I hasten to withdraw the note which I wrote in the Appendix to my edition of the Rudens (ed. Minor, 1901, p. 169).

It seems then that the term "Breves Breviantes" might well be replaced by some term which should suggest a more adequate explanation of the facts of morphology to which it is applied.

But what I am concerned with here is not morphology but Old Latin prosody, the most prominent feature of which is in current theory brought into connection with the above-mentioned morphological facts. It is supposed that the shortenings found in beně, malě, etc., find their counterparts in OL verse, which is said either to shorten long syllables following a short, or to "reflect the pronunciation of everyday life" (Lindsay), or to treat long syllables following a short as "metrically short." Take Skutsch's statement of the law:

An iambic sequence of syllables which has the stress (Ton) on the short syllable, or which is immediately followed by the stressed syllable, becomes pyrrhichic (wird pyrrhichisch).1

The last two words are, probably intentionally, ambiguous; "becomes pyrrhichic" may mean either "becomes a pyrrhic" or "becomes equivalent to a pyrrhic"; the latter might be identical with "counts as a pyrrhic"-which was all that the originator of the law meant when he used the word gelten in his formula.2 The word Ton in Skutsch's statement is also ambiguous, and intentionally so; it is selected, as the author says, so as to cover ictus (verse-stress) as well as accent (word-stress). Brix expressed the same thing more explicitly when he said that the ictus metricus is very frequently equivalent in verse to the word-accent. Lindsay, regarding the ictus metricus as inoperative in this connection, words the law as follows, and attempts (not very successfully in my opinion) to explain all the phenomena by an appeal to the word-accent or the sentence-accent:

After a short syllable an unaccented syllable, which was long by nature or (more frequently) by position, was pronounced half-long in ordinary

11891; in the Jahresbericht über die Fortschritte der romanischen Philologie 33; repeated in the author's Forschungen (Plautinisches und romanisches) 6f. See also Sommer Handbuch der lateinischen Laut- und Formenlehre (1902) 141, 161.

2C. F. W. Müller Plautinische Prosodie (1869) 85.

3 Einleitung to the Trinummus, 1st ed. (1864), p. 14; 5th ed. (1907, by Niemeyer), p. 16.

speech, and scanned by the dramatic poets either long or short, when the accent fell on the following syllable or on the preceding (short) syllable.'

There is also another important difference between Lindsay and his immediate predecessors (Müller, Klotz, and Skutsch). They say nothing about these shortenings being based on the pronunciation of ordinary life; Müller's phrase "count as short" seems, indeed, to imply something different from actual shortening; and Klotz went so far as to declare the shortening to be a purely metrical phenomenon: thus he called the law a metrisches Kürzungsgesetz and said explicitly that the syllables which were treated as short by the OL dramatists were long in actual speech-a doctrine which in this form has found few supporters; for "metrical shortening" is only another name for conventional shortening, and verse so written is verse for the eye, not for the ear. Skutsch adopts an intermediate position. While, on the one hand, he denies (see below, p. 7) that these shortenings reflect the pronunciation of ordinary speech, he maintains that they are analogous to certain well-known phenomena of actual speech (beně, malě, etc.) and in accordance with the physiology of sound, i.e., that the effect of the ictus metricus in cases like turbines is essentially similar in character to the effect of the accent in cases like béně, though the former was never heard in ordinary speech; for the ictus metricus, like the word-accent, was an expiratory stress, and where the two come into conflict he holds that the accent must give way in order that the rhythm of the verse may not be sacrificed.

But, different as these statements are in some respects, they have one element in common-the doctrine that the syllables in question were in some sense shortened in OL verse. The very names Iamben-Kürzungsgesetz, Breves Breviantes, admit of no other interpretation. I admit that the phrases "counting as short," "scanned as short," and "metrically short," strictly interpreted, only mean that these syllables, though really long, were treated as if they were short. But Skutsch and his followers did not mean, when they spoke of iambic shortening, to imply that no real shortening takes place; and Lindsay definitely commits himself to the statement that the syllables in question were pronounced half-long (i.e., half-short) in 1Jahresbericht über Plautus (1907) 171; cf. Intr. to his edition of the Captivi 30.

ordinary speech; nor can he mean that the pronunciation in verse was different; for he tells us elsewhere that "Plautus scanned as he pronounced." Now it seems not to be generally realized that this doctrine of shortening involves phonetic difficulties of the gravest kind. No difficulty of pronunciation is involved in shortening the final syllable of words like abī, malō, bonās, manūs, or the second syllable of verēbamini, vidēbatur, or the first syllable of (bene) ēvenisse, (id) ēventurum, etc. But appreciably to shorten a syllable which is long by position (i.e., closed by a consonant and followed by another consonant) is in some cases physically impossible except by dropping one or more of the consonants.' Are we then to suppose that the word apstulisti was pronounced a-stulisti in ordinary speech and in the Aulularia of Plautus, 1. 645 (Quid apstulisti hinc? beginning a trochaic septenarius)? In some cases, no doubt, a different division of syllables would get over the difficulty: a-psurde, Ale-xander, voluptatem, senectutem, o-mnes, etc., are conceivable, though hardly plausible. But a-pstulisti is unpronounceable, and so is e-xprobras. The only way of shortening the first syllable of this word in Trin. 318-quid exprobras?—is to turn the x into an s and attach it to the next syllable (e-sprobras), neither of which measures exactly commends itself: the word has its ordinary quantities in Most. 300— quor exprobras? These are not isolated examples: we have the same problem to face in connection with the supposed shortening of the first syllable of a host of words like inclamare, indignus, indiligenter, invocatus, argentam, hercle, and of the second syllable of words

1This may sound like an overstatement in view of the fact that the law of Breves Breviantes is accepted by many scholars who are eminent phoneticians; but it is confirmed by the great authority of Mr. Henry Sweet. In reply to a question addressed to him by the present writer as to whether he considers it physically possible to shorten fenestra, uoluptatem, abstuli, supellex, he says "I cannot conceive of any quantitative prosody making such groups as str, bst, ll short without omissions: pt could be made short by pronouncing the two consonants simultaneously as in the English exactly." And again "Such a group as str could not possibly be made or felt to be short even under the weakest accent-not in any language. It could not even be regarded as half-short." If anyone doubts whether bst takes time to pronounce let him articulate the German word 'pst' (=hush).-These difficulties, or difficulties like them, seem to have been felt by Sommer, who in a note on p. 142 of his Handbuch der lat. Laut- und Formenlehre says "The phonological explanation (lautphysiologische Ratio) of this is not yet clear. In itself a syllable consisting of a vowel + consonant can never be short in pronunciation." The name Abdulhamid has its first syllable long whether it be pronounced Abd-ul-hamid or Ab-dul-hamid.

like quadringenti (which Lindsay would write quadrigenti), tabernaculum. The necessity of dropping a consonant in order to produce shortening is especially clear where there is a doubled consonant-unpronounceable at the beginning of a syllable; e.g., accepisti, ecce, occulto, supellectilis, annona: yet it is generally supposed to be a characteristic of Latin, as of modern Italian, that both the consonants were pronounced in such cases. Nor is this the end of the impasse into which we are led by a strictly phonetic interpretation of the doctrine of Breves Breviantes. For there are numerous examples in Plautus of words in which a syllable closed by a consonant and also containing a naturally long vowel has, according to this law, to be somehow shortened: e.g., insidiae. Here, even if the n could be taken over to the next syllable (a phonetic impossibility) we should still be left with a long i. How the syllable incan have been short or even half-short in ordinary speech passes my comprehension, especially as there are plenty of instances in Plautus in which the first syllable of this word is clearly long (e.g., Curc. 25). Yet we are asked to believe that in two successive lines of the Pseudolus (anapaestic meter) Plautus scanned and pronounced the first syllables of ignobilis and insidias short: Pseud. 592 f. Other cases of this kind are ignavus (Ter. Eun. 777), īgnorabitur (Plaut. Men. 468), înfimatis (Stich. 493), înfuscabat (Cist. 19), ōrnatu (Trin. 8406). Even more conclusive, as not needing any ancient testimony as to their quantity, are cases like vides quae sim (Most. 199), iuben mi ire (Amph. 929). These syllables in this position were undoubtedly long (at least as long as a syllable containing a final long vowel, e.g., amō) in ordinary speech. This was fully recognized by Skutsch on one occasion; he called it a "prinzipielle Unrichtigkeit" of Anton Marx that he took the "shortenings" of Plautus and Terence as evidence of the quantities of the vowels in ordinary speech. Yet Skutsch still believes in real "shortening" in OL verse, ascribing it to conditions analogous to, though not identical with, those prevailing in ordinary speech (see above, p. 5) Apparently he does not realize the phonological difficulties to which I

1 Berliner Philologische Wochenschrift XXII (1902), 1238: review of Anton Marx's Hülfsbüchlein für die Aussprache der lat. Vokale in positionslangen Silben, 3d ed.,

1901.

have alluded above. It is easy to say that the syllables in question were shortened, but difficult to suggest how they were shortened or to conceive of the form which the words would assume result of such a process.

as the

In comparison with the difficulty just discussed all other difficulties involved in the doctrine of Breves Breviantes shrink into insignificance. Yet it must be added that the theory of shortening is exposed to the old criticism of Bentley-"Mutantur tantum rei; crimen ipsum non eluitur: aut hi aut illi sunt culpae damnandi" (Schediasma 12 f.). This might be otherwise worded nowadays, in view of Lindsay's appeal to a "half-short" pronunciation in ordinary speech. But there is nevertheless a difficulty. Apart from the fact that the syllables in question are very often to be scanned as long in OL verse (e.g., Capt. 452 abī, 843 iubē; Men. 215 venī; Bacch. 777 tacēs; Rud. 442 and Merc. 359 voluptatem; Trin. 3 adest), what induced the dactylic poets to go back to the long pronunciation? The answer ordinarily given-that the usage of Ennius and his followers was of the nature of an artificial reaction against the popular pronunciation of their times-does not satisfy me, for various reasons. For one thing, poetry made up of a succession of words pronounced in a manner corresponding to the pronunciation of the English "wind" in verse1 would surely have been an offense to the ears of Roman readers. Besides, Latin was not so rich in short syllables and poor in long ones that the poets would have deliberately chosen to make syllables long which they might have made short. On the contrary they adopted various devices in order to secure short syllables."

There still remains to be considered the evidence of artistic prose. There is no trace of the operation of a law of Breves Breviantes in the clausulae of Cicero's Orations, so far as I have been able to discover from Zielinski. On the contrary the Clauselgesetz postulates a long syllable in such cases. Thus the words mihi, tibi, sibi, ibi, ubi, have the second syllable long (Zielinski,3 p. 183); so too the adverbial modo, and the words ego, lego (ibid.). And among 1Cf. Lindsay Intr. to Captivi 12.

2 See Bednara "Aus der Werkstatt der dactylischen Dichter" (Archiv für lat. Lexicogr. und Gram. XV, 223-32).

$ Das Clauselgesetz in Cicero's Reden (1904).