² In S.El. 653 Benedict proposed τκνοιϲ for τκνων, and this has now some slight manuscript support: see Dawe, , Studies i. 182Google Scholar, and the apparatus criticus to his Teubner edition (1975).

³ loc. cit. (part I, p. 357 n. 73), 107–8.

⁴ Roux's ‘en fuyant la societe des citoyens’ and Ussher's ‘in avoidance of the townsfolk’ do not do justice to the meaning of the word Φυγή.

⁵ πολίταιϲ (V³) is not necessarily a genuine variant or even conjecture but may owe its origin to the scholium in V (τ δ πολιτνντ δοτικϲ, τοȋϲ πολίταιϲ); cf. B^gl τοȋϲ πολίταιϲ.

⁶ Elmsley, (Annotatio, p. 4)Google Scholar, Kvičala(Denkschriften der Kais. Akad. der Wiss. Wien, Phil.-Hist. Cl. 29 [1879], 238)Google Scholar, Verrall.

⁷ RhM 21 (1866), 148–9Google Scholar= Opusc. 1 (1866), 749–50Google Scholar.

⁸ In Classical Studies in Honour of Henry Drisler (1894), p. 11.

⁹ Known to me only from Bursian (1892), 256.

¹⁰ Cf. Andr. 642–3 ϲμικϲ π' ρϰϲ νεȋκοϲ νθρώποιϲ μέγα/γλϲϲ' κπορίζει, fr. 32 κακϲ <π'> ρϰϲ τέλοϲ κακν [Hes.] fr. 43(a). 61 M. -W. ι ρϰϲ λίγηϲ, Sol. fr. 13. 14 West ρϰϲ δ' ι ρϰϲ (West: ρϰ δ' ι λίγου cod.: ρϰν δ' ι λίγου Arnott ap. West) γίγνεται ϲτε πυρϲ, Bacchyl. 11. 64–5 νεȋκοϲ γρ μαιμκετον βληϰρϲ νπαλτο καϲιγνητοȋϲ π' πρϰϲ. Paley, indeed, claims that ‘Had the poet written ϲμ7iota;κϲ πρϰϲι., no one would have found the slightest difficulty. But ϲμ7iota;κϲ is virtually implied in μείζονι θυμι The first claim is right, the second wrong.

¹¹ Cf. Hermann: ‘nubesque diceretur sublata ex suo initio, quod dici absurdum esset, quia nihil est, quod non nascatur ex suo initio’.

¹² This is the sense demanded by Hermann (‘Apertum vero, hoc dici debuisse, ex initio, quod fecisset Medea, cognosci, multo earn graviora nubila concitaturam esse’), though he attempted to extract this sense from the bare genitive ρϰϲ. Note also Σ^v π' ρϰϲ δ τ ν Φοϲ τϲ οἰμωγϲ ξαιρμενον δλν ϲτιν ὃτι κτλ.

¹³ We could retain the δ' by reading δλα δ' π', as Dr Dawe suggests to me. For the neuter plural see e.g. Barrett, on Hi. 269Google Scholar, Bond on Herc. 1058. But the asyndeton is not only natural; it is, I think, preferable. For a very similar asyndeton see 119.

¹⁴ See above, n. 9.

¹⁵ Michelangeli, L. A., Saggio di note critiche al testo delta Medea (1878)Google Scholar, known to me only from Wecklein, , BPhW 21 (1901), 1284Google Scholar(‘ein unnützes Wort’) and Levi, A., Boll, di Fil. Cl. 5 (1898–1899), 211Google Scholar(‘felice e prudente’).

¹⁶ Note also Theogn. 425 (u.l.), Sol. fr. 13. 14 (cj., cited above, n. 10).

¹⁷ It is accepted by Hartung, , Wecklein, , Méridier, , von Arnim, , and Wilamowitz, (Hermes 15 [1880], 506Google Scholar= Kleine Schriften 1 [1971], pp. 41–2Google Scholar, and on Herc. 1140); and Paley and Weil have a good word for it. For the confusion νΨει/νιει see Or. 609, A. Ch. 131; similarly Med. 1382 προϲΨομεν (uel -ομαι) codd.: ξομεν Choerob. epim. p. 112, Et. Ma. p. 750. 44.

¹⁸ The Lucretian passage is cited by Regenbogen, O., Eranos 48 (1950), 24Google Scholar, who, however, derives the idea from Anaximander (DK A 23), Empedocles (DK A 63), and Heraclitus (DK A 14). But the two former explain thunder in terms of the effect of wind, not fire, on a cloud; and Heraclitus is explaining lightning, not thunder. Anaximander's theory is reproduced in Ar. Nub. 404–7; cf. Kahn, C. H., Anaximander and the Origins of Greek Cosmology (1960), pp. 100–2, 108–9Google Scholar.

¹⁹ Statements that A has μολὼν for βον(–ν) are false.

²⁰ Fraenkel, , RhM 72 (1918), 177Google Scholar = Kleine Beitrdge 1 (1964), pp. 181–2Google Scholar. Similarly Valgiglio, following Wilamowitz, , Griechische Verskunst, p. 538Google Scholar, but with ἂλγεϲι δᾃματοϲ, which gives a pure dactylic hexameter for the second line (on this line see below, n. 33). Wilamowitz had earlier suggested μελθροιο(Hermes 15 [1880], 510–11Google Scholar= Kleine Schriften 1 (1971), pp. 46–7)Google Scholar, on which see Page.

²¹ See Dale, , Lyric Metres, pp. 29–30, 117, 167Google Scholar.

²² See Parker, L. P. E., CQ n.s. 8 (1958), 86CrossRefGoogle Scholar.

²³ WS 77 (1964), 26–7Google Scholar= Collected Papers (1969), pp. 198–9.

²⁴ Page's note illustrating correption in non–lyric anapaests is strangely irrelevant, since these anapaests are lyric. His discussion is supplemented by Barrett on Hi. 246 (Addenda, pp. 432–3).

²⁶ I doubt whether π μελθρου can mean ‘in the house’ (phrases like π' οἰκ7alpha;τοϲ καθϲθαι, ‘work in a brothel’, cited by LSJ s.u. π A. i. 1, are of a specialized kind). Even if it could, it would have to be referred to Medea, which would give no sense; her cry could not be described as being π μελθρου It certainly cannot mean ‘at (near) the house’ and so be referred to the chorus.

²⁶ Page's statement that the scholia take it as an adjective is mistaken. They offer three different explanations: (i)Σ^hbv τ ἒϰον δᾁο πᾁλαϲ κα εἰϲδουϲ, μ7alpha;ν μν αὐθεντικν, τραν δ, ἤν Ỏμηροϲ (Od. 22. 126, 132, 333)λγει ρϲοθᾁρην (i.e. a house or room with two doors), (ii) Σ^hbv π τοû μΦιπᾁλου … τουτϲτιν π τοȗ πυλνοϲ (i.e. ddorway) (iii) Σ^bv περ τϲ θᾁραϲ τποϲ (i.e. the area around the doors). For Σ^h see Daitz, S. G., The Scholia in the Jerusalem Palimpsest of Euripides (1979), p. 96Google Scholar.

²⁷ The prepositional phrase ἒϲω μελθρου then depends on βον, ‘a cry within the house’. Wecklein, and Page compare Hi. 234–5Google Scholar; see also Diggle, , Studies, pp. 28–9, 69Google Scholar.

²⁸ For references see Gow on Theoc. 14. 42. Dawe, (Studies iii. 123–4)Google Scholar has not convinced me that Philoctetes' cave did not have a rear entrance. For other caves with two entrances see Kannicht, and Snell, on TrGF II 646a 21–2Google Scholar.

²⁰ loc. cit. (above, n. 23), p. 27 n. 20 = p. 199 n. 1.

³⁰ Proposed in hisedition of Plato, , Philebus (1855), p. 92Google Scholar; also proposed by Allen, F. D., HSCP 9 (1898), 42–3Google Scholar.

³¹ Mnem. n.s. 5 (1877), 23Google Scholar.

³² Verrall prints 133 as I do; Page's objection that this ‘breaks the symmetry of the two lines of lyric dactyls which follow’ is no longer valid, once π' is deleted and γον accepted.

³³ Weil, von Arnim, and Méridier, like Koster, W. J. W., Traite de metrique grecque (ed. 2, 1953), p. 197Google Scholar, join 137 to 136 to form a hexameter of six dactyls. In view of the dactylic, rather than spondaic, end, it is more natural to keep the dimeter discrete. Line 133 Weil and Meridier print (remarkably) as a hexameter ending in two spondees κολϰδοϲ, οὐδ πω ἣπιοϲ λλ, γεραι, λξον).

³⁴ With the exception of Verrall and of F. D. Allen, loc. cit. (above, n. 30).

³⁵ This tempts one to suggest ηὐδτο [γρ] ταὐτα κοὐδπω, but the temptation should probably be resisted. For κοὐδπω see Med. 60, A.Pe. 814, Ar. Equ. 168, Au. 1227.

³⁶ Some editors print οὐδ πω. We do not need a connective here, and I am not sure whether editors, by so printing, mean to imply that οὐδ is connective (Denniston and Page show by their translation that they do not).

³⁷ But not S. Ph. 446.

³⁸ As for the metre, if δᾃματοϲ is taken with this line, we have iambus+cretic+bacchiac; if with what precedes, we have the colon ∪––∪–∪––, discussed by Stinton, T. C. W., BICS 22 (1975), 84–8Google Scholar.

³⁹ See Fraenkel, on Ag. 369Google Scholar. At Hec. 740 I am sorry that I can no longer agree with DrMatthiessen, (GRBS 10 [1969], 302)Google Scholar, with whom I formerly agreed (JHS 95 [1975], 198)CrossRefGoogle Scholar, that κρανθν should be preferred to πραϰθν. Bain, D., Actors and Audience (1977), p. 14Google Scholar n. 1, has convinced me that the verb is less suitable.

⁴⁰ This point is made by Schoemann, G. F., Opuscula Academica 3 (1858), p. 193Google Scholar and by Verrall.

⁴¹ Cf. the reading of C at Hi. 746: κυρν DLP et B²γρ: ναων ΩVE: γυιν C. But possibly the answer here is κρανων (Wecklein): cf.Hec. 219 κρανθεȋϲαν] κυρωθεȋϲαν M^glO^gl.

⁴² As Herwerden, says (RhM 58 [1903], 143)Google Scholar, ‘auro non habet quo referatur’. What precedes (ᾼα δ γον/τ τ' εὐγεϲ κα τ δυϲγενϲ) makes no metrical sense. Wilamowitz, (De tragicorum graecorum fragmentis commentatio [1893] = Kleine Schriften 1 (1971), p. 205Google Scholar; see also Hermes 62 [1927], 289)Google Scholar suggested τ τ' εὐγενϲ <πΦυκε> κτλ. (cf. Snell, , Hermes, Einzelschriften 5 [1937], 14)Google Scholar. Körte (ap. Luria, , Hermes 64 [1929], 493Google Scholar n. 2) suggested δυϲγενϲ <πλει>. I should prefer τ τ' εὐγενϲ κα [τ] δυϲγενϲ,like τ τ' ὃντα κα μλλοντα (Ion 1 Hel 14 923 S El. 1498).

⁴³ I am glad to find that Stinton, loc. cit. (above, n. 38), 85, calls it ‘attractive’; hardly anyone e\se has ever been attracted by it.

⁴⁴ Cf. Mastronarde, D. J., Contact and Discontinuity: Some Conventions of Speech and Action on the Greek Tragic Stage (1979), pp. 60–1Google Scholar.

⁴⁵ There remains the anomalous genitive at IA 1242 ἱκτευϲον πατρϲ. But I am not as confident as Page (Actors' Interpolations in Greek Tragedy [1934], p. 185) that Euripides wrote these lines.

⁴⁶ Kritische Beiträge zur Erklärung der griechischen Tragiker (1883), known to me only from Bursian (1884), 105, Philologus 43 (1884), 710Google Scholar.

⁴⁷ See Dawe, , Collation and Investigation, p. 283Google Scholar.

⁴⁸ For the confusion of πι– and πο– see p. 61.

⁴⁹ To the articles referred to here and in what follows may be added Weil, H., Jahrb.f. cl. Phil. 13 (1867), 381–2Google Scholar, R. Rauchenstein, ibid. 21 (1875), 838–9, J. Kvičala, loc. cit. (above, n. 6), 274, Schwartz, E., Hermes 32 (1897), 493–5Google Scholar, Harry, J., The Greek Tragic Poets (1914), pp. 217–21Google Scholar, O. Regenbogen, loc. cit. (above, n. 18), 49–50, G. Müller, loc. cit. (part i, n. 43), 81–2, Willink, C. W., CQ n.s. 16 (1966), 240Google Scholar n. 3.

⁵⁰ Page's statement that ‘the δρμοϲ at the stadium at Olympia is 210 yards’ is a slip (it would be right if by δρμοϲ he meant ϲτδιον but a moment before he used δρμοϲ for δαυλοϲ). On the length of Greek stadia see Gardiner, E. N., Athletics of the Ancient World (1930), p. 128Google Scholar.

⁵¹ The ἂν is indispensable, whether it is restored here or in the next line (so Davison and Bergson, against Levitt). Its position is perfectly natural in either conjecture: with ἒλκων it owes its position to its tendency, snared with enclitics, to be placed as early as possible in the sentence; with νθπτετο it owes its place to its tendency to stand next to the main verb (for its late position, again with a participial clause preceding, see 190–1 ϲοΦοᾀϲ/τοᾀϲ πρϲθε βροτοϲ οὐκ ἂν μρτοιϲ). Cf. Page on 250, Barrett on Hi. 270 – I observe that no parallel has been adduced for the sense which must be ascribed here to the compound verb, ‘reach, attain'; for the simple verb in this sense see LSJ s.u. in. 6.

⁵² The accusatives in Herc. 662 διϲοᾀϲ ἂν ἒβαν διαᾁλουϲ, 1102 δαυλον … μολᾃν are perhaps better explained as internal (δαυλοϲ being treated as a nomen actionis), like δν (et sim.) λθεȋν: cf. K-G i. 303–11, Schwyzer ii. 74–8.

⁵³ On these expressions see Mastronarde, op. cit. (above, n. 44), p. 67 n. 46.

⁵⁴ Cf. Moschion, , TrGF 97 F 6 1–2Google Scholar πρτον δ' ἂνειμι κα διαπτᾁξω λγωι/ρϰν βροτεου κα κκατϲταϲιν βου.

⁵⁵ On the length of the hippodrome at Olympia see Gardiner, op. cit. (above, n. 50), p. 225, Harris, H. A., G&R n.s. 15 (1968), 113–26Google Scholar.

⁵⁶ The connection of thought was apprehended by Σ^bv (c): τ κατ ϲ οὐκ ξεργϲομαι τι λγωι αὐτ γρ γνᾃϲηι κτλ Σ^bv (a) and(b) misunderstand 1222.

⁵⁷ See p. 5 7 for some instances of Φ- and ἂΦ- confused. Note also 552 Φλκων] ὐΦ– Ο: Φ- α.

⁵⁸ Quaestionis de scholiorum Euripideorum in poetae verbis restituendis auctoritate et usu specimen (1872), pp. 38–9. The same proposal was made by Stadtmüller, , Beiträge zur Texteskritik der Eur. Medea (1876), p. 34Google Scholar n. 1. Barthold's elaboration ϲυμΦορϲ ναϲτροΦν, where ϲυμΦορϲ is taken from the scholia, is needless.

⁵⁹ On ϲυμΦορϲ50;πϲυμΦορ?οϲυμΦορϲτρϲυμΦορϲ74;ϲυμΦορ?Φων at S.Tr. 221 see Dawe, , Studies iii. 81–2Google Scholar, Bond on Herc. 736.

⁶⁰ De Versibus Dochmiacis (1811–12), p. 291.

⁶¹ Scribes tend to resolve compounds into their component parts: e.g.Cycl. 610 ξενοδαιτυμνοϲ (ξνων δ–), El. 735 πειρδροϲοι (ἂπειροι δρϲου), Herc. 883 κατογκεΦλοιϲ (κατνκεΦαλαȋϲ), Tr. 536 μβροτοπᾃλου (μβρτα π-), 603 ρημπολιϲ (ἒρημοϲ π-), IT 405–6 περικοναϲ ναοᾀε (περ κοναϲ ναοȗ). Cf. Fraenkel, , Ag. ii. 376Google Scholar n. 1, Johansen, Friis and Whittle, on A. Su. 198Google Scholar.

⁶² ‘ein nichtssagendes Epitheton’ Nauck, loc. cit. (part i, n. 73), 135, ‘épithète faible et insignifiante’ Weil.

⁶³ p. 210 ed. Elmsley, ii. 48 Christodoulou.

⁶⁴ For references to Lewis and Snell, see below p. 63.

⁶⁵ ‘Φνοϲ μεβεται ѕϲ. Φνωι, caedes caede permutatur, caedem caedes sequitur’ Matthiae, ‘Nescio an Anglice signified to ensue’ Elmsley, ‘Why, instead of love and duty, is murder adopted as a new and sudden impulse?’ Paley, ‘“comes in the place of, succeeds” (supply αὐτοȗ ѕϲ ϰλου)’ Verrall, ‘Avec Arnim nous sous-entendons après μεβεται quelque chose comme ντ τϲ πρϲθεν Φιλαϲ’ Méridier.

⁶⁶ Σ^bv pertinently observe that Medea calls upon ‘Earth’ in 1251 ὡϲ μλ*lambda;ουϲαν δϰεϲθαι τϲ μιαιΦοναϲ τ αἷμα. – It would be just possible to take αἷμα as ‘offspring’, a sense inadequately illustrated by LSJ s.u. ill. 2: see Pi. W. 6. 35, thereafter not apparently before Hellenistic poetry (Pfeiffer on Call. fr. 67. 7, Gow on Theoc. 24. 73). But the verb πτνειν makes such a sense unlikely here.

⁶⁷ The genitive is of the same type as in IT 397–8 ‘αϲιτιδα γαȋαν/Eὐρᾃπαϲ διαμεΨαϲ, Hel. 1186–7 ππλουϲ μλαναϲ … λευκν μεΨαϲ, S. Tr. 736–7 λᾃιουϲΦρναϲ/τν νȗν παρουϲν τYF6;νδ' μεΨαϲθαι, E. Med. 967–8 τν δ' μν παδων Φυγϲ/Ψυϰϲ ἂν λλαξαμεθ' οὐ ϰρυϲοȗ μνον, Andr. 1029 ναλλξαϲα Φνον θαντου (Stevens: θαντωι codd.). Cf. K -G i. 378, Schwyzer ii. 127.

⁶⁸ It is perhaps best to take Φνωι Φνοϲ as an independent phrase (‘murder upon murder’, as in Ph. 1495 οὐκ ἒριϲ λλ Φνωι Φνοϲ see K-G i. 444 Anm. 4, Schwyzer ii. 156, Kannicht, on Hel. 366)Google Scholar, and to take ξαμεβων with δι' αματοϲ (‘passing through bloodshed’, as at Ph. 1051–2 δι' αμτων δ' μεβει/μυϲαρν εϲ γνα).

⁶⁹ ἒτερον (West, M. L., BICS 28 [1981], 69)Google Scholar should perhaps be preferred.

⁷⁰ So also Weil, , Jahrb.f. cl. Phil. 13 (1867), 383Google Scholar(‘Φνωι, oder vielmehr Φνον, ist vor Φνοϲ ausgefallen’).

⁷¹ For a different interpretation see Kells, J. H., CQ n.s. 10 (1960), 129–30CrossRefGoogle Scholar.

⁷² Humanitas 4 (1952), 15–18Google Scholar. I cannot approve of Dodds' other proposals in this passage: to place the question-mark after ζαμενϲ instead of after μεβεται, and to read αἷμα <μταν> in 1256.

⁷³ Similarly <Φνωι> would give the meaning ‘murder is repaid by murder’ (LSJ s.u. μεβω B.3).

⁷⁴ See part I, n. 15.

⁷⁵ Assuming that [ινα ωϲ and αρα ηϲθ] were not elided. The general practice of the papyrus (there are a few exceptions) is to elide before the same vowel (e.g. δ(ε)εκ) but not before a different vowel. In the third line we may compare ο[τε η with 1102 τε οποθεν.