^{page 66 note 2} Observations de Graecae linguae dialectis, 1870, §9 (in Opuscula, i, 1827, p. 136): ‘Patet vero, vicissim e diligentia poetae vel negligentia aetatem fabulae elucescere’.

^{page 66 note 3} Eur. Supplices, 1811, p. iv: Elementa Doctrinae Metricae, Leipzig, 1816, p. 115Google Scholar; Glasgow, , 1817, p. 79: Epitome Doctr. Metr., 1818, p. 59Google Scholar; 2nd ed. 1844, p. 54: Eur. Bacchae, 1823, p. xxxix (also pp. xli and xlii).

^{page 66 note 4} Eur. Medea, Oxford, 1818, p. 70Google Scholar, n. h; Leipzig, 1822, p. 55, n. h; ‘melius de ea re judicare possemus, si pauciores essent Euripidis tragoediae, quarum aetas nobis prorsus ignota esset’.

^{page 66 note 5} ‘Notice of Mr. Elmsley's Edition of the Medea of Euripides’, Classical Journal, vol. xix (no. 38), 1819, pp. 271–2Google Scholar, reprinted under title of ‘Annotationes Hennanni’ in Elmsley, Medea, Leipzig, 1822, p. 329, and in Elmsley, Heraclidae et Medea, 2nd ed., 1828, p. 486: ‘nobis quidem minime dubia videtur haec ratio tempora tragoediarum ex scribendi incuria constituendi’.

^{page 66 note 6} If he had done so, he would hardly have made the error of stating, with regard to the Ion, ‘quantum ex numeris versuum colligi potest, scripta haec fabula nee post Olymp. LXXXIX [424–421] nee multo prius’ (Ion, 1827, p. xxxii).

^{page 66 note 7} De chronologia fabularum Euripidearum disp., Marburg, 1839Google Scholar. Zirndorfer's figures are far from correct, especially in the case of the Hec.

^{page 66 note 8} De pedibus solutis in dialogorum senariis Aeschyli, Sophoclis, Euripidis, Berlin, 1866, p. 12Google Scholar, ‘ne indicium illud nimis sit fallax atque haud raro deceperit Zirndorferum, vereor quam maxime’; cf. p. 43. Müller's totals, given in a footnote to p. 44, are commendably accurate.

^{page 66 note 9} ‘Die Auflösungen im Trimeter des Euripides’, Philologus, xxiv (1866), pp. 407–21Google Scholar. His long lists of references are fairly complete, but need to be analysed and tabulated before being of any practical value. As Descroix, J. (Le Trimètre iambique, p. 109)Google Scholar remarks, ‘les pourcentages ne sont pas marqués qui seuls permettraient une comparaison facile et efficace’. Such a tabulation of Rumpel's lists was in fact provided by Tanner, R. H. in ‘The Οδνσσς of Cratinus and the Cyclops of Euripides’, Trans. Proc. Amer. Philol. Assoc. xlvi (1915), p. 185Google Scholar, n. 34; in it a few of Rumpel's minor errors are corrected (p. 186, n.). Other faults in Rumpel's work are criticized by Descroix, pp. 109, 120–1, 172.

^{page 66 note 10} ‘Die Abfassungszeit der Andromache des Euripides’, Hermes, xviii (1883), p. 496Google Scholar, n. 2: ‘dieses formale Kriterium ist ein wohl zu beach tendes Moment, auf dessen Bedeutung schon lange vorher G. Hermann hingewiesen hatte, aber durchaus kein untrüglicher Massstab. Noch weniger durfte Zirndorfer mit diesem metrischen Kanon die dramatische Composition der euripideischen Tragödien unmittelbar in Verbindung bringen.’

^{page 67 note 1} Herakles, i² (1895), p. 143, cf. i¹ (1889), p. 348.

^{page 67 note 2} The Tragic Drama of the Greeks, 1896, p. 283.

^{page 67 note 3} ‘Sophocles, Statistics, and the “Trachiniae”’, Amer. Journ. Philol., lx (1939), p. 178Google Scholar, n. 3.

^{page 67 note 4} The great change between Hippol. and Andr. (see below) must have been entirely deliberate (whether or not lost plays intervened): the subsequent general tendency for the resolutions to increase was probably due in part to conscious intention, in part to unconscious development of style. But in most other respects the development seems to have been usually subconscious.

^{page 67 note 5} It would of course be quite incorrect to claim that nowhere in the plays do passages occur where Euripides consciously varied the frequency of resolutions to express special emotions or to create special effects: Descroix (pp. 236–8) cites several possible examples, some of them in the Bacchae: ‘s’il est question de Dionysos, de son culte, des Ménades, le rythme devient agité et comme secoué par le délire bachique; les séries de brèves se pressent. Mais la situation se fait pathétique, les résolutions disparaissent….’ There are without doubt certain other similar instances: but on the whole they appear to be relatively few in number, and to make little or no difference to the figures for the cornconscious plete plays.

^{page 68 note 1} The percentages used by Haigh (p. 283, n. 1) were calculated (without direct acknowledgement) from Zirndorfer's totals, and are consequently quite unreliable; the figure for the Hec. (cf. also p. 295, n. 1) is exceptionally inaccurate.

^{page 68 note 2} “The Chronology of the Dramas of Euripides’, C.R. xiv (1900), p. 438Google Scholar. The smallness of the resolution-frequencies attributed by Church to some plays shows that a very large number of trisyllabic feet must have been entirely overlooked. The figures are in consequence extremely misleading: e.g. in the list of increasing resolutions, the positions of Med. and Alc., Hec. and Andr., Phoen. and Ion are all inverted out of their true order, while Heracl. and Herc. are quite out of place. It is unfortunate that Rose, H. J., in his Handbook of Greek Literature, 1934, (see p. 184Google Scholar, n. 17) accepted such an inaccurate set of statistics.

^{page 68 note 3} ‘Verse–Weight’, C.Q. viii (1914), pp. 209–10Google Scholar; ‘Aristophanes, Frogs, 1203’, C.R. xxxvii (1923), p. 13Google Scholar, cf. Proc. Cambr. Philol. Soc. xciv–xcvi (1913), p. 18Google Scholar.

^{page 68 note 4} Firnhaber, C. G., Iphigenia in Aulis, 1841, pp. 294–5Google Scholar, gave, for both Bacch. and I.A., totals of resolved feet in each position and classified references to lines containing two or three resolutions. His figures are unreliable, as he himself admitted (p. 294, n.). Similar totals (fairly complete) were provided for Tro., I.T., Ion, and Hel. by Enthoven, L. (see below, p. 79Google Scholar, n. 6).

^{page 68 note 5} Tragodumenon Libri Tres, Cracow, 1925, Liber II, pp. 133–240Google Scholar.

^{page 68 note 6} Le Trimètre iambique des iambographes à la comédie nouvelle, Mâcon, 1931Google Scholar.

^{page 68 note 7} Except in the totals published by C. F. W. Müller and Enthoven, and in the inadequate percentages of A. Church.

^{page 68 note 8} All references to ‘proper names’, both here and below, should be understood as also including adjectives formed from proper names.

^{page 68 note 9} Except perhaps in some of the declined cases of πατρ, in ἱερóς, πόλεμος (and compounds), πότερον, θάνατος, etc., common words without suitable Greek equivalents or alternatives, which cause resolution whenever used: cf. p. 73, below.

^{page 68 note 10} It is true that, as a natural development parallel to the increase in frequency of ordinary resolutions, Euripides tended to use proper names which involve resolution more often in the later than in the earlier plays. Thus lines like 'Eλένην Mενέλεως ἳνα λάβῃ. καλν γνος (I.A. 1168) do not normally appear in his earlier composition. But in spite of this increasing tendency to repeat resolution-causing proper names frequently and sometimes even superfluously, the fact remains that his employment of them was never anything like free (in the sense that his employment of ordinary resolutions was free), since at all times he was clearly unable either to avoid them (beyond a certain degree) if the plot required them, or to repeat them if the plot did not require them. Consequently such resolutions should be considered in a class apart. (For a discussion of some of them see Harrison, in C.R. liv (1940), p. 154Google Scholar.)

^{page 68 note 11} Kitto (p. 184, n. 7) suggests, with reference to Sophocles, that ‘so evidently careful a craftsman would obviously avoid ordinary resolutions in proportion as he could not avoid the others’ [i.e. resolutions involving proper names]; but the same would hardly be applicable in the case of Euripides.

^{page 69 note 1} Harrison, p. 209, n. 1; Zieliński, pp. 144–5; Descroix, p. 171.

^{page 69 note 2} Reference to the last columns of Table I shows that the set of figures which includes trisyllabic feet resulting from proper names inverts the order of Med. and Heracl., Hec. and Suppl., Herc, and Tro., I.T. and Ion.

^{page 69 note 3} They were rightly left out of consideration by Yorke, E. C., ‘Trisyllabic Feet in the Dialogue of Aeschylus’, C.Q. xxx (1936), pp. 116–19Google Scholar, and by Kitto, op. cit., p. 184, n. 7.

^{page 69 note 4} Zieliński's defence of the inclusion in his statistics of proper names fails to answer any of the above objections (p. 154): ‘etsi enim concedendum est nominis proprii inevitabilem usum solvendi necessitatem poetis imposuisse, tamen cur haec solutio cum alia cumularetur nil fuit causae, nisi jam ita assuefactae erant aures solutionibus conduplicatis, ut eas sine molestia ferrent.’

^{page 69 note 5} Cf. Denniston, J. D., ‘Lyric Iambics in Greek Drama’, in Greek Poetry and Life, p. 127Google Scholar, where the different rules governing lyric trimeters are enumerated.

^{page 69 note 6} To define more precisely, a resolution is taken as invoving a proper name if the proper name occupies either one or two of the last two syllables in a dactyl or tribrach, or of the first two syallables in an anapaest: consequantly in such a line as γάμοις 'Iάσων βασιλικοῖς εὐνάζεται (Med. 18) the resolution is not regarded as invoving the name. (It may be remarked that there are three instances in the Cyclops (154, 558, 560) of an anapaest in which the only syllable to involve a proper name is one of the first two, but that no such anapaest is found in tragedy proper; there seems to be no instance at all of an anapaest in which only the two short syallables involve a proper name.)

^{page 69 note 7} Many of the discrepancies between previous sets of figures were largely due to divergent treatments of the trimeters (real and apparent) which occur in lyric surroundings: as the problem admits so much subjectivity, the only exact method is to give a full list of the lines that are (rightly or wrongly) included here as having been spoken amidst lyrics: Alc. 246–7, 251–2, 257–8, 264–5, 404–5; Med. 1271–2, 1277–8; Heracl. 77–9, 84–5, 88–9, 93–4, 97–100, 105–6; Hippol. 565–8, 570, 575–6, 581–2, 589–90, 834–5, 856–65, 871–6, 881; Andr. 828, 832, 836, 840, 845, 851–2, 1184–5, 1203, 1208, 1218, 1221; Hec. 688, 693, 698, 701, 709, 713, 1085–6, 1093; Suppl. 1009–11, 1073, 1076; Herc. 740–1, 747–8, 754–6, 760–2, 816–17, 819, 916, 1039–41, 1179, 1181, 1191, 1195, 1198, 1202; Ion, 771–5 (= 4 ll.), 778–81, 785–8, 792–5, 1443–4, 1450–1. 1456–7. 1462, 1468–9, 1473, 1477, 1485, 1488; Tro. 235–8, 240, 243, 246, 248, 251, 255, 259, 261, 264, 268, 270, 273, 277, 1218–25, 1232–4, 1240–50; El. 866–72, 1165, 1168, 1172–6; I.T. 646, 650, 831, 841, 850–1, 855, 863, 866; Hel. 625–6, 630–1, 646–7, 652–3, 660, 663, 665, 669, 672, 675, 679, 683, 688; Phoen. 106–8, 112–13, 117–18, 123–6, 131, 133–4, 138–40, 150, 154–5, 158–62, 170–4, 179–81, 193–201, 1342–4, 1347–9 Or. 1251–2, 1258–60, 1271–2, 1278–80, 1286–7, 1291–2, 1296–8, 1301, 1380, 1393, 1425, 1452, 1473, 1503–5, 1539–40, 1543–4; Bacch. 1024–7, 1029–30, 1032–3, 1039–40, 1165–7; I.A. 1336–7; Rhes. 704–5, 722–3, 736–7, 745–6, 890–4, 904–5. All other trimeters in the lyric sections are ignored.

^{page 70 note 1} In Cycl. 115 and 231, however, scansion by synizesis has been preferred to the assumption of an anapaest, by satyric licence, in the fourth and third feet respectively.

^{page 70 note 2} All figures are based on Murray's Oxford text. I.A. 1578–614 and 1621–6, which are bracketed, are omitted, as well as the fragments of the lost parts of the Heraclidae and Bacchae. Lines bracketed as corrupt are uniformly excluded: lines completely obelized are also excluded, except for Cycl. 395, 440; Med. 738, 741, 910; Heracl. 396–7, 785; Hippol. 468, 952, 1459; Suppl. 1089–91; Herc. 845, 1417; Ion, 2, 828; Tro. 862–3; El. 546, 929; Or. 497; I.A. 521, 1022–3. Lines only partly obelized are included, but with the following exceptions, in which the sense or metre seems unsatisfactory: Med. 1077; Heracl. 223, 513, 969; Suppl. 1101; Herc. 1003, 1159, 1304; Ion, 286, 602; Or. 1236; I.A. 971, 1185, 1207, 1573. Although deficient to the extent of one foot, Heracl. 710 and Ion, 1171 have been retained. The undermentioned common words have been counted as ordinary resolutions (they are printed in the Oxford text as proper names): Alc. 24, 844, 1141 θάνατον; Andr. 603 Φίλιον; Hec. 345 ίκέσιον; Herc. 615 ϰθονίας; Ion, 1130 γενέταιο, 1147 οὐρανός, 1149 ἤλιος; Tro. 769 θαντον; I.T. 1161 ὃσίᾳ; Hel. 570 ἒνοδίας; Phoen. 3 ἢλιε, 532 Φιλοτιμίας, 536 ἰσότητα, 542, ἰσότης; Bacch. 1026 ὂΦεος.

^{page 70 note 3} It has not been previously suggested that in this first period the proportion of trisyllabic feet underwent a small but definite decrease, although some writers have implied that it remained approximately static at this time (Descroix, p. 58, cf. Harrison, p. 210). In spite of Hermann's repeated statements (locc. citt.) that the increase in the frequency of resolution did not commence until about Olymp. 89 (424 B.C.), it has usually been assumed that the tendency to increase started from the very beginning of Euripides' dramatic composition: consequently the fact that (for instance) the Hippolytus (428) contains a smaller proportion than the Alcestis (438) has been sometimes taken, by those who make this incorrect assumption (e.g. by Owen, A. S., Euripides' Ion, 1939, p. xxxvi)Google Scholar as a proof that the evidence of resolutions for dating the plays is neither exact nor reliable.

^{page 71 note 1} Too much stress should not be laid upon the slight rise in frequency from Alc. to Med. The decrease was apparently so gradual that small irregularities are not surprising.

^{page 71 note 2} Haigh, p. 373, ‘Euripides, after beginning his career by the imitation of Sophocles,…’ (cf. p. 283): Descroix, p. 58, ‘dans ses quatre premières tragédies, Euripide se montre le disciple docile de ses illustres devanciers et se conforme à la tradition métrique’. Both these statements, however, are of too sweeping a character, and need qualification.

^{page 71 note 3} Zieliński (p. 140) calls the plays of the first period ‘severioris styli’; all the others he divides into three groups, ‘semiseveri styli’, ‘liberi styli’, and ‘liberrimi styli’. These groups, although perhaps not without usefulness as an indication of the changes in the poet's metrical style, are quite artificial, since his dividing-lines are the arbitrary figures of 15% and 30% respectively of resolutions (according to his statistics).

^{page 71 note 4} Some of them were noticed by Zieliński (pp. 144–6), Descroix (pp. 171 ff.).

^{page 71 note 5} Rhes. 804; Alc. 802.

^{page 71 note 6} In the Alcestis, however, third-foot dactyls form merely 32% of the resolutions. The smallness of this figure may just possibly be due to the play's pro-satyric nature (the satyr-play Cyclops contains only 23%).

^{page 71 note 7} Murray accepts μβαλο (Nauck) for ναβαλο (MSS.) in Alc. 526, but in Andr. 444 retains ναμένει (MSS.) where Nauck proposed μμένει. In Hec. 1281, however, he reads μμένει (l) against the ναμένει of the MSS.

^{page 72 note 1} Included in this total, but not shown in the columns of the table, are 21 anapaests admitted by satyric licence into feet other than the first, i.e. 8 (= 4%) in the second, 1 (= ·5%) in the third, 6 (= 3%) in the fourth, and 6 (= 3%) in the fifth: see p. 70, n. 1, above.

^{page 72 note 2} The term ‘internal evidence’, whenever here used, is of course exclusive of the evidence of resolutions in iambic trimeters.

^{page 72 note 3} The play's genuineness has been upheld by several scholars: cf. Vater, F., Vindiciae Rhesi Tragoediae, 1837Google Scholar; Paley, F. A., Eur. i, 1857, p. 7Google Scholar; Porter, W. H., Rhesus, 1916, pp. xxxv–liiGoogle Scholar, Hermathena, xvii (1913), pp. 348–80Google Scholar; Bates, W. N., ‘Notes on the Rhesus’, Trans. Amer. Philol. Assoc. xlvii (1916), pp. 5–11Google Scholar, Euripides, 1930, pp. 189–91.

^{page 72 note 4} Cf. Rolfe, J. C., ‘The Tragedy Rhesus’, Harvard Studies in Cl. Philol. iv (1893), p. 91Google Scholar. A similar conclusion was reached by Lachmann, K., De choricis systematis tragicorum Graecorum, 1819, p. 116Google Scholar, in regard to the choral odes: ‘is poeta, qui Rhesum scripsit, carmina chorica ad eandem plane rationem contexuit, quam Sophocles et Euripides in antiquissimis earum fabularum, quae exstant, secuti sunt.’

^{page 73 note 1} Menzer, O., De Rheso tragoedia, Berlin, 1867, p. 30Google Scholar; Richards, G. C., ‘The Problem of the of the Rhesus’, C.Q. x (1916), p. 195Google Scholar, n. 1; Zieliński, p. 142. Menzer (pp. 27–8) gives the total of resolutions in the Rhesus accurately; cf. Spengler, , De Rheso tragoedia (Programm d. Gym. z. Düren, 1857), p. 12Google Scholar, and Hagenbach, F., De Rheso tragoedia, Basle, 1863, pp. 26–7Google Scholar.

^{page 73 note 2} See Tables 4A and 4B below.

^{page 73 note 3} A similar opinion is expressed in ‘L'Authenticité du “Rhésus” d'Euripide’, Antiquité Classique, ii (1933), p. 109Google Scholar, by H. Grégoire, who wishes to assign the play to 424; cf. Goossens, R., ‘La date du Rhèsos’, Ant. Class, i (1932), pp. 93–134CrossRefGoogle Scholar.

^{page 73 note 4} The placing of the Rhesus at the beginning first period involves the assumption that the above-mentioned tendency to decrease was already in operation before the earliest dated play in the period (i.e. Alc.): such an assumption, however, seems quite reasonable. There is nothing in the bare proportion of resolutions (8·1%) itself to prevent the Rhesus being placed between the Hippolytus (4·3%) and the Andromache (11·3%). But Table 2 shows that it contains the lowest proportion (5%) of second-foot tribrachs and the highest (67%) of third-foot dactyls (as already mentioned)—both strong indications that the Rhesus may be the earliest play.

^{page 74 note 1} It is not always fully realized that, if the Rhesus is genuine, many of the peculiarities which are commonly considered proofs of its spuriousness may be in fact indications of its early composition at a time when Euripides' style combined experimental innovations which close imitations of his predecessors. See schol. on Rhes. 528. The extent of this imitation in certain respect was noticed by the writer of the ὑπόθεσις, who mentions a Σοφόκλειος ϰαρακτήρ in the play.

^{page 74 note 2} The same conclusion was reached (p. 152) by Zieliński, who assigned the play to 442 (p. 238).

^{page 74 note 3} Hartung, J. A., Euripides Restitutus, 1843, vol. i, pp. 8–9Google Scholar.

^{page 74 note 4} C.Q. x, p. 195: ‘if there are any two plays which from metrical considerations we can put closely together in time’, he wrote, ‘it is surely these.’ Even if there were the smallest evidence of a special metrical similarity (and there is not, cf. Rhes. in Table 2 with Ajax in Table 4B), it would not necessarily prove the two dramas to have been contemporary, since, however faithfully Euripides in his early tragedies may have followed many of Sophocles' metrical traditions, it is extremely improbable that at any one time their styles were identical.

^{page 74 note 5} This fact and the internal evidence dispose of the possibility admitted by the frequantly of resolutions (5·7%) that the drama was composed between Hippol. (4·3%) and Andr. (11·3%), althougth it must be conceded that the figures in Table 2 for second-foot tribrachs and third-foot dactyls might be slightly more suitable to the later date.

^{page 74 note 6} Analecta Euripidea, 1875, p. 152.

^{page 74 note 7} The Chronology of the Extant Plays of Euripides, 1905, p. 28.

^{page 74 note 8} Op. cit., p. 148. In a later work, Herakles, i² (1895), p. 143 (i¹, p. 348), he grouped the Andr. with the first period in the order Alc., Med., Hippol., Andr., Heracl., all of which he dated before 425. But in metrical style at least it is clearly very different from the plays of the first period, and is two or three years later than the latest of them.

^{page 74 note 9} If the Hec. was acted in 425, the Andr. should be placed in 426.

^{page 74 note 10} Eur., Budé, ed., vol. ii, 1927, p. 106Google Scholar: ‘On sera done conduit à le placer avant l'été de 424’ and ‘nous serions tenté de dire qu' Andromaque a dû être jouée entre 427 et 425’.

^{page 74 note 11} Handbook of Gk.Lit., p. 185: ‘not long after the beginning of the Peloponnesian War, perhaps not far from 425.’

^{page 75 note 1} ‘Euripides and Tharyps’, C.R. xxxvii (1923), pp. 58–60Google Scholar: cf. Klotzsch, C., Epirotische Geschichte, 1911, p. 221Google Scholar.

^{page 75 note 2} Thuc. ii. 80 Θάρυπος το βασιλέως ἒτι παιδς ὃντος.

^{page 75 note 3} Even if the play was produced at Argos (cf. Bergk, , Hermes, xviii (1883), pp. 487–510Google Scholar, and Page, D. L., ‘The Elegiacs in Euripides’ Andro mache', in Greek Poetry and Life, 1936, pp. 223–30)Google Scholar, Tharyps' arrival in Athens might well have suggested the subject of the drama.

^{page 75 note 4} Die griechische Tragödie, 1930, ii, p. 84 (note on i, p. 304). Cf. also Kitto, , Greek Tragedy, 1939, p. 235Google Scholar, n. 2.

^{page 75 note 5} Geschichte der gr. Litt., 4th ed. 1882, i, p. 605Google Scholar, n. 2 (English trans., 1840, p. 369, n. 1; 1858, i, p. 486 n.).

^{page 75 note 6} Another consideration is that, if the Hec. belongs to 425 instead of to 424, the Andr. presumably must be placed in 426, which allows a year less for the remarkable change in the poet's metrical style between the Hippol. and the Andr.

^{page 75 note 7} As regards metre, 422 and even 421 cannot be excluded as possible alternatives, although they are less likely.

^{page 75 note 8} Cf. Müller, K. O., G.g.L. ⁴, i, p. 608Google Scholar, (Eng. trans., 1840, p. 371; 1858, i, p. 489), and Croiset, , Hist, de la litt. gr. ³ iii, p. 319Google Scholar. Grégoire, H., Eur., Budé, ed., vol. iii, 1923, pp. 97–8Google Scholar, advocates 422 but admits that 423 would suit as well.

^{page 75 note 9} But cf. Kitto, , Greek Tragedy, p. 225Google Scholar: ‘the theme demanded that Creon should turn the potential sacrifice into a real one, a point to remember when one is considering whether the play reflects the refusal of the Thebans to restore the Athenian dead after Delium…. If there were no dramatic reason for their refusal to yield here, the inference that Euripides was thinking of Delium would be irresistible; as there is such a reason Delium may be coincidence.’ But Kitto suggests that II. 738 ff. may be an allusion to the Athenian rejection of the terms offered by Sparta in 425.

^{page 76 note 1} Cf. Thuc. iv. 97–9.

^{page 76 note 2} Oxford text, vol. ii, introd. to Suppl.

^{page 76 note 3} A point noticed by Descroix (p. 171), who concludes, ‘nous inclinons done à croire que la composition, sinon la représentation d'El., est antérieure de plusieurs années à la date de 413’ (cf. p. 58).

^{page 76 note 4} The earlier periods which the same percentage of third-foot dactyls in El. could also suit (cf. Table 2) are clearly inadmissible.

^{page 76 note 5} In view of the above figures, Murray's statement (Oxford text, vol. ii, introd. to El.), ‘post Troadas (A.C. 415) eam numeri et stilus ponunt’, is rather strange—at least as far as the iambic trimeters are concerned.

^{page 76 note 6} Op. cit., p. 68. In the words μνσαροȋς and μηδ' πιόρκων μέτα (1350, 1355) he found aspersions on Alcibiades' perfidy in 419 (Thuc. v. 56 ‘Aλκιβιδον πείσαντος… ὃρκοις).

^{page 76 note 7} According to Nauck, , Eur. 1889, i, p. lxxvGoogle Scholar, the whole epilogue (1233–1359) is spurious: if this is so, Zirndorfer's date for the play's per formance may be correct; but the passage is usually considered genuine.

^{page 77 note 1} Paley, , Eur. ii, p. 387Google Scholar(n. on El. 1347); ‘the words could only apply to the sailing out of the expedition.’

^{page 77 note 2} G.g.L.⁴ i, p. 614, n. 1 (Eng. trans., 1840, p. 374; 1858, i, p. 493): he states that the passage ‘clearly refers to the fleet which sailed from Athens to Sicily’ in 415.

^{page 77 note 3} Aelian, , Var. Hist. ii. 8Google Scholar.

^{page 77 note 4} By this period, however, tragedy was performed at the Lenaea also, at which Agathon had gained his first victory in 416; cf. Haigh, , The Attic Theatre, 3rd ed., 1907, pp. 25–6Google Scholar. Eur. might have competed at both festivals in 415.

^{page 77 note 5} Thuc. vi. 6–7.

^{page 77 note 6} Cf. Parmentier, L., Eur., Budé, ed., vol. iv, 1925, p. 189Google Scholar: ‘l'allusion doit done viser la grande flotte de secours envoyée à Nicias en l'année 413.’

^{page 77 note 7} Lines 1280–3 have often been urged in vindication of this date, as being a preparation for the Helen in 412, the following year. But (as pointed out by Naber, S. A., ‘Euripidea’, Mnemosyne, Nova Series, x (1882), p. 271)Google Scholar this assumption is not necessarily valid, since the alternative version of the legend of Helen was known ever since Stesichorus' recantation, and was in common circulation; cf. Plato, , Phaedrus, 243 aGoogle Scholar.

^{page 77 note 8} pp. 110–11: cf. also Bergk, , Gr. Lit., vol. iii (1884), p. 553Google Scholar n.; ‘der Dichter auf die Elektra mehr Sorgfalt verwendet hat.’

^{page 78 note 1} A. W., de Groot, A Handbook of Antique Prose-Rhythm, 1919, p. 102Google Scholar: ‘metrical tendencies are partly unconscious, partly conscious.… As far as they are conscious, it is doubtful whether the artist was able to formulate them himself.’

^{page 78 note 2} Tragic Drama, p. 301 n. Zieliński (p. 141), too, assumed an accidental metrical ‘recession’ to account for the case of the El.

^{page 78 note 3} In spite of this early date of composition it is still quite possible that the Electra of Sophocles preceded it.

^{page 78 note 4} The fact that the 21·5% of Herc. is slightly in excess of the 21·2% of Tro. is without special significance: the cause may be a very small ‘accidental’ variation; but it is possible that the greater part of Tro. was composed somewhat before Herc. (although not performed until after it). On the other hand, cf. p. 77, n. 4, above.

^{page 78 note 5} Cf. Müller, K. O., G.g.L. ⁴ i, p. 610Google Scholar(Eng. trans., 1840, p. 372; 1858, i, p. 490).

^{page 78 note 6} Herakles, i², p. 135 (cf. Herakles, i¹, p. 349).

^{page 78 note 7} He proposed 421 in Analecta Eur., 1875, p. 153, but gave 422 as an alternative in Der Mütter Bittgang (Hiketides), 1899, p. 26.

^{page 78 note 8} Cf. Macurdy, pp. 95–107, who gives the internal and other evidence confirming this date.

^{page 78 note 9} Analecta Eur., p. 153.

^{page 78 note 10} The statistics in Table 2 for first-foot dactyls do not favour either an earlier or later date; the proportion of third-foot dactyls is smaller than might be expected, but this very fact may be taken as a slight piece of evidence against any earlier dating.

^{page 79 note 1} e.g. Boeckh, A., Gr. Trag. Princ., 1808, p. 191Google Scholar, gave 427; Hermann, J. G., Ion, 1827, p. xxxiiGoogle Scholar, circa 424–421; Hartung, op. cit., vol. i, p. 451, Olymp. 88, 2, i.e. 427; Fix, T., Eur. (Didot, ed.), 1840, p. xGoogle Scholar, Olymp. 90, i.e. circa 420; Bayfield, 1891, p. x, ‘the commonly assigned date is accordingly about B.C. 425’.

^{page 79 note 2} Analecta Eur., p. 154; Herakles, i², p. 144.

^{page 79 note 3} ‘Die beiden Elektren’, Hermes, xviii (1883), p. 242Google Scholar, n. 1: ‘will man also die letzte Möglichkeit, so rücke man Ion zu Elektra 413.’

^{page 79 note 4} Eur. Ion, 1926, p. 24: cf. also Kranz, , De forma stasimi (Diss. Berol.), 1912, p. 43Google Scholar.

^{page 79 note 5} Die attische Autochthonensage, Berlin, 1897, pp. 138–9Google Scholar.

^{page 79 note 6} De Ione fabula Euripidea quaestiones selectae, Bonn, 1880, pp. 7–19Google Scholar.

^{page 79 note 7} Lys. 909–13.

^{page 79 note 8} Helen, 243–4, it is suggested, is a reminiscence of Ion, 887–90.

^{page 79 note 9} Thέorie des formes lyriques de la tragédie grecque, 1895, p. 257.

^{page 79 note 10} The figures for Phoen. in Table 2, however, lend this suggestion no support, and the proportion of first-foot dactyls weighs against it.

^{page 79 note 11} Phoenissae of Eur., 1911, pp. 7–32; p. 7, ‘We have seen that the ύπόθεσις complains that it is “padded” (παραπληρωματικόν). The suspected places, apart from single lines, are … (1) 88–201 (Schrader, Verrall). (2) 1104–1141 (Wecklein). (3) 1221–1263 (Paley). (Add 1264–1282, Paley and Verrall.)’

^{page 80 note 1} The section of the dialogue between Jocasta and Polynices which refers to the latter's return from exile (387–99) has been interpreted as referring to Alcibiades’ return to Athens in 411 (Thuc. viii. 81, 97) by Zirndorfer (pp. 82–3), who accordingly gives 410 as the date of performance—a suggestion approved by Hermann, J. G., Phoenissae, 1840, p. xvGoogle Scholar.

^{page 80 note 2} Evidence serving to justify this conclusion is afforded by an examination of the lines in the Oxford text which are bracketed as interpolations. In the 90 lines so bracketed in the last five plays (Hel., Phoen., Or., Bacch., I.A.) there are only 18 resolutions, or 20%, compared with an average of 32·5% in the genuine lines (1,720 res. in 5,285 trimeters, cf. Table 1). The rarity of resolutions in the interpolated lines was probably greater even than appears, as a large proportion of the 20% may be attributed to the resolutions occurring in the not negligible number of genuine lines which must inevitably have been wrongly bracketed. For all the plays, the figures are 27 res. (also 12 others involving proper names) in 168 bracketed lines, or 16·1%, compared with the general average of 20·12% in the genuine lines. Again, resolutions in lines wrongly suspected are doubtless responsible for a considerable part of the 16·1%.

^{page 80 note 3} If this explanation is the correct one, the figures for Bacch. and I.A. in Table 2 confirm that the lines subsequently added contained very few resolutions: except for a small diminution in the frequency of first-foot dactyls in Bacch., the tendencies of the first-foot and third-foot dactyls continue regularly, which would hardly be the case if large numbers of trisyllabic feet, used by a different poet in different positions, had been admixed.

^{page 80 note 4} Cf. Cantarella, R., ‘L'influsso degli attori su la tradizione dei testi tragici’, Rivista Indo-Greco-Italica di Fil., Ling., Ant., xiv (1930), p. 203Google Scholar [Fasc. iii–iv, p. 67], ‘La tragedia [i.e. Bacchae] ebbe certamente a subire dei mutamenti, che oggi però è molto difficile definire’.

^{page 81 note 1} Included in this total are 11.376–7 and 1336–7, in which no resolutions occur.

^{page 81 note 2} Philologische Beyträge, Zürich, 1819, pp. 143–55Google Scholar.

^{page 81 note 3} Actors' Interpolations in Greek Tragedy, 1934, p. 138.

^{page 81 note 4} The possibility that the infrequency of the resolutions in these scenes is due to conscious changes of style (cf. supra, p. 67, n. 5) or to accidental variations should not be overlooked; but interpolation seems the more likely explanation.

^{page 81 note 5} Eur., vol. iii, 1839, pp. 440–1Google Scholar, n. c: ‘Itaque sic statuendum censemus, Euripidemduas elaborates quidem, sed non coagmentatas, reliquisse exordii partes, alteram v. 1–48. complexam, alteram 117–163. quae ut inter se cohaererent, ab alio poeta, sive is Euripides minor sive alius fuit, inserti sunt v. 49–116.’; Poetae Scaenici Gr., 5th ed., 1869, Eur., p. 264.

^{page 81 note 6} It must be stressed, however, that the variations are only noticeable when individual short passages are examined. When large numbers of resolved feet in a single play are considered en bloc, they present an aspect of great regularity. Thus if the resolutions in most of the plays were counted by successive periods of ten lines, the totals would vary widely; but if they were counted by periods of a hundred lines, the results for each hundred lines would usually be fairly constant.

^{page 81 note 7} De Iphigeniae Aulidensis forma ac condicione, Berlin, 1870, pp. 176–7Google Scholar.

^{page 81 note 8} Examples of such lines, with the two resolutions in various possible positions, are given by Taccone, A., ‘Il trimetro giambico nella poesia greca’, Memorie della Reale Accademia delle Scienze di Torino, Scienze Morali, Storiche e Filologiche, Serie IIa, liv (1904), pp. 48–93Google Scholar.

^{page 82 note 1} All resolutions involving proper names are left out of account both here and in the figures that follow.

^{page 82 note 2} The conspicuously small number of 23 in the I.A. may be another testimony to the scarcity of resolutions in the interpolated passages.

^{page 82 note 3} Zieliński (pp. 213–40) attempts to establish chronology of the lost plays of Euripides by examining the frequency of resolutions in their fragments. His results are interesting and often valuable: but it will be realized that deductions from small fragments of merely four or five lines can sometimes be very deceptive (cf. p. 81, n. 6, supra).

^{page 82 note 4} In Cycl. are the following stretches: 161–4, 239–43 (five), 274—7, 433–6, 557–62 (six), and 587–90.

^{page 82 note 5} Cf. supra, p. 70, n. 2.

^{page 82 note 6} The extent of the influence of the early iambus on the tragic trimeter is discussed by Knox, A. D., ‘The Early Iambus’, Philologus, lxxxvii (1932), pp. 32–5Google Scholar.

^{page 82 note 7} Cf. Maas, P., Griechische Metrik (Gercke-Norden, Einleitung in die Altertumswissenschaft, the 3rd ed., 1927, vol. i, part 7), p. 24Google Scholar, § 102, ‘Teilung der longa ist sehr selten und ist beschränkt auf längere Wörter, in denen sich Kürzen häufen’; Koster, W. J. W., Traité de métrique grecque Leyden, 1936, pp. 83–4Google Scholar.

^{page 82 note 8} The jambographers are quoted from Diehl, E., Anthologia Lyrica Graeca, vol. i, 2nd ed., 1936Google Scholar. References in brackets come from Bergk, Poetae Lyrici Graeci, 4th ed. The iambic fragments of Archilochus total 43 lines, giving a ratio of one resolution in 14⅓ lines. There are also two resolved feet involving proper names, frs. 20 (19) and 26 (23). The 28 (or so) trimeters in the frs. of the epodes are not taken into consideration (as being possibly subject to different metrical rules), but in any case there is no special feature in their evidence. On the metre of Archilochus, cf. Usener, H., Alt griechischer Versbau, Bonn, 1887, p. 115Google Scholar, n. 9. Wilamowitz, , Griechische Verskunst, 1921, p. 289Google Scholar, postulates a certain amount of development in iambic poetry before Archilochus, but this is questioned by Knox (op. cit., p. 35, n. 26).

^{page 83 note 1} These instances occur in 180 lines of fragments. Cf. Rumpel, , Philologus, xxv (1867), p. 55Google Scholar, n. 1. In fr. 9, the source (schol. on Eur. Phoen. 207) reads δι μακρν in fr. 15. 2, the source (Et. M. 634.1) reads ρσοθύρης.

^{page 83 note 2} P. Maas in Pauly-Wissowa, R.E., s.v. ‘Simonides (I)’ writes, ‘Die Iamben sind ganz streng, keine Auflösungen (7, 1. 7 θεός, lies θεύς oder Zεύς; Iο δι μακρν zerstört auch die Zäsur, Umstellung beseitigt nicht die Auflösung, bleibt also unsicher)’. Descroix (p. 112), however, counts five resolutions in Sem.—three third-foot dactyls, one first-foot dactyl, and one first-foot tribrach.

^{page 83 note 3} This is the only resolved foot in 42 iambic trimeters.

^{page 83 note 4} In the five lines of τρίμετρα ρθά among the frs. of Hipponax (frs. 66–9 Diehl) occur a first-foot tribrach, λίγα Φρονέονσιν οί ϰάλιν πεπωκότες (fr. 66), and a first-foot anapaest involving a proper name (fr. 67, 1.2). The Fragmenta ianibica adespota in Diehl (i², fasc. 3, pp. 68–73) afford one instance of a second-foot and one of a fourth-foot tribrach (frs. 5 and 23).

^{page 83 note 5} References to tragic fragments are quoted from A. Nauck, Tragicorum Graec. Fragmenta, 2nd ed., 1926.

^{page 85 note 1} Cf. Müller, K. O., G.g.L⁴ i, p. 545Google Scholar(Eng. trans. 1840, p. 327; 1858, i, p. 432); ‘zu den letzten Werken des Äschylischen Genius gehört aller Wahrscheinlichkeit nach sein Prometheus…’; Harrison, ‘Aἰσϰύλος ΣοΦοκλΐζων’, Proc. Cambr. Philol. Soc. cxviii–cxx (1921), pp. 14–15Google Scholar; Schmid, W., Untersuchungen zum gefessellen Prometheus, Tübinger Beiträge zur Altertums wissenschaft, ix, 1929, p. 19Google Scholar (where the Pr. is considered spurious but dated between 458 and 445); Thomson, G., Prometheus Bound, 1932, pp. 38–46Google Scholar; Denniston, , The Greek Particles, 1934, pp. lxvii–lxixGoogle Scholar; ‘Pauses in the Tragic Senarius’, C.Q. xxx (1936), pp. 76–7Google Scholar; Yorke, , op. cit., C.Q. xxx, p. 117Google Scholar; ‘The Date of the Prometheus Vinctus’, C.Q. xxx, pp. 153–4Google Scholar; Knight, W. F. J., ‘Zeus in Prometheia’, J.H.S. lviii (1938), p. 53Google Scholar; Robertson, D. S., Proc. Cambr. Philol. Soc. clxix–clxxi (1938), pp. 9–10Google Scholar; Mullens, H. G., Greece and Rome, viii (1938–1939), pp. 160–71Google Scholar.

^{page 86 note 1} The only instance of it found in the frs. involves a proper name, Kριτίης Ξȋος ν κασωρικ ῷ δόμῳ (Hipponax, fr. 67, l. 2).

^{page 86 note 2} ὦ θεομανές … (Sept. 653), κα τνα σνοισθα… (Cho. 216), ἢλιος ἅναγνα … (Cho. 986). (στρας, ὃταν … in Ag. 7 is bracketed in the Oxford text.) There seems no reason for suspecting Aeschylus' use of the first-foot dactyl, in spite of the doubts of Yorke, (C.Q. xxx, p. 117)Google Scholar. Cf. Housman, A. E., Journ. Philol. xvi (1888), p. 245Google Scholar. Cases of this resolution involving proper names occur at Sept. 450, Ag. 1312, and Pr. 730. It will be noticed that the resolution is found in the later plays only.

^{page 86 note 3} The smallness of this figure is only in a the negligible degree due to the absence of such resolution in the plays anterior to 425 (Andromache). The average even in those which do contain this resolution (i.e. those which were written in 425 and after) is very much less than 2%.

^{page 87 note 1} The subject of trisyllabic feet in the minor tragedians has been treated by Müller, C. F. W., De pedibus solutis in tragicorum minorum tritnetris iambicis, Berlin, 1879Google Scholar.

^{page 87 note 2} The following passages from frs. of satyrplays belong to this period: Aristias, 5 ll., 2 res., frs. 3 (l. 2), 4; Ion, 11 ll., 4 res., frs. 21, 23, 29 (ll. 1, 2), also fr. 24 (l. 3) involves a proper name; Achaeus, 29 ll. (both ll. of fr. 8 included, fr. 27, l. 2 excluded), 3 res., frs. 7 (l. 2), 8 (l. 1), 14 (l. 1); Iophon, 2 ll., no res. On the resolutions in Eur. Cycl., Soph. Ichneutae, and the satyr-plays of the minor tragedians, see Walker, R. J., The Ichneutae of Sophocles, 1919, pp. 189–269Google Scholar.

^{page 87 note 3} The references are: Neophron, fr. 2 (l. 15); Aristarchus, frs. 2 (ll. 4, 5), 3; Ion, frs. 2, 54 (l. 1), 55 (l. 2), 56, 60, 63 (l. 2), also fr. 1 (l. 3) involves a p.n.; Achaeus, frs. 3 (ll. 1, 3), 37 (l. 4), 42 (l. 2), 44; Sthenelus, fr. 1; Agathon, frs. 4 (ll. 2, 3, 4, 5, 6), 5 (l. 2), 13 (l. 1), 15, 21, 22, 25, also fr. 4 (l. 3) involves a p.n.; Critias, frs. 1 (ll. 1, 6, 19, 31, 40), 2, 5 (l. 1); Dicaeogenes, frs. 2 (l. 1), 4 (l. 2); Diogenes (Atheniensis), fr. 1 (ll. 3, 7, 10, 11), also fr. 1 (ll. 1, 2, 7) involve p.ns.; the following ll. contain two res. each, Agathon, fr. 4 (ll. 4, 5), 21. Not included in the table are the frs. of Melanthius (1 l.), Iophon (3 ll.), and Xenocles (1 l.), which possess no resolution. The Incertarum fabularum fragmenta have been counted with the tragic frs. in Tables 5 and 6, although a few of the lines, e.g. Achaeus, frs. 42, 44, may be satyric.

^{page 87 note 4} The ratio for Neophron (1 in 24) may favour a fairly early date if the lines are genuine. Their authenticity has been impugned by Wilamowitz, , Hermes, xv (1880), p. 487Google Scholar, and by others, but defended by Haigh, , Tragic Drama, pp. 290–1Google Scholar.

^{page 87 note 5} It is noteworthy that, of the 14 res. occurring in 39 ll. of Agathon, 7 are in the 6 lines of fr. 4. (If this fr. were omitted, the frequency would be 1 in 4·7 (7 res. in 33 ll.), a figure closer to those of his contemporaries.) Of these 7 res., 4 are second-foot tribrachs, a form of resolution found much more commonly in the fourth cent, (see infra). If rightly attributed to Agathon, the passage was probably written near the end of his life, when fourth-cent, tendencies were already commencing.

^{page 87 note 6} For the dates of the minor tragedians cf. Haigh, , Tragic Drama, Appendix I, pp. 463–72Google Scholar. On the date of Dicaeogenes cf. Dieterich, A., R.E. v. 1, 563Google Scholar, ‘nach Schol. Aristoph. Eccl. I scheint er ein Zeitgenosse Agathons gewesen zu sein.’

^{page 87 note 7} The references are: Chaeremon, frs. 1 (ll. 4, 7), 4, 5, 6, 7 (ll. 1, 2), 8, 9 (l. 2), 11 (l. 1), 13 (l. 2), 14 (ll. 2, 3, 9, 12, 13, 16, 17), 15, 17 (ll. 1, 2), 20, 25, 33, also fr. 12 (l. 1) involves a p.n.; Antiphon, fr. 2 (ll. 1, 2), also fr. 2 (l. 2) involves a p.n.; Dionysius, frs. 2 (ll. 2, 3), 4, 6, 8, also fr. 9 involves a p.n.; Carcinus, frs. 1, 4 (ll. 1, 2), 5 (l. 2), also fr. 5 (ll. 6, 9) involve p.ns.; Theodectes, frs. 201. 1, 2, 4), 6 (ll. 2, 3, 5, 6), 7 (l. 1), 8 (ll. 3, 5), 9 (l. 2), 10 (l. 2). 13 (l. 1), 14 (l, 1); (μαλακόφθαλμος in fr. 6, l. 1 is not listed; ‘corruptum’, Nauck;) Cleaenetus, fr. 2 (l. 2); Crates, frs. 2 (l. I), 3 (ll. 2, 3, 4), 4 (l. 2); the following ll. contain two res. each, Chaeremon, frs. 1 (l. 7), 15, 17, (l. 2) 33, Theodectes, fr. 6 (ll. 2, 3, 5), Crates, fr. 3 (l. 4). Not included in the table are the frs. of Astydamas (9 ll.) and Diogenes Sinopensis (4 ll.), which possess no resolution.

^{page 88 note 1} The above applies to tragic frs. only; the satyr-plays of the fourth cent. retained their usual metrical freedom, as their frs. show: Astydamas, 4 ll., 4 res., fr. 3 (ll. 2, 4); Python, 18 ll., 12 res., fr. 1 (ll. 1, 4, 6, 10, 11, 12, 13, 14, 16, 18), also fr. 1 (ll. 8, 17) involve p.ns.

^{page 88 note 2} Nauck, p. 816, ‘trisyllabis enim pedibus sicut Sositheus et alii quidam ita Moschion abstinuit’. Fr. 10, which contains resolved feet, was rejected by Meineke.

^{page 88 note 3} Sositheus' remains total 27 lines, of which fr. 4 (1 l.) seems satyric. On frs. 2 (21 ll.) and 3 (3 ll.), both from the Daphnis, Nauck notes (p. 821), ‘sine dubio Daphnis Sosithei satyrica fuit fabula haud dissimilis Alcestidi Euripideae’, i.e. the drama appears to have been a prosatyric tragedy like the Alcestis. It is unlikely that the banishment of resolutions affected satyr-plays proper; cf. Lycophron's satyric frs., 14 ll., 8 res., frs. 1 (l. 5), 2 (ll. 1, 2, 3, 4), 4 (l. 2).

^{page 89 note 1} There is no resolution in the tragic frs. of Lycophron (5 ll.) or of Philiscus (2 ll.), both known to be contemporaries of Sosiphanes and Sositheus. The thirteen dramatists of unknown date (several of them probably very late) listed by Nauck (pp. 824–32) present only 4 res. in their combined total of 60 ll. It is possible that these 4 cases, Apollonides frs. 1 (l. 3), 2 (l. 1), Isidorus, fr. 1 (l. 4) and Pompeius Macer, fr. 1 (l. 3), are only apparent exceptions, which might well be explained by the assumption of either early fourth-cent, composition or satyric origin.