(j) Verses arranged symmetrically, as regards the number of syllables in the words and the order of the words, are indicated in part on p. 286. Other examples are:

votiva paries indicat uvida (3 syl., 3 syl., 3 syl., 3 syl.), (I, 5, 14). est qui nec veteris pocula Massici (1 syl., 1 syl., 1 syl., 3 syl., 3 syl., 3 syl.), (I, 1, 19).

prudens Oceano dissociabili (2 syl., 4 syl., 6 syl.), (I, 3, 22). quam lentis penitus macerer ignibus (1 syl., 2 syl., 3 syl., 3 syl., 3 syl.), (I, 13, 8).

certat tergeminis tollere honoribus (2 syl., 4 syl., 2 syl., 4 syl.), (I, 1, 8).

quassas, indocilis pauperiem pati (2 syl., 4 syl., 4 syl., 2 syl.), (I, 1, 18).

Censorine meis aera sodalibus (4 syl., 2 syl., 2 syl., 4 syl.), (IV, 8, 2).

sollers nunc hominem ponere nunc deum (2 syl., 1 syl., 3 syl., 3 syl., 1 syl., 2 syl.), (IV, 8, 8).

VI. Other Sound-Effects.

A. Inter-verse hiatus. When one verse ends with a vowel or m and the next verse begins with a vowel or h, there results what may be termed inter-verse hiatus. Although Horace seems in general to have avoided this sound-effect in his best lyrical work, still 36 times it occurs between a Lesser Asclepiad and a following verse, affecting

5 per cent of the Lesser Asclepiads in I, I.

8 per cent of the Lesser Asclepiads in III, 9.

III, 10. "III, 13. III,

" III, 16.

15.

" III, 19.

" III, 24.

པ

39

66

66

་

In a general way the poems having much inter-verse hiatus show signs of early workmanship, carelessness, or want of recent practise in writing lyric poetry. Horace indirectly states the last named fact in IV, 1 and inter-verse hiatus is there abundant. Much inter-verse hiatus marks I, 3, which is one reason for assigning it to an early period. In the same direction points perrupit (v. 36) with the last syllable long by diastole-a trait that is wanting in his later work. Where much inter-verse hiatus appears we generally find remarks like the following among the commentators: I, 15-"In this perhaps youthful experiment Horace attempts, as Quintilian says of Stesichorus, to support the weight of an epic theme on the lyre." I, 21-"The poem may be a sketch for a carmen seculare." In fact, the metrical art of a Greek or Roman poet sometimes undergoes such orderly and systematic development, that having plotted the curve of his growth, so to speak, we are able to locate chronologically a selection from his poetry simply by noting the characteristics of its form. III, 15 for example shows so much inter-verse hiatus that one is led to suspect the poem is an early effort. This view receives confirmation in Kiessling, who from another point of view says it "gehört mit I, 25 und IV, 13, sowie den Epoden 5 und 8 zusammen, den der alexandrinischen Dictung geläufigen Typus der alten noch immer mannstollen Vettel zu zeichnen."

B. The question arises, why the second syllable of a Lesser

Asclepiad is irrational. The answer may be in part as follows. The part of a word used to fill this place was already familiar to the ancient in its long form. When he heard this sound compressed into the time of a short standing between two long syllables at the outset of the verse, a peculiar musical effect-rich and full-was produced. A trochee beginning the verse would have given quite a different effect.

UNIVERSITY OF CALIFORNIA.

LEON J. RICHARDSON.

V. THE UNREAL CONDITIONAL SENTENCE IN

PLAUTUS.
I.

THE USE OF THE IMPERFECT SUBJUNCTIVE FOR THE
PRESENT UNREAL.

It is assumed with reason that the present subjunctive was the main, if not exclusive, expression for the present unreal sentence at some time in the pre-literary period of the Latin language, and that the imperfect subjunctive, at this early period, had its normal past-tense force, and shared with the pluperfect subjunctive the province of the past unreal conditional sentence.' This view is supported by the actually existing state of affairs in Homeric Greek; for there the present optative is used for the present unreal, while the past unreal finds expression in the imperfect and aorist indicative (with sporadic cases of the optative). Goodwin' denies that the imperfect indicative has yet begun to take on the function of the present unreal condition, which is its province at a later period. Further evidence looking in the same direction may be found in the development of the idiom in Latin. After Plautus, the present subjunctive rapidly drops out of use as the expression of the present unreal condition, and is replaced by the imperfect subjunctive, which, in turn, gives up the hold that in the early writers, it still has on the past unreal condition. This development seems to imply that Plautus is in a stage of transition-that, before his time, the present subjunctive was used more, and the imperfect less, as the expression for the present unreal conditional sentence. Finally, the Homeric use of the present optative and the Sanskrit use of both that mode and the subjunctive' for this type of sentence would seem to indicate that the present subjunctive was the most natural speech-form at the command of the early Roman for the expression of the present unreal condition.

1Cf. H. Blase, Geschichte des Irrealis (Erlangen, 1888), p. 1. 'Greek Moods and Tenses, §435.

3 Goodwin, Greek Moods and Tenses, $438. Whitney, Sanskrit Grammar, §581, and b.

The change of function on the part of the imperfect subjunctive in Latin, and of the imperfect indicative in Greek, presents an interesting and perplexing problem, and one for which, I think, no final answer has yet been proposed.' At least a part of the trouble with the solutions that have been advanced is the neglect of one or more of the following indispensable conditions of a satisfactory theory:

(1). It is absolutely essential to clear thinking on this subject that the grammatical and the psychological aspects of the question be sharply distinguished in the mind of the investigator. It is one thing to determine when and how men came to think in the unreal form, and quite another to explain the history of the form of words used in the expression of that class of conditional thought. As I have elsewhere shown, the Romans were thinking their unreal conditions clearly enough, and were able to make the hearer catch the meaning even when the present subjunctive was used, long before the imperfect subjunctive was settled upon as the exclusive speech-form of the present unreal condition. Aside from the proof there given, the same thing may be assumed on general principles, for the adaptation of these past tenseforms to this type of conditional sentence is a late process in language, and we can hardly assume that thought was crude and undeveloped at the time the change took place. It is probable then that, in attempting to answer the question under discussion, we should think of the present and past unreal thoughtforms as fixed, and of the imperfect subjunctive as leaving to the pluperfect the old function which it had shared with it, and passing over to the expression of the present unreal conditional sentence. Any theory that attempts to explain the new use of this mode and tense by a concomitant evolution of the unreal thought categories, stands upon a very unstable base, for Latin at any

rate.

(2). In proposing a theory to explain the change in the use of the imperfect subjunctive in Latin, some attention must be paid to the similar change found in Greek and in the Germanic

Cf. Blase, Geschichte des Irrealis; Greenough, Harvard Studies, Vol. VII; E. H. Miles, Comparative Syntax of Greek and Latin (Cambridge, 1893), additional note, p. cxxvii. A summary of other views may be found in an article on the Spanish conditional sentence by E. Gessner, Zeit. für Roman. Phil., xiv. p. 23 ff.

Class. Rev., xv., p. 51.