, 2). In Caesar, however, the associative elements do not play a very important part. He is somewhat akin to Quintilian. This is not true, however, of Book viii, written by Hirtius.

This rapid and rather sketchy review of the essential qualities of the sentence structure of five writers who possessed remarkably strong individualities will perhaps show in a general way what the method of study here followed would accomplish. Within the general field a number of subordinate questions arise. Some of these, both because of the inadequacy of our sources of information concerning the

lives of the writers and also because of the small part of their writings that has been preserved to us, can never be solved. Others have been briefly suggested or touched upon in the course of this paper. Do different schools of stylists show general differences in sentence structure? Did any general development of the type of sentences employed take place in the centuries between Plautus and Isidor? Such a development can be clearly traced in German literature. Does the type of sentence structure show any development of the mode of thought, in the case of those authors whose extant works extend over long periods of time? Does the test of sentence structure throw any light on the authenticity of writings of questioned authorship? Lastly, it should be the main object of one who deals with this subject to determine so far as possible whether and in how far the qualities of mind shown in an author's sentences can be correlated with those shown in his actions otherwise, thus bringing the style in relation to the man himself.1 This is, of course, difficult in the cases of writers concerning whose lives we know little. The great advantage, however, of the general method followed in this paper is that it is individualistic — that the style of each author is approached as a distinct and different problem. The method does not seek, as does the Nägelsbachian, to impose a group of artificial (rhetorical) forms upon a writer, but on the contrary recognizes an unlimited number of types which it would abstract from the writer's sentence structure.

1 Professor Fred Newton Scott, of the University of Michigan, has undertaken the study of the rhythms of Walt Whitman from much the same point of view, and has obtained interesting and valuable results.