t(u)om vídi; Hec. 516 viró m(e)o respondébo.

3

As has

already been stated (p. 192), the possessive usually precedes pater, and there can be little doubt that in the nom. sing., when a pyrrhic pronoun is prefixed, méu(s) pater was usually pronounced under one accent, according to the preferred rhythm (rhythmic grouping) úv, ~— (A.J.P. XXV, 160, 260, n. 1; cf. also Lindsay, Captivi, 369, and the frequent mé(um) erum, etc.). In the case of the iambic possessive, however, an accent meúm patrem arises much less easily, and it is not clear that any regular word-order exists in the case of the possessive;1 the assumption of a recessive accent, however, is wholly unnecessary, since syncope is produced equally by the original accent meum pátrem. It is quite probable that the accent recedes freely upon quasi-monosyllabic m(e)um, but, in any case, the verse-treatment is m(e)úm-patrém, never meum pátr(em). Thus the dramatists have m(e)úm patrém, m(e)ó patri, m(e)í-patrís, m(e)ó-patré with synizesis thirty

1 Nor is patrem meum a traditional word-order in O. Lat. ; hence I agree thoroughly with Wallstedt, who in his good article, Zur Betonung des Possessivums, Lund, 1906, concludes from the usual avoidance of pătrém-meum, mănúmea, etc., in verse-closes that such phrases were not regularly accented as single words; for a similar hint, cf. Trans. Am. Phil. Assoc. XXXV, 49. (I may add that Wallstedt (p. 27 ff.) evidently accepts all the results of my own special articles, although the criticism made on p. 28 shows that in part he misunderstands my actual views; thus I do not deny the existence of an actual but less usual accent quod fácis, etc., cf. A.J.P. XXV, 411 ff.; for coalescence in pronunciation, he should see ib. 160 ff.) As further confirming Wallstedt's conclu

2

3

sions, I should like to point out that spondaic words, followed by the possessive, are not allowed in the inner feet or in the critical feet of iambic verse, i.e. we do not find fratrém-meum nor fratrém-meum, as we should most certainly do, if meum were regularly enclitic. Especially may one object to the word-group voluptás-mea, which is so often confidently assumed (Klotz, Grundz. 92; Lindsay, Capt. 367); this accent could only arise from a usual word-order, but the usual order with the vocative case, in fact the invariable prose order, is mea voluptas (cf. mi pater, mi patrone, etc.), which Pl. retains 12 times, using volup tas mea 7 times only for metrical convenience (cf. Ferger, De vocativi usu Pl., Strassburg, 1889, p. 17). For the well-known rule of order with the vocative, see Ferger, .., 18, 14 (e.g. "pronomina, si per metrum licet, ante vocativum ponuntur, neque nisi metro repugnante vocativus praecedit "), and Nilsson, I., 12, 34. In fact, voluptás, molestae, with accent on the ultima were probably not much more displeasing to a Roman ear than rēgnúm; much harsher seems the rare accentuation of the initial syllable, as in dédisse, bibisti.

three times, měúm patrém without synizesis only once (Ba. 380). This thoroughgoing synizesis is by no means wholly due to the avoidance of the double iambus; for, in the order patrém meúm, patrí meó, they employ the diiambus five times (Men. 747, 750; Tri. 280; St. 10; Cap. 1012), and the treatment with elision, e.g. patrí me(o)≤, twice (Cap. 318, 377). Phrases like měúm-patrém do not, however, occur in versecloses, since they almost never occur without synizesis in any case, and also because the group-accent m(e)úm-patrem arises only through synizesis; cf., however, Naev. tr. fr. 13 quin měá-manú | Moriáre, where the correction of R3 (mea móriarís manú) seems unnecessary. Similarly the dramatists have m(e)ám-fidem six times and m(e)úm-virum nine times, always with synizesis; in the opposite order the diiambus virúm suóm occurs perhaps St. 284. The influence of metrical convenience or of a preferred rhythm in giving value to the weak vowel is well seen in meàm-sententiam, which shows synizesis six times, non-synizesis twice, and in adměàm-sententiam, which retains its full form twice (Au. 383; Poe. 1126).

5

6

My conclusion is that thoroughgoing synizesis is certain in meo, suo, fui, deo, diu, and occasional synizesis in scio, die ; and I accept fully the views of Götz and Schöll (ed. min. fascic. II, vi): "non méo, súð sim. probamus, quod non una re firmari putamus," and of Corssen, Ausspr. II2, 750: "Die Meinungsäusserungen von C. Müller über die Synizese stehen im Widerspruch zu den Lehren der Grammatiker, zu dem Zeugniss der lateinischen Schrift, zu den Lautgesetzen der lateinischen Sprache." (Cf. Leppermann, .., 81.)2 l.l.,

1

1 I omit entirely Trag. fr. inc. inc. 216 (sŭúm patrém), because of its uncertain date. The citations for synizesis are as follows: M(e) um patrem: Cap. 238, 1024, 1032; Ep. 349, 374; Men. 736; Mer. 787; Mo. 979; Tri. 1178; Phor. 874; Pacuv. tr. fr. 139. — M(e)ó patri: Am. 144; Ba. 685, 731, 734; Cap. prol. 21, 237, 588, 923, 979, 987; Mer. 80, 631, 954; Hau. 259; Hec. 820, 865. — M(e){ patris (final s making position): Am. 31; (Mo. 1125); Eu 1048; Phor. 788. — M(e)8 patre: Ba. 931; Men. 1079; Hau. 823.

2 It will be observed that it is quite possible to treat Plautine synizesis, as I have in fact treated it throughout this whole section, in entire dependence upon the teachings of the metrici respecting the value of syllables, since they have, in

Later Usage. — The use of synizesis had been very extended in O. Lat., but a complete change of attitude took place in the later literary language. A more precise system of pronunciation was accepted both by writers of verse and of rhythmical prose, and a consistent body of theoretical principles was formulated. Hence the Augustan poets sought to reduce both shortening and slurring within the narrowest limits possible, and to confine these processes to a limited circle of words. But of the two processes they naturally viewed with much greater toleration that which involved the actual shortening of syllables; as refined and tasteful artists, they could not but regard the frequent slurring of syllables. as vulgar and incorrect. Hence while they admitted (at least theoretically) the consonantization of the vowels i(e) and u, they rejected almost entirely the slurring of these sounds, and preferred, through a strengthened pronunciation of the first vowel, to introduce scio, nescio-quis (Cat. 6, 4; Hor. C. iii, 24, 64), drỡ, diuturnus, diutius, etc., in place of O. Lat. sc(i)ō, nesc(i)ōquis,1 dvō or d(u)ō (cf. Gk. duwdeкa and dwdeka),2 diutius (cf. d(i)ūdum); cf. also the introduction of the forms děi, děis in this period (Sturtevant, .., 21), and the restoration of nihil, prehendo, hercule, mehercule in place of O. Lat. nil, prendo, hercle, mehcrcle. Moreover, the poets of the empire, such as Seneca, Martial, Juvenal, and Statius, followed the guidance of analogy and the trend of careful pro

fact, provided for exceptionally short syllables through the prosodical figure Synizesis. For this reason some of the preliminary remarks contained in the first section of this paper seem to me now somewhat unnecessary, although the ancient metricians may be justly criticised for not distinguishing more clearly between the natural and the artificial forms of synizesis.

1 But see above, p. 174, n. 2 (end); to examples of the restored vowel should be added the frequent duellum of Horace (C. iii, 5, 38; 14, 18, etc.).

2 According to Studemund, A.L.L. III, 550 f., duō is very nearly duō in the dramatists, that is, in the nom. masc., where it is the sole form, Pl. allows it to end an iambic line (duo), but in the acc. masc., where the form duos also exists, he treats it as a monosyllable (dvo) and uses the form duōs instead at the end of a line. From this use it is probable that duo was more nearly one syllable than two in Plautus's time; cf. also Lindsay, L.L. 411. There are two exceptions to the rule, i.e. two cases of acc. duo in verse-closes: Ep. 187; Ps. 1000. An original duo, which would not admit synizesis, is improbable; cf. late duo.

nunciation, as it existed throughout this entire period, when they resolved the synizesis-diphthong ui in cui, cuicumque (earlier quoii, quoiicumque), and huic into cut, cuicumque, and huic; see examples in L. Müller, R.M.2, 318 f.; Neue, II3, 454. Hence, after the period of Lucilius, Lucretius, and Varro (cf. L. Müller, .., 546), we find only those forms freely slurred by the dactylic poets which present especial difficulty in hexameter verse, e.g. eidem (dat. sing.), eodem, eaedem, eosdem,1 and of other slurred forms we find only a few isolated examples, viz. dat. sing. (e)i regularly in Cicero's clausulae (Zielinski, Clauselgesetz, 176),2 and once in Catullus (82, 3), (e)is (Sen. Troad. 191), s(u)āpte (id. Agam. 250), (i)ō (Cat. 61, 124 ff.; Mart. xi, 2, 5),3 v(i)ētis (Hor. Epod. 12, 7,- dactylic), (I)ule (id. C. iv, 2, 2-sapphic; cf. L. Müller, .l., 307), S(u)ēvo (Prop. iv, 2 (3), 45), etc. Similarly, there is reason to believe that the more dignified poets entertained some prejudice against the forms of the pronoun is, on account of their frequent slurring, and although they accepted fully the contract forms di, dis, idem (nom. pl.), isdem, they seem to have especially avoided the nom. and abl. pl. of is. According to Meader-Wölfflin, A.L.L. XI, 373, dissyllabic ei and eis are represented in the poets only by the group-form in-ĕis, Manil. ii, 744,5 and whether monosyllabic i and is were allowed at all, is a question still under discussion (Meader, Latin Pronouns, 23).

In the remains of popular poetry, however, and in later poets like Terentianus Maurus and Ausonius (L. Müller, 322), colloquial synizesis forms occur much more freely, and it is evident from late inscriptional forms like so, tis, quescas, etc., that they were always retained in vulgar Latin, although the distinction between consonant and vowel i and u probably

1 All the examples are collected by Skutsch, Tépas, 148.

2 Dat. sing. ei was almost as much a monosyllable in Priscian's time as huic and cui; cf. Keil, III, 10, 2 ff.

3 Cf. Munro, Criticisms and Elucidations of Catullus, Cambridge, 1878, 135 ff. 4 Cf. L. Müller, l.l., 297.

5 Similarly trisyllabic eidem and eisdem are represented only by the rhythmical groups se1d-ĕidem, Pl. Mi. 758 (Sturtevant, .., 25), and dăt-čisdem, Juv. xiv, 30 (L. Müller, .., 297).

became more marked as time went on.1 Especially well known is the late Latin tendency for di, when followed by a vowel, to assume the spirant sound of y (Lindsay, L.L. 49, 84), and to be written at times z or d, e.g. des CIL. V, 6244; 2(es), ib., 1667; do, dae (Schuchardt, Vok. des Vulgärlat. II, 463; III, 289; Bücheler, Lex. It. vii; Seelmann, Ausspr. d. Lat. 187); cf. Oscan zicolom for *dieculom. The beginnings of this tendency, especially in hiatus before a long vowel, may be recognized in a few O. Lat. words, e.g. Iūturna from Diuturna, Iovem from O. Lat. Diovem, and, to a limited extent, also, in the Plautine scansions d(e)o, d(e)ae, d(i)e,3 d(i)erectus, D(i)espiter, d(i)u, d(i)utinus, d(i)utius, d(i)udum; cf. also the scansions dvellum, d(u)o, d(u)odecim, d(u)im, and the tendency of du in some of these forms to pass into d, and later, into b (Stolz, Müller's Handb. II3, 2, 82).

III. ADDITIONAL PRINCIPLES OF SYNIZESIS.

In the preceding section the vague term synizesis has been more exactly defined through the law of Brevis Coalescens. It remains to state very briefly, as the limits of the present article require, a few additional facts which are simple corollaries of this law, but which, in several cases, have a bearing upon the general subject of classical synizesis and the limitations to which it is subject.

(1) Two Short Vowels Preserved Intact. Word-forms in which two short vowels come together, e.g. měŭ(s), měă, făĕrit, are not subject to synizesis, except in those cases where the second short vowel is elided before a following long vowel, and an iambic sequence results, as in Mo. 98 m(e)ă, haud áliter. Hence it is scarcely necessary to discuss seriously the assumption of wholly unnecessary and gratuitous syni

1

1 Cf. L. Müller, R.M.2, 308: "Contra in illis, quae sunt tuus, suus, cum initio valde fuisset exili sono littera eadem, corroboratast sensim."

2 Cf. Bull. d. ist. arch. 1871, 136 f. The two forms, however (like Diāna and Iāna, Diānus and Iānus), are perhaps connected only by popular etymology; cf. Stolz, Hist. Gramm. 305. For the treatment of di in hiatus in the Pelignian dialect, cf. Lindsay, L.L. 49.

3 Cf. dibus, CIL. VI, 25540, for diebus.