² The Concordance of Allen-Italie s.v. πνοϲ lists H.F. 22, 89, 127, 357, 427, 575, 597, 729, 937, 1275, 1279, 1353, 1410; cf. 259, 388, 501 (πονϲῖν), 581 (κπονεῖν), 1190 (πολπονοϲ). Note the gap between 937 and 1190.

³ Held. 625. My attention has been drawn also to Pindar, Ol. 11.4, Nem. 6.24, Isth. 1.42, etc., and to Parry's, H. article in AJPh 86 (1965), 363ffGoogle Scholar. for the ‘epinician’ aspect ofH.F.

⁴ Cf. P. E. Easterling on S.Tra. 20–1.

⁵ 1196 οἄκ ἄν L; an easy correction, not mentioned by Diggle or Bond, which at once enhances the dialogue and removes a metrical anomaly.

|… is consistent with the ‘enoplian dochmiac’ context (here dactylic); by contrast,

is inescapably an inappropriate chor. dim. Ion 1478 KP. ἲϲτω γοργοφνα…ιΩ. το τοτ' ἔλεΞαϲ should not be cited as a parallel instance of aeolic (phalaecian) in a similar context. Better lineation there treats τ τοτ' ἔλεΞαϲ KP. ἄ ϲκοπελοῖϲ π' μοῖϲ as an iambelegus, following a contracted ‘D’ but there are issues in that context which need to be pursued elsewhere.

⁶ Cf.πολυπλανϲ (of Menelaus) at Hel. 203 and πολυπλνητοϲ (of human life in general) at Hipp. 1110. The overtone of mental ‘wandering’ in the case of Heracles seems inescapable (cf. Hipp. 240), but it is only an overtone.

⁷ ἔτλην τλαϲ, cf. comm. Or. p. 86 on τλμων 'Oρϲτηϲ. λοίϲθιον ‘last and worst', cf. Page on Med. 1105. For the θριγκϲ-metaphor Bond compares A.Ag. 1283.

⁸ Bond, after Dawe and Diggle, mentionsCyc. 471, I.T. 1046, S. Aj. 61, O.C. 542; see also Kells, J. H., CQ n.s. 16 (1966), 51CrossRefGoogle Scholar, on E. El. 100, who adds Or. 1544, and further in my comm. on Or. 816–18.

⁹ Text and lineation as Dale (Lyric Metres ², 118) and Webster, not Dawe.

¹⁰ Cf. Denniston on El. 1141.

¹¹ See Bond, p. 332, and comm. Or. p. 104; note also Or. 161 φε, μχθων.

¹² Studies 54–7; as to the hiatus, see also his further observations in Illinois Classical Studies 6 (1981), 96Google Scholar.

¹³ For the interpolation of an explanatory participle (as in Med. 981) Dr Diggle has drawn my attention also to the note of Friis Johansen-Whittle on A.Su. 568 (p. 453).

¹⁴ Cf. Conomis, N. C., Hermes 92 (1964), 23ffGoogle Scholar.

does not occur at all (Diggle justly obelizes Ion 782f. πῶϲ φιϲ; ἄφατον ναδητον , which cannot be scanned as 2δ; IA 1307 εὐναῖιι βαιιλίιιν (sic) is a syncopated iambo-trochaic dimeter).

is attested only at S. Ant. 1273 θεϲ τοτ' ἄρα ττε (read ττε θεϲ ἄρα ττε?). It is hard to see why

was thus eschewed; but we cannot escape from the evidence.

¹⁵ Diggle leans on Ale. 201 κλαει γ' ἂκοιτιν… in reply to 199–200 ἥ πιου ϲτνζει…; But the long (presumptive) question there is quite different from the one-word question εὒδει; and there is no να before κλαει γ'…The reply ‘Aye, he sleeps a bad sleep…’, if that is the general sense, introduces a new point after the affirmative answer, a point not implied by the questioner and contrasting with what Amphitryon himself said in 1043–4 and 1048. For such a sequence of thought Bond rightly looks for a pattern like θϲω, φνον γε θῆλυν.

¹⁶ ‘He sleeps a ruined sleep’ may be an intelligible hypallage in English, but no convincing Greek parallel has been cited, or for the rendering ‘deathly’ (Gray and Hutchinson). Wilamowitz explained this and other exx. as damnatory, implying the optative ὂλοιτο (cf. Eng. ‘perishing’, ‘perisher’); but Amphitryon has no good reason for ‘cursing’ the oblivious sleep which he wishes to continue. At least Wecklein was aware of the problem when he proposed λμενοϲ.

¹⁷ Cf. also Or. 1374 βαρβροιϲιδραϲμῖϲ, , Ba. 77 ϲοιϲ καθαρμοῖϲιν (-μϲ nouns like φαλμϲ), Med. 865 τλμονι θυμι, Hel. 374 φοναῖϲ, 629, 693, 697, Ph. 225, 346, 656, 832, Or. πολυπνοιϲ νγκαιϲ, etc.

¹⁸ As to ἂυπνον, that can be understood as implying οὐκ εὐδαμονα (cf. 1013–14). But, apart from the metrical problem, it is not obviously appropriate in this context that Amphitryon should disparage by ‘negation’ the one feature of the situation which (as ‘calm oblivion’) he wishes to preserve, and which (in reply to εὔδει) he is concerned to affirm. Bond allows that S. Phil. 848 is different, where the chronically ill sleeper is more literally ‘sleepless’ (and consequently εὐδρακϲ). He should not have speculated here about ‘nightmares’, inconsistently with what Amphitryon has said in 1042 and 1049–50.

¹⁹ The traditional line-numeration, reflecting the lineation of LP and the Aldine Edition, is 1061 εὔδει;… | 1062 ϲ… | 1063 τοξρει…

²⁰ Cf. Barrett, on Hipp. 752–7Google Scholar, and Diggle, in Dionysiaca: Nine Studies…Presented to Sir Denys Page… (Cambridge, 1978), pp. 171–2Google Scholar. A typical ex. of the ‘integral’ type of int. ace. phrase, as classified by Barrett and further illustrated by Diggle, is Hel. 77 πλαυϲιν εἰκοϲἒθανεϲΔιϲ κρηϲ. Predicative advancement in the word-order is a natural feature of such idiom. For the Or. passages cited, see comm.ad locc.

²¹ Comm. Or. p. 111, where sympathetic consideration is given to the alternative interpretation ‘free from blood(-guilt)’ (Hermann, Verrall). It is surprising that A. Dihle does not mention πφονοϲ, even in a footnote, in his survey of ‘πο- privatives’(Glotta 63 [1965], 137–9)Google Scholar.

²² The implicit antithesis of φλοι (a fortiori φλτατοι) and πολμιοι/χθρο seems t o have been missed by commentators on 1133. Tragedy offers countless variations on this ‘friend/foe’, ‘love/hate’ dichotomy (e.g. Med. 16; cf. comm. Or. p. lxiv).

²³ Here and elsewhere I pass over in silence some corrections of L adopted by Diggle and accepted with convincing arguments by Bond, as 1016 πτρα (Bothe, for πτρα) and 1017 ἄπιϲτοϲ (Reiske, for ἄριϲτοϲ).

²⁴ Comm. Or. p. 288. Though he mentions the alternative colometry in PCPhS n.s. 20 (1974), 13–16Google Scholar, Diggle follows Murray here, but not atHec. 1067–8 in Euripidis Fabulae I.

²⁵ See Diggle, PCPhS art. cit., and comm. on Or. 1302[φονεετε] κανετε (φονεετε del. Hermann). Diggle is evidently on the defensive as to τδε ' ὑπερβαλεν | παρμεν | … in the light of his own evidence, and he goes on to propose a doubling of τλανε (unwanted in this ‘quiet’ context). An alternative excision, of course, would be τδε ' ὑπερ[βαλε παρ]δραμεν.

²⁶ The only parallel (Rh. 832 παραιτομαι) is vitiated by faulty responsion and doubtful Euripidean authorship; but in itself it is quite easily emendable to παραιτομεθα (the same error, I believe, as at Hel. 664 ϲοϲομαι for ϲοϲομεθα).

²⁷ <ϲι> μοραι Kirchhoff; <νιν> Paley, <δ> J. H. H. Schmidt.

²⁸ It should not be objected that ϲυγκατεργζεϲθα τιτ τινι ought to mean ‘to collaborate with A in destroying B’, rather than ‘to destroy B in conjunction with (destroying) A’. There may ! be a theoretical ambiguity, but cf. A.Ag. 1605–6 μ' …θλ7omega;ι πατρ ξυνεξελανει (like Th. 6.88.8 ξυναπϲϲτελλον αὐτοῖϲ πρϲβειϲ), and ϲυνθπτιν τινι (Alc. 149, etc.).

²⁹ There is n o instance in Ion or Or. As to H.F., 1078 is dealt with below. At 888 κακοῖϲινκπετϲουϲι ends a compoundly corrupt sentence. A t 890 and 893 there is flawed responsion; the latter should be βοτρων π χεμαϲι <ν> λοιβϲ (the same

verse as 1190, An. 857, 862, El. 586, 588,590, Ion 1486,1494, Hel. 657,680,681, Hyps. fr. 64.94), and the notorious problem in the former (as I hope to argue elsewhere) can then be solved by writing [οὐ] Βρομουκεχαριϲμνων (for μνα) θρϲωι.

³⁰ See comm. on Or. 94 τφον…πρϲ καϲιγντηϲ μολεῖν.

³¹ Kannicht's emendation ἔρν' is wrong for the sense (there is no room here to elaborate other objections to his text): the laundered crimson ππλοι were being spread ‘on standing reeds’ (lit. ‘saplings of reed') as in Hipp. 128 they were spread on a warm sunny rock, μφ' ὔδωρ) makes a local point (‘by, near’, cf.I.T. 6; Diggle, , Studies 80)Google Scholar, and ν χλαν is also ‘environmental’ (comm. on Or. 329–31). By contrast, the concluding μφ-phrase there goes closely with θλπουϲα, as κοϲιν μφ' οἴκων here goes closely with νημμνα.

³² Cf. Diggle, , Studies 66–7Google Scholar, and (for Sophocles) Easterling, , Hermes 101 (1973), 14–34Google Scholar.

³³ See Stinton, T. C. W., ‘Pause and Period in the Lyrics of Greek Tragedy’, CQ n.s. 27 (1977), 27–66CrossRefGoogle Scholar. To the rule enunciated on p. 40 (‘pendant close followed by short or anceps can never coincide with word-overlap or elision’) he admitted some exceptions, but none remotely supporting an exception here. The reizianum is rightly denned as

by M. L. West on pp. xii and 199 of his Greek Metre (Oxford, 1982), though not on p. 30Google Scholar (cf. my review in JHS 104 (1984), 227CrossRefGoogle Scholar).

³⁴ Reiske's is not the only published emendation, but I refrain from giving fresh currency to the others listed by Prinz-Wecklein.

³⁵ The emphasis is on the paired n. pi. words, which together characterize Heracles' condition as ‘calm’ and ‘sleep-like’. To govern them, the colourless verb γειν is at least as suitable as ἰαειν (properly ‘pass the night’; Ph. 1538 is a more natural extension from that, pace Bond, than the supposed use here). Note that, with ἰαονθ', the n. pls are adverbial, not substantival. Is there a parallel for an ώδηϲ adjective used thus?

³⁶ The favourite unit

(related to the telesillean, but also behaving like

and

) is T in my notation; see comm. Or. pp. xxi, 113, 288. 1080 ταφων πεκλυϲτον ἂ-|ϲτυ πρϲαϲ is T ba (clausular), analogous to A ba at 1197 πολυμοχθτερον πολλυπλαγκττερ τε θνατν.

³⁷ For the pattern of this exchange, Mastronarde, D. J. in Contact and Discontinuity (Berkeley, 1979), p. 61Google Scholar, compares Alc. 872–5/889–92 and Tr. 1229–30.

³⁸ Comm. Or. p. 288; cf. also Hec. 1084, and Barrett, on Hipp. 593Google Scholar.