2. Mastronarde, D. J., Euripides Phoenissae (1988)Google Scholar includes an impressively long ‘Conspectus versuum suspectorum’ (139–43); see also his historical survey in Phoenix 32 (1978) 105–28CrossRefGoogle Scholar, and Mueller-Goldingen, C., Untersuchungen zu den Phönissen des Euripides (1985)Google Scholar passim. As to ‘passages’, Mastronarde surprisingly brackets only 1013–18 and 1737–66, but he admits to suspicion elsewhere (even of retractatio at 1595–1607); Mueller-Goldingen adds 774–7 and 1104–40, variously assigning these and numerous shorter interpolations to ‘Schauspieler’ and ‘Herausgeber’. Both can be called conservative (notably as to the Finale), though less so than some other recent critics (n.4 below). Much less conservative, after Fraenkel (n.6 below), is Reeve, M. D., ‘Interpolation in Greek Tragedy’, GRBS 13 (1972) 247–65, 451–74Google Scholar, and 14 (1973) 145–72 (the second section is the part devoted to Pho.). On the papyrus evidence, see Haslam, M. W., CQ n.s. 26 (1976) 4–11CrossRefGoogle Scholar.

3. Cf. my Euripides Orestes (1986, with corrections and additions 1989), lxii.

4. Too many critics are prejudicially committed to defence, regardless of the strength of the internal evidence, taking the line that all arguments based on ‘literary’ considerations are ‘subjective’. Craik, E. M., Euripides Phoenician women (1988)Google Scholar comes too close to that position in her Introduction (50) and in the niggardly use of her sigla for athetesis and ‘suspicion’ (cf. my forthcoming review in JHS). Tuilier, A. in ‘Nouvelles remarques sur l'exodos des Phéniciennes’ in Studi… Cataudella I (1972) 347–56Google Scholar, demonstrates the respectable antiquity of the Finale, substantially as we have it; but he is blind to internal evidence, and seems quite unable to see that his arguments for authenticity have little bearing on what may have happened to the play soon after its creation. Similarly blinkered in approach is van der Valk, M., Studies in Euripides: Phoenissae and Andromache (1985)Google Scholar. It is proper to resist the more speculative atheteses, insubstantially based on ‘Stilgefühl’, that have clouded criticism of the play. Not much would remain if they were all accepted. But it does no service to Euripides to argue (in effect) that he was capable of shoddy workmanship and self-contradictory dramatic structure.

5. He also permits me to say that in general his atheteses will be far more extensive than those of Mastronarde and Craik, and will in several places go beyond those of Fraenkel and Reeve.

6. Fraenkel, E., Zu den Phoenissen des Euripides (SBAW (1963) Heft 1)Google Scholar; see also his earlier article in Eranos 44 (1946) 81–9Google Scholar. For the most recent studies see nn.2, 4 above. My knowledge of dos Santos Alves, M., Euripides, As Fenicias, introd., trad, e notas (1975)Google Scholar is derived solely from Mastronarde; but he appears to have broken some new ground in the right direction (cf. nn.31, 46, 47 below).

7. I hope to pursue elsewhere the related questions whether 1582–1766 contains any authentic passages and how the Euripidean play may have ended. Less directly relevant is the much-discussed relationship between the ‘Teichoskopia’ in the Prologue (103–201) and the inorganic ἔκφρασις in the First Messenger's speech describing the Seven and their armour (1104–40). Since, however, I do allude to the Prologue below, perhaps I should say here that I have no doubt whatever of the essential authenticity of 103–201, with its highly characteristic, partly ‘enoplian dochmiac’, dialogue (though it is not free from minor interpolation and is somewhat corrupt), and I agree with Mueller-Goldingen (against Mastronarde, 1978) in rejecting 1104–40.

8. A short half, as things stand; but the Urtext is likely to have been very much less than 1766 verses in length.

9. See my comm. on Or., 211–15, for lines 804–6 and the following ode; also comm. on Or., 77, 144, for lines 348–806, beginning with the expected entry of Menelaos, as constituting ‘Act II’ in the structure of that play. Pho. 261–783, beginning with the expected entry of Poluneikes, is differently articulated, with the central ἀγών 355–637 set between musical passages (292–354 Xo. Ιοκ.; 638–89 Xo.). But it has an otherwise similar kind of continuity (especially if Eteokles remains on stage during 638–89, see below), with a chiastic balance between 261–354 and 638–783, framing the Ιοκ. This view of the structure accounts for the shortness of the ‘2nd episode’, which could in fact be even shorter without detriment to the design.

10. Reeve's athetesis of Or. 804–6 was, in this instance, misconceived (see comm.).

11. So Bond on H.F. 166ff., and similarly Mueller-Goldingen, 130.

12. Cf. Or. 399, and Kannicht on Hel. 559–60.

13. Cf. comm. on Or. 638–9, etc.

14. Lines 684–7 should, I think, be trimmed thus: followed then by either ἄμυνε τᾶιδε γαίαι or (cf. El. 679). Full identification of Demeter and Ge is typical late Euripidean syncretism. Interpolation of names is a common type of error (cf. Or. 964); and ἐκτίσαντο (v.1. -κτήσ-), in the same sentence as κτίσαν (882), looks like another gloss, prima facie.

We now have characteristic dimeters with one overlap (for the syncopated sequence ba cr ba ia, cf. Or. 1442, 1463–4, etc.).

15. Lines 524–5 . There is no need for a ‘personifying’ capital letter either in 524 or (as Craik) in 523; Eteokles stops well short of deifying Monarchy (unlike Εὐλάβεια). In her note on 767 Craik overstates Eteokles' ‘generally blasphemous stance’, overlooking the words in these lines. As to 772–3, a normal Athenian would not regard μομφή concerning τέχνη μαντική as δυσσεβής, though it might be unwise (especially in tragedy).

16. Thgn. 117–18 West ; but cf. also Bias (D–K 1.45.11) , which is close to Ad. el. 23 West , and ‘Pythagorean’ uses of the word (see D–K 3.182b). The adj. appears first (with a similar gnomic connotation) in Demokritos B.91 . For the verb, LSJ first cites two frs. of Aesch. (both imperative).

17. Cf. the association with ‘standing’ and ‘falling’ in Soph., O.T. 46–50, 616Google Scholar. Wrestling featured in the education of a kalokagathos, and Greek gnomic thought is shot through with ‘agonistic’ metaphors derived from that (as, e.g., in Hipp. 426–7 ).

18. See comm. on Or. 682–716, 696–703, 699, 748, 793, 1058–9.

19. Cf. Bond (n.ll), and Collard on Supp. 324–5 (other such topical words are σωτηρία, ἡσυχία, ἑταιρία, cf. comm. on Or., xxiv, xliii, etc.). But the evidence (especially of Ar. and Soph.) does not suggest that the ‘normal Athenian’ in all circumstances regarded εὐλάβεια as ‘suspect policy’; still less (pace Craik) does it suggest a word associated with ‘repressive régimes’. It remains one of a number of gnomic ‘goods’, and is as apt on the lips of Chrusothemis and the blind Oidipous as it is on the lips of the ‘tyrants’ Lukos and Eteokles. (Lukos, incidentally, had earlier demonstrated that he has the normal high regard for εὐανδρία by impugning Herakles' courage in battle). Eteokles is not a ‘tyrant’ in the bad sense; what he sets store by is ‘monarchy’ (523–5, n.15 above).

20. Cf. comm. on. Or., xxxvi, for the similar disappearance of Helen and Apollo (prominent early on and at the end) from the foreground of attention in the middle of that play. One wonders at the ancient critic (Hypoth. 3) who regarded the verbal ἀγών with Poluneikes as περὶ οὐδενός in the drama.

21. Note the characteristic element of misdirection there, suggesting (a) an antitaxis in the Aeschylean manner (b) an initiative taken by Poluneikes (a suggestion reinforced by 635 in P.'s final line). There is no misstatement, and the fatricidal ἔρως has been appropriately enunciated; but the audience have been subtly misled as to what will in fact happen. The theme of fratricidal ἔρως reflects Septem 688, 692 (where the Chorus associate it with ἄτη). Eur. takes it further (a) by making the brothers actually voice the shocking ‘desire’; (b) by thematic association of this ἔρως with ἐρινύς (255, 624, 1029, 1306, 1503) and ἔρις (81, 351, 500, 798, 811, etc.); cf. I.A. 585–9 (comm. on Or. 253).

22. Line 756 (cf. nn.25, 40) is not covered by P. Merton 2, 54; but the chances are that both 778 σοὶ (del. Kirchhoff, Haslam, Mastronarde) and 781 (n.38 below) are relatively late interpolations, still absent from at least part of the tradition in the second century AD (unless of course their omission in a single papyrus is merely fortuitous).

23. Reeve (1972) 454–5. There is no need to repeat his evidence here.

24. -αργ- of course implies ἔργ- (sc. α-εργ); for ‘hand’ (χείρ) as antithetic to ‘speech’ (λόγοι, etc.), cf. comm. on Or. 1027–8, 1047–8.

25. Lines 754–6 (756, cf. nn. 22, 40).

26. Burges proposed μηκέτ᾽ ἀργῶμεν (with χέρα acc. of respect); but both the semi-transitive κατα- compound and the semi-abstract χέρα are highly characteristic of Euripidean idiom, in which there is often an element of neologism.

27. Cf. Eteokles' entry-speech at 446–7, which is notably brusque and impatient (under pressure of business elsewhere), eliciting from his mother the response (452). The confused intervening lines 448–51 are rightly deleted by Diggle, (SIFC 7 (1989) 200)Google Scholar.

28. ‘It could be then that 753 is genuine’ (Reeve). Walter transposed 753–5 to follow 773 (deleting 774–8), but he needed to delete 754–5 also. No transposition is required in the solution (keeping 753) proposed below.

29. Lines 774–7 (775) , τις ἧι (del. Walter, Wecklein, alii; Reeve, 456).

30. Reeve aptly refers to ‘the man who elsewhere anticipates the “Wechselmord”’, mentioning 765 (n.31), 880 (n.41), 1263 and 1269. I cannot agree with a reader who regards this objection to 774–7 as ‘pedantic’; moreover, as Reeve has shown, there is more than one objection to these lines.

31. Lines 754–6, see n.25; 757–65 (760) (765) . The writer evidently assumed that the audience would understand the ambiguous ἡμᾶς in 765 as ‘both of us’ (cf. n.30). All these lines are excised by dos Santos Alves (757–62 for the first time, apart from Schumacher's deletion of 759b–67a); but he should not have cut 753 as well.

32. In 748 Mastronarde accepts στόμα (Jackson, after Soph., Ant. 119Google Scholar ἑπτάπυλον στόμα). Musgrave's κύκλον (a word which occurs adjacently in Ant. 118) seems much more natural with -πυργον, and in a spoken context where there is no reason to look for anything fancy. O. Taplin takes a similar view in The stagecraft of Aeschylus (1977) 143 n.2Google Scholar, but adds όδόν as an alternative. Craik calls ἐς πόλιν ‘unexceptionable’, but offers no parallel for its use by a person going from within a city to its perimeter.

33. ἔσται τάδε is standard in response to an imperative; but the implication πείθομαι may indicate that the speaker is deferring in a kind of volte-face (like ‘very well, if you say so’ in English); cf. I.A. 1033 ἔσται (Markland, for ἔστιν) (Klytaimestra to Achilles). So here Eteokles, previously aggressive, is deferring to Kreon's defensive εὐβουλία.

34. A reader opines that ‘766 reads awkwardly after 753 (καταργῶμεν-ἀργόν)’. On the contrary: Eteokles explicitly wishes to avoid the reproach of ἀρχία. That very thought puts him in mind of the need to consult Teiresias, failure to do which might invite such a reproach. There may well be a reminiscence of Soph., O.T. 287Google Scholar (in reference to the summoning of Teiresias) .. Eur. undoubtedly knew O.T. intimately (cf. also comm. on Or. 46ff).

35. ἄξοντα (Markland, Valckenaer) for λαβόντα gives more natural idiom in 770: David Kovacs draws my attention to Hcld. 136f. . Or one might consider ἄγειν for the unnecessary voc. Κρέον (‘I shall send M. to get and bring T. here’; for inf. ἄγειν after ‘send’, cf. Od. 24.419, Th. 4.132, 6.50 (LSJ πέμπω 1.3, K–G 2.17). The text is defensible (λαβόντα simply ‘with’); but it goes somewhat against the grain to construe Τειρεσίαν, immediately preceded by πέμψω δεῦρο, as the object only of λαβόντα.

36. Cf. Craik, 46. Doubling of these roles is likely, even if we do not accept (as Craik does), or accept only as evidence of later practice, the inference from Σ (on 93) that the roles of Iokaste and Antigone were doubled in the first half of the play (despite their presence together on stage in 1270–83).

37. Cf. n.15 above. For ἡδύς = εὔφρων, cf. Hipp. 289–90 ἡδέως is commoner in this sense (LSJ ἡδύς III); but cf. ταχύς = ταχέως, which occurs (in Eur.) only at Or. 254. Craik detects an ironic ambiguity: T.'s coming will not be ‘pleasing’ to Kreon. But such a cryptic double point cannot be taken by the audience as Eteokles speaks. μομφάς after ἐμεμψάμην is pointed, not casual repetition: the μομφή is mutual. That, together with the context (‘me, not you’), seems to favour ἐμοί (vulgo -ε μοι) in 773. There is no need for West's ἔχει (BICS 28 (1981) 66–7Google Scholar); for the ellipse of the subject, cf. Andr. 153, Hec. 248, Supp. 204, H.F. 235, Or. 428, 1121–2 (refs. mostly contributed by J. Diggle).

38. Line 781 (om. Π, del. Haslam); a probably late interpolation (n.22), suspected even by Craik, but retained by Mastronarde. For ώς ἐπί not introducing a clause, cf. 1363 (also εἰς ἀγῶνα), Hcld. 672 , and similarly ὡς ἐπί El. 1326, Tro. 326, H.F. 882, etc. Completion of (putatively) incomplete sentences is perhaps the commonest kind of one-line interpolation; cf. comm. on Or. 1022–3[–4].

39. Cf. comm. on Or. xlvi; for the ‘sinister’ sense of ἀγών as an ironic undertone at an exit, cf. especially Or. 1342 and Ba. 975.

40. Cf. nn. 22, 25, 31 above. Interpolators often include in their compositions lines gleaned from elsewhere (756 = 1376). But the authenticity of 1376 is by no means assured (del. Valckenaer, Mastronarde), and it is quite possible that the verse is an interpolation in both places (added later at 1376, where it seems not to have featured in Diodorus' text).

41. Note that, when Kreon at 1590–1 exiles Oidipous with a reference to Teiresias, he is alluding to what Teiresias has said to him about Oidipous and his family in another suspect passage (869–80, 886–90 del. Fraenkel, 868–80 del. Reeve, cf. n.30 above). As Reeve argues (458), Teiresias' opening speech is far too long for one at first reluctant to speak, and the explanation of 867 (, Κρέον) in 868–80 conflicts with the explanation in 931–41 of the need for an expiatory sacrifice.

42. The Theban myths were still in fluid growth in the fifth century; see Brown, A., Sophocles: Antigone (1987) 3–5Google Scholar (with bibl.). Euripides will not have considered himself bound to accept Sophocles' inventions (which surely included Haimon as Antigone's fiancé and may well have included the burial-prohibition); cf. his resurrection of Iokaste (against O.T.) to play a major role in Pho., and his suppression of Ismene; also Soph.'s own suppression of Haimon in O.C. Only gradually, as with Arthurian myths, did the most successful and popular inventions become canonical as the basis for further inventions; and the known later popularity of Antigone may well have been triggered by posthumous revival.

43. See Hutchinson, G. O., Aeschylus: Septem contra Thebas (1985) 190–3, 209–31Google Scholar (and Brown (n.42) 4).

44. The passages excised by Fraenkel but defended by Mastronarde are: 1586b–90a, 1592–5, 1597–1614, 1621, 1627–33, 1635–6, 1639–82, 1689–90, 1703–9. If (ex hypothesi) all these passages were composed by Ψ, it will prove difficult to resist extension of the hypothesis to embrace some at least of the passages framing and intervening between these items.

45. Text as Mastronarde, apart from my brackets and the colon after Ψἄπο 943.

46. I have not been able to check the atheteses of dos Santos Alves (n.6), cf. nn.31, 47. According to Mastronarde's ‘Conspectus’ (n.2) he is the first and only person to have excised 757–60; but he is not reported as having correspondingly athetised the other ‘Haimon’ passages.

47. Leidloff and Norwood suspected the whole scene (690–783). Zipperer deleted 691. 696 was suspected by Geel and deleted by Herwerden. These may be relevant indicators of critical discomfort; but dos Santos Alves, excising 692 consistently with his excision of 753–65 (n.31), seems to be the only critic to have cut anything here for a reason other than vague ‘Stilgefühl’.

48. Mastronarde records the following conjectures for ἐπῆλθον 697: ἐμόχθουν Valckenaer, ἐμόχθσ᾽ Geel, ἐπλήρουν Schoene, ἀπήντλουν anon. (CJ 1813), ἐφοίτων Stadtmüller, ἐπῆισσον Polak. I agree that the fault probably lies here rather than in 699, and I venture to suggest ἧ πολλά γ᾽ ἤθλουν (or ἤθλησ᾽), making the point that the elderly Kreon has ‘taken a lot of exercise’ in his ‘circuitous’ search. For the ‘athletic’ metaphor in connection with an entry, cf. Or. 456; for the idiom, cf. Hcld. 448 πολλὰ μοχθήσας, Supp. 317 , etc.

49. Kreon is given advance identification as son of Menoikeus and brother of Iokaste in another place, viz. line 11 (generally excised) in the Prologue. It rather looks as though Eur. expected his audience to know this much about the Theban royal family, whereas clarification was later considered desirable.

50. The text of 701–2 is variously uncertain. (1) πολλῶι can hardly stand where ἐνδεεῖς is not, as in Hcld. 170, equivalent to a comparative (‘inferior to’ + gen.), and many edd. accept πολλῶν (Ad, J, conj. Burges); (2) ὡς, is a little odd (prima facie one might have expected ὅτ᾽ for ‘when’). Markland proposed ὧν (‘for discussion of which’), and Wecklein ἅς. The latter is superficially tempting, since E.'s favourite συνάπτειν (comm. on Or. 1481–2) is usually transitive; but διαλλαγὰς συνῆψα (aorist) could be said only of a successful reconciliation. Perhaps we should look for a different line of conjecture: e.g. (‘falling far short of διαλλαγή’) / (‘all my dealings with P. when I came to talk with him’). πολλῶι thus comes into its own, and συνῆψα is duly transitive. διαλλαγή (another topical word, cf. Ar. Ach. 989) properly means ‘reconciliation, cessation of hostilities’, from a root sense ‘change’, rather than ‘truce’; elsewhere pl. in this play ([375], 443, 515), but cf. καταφυγή Or. 724 after the pl. in 448 and 567. There may well be other possibilities on similar lines (with πολλῶι and διαλλαγῆς).

51. Insufficiently recognised; cf. Taplin (n.32) 110–13, who refers to the numerous instances in Soph, and Eur. collected by Ritchie, W., The authenticity of the Rhesus of Euripides (1964) 116Google Scholar n.2. Note that Kreon remains on stage during 784–833, and Teiresias duly speaks first in the following scene.

52. 446–7[–51], cf. n.27 above. The walls may not have been mentioned there, but that is where Kreon expected to find him (698–9).

53. See above (with n.9). A reader objects to Eteokles ‘loitering’ on the stage rather than taking himself off; but why should his attitude be that of a loiterer? One can visualise more than one appropriate static posture: e.g. brooding thought, or arrogant defiance (‘what I have I hold’), or anger still directed at the parodos by which Poluneikes made his exit. One might even allow him briefly to follow Poluneikes offstage (as if to make sure of his departure), returning before the end of the ode.

54. For a schematic outline of the plot, see page 186. As to overall length, my excisions in 1–1541 still leave more than 1400 lines (even if I excise passages of which I am merely suspicious), to which we may add whatever originally stood after 1541 (cf. n.7); and the mathematics, though admittedly speculative, are consistent with a literally central position for the ode 784–832, as for Or. 807–43 (cf. n.8).

55. For the ‘Teichoskopia’ in the Prologue (the more important as exposition if 1104–40 is excised), cf. n.7. For the prominence of Antigone only in the Prologue and Finale, cf. Helen and Hermione in Or. (n.20 above).