^{page 3 note 1} The following editions are referred to: G. Hermann (Leipzig, 1810), F. A. Paley (London, 1874²), A. J. E. Pflugk and N. Wecklein (Leipzig, 1877), A. Gray and J. T. Hutchinson (Cambridge, 1886²), U. von Wilamowitz-Moellendorff (Berlin, 1895²), N. Wecklein (Leipzig, 1899), G. Murray (Oxford, 1913³), L. Parmentier (ed. Budé, Paris, 1923).

^{page 3 note 2} Wecklein's list may be supplemented by the following: Schenkl, K., Disputatio de locis aliquot Euripidis Herculis, in Gratulationsschrift… G. Curtius (Prague, 1874)Google Scholar (known to me only through Bursian iii (1874/1875), 450Google Scholar), Hartman, J. J., Mnemosyne n.s. xi (1882), 127Google Scholar, Apelt, O., Philologus lxiii (1904), 251–2Google Scholar, Harry, J.E., BPhW xxxiii (1913), 733–5Google Scholar ( = The Greek tragic poets (Cincinnati, 1914), pp. 118–20Google Scholar), Könnecke, O., WKlPh xxxii (1915), 1170Google Scholar.

^{page 3 note 3} CQ n.s. xxii (1972), 242Google Scholar. A further passage to add to the list is Hipp. 803 , .

^{page 3 note 4} They are confused in this play at 470, 604, 1098 bis (Canter's τε is confirmed by P. Heid. 205 [Siegmann, E., Literarische griechische Texte der Heidelberger Papyrussammlung (Heidelberg, 1956Google Scholar) = Pack, R. A., The Greek and Latin literary texts from Greco-Roman Egypt² (Ann Arbor, 1967)Google Scholar, no. 392]), and 1377.

^{page 4 note 1} But I include below the single instance of θύγατερ in Euripides' trochaic tetrameters, at IA 886.

^{page 4 note 2} On whose identity see Zuntz, G., An inquiry into the transmission of the plays of Euripides (Cambridge, 1965), p. 289Google Scholar.

^{page 5 note 1} Sophocles uses the vocative only once in iambic dialogue (OC 398 , θύγατερ;), Aeschylus never.

^{page 5 note 2} Marginalia scaenica (Oxford, 1955), pp. 1–3Google Scholar.

^{page 6 note 1} Another ⟨νῶιν⟩ has recently crept into the text of Euripides. S. G. Daitz (Teubner edition 1973) prints at Hec. 162–4 . The supplement is Musgrave's; but the wrong part of the sentence has been amended. The fault, whose correction may be expected to restore the missing syllable, lies in ποῖ δ' ἥσω, where the verb ἥσω has to be given, against analogy, an intransitive sense (‘whither shall I speed?’). In PCPS n.s. xv (1969), 44Google Scholar I dismissed this as ‘impossible syntax’ (I might have added that speeding will not come easily to an old lady who cannot raise herself from the ground without assistance). But Daitz has come to the defence of ἥσω with the annotation ‘cf. fr. 779.4, Rhes. 291–292, Aesch. Pers. 470’. Non tali auxillo nec defensoribus istis. At fr. 779.4 (Phaethon 171) , the verb means what it means at S. Ai. 154 , that is ‘aim for’ (LSJ s.u. 1 3 b). At Rh. 291–2 , the meaning is ‘we sent our flocks to the uplands’, not ‘we sped to the edges of the flocks’. At A. Pe. 470 , I transcribe D. L. Page's critical note (Oxford, 1972): ‘ἴησ' ΜΙΥΟ^acΡ^γρGF, ἤϊσʾ A, ἤιξʾ fere rell.; ἤϊσσ᾿ G. C. W. Schneider'. The text is quite uncertain: see H. D. Broadhead's edition (Cambridge, 1960), p. 272.

^{page 6 note 2} Classical Journal viii (1813), 207Google Scholar.

^{page 6 note 3} RhM n.F. xxvi (1871), 220Google Scholar.

^{page 7 note 1} The misinterpretation of ἐπινωτίσας by LSJ s.u. is corrected by LSJ Suppl. Compare Ph. 654, Rh. 208–9, Longus 1.20 , and note also (‘that which covers the back, e.g. wings’ LSJ) of the Sphinx in E. Oedipus (fr. trag. adesp. 541 N = Pap. Oxy. 2459 = Austin, , Noua fragmenta Euripidea in papyris reperta (Berlin, 1968)Google Scholar, fr. 83.8). And see Denniston and Page on A. Ag. 286.

^{page 7 note 2} At Hel. 71–2 , I should accept Dingelstad's ἐχθίστης.

^{page 8 note 1} The same noun and verb are combined at Su. 1146, El. 42, S. El. 475–6. For illustration of this use of ἴτω see the passages quoted in my note on Phaethon 101 (Cambridge, 1970)Google Scholar. I will say a little about one of the passages quoted there, since almost everyone has misunderstood it. At Su. 1052 Evadne cries . The usual interpretation is this: ‘ualeat lux, nuptiaeque’ Portus (similarly Markland), ‘ualeant faces et nuptiae’ Hermann, ‘ἴτω, ualeat facessat’ Bothe, ‘ualeant taedae et nuptiae’ Wilamowitz, , Analecta Euripidea (Berlin, 1875), p. 121Google Scholar, ‘Adieu, lumière, adieu mes noces!’ H. Grégoire (Budé edition, Paris, 1923), ‘Addio (‘se ne vada’) luce (quale simbolo di vita) e nozze’ G. Ammendola (Turin, 1956²), ‘Vaarwel, licht en huwelijk’ G. Italie (Groningen, 1951). The meaning is precisely the opposite: ‘ueniat lux nuptiaeque’ C. P. Buttmann on S. Ph. 1092 (1822), p. 159 (cited by Matthiae), ‘Willkommen Licht, willkommen Ehe’ Wilamowitz, , Griechische Verskunst (Berlin, 1921), p. 554Google Scholar. Evadne is looking forward to the renewal of her marriage with her dead husband in Hades. Her intention and her sentiments are those of Brünhilde: ‘Helles Feuer / das Herz mir erfaßt, / ihn zu umschlingen, / umschlossen von ihm, / in mächtigster Minne / vermählt ihm zu sein!… Siegfried! Siegfried! Sieh!/Selig grüßt dich dein Weib!’

^{page 8 note 2} See Murray's edition, vol. ii p. iv.

^{page 8 note 3} Jackson, 's proposal, Marg. scaen., p. 149Google Scholar, is more plausible than any other. In 807 Murray ought not to have printed Musgrave's γᾶς ὃς, for which there is no metrical necessity. He was probably swayed by Headlam, W.'s tendentious remarks in CR xvi (1902), 245Google Scholar. Equally unnecessary is Headlam's for at 1289: see PCPS n.s. xv (1969), 59Google Scholar for equally uneuphonious collocations of syllables.

^{page 9 note 1} Corrupted to ὡς or πῶς in various manuscripts.

^{page 9 note 2} See also below, p. 25.

^{page 9 note 3} See Conomis, N. C., Hermes xcii (1964), 43–5Google Scholar.

^{page 9 note 4} See Wecklein, , JfPh Suppl. vii (1874), 382Google Scholar. It is conceivable that P's πόλεως, although in itself merely a mistranscription of L's πόλεως, may nevertheless be right. According to Conomis, loc. cit. p. 23, there are 281 instances in Euripides of the dochmiac shape but only 13 of the shape .

^{page 10 note 1} I assume that no one will wish (having punctuated with a stop after ἔκγονος) to take ἀποβαλεῖς as the limit of the relative clause.

^{page 10 note 2} I omit instances of anadiplosis, like φέρετε φέρετε, and of pairs of compound and simple verb, like . The latter I have spoken of in GRBS xiv (1973), 265Google Scholar.

^{page 10 note 3} Text doubtful: post Blomfield Wecklein, ἐξορμάσης L.

^{page 10 note 4} πόλι Musurus, πόλις L; Θεσσαλίας Hermann, -ία L. On Hermann's conjecture, neglected by editors, see Housman, , CQ xxviii (1934), 127CrossRefGoogle Scholar = Classical Papers (Cambridge, 1972), p. 1224Google Scholar.

^{page 10 note 5} I mention in passing 1082, which ceases to be relevant when Wakefield's conjecture in 1081 is accepted, as it must be.

^{page 10 note 6} It would be futile to allege that is balanced or outweighed by the whole phrase .

^{page 11 note 1} Euripides, Hippolytos (Oxford, 1964), p. 434Google Scholar. For fuller discussion see Conomis (supra p. 9 n. 3), pp. 34–8.

^{page 11 note 2} Marg. scaen., p. 213.

^{page 11 note 3} Apud Musgrave, S., Exercitationum in Euripidem libri duo (Lugd. Bat., 1762), p. 170Google Scholar.

^{page 11 note 4} Classical Journal viii (1813), 213Google Scholar.

^{page 11 note 5} ‘“Lyreless” in these contexts means in effect “gloomy”, for dirges were regularly sung not to the lyre but to the flute, an instrument deemed most suitable to songs of lamentation’ Denniston and Page on A. Ag. 990; ‘ἄλυρος negatives the idea of joy, dance, festivity, usually associated with music…“Lyreless” thus commonly means “unfit for music”, “anything but festive”’ A. M. Dale on Alc. 447.

^{page 11 note 6} I leave out of account S. fr. 699 P (Bergk, ἄνανδα cod., ἄνανδα alii) καὶ ῥακτήρια. We have no context to help; but the expression looks like a simple oxymoron, “unmusical strains”, as Pearson says. And it would be unwise to found any argument on the possibility that Aeschylus wrote anything so odd as , “fluteless forehead”, in a fragmentary context at Pap. Oxy. 2251.8 = A. fr. 496.8 M (ἄναυδον Stinton apud Lloyd-Jones, Loeb edition, p. 572).

^{page 11 note 7} The sense of ἄναυδον here was first explained by Lloyd-Jones, H., CQ n.s. iii (1953), 96CrossRefGoogle Scholar.

^{page 12 note 1} Jackson, , Marg. scaen., p. 34Google Scholar n. 1, conjectured for , perhaps rightly.

^{page 12 note 2} See E. R. Dodds on Ba. 126–9, and The Greeks and the irrational (Berkeley, 1951), p. 273Google Scholar. For the flute as an accompaniment to a non-Dionysiac κῶμος see W. Headlam on Herodas 2. 34–7.

^{page 12 note 3} Aduersaria II (Cambridge, 1831), p. 118Google Scholar.

^{page 12 note 4} For the ‘orgiastic character’ of flute-music see Arist. Pol. 1341^a21. The flute was used in the worship not only of Dionysus but also of Cybele: Hel. 1351, Ba. 128, Pl. Crito 54 D, Men. Theoph. 27–8 (on which see Handley, E., BICS xvi (1969), 92–3Google Scholar), Cat. 63.22, Lucr. 2.620. See also Dodds, , The Greeks and the irrational, pp. 78, 273Google Scholar.

^{page 12 note 5} While I am speaking of the noun φόβος, let me commend, and assign to its correct author, its restoration at one other place in our play: 1218 . Wilamowitz's φόβον, which editors have ignored, is certain and excellent. But Wilamowitz was anticipated by Schmidt, F. W., Kritische Studien zu den griechischen Dramatikern II (Berlin, 1886), p. 198Google Scholar. And Schmidt was anticipated by Blomfield, C. J., Quarterly Review ix (1813), 353Google Scholar; this review (of Elmsley's Heraclidae) is anonymous, but is assigned to Blomfield by Luard, H. R., Journal of Classical and Sacred Philology iv (1858), 199Google Scholar.

^{page 12 note 6} See above, p. 11 n. 5, and Page, D. L. in Greek poetry and life (Oxford, 1936), pp. 206 ffGoogle Scholar.

^{page 12 note 7} See above, n. 2.

^{page 13 note 1} See Wilamowitz, p. 196. A contrast between the true Dionysiac madness and Lyssa's perversion of it is introduced in 892–5: (Hartung, -ιω L: for a different view see Jackson, , Marg. scaen., pp. 33–4Google Scholar) , (Scaliger, not Barnes, as Mr C. Collard shows in an article forthcoming in CQ, λώβας L). It must not be suggested, as it was by Pflugk, that ἀναύλοις may equally denote the reverse of the beneficent flute-induced trance of the votary of Dionysus. This is impossible, since the Dionysiac imagery has not yet been introduced by line 879, andἀναύλοις cannot therefore convey a negative allusion to it. I have learned nothing relevant to the present inquiry from Duchemin, J., ‘Le personnage de Lyssa dans l'Héraclès Furieux d'Euripide’, REG lxxx (1967), 130–9CrossRefGoogle Scholar.

^{page 13 note 2} Throughout this discussion I have translated αὐλός ‘flute’, for the sake of simplicity. The αὐλός differed in one essential respect from the modern flute and was more akin to the oboe or clarinet (see, for example, the article ‘Music’ in the Oxford Classical Dictionary (Oxford, 1970 ²), p. 710)Google Scholar; but this does not affect my argument.

^{page 13 note 3} Dale, A. M., The lyric metres of Greek drama (Cambridge, 1968 ²), p. 172Google Scholar adds Hel. 644. But uncertainty over the papyrus reading has complicated the issue here: see Zuntz (supra p. 4 n. 2), pp. 219, 225. For anapaestic monometer and iambus in a dochmiac context see El. 586 and 588.

^{page 13 note 4} For further instances see Dale, pp. 168, 175.

^{page 14 note 1} Aduersaria in Sophoclis Philoctetam (Leipzig, 1823), p. 71Google Scholar.

^{page 14 note 2} CQ n.s. xviii (1968), 267Google Scholar.

^{page 14 note 3} Robertson, D. S., PCPS clxvi (1937)Google Scholar, 1 also speaks of ‘the awkwardness of the datives’ and conjectures (with τὰ σὰδ᾿ for τάδ᾿ in the line above).

^{page 14 note 4} Parmentier compares 917–18 , ‘the father's destructive madness against his children’. Here ἄταν possesses a verbal force which φόνον does not have, and so it may justly be linked with παισί. The construction is similar to IT 387 (quoted by Wilamowitz); see also Kühner–Gerth 1, p. 428. No more effective is Pflugk's plea: ‘datiuum commodi διογενεῖ κόρωι posuit poeta pro gen. a nomine φόνον apto, ne genetiui cumularentur’.

^{page 15 note 1} Aduersaria Critica I (Copenhagen, 1871), p. 248Google Scholar.

^{page 15 note 2} Parmentier compares Tro. 1244–5.

^{page 15 note 3} With Wilamowitz's, not Murray's, colometry in these two passages.

^{page 15 note 4} Jackson, , Marg. scaen., p. 38Google Scholar, alleges that HF 1190 is another ‘clear case’. No so,since ὕδρας need not be scanned as a spondee (see Ion 191). Kannicht, R., Euripides, Helena (Heidelberg, 1969), II, p. 182Google Scholar, creates another example at Ion 1486, but only by taking κύκλωι as a spondee and by improbable colometry (the colometry of Wilamowitz and Murray is right here).

^{page 14 note 5} So Denniston, J. D. in Greek poetry and life (Oxford, 1936), pp. 137, 141–2Google Scholar; Conomis (supra p. 9 n. 3), 34–5.

^{page 14 note 6} So Wilamowitz, , here and in Griechische Verskunst, p. 407Google Scholar; Bond, G., Euripides, Hypsipyle (Oxford, 1963), p. 128Google Scholar; Barrett (supra p. 11 n. 1), p. 318.

^{page 14 note 7} So Jackson, , Marg. scaen., p. 39Google Scholar. The correption - is limited in tragedy to δείλαιος . For tragic and non-tragic instances see Denniston on El. 497, Headlam on Herodas 7. 39, M. L. West on Hes. Theog. 15, Gomme and Sandbach on Men. Epitr. 348.

^{page 14 note 8} Marg. scaen., pp. 37–9.

^{page 15 note 1} See Ritchie, W., The authenticity of the Rhesus of Euripides (Cambridge, 1964), p. 313Google Scholar. For aeolo-choriambics as clausulae to dochmiac passages see Conomis (supra p. 9 n. 3), 46.

^{page 15 note 2} I say that this is no change because L regularly omits adscript ν even where metre requires it. Triclinius restores it when he sees that it is required (1, 78, 285, 305, 595, 635, 852, 981, 1000, 1174, 1289, 1306, 1370). But he has failed to notice its omission at 366, 564, 651, 759, 885, 966, 1193.

^{page 15 note 3} 744 ∼ 758 appear to show a cretic (fully resolved) interposed between dochmiac dimeters; but the text is not certain. For further instances of cretics among dochmiacs see Conomis (supra p. 9 n. 3), 48.

^{page 15 note 4} The following editions are referred to: G. Hermann (Leipzig, 1827), C. Badham (London, 1851, 1853, 1861), F. A. Paley (London, 1874²), A. W. Verrall (Cambridge, 1890), M. A. Bayfield (London, 1891²), C. S. Jerram (Oxford, 1896), N. Wecklein (Leipzig, 1898 and 1912), G. Murray (Oxford, 1913³), H. Grégoire (ed. Budé, Paris, 1923), U. von Wilamowitz-Moellendorff (Berlin, 1926), A. S. Owen (Oxford, 1939).

^{page 15 note 5} The text of this line is sound. Elmsley's argument (on Med. 416 [426]) that Attic poets do not elide third-person -ε before ἄν is invalidated by Hec. 1113 παρέσχ᾿ ἄν (Heath, παρέσχεν ἄν uel παρεῑχεν codd.), Tr. 397 (Schaefer, P Chr. Pat., ), 399 εἷχ᾿ ἄν (Schaefer, εἷχεν uel εἷδεν codd.), Ba. 1312 (Heath, -βανεν P), S. El. 914 (Heath, -θ;ανεν codd.), Ar. Ran. 946 εἷπ᾿ ἄν. See also Harrison, E., PCPS clvi (1933), 10–11Google Scholar. For the word-division εἷχ᾿ ἄν in the fifth foot see S. El. 413, E. Hcld. 456, Andr. 935, 1184, Ph. 1619, 1626, Ba. 1271, IA 523 (Markland), fr. 362. 2 (Porson), and Porson, , Suppl. ad Hecubae praef. (1802), pp. xxxii–xxxiiiGoogle Scholar. [See now Daitz (supra p. 6 n. 1) for the mss. at Hec. 1113.]

^{page 16 note 1} This use of τί δ᾿ εἰ (ἤν) is illustrated from tragedy and comedy by Stevens, P. T., CQ xxxi (1937), 184Google Scholar. I should add to his list El. 978 τί δ᾿ ἤν (Nauck, iam τί δ᾿ ἄν Musgrave, τῶι δ᾿ ἤν Camper, τῶ δ* [δαὶ l] L) (Barnes, -θίηςL) τιμωρίαν, which is preferable to Porson's τῶι (on Ph. 1638 [1622]). A closely related locution is a question introduced by ἀλλ᾿ εἰ. Instances are given by Jackson, , Marg. scaen., p. 88Google Scholar, who rightly restores this locution at HF 1202 (Seidler, L) (Jackson restored the question mark after ἧλθον, where Seidler had placed only a comma). But a further instance in HF has gone undetected: 1072–6 (Amphitryo) (Wilamowitz, L) ἐπὶ (ed. Hervag., ὧ L) (l, τ- φ- L), . So editors print; but a stop is wanted after τάλας and a question mark after ἕξει.

^{page 16 note 2} This line has been the object of some very unhappy interpretations and emendations. For the most recent see Broadhead, H. D., Tragica (Christchurch, 1968), pp. 162–3Google Scholar.

^{page 16 note 3} Studien zu Euripides (Vienna, 1895), p. 93Google Scholar.

^{page 17 note 1} These conjectures are certain, though Verrall and Bayfield attempt to retain ὅ. The conjecture οὔ is ascribed by Wecklein and others to Hermann, but it should more correctly be ascribed to Dobree (Aduersaria II (1831), p. 111Google Scholar), since Dobree died in 1825. οὗ is similarly corrupted to ὅ at the beginning of the line at 245, and οὐ φιλῶ to ὀφείλω at the beginning of 526.

^{page 17 note 2} Adnotationes criticae et exegeticae in Euripidem (Amsterdam, 1874)Google Scholar, known to me only from Bursian iii (1874/1875), 438Google Scholar.

^{page 17 note 3} This line means simply ‘She has lost a son; I have lost a mother’. It cohered well enough with the old arrangement; it coheres no less well with the new.

^{page 17 note 4} See also Denniston, , Greek particles, p. 424Google Scholar.

^{page 18 note 1} Line 323 must be answered by 326. There are then only two plausible positions for 324–5. They must go either after 329, where Jacoby, C. placed them (JfPh cxiii (1876), 188–9)Google Scholar, or after 327, where they were placed by Wecklein, (Ars Sophoclis emendandi (Würzburg, 1869), p. 193)Google Scholar, Madvig, , Aduersaria I, p. 264Google Scholar, and Wecklein, , JfPh Suppl. vii (1874), 337Google Scholar. The latter transposition is adopted by Wilamowitz. But I should follow Jacoby, because Ion's suspicion in 325 that he may be the child of some wronged woman is the perfect spark for Creusa's ejaculation of sympathy in 330 and her introduction of the friend who is alleged to have suffered a similar mischance. His remark in 329 that he has no clue to the identity of his parents gives no such excuse for Creusa's φεῶ or for the introduction of her fictitious friend.

^{page 18 note 2} Cf. Dale, p. 171.

^{page 18 note 3} For this responsion see Phaethon 69 ∼ 77 ( = fr. 773.25 ∼ 33). Other instances of irregular responsion in aeolo-choriambics are given by Page, , Sappho and Alcaeus (Oxford, 1955), p. 81Google Scholar.

^{page 18 note 4} Griechische Verskunst, pp. 568–9.

^{page 18 note 5} BPhW xvii (1897), 1348Google Scholar.

^{page 19 note 1} For this possibility see G. Zuntz (supra p. 4 n. 2), pp. 193–201.

^{page 19 note 2} Editors' translations of καρποτρόφοι are often very lax. According to Badham (and, in the same words, Owen) it means that ‘these children will in their turn be fruitful’; similarly Verrall, ‘productive, because they reproduce themselves’; and Bayfield, ‘with hope of fruitfulness, i.e. giving hope that when grown up they too will have children’. This is what the context requires; but I do not see how the word can mean ‘fruit-nourishing, not now but in the future’. Wilamowitz saw the real implications of the word: ‘καρποτρόφοι… also selbst schon die Frucht von Kindern gebracht haben’.

^{page 19 note 3} Murray should not have printed κουροτρόφον here.

^{page 20 note 1} But not 922 (adduced by Owen and others), where καρποῖς is certainly a mistake for κάποις.

^{page 20 note 2} Contrast 735–6 . See also Denniston on El. 337 and my note on Phaethon 94.

^{page 20 note 3} Wecklein (in his 1912 edition) saves νέων by changing νέος to νέοι; but the plural is unlikely after a singular verb.

^{page 21 note 1} (a) 894–6 . The first two lines are regular dochmiacs; the resolution of the initial anceps in the third line should make us hesitant to take this line as dochmiac too (as it is taken by Dale, , Lyric metres, p. 60Google Scholar). We have the same sequence at 147–50; two dochmiacs and two lines of the length in an otherwise wholly anapaestic context. Perhaps iambic scansion is justifiable (). (b) On the metre of 906–9, which have been taken as dochmiacs, see below, p. 25.

^{page 21 note 2} (a) 883 is interpreted as a resolved paroemiac by Dale, , Lyric metres, p. 59Google Scholar n. 2. But tragedy provides no instance of the epic scansion κέρᾰσιν (A. fr. 185N [307a Mette] κέρᾱσι, S. Tr. 519 , E. Ba. 921 κέρᾱτα). Furthermore, as Professor Page points out to me, there is no analogy in tragedy either for word division between the two syllables of a resolved long in anapaests () or for a resolved paroemiac (or full dimeter) starting with resolution and continuing normally. We should perhaps restore a regular dimeter by writing κεράεσσιν (Madvig, Aduersaria I, p. 265Google Scholar). For the termination see 206 (τείχεσσι Murray, -εσι L), Alc. 756, Tr. 280, Ba. 76, 560, Phaethon 80, A. Pe. 553, Su. 95, 780, 1043, S. Ai. 374, Ant. 116, 976, 1297, OT 1100, Kühner–Blass I, p. 270. (b) 889–90 should be obelized. The rhythm of 889 is inscrutable (iambic or trochaic is alien to the context; not a resolved paroemiac, for the reasons given by Dale pp. 63–4, whose own analysis, however, can hardly be right). Burges's (Troades (Cambridge, 1807), p. 138Google Scholar, comparing for the middle Hel. 243) would restore a resolved anapaestic dimeter. But several linguistic difficulties remain. If, as we should expect, ἀνθίӡειν means ‘adorn with flowers’, then it must have an object. Wilamowitz's (Verskunst, p. 267 n. 3), which gives ἀνθίӡειν the unexampled sense ‘cull flowers’, carries no conviction. Nor does Musgrave's ἀνθισμόν. Since Appian (cited by LSJ) uses the middle of ἀνθίӡειν in the sense ‘cull flowers’, we might consider ἀνθιӡομένα. Further, the construction of φάρεσιν doubtful; and in place of χρυσανταυγῆ (here only) Paley plausibly suggested χρυσαυγῆ, comparing S. OC 685 . ‘Golden flowers’ do not appear only (as Owen asserts) in Sappho, but also at Pi. Ol. 2. 72. The association of flower-picking with abduction goes back to the Hymn to Demeter, 6 sqq. (for further instances of this motif see N. J. Richardson ad loc. (Oxford, 1974)). (c) 900 is another paroemiac with illicit resolution (see above on 889). Heath's slight change is much superior to Wilamowitz's (Verskunst, p. 368 n. 2).

^{page 22 note 1} Lyric metres, pp. 60 sqq.

^{page 22 note 2} Murray's colometry destroys the balance (as Dale observes, ad loc.). Read either . or (sol)/ στυγναὶ κτλ. or .

^{page 22 note 3} Cf. Collard, C., CQ n.s. xiii (1963), 187Google Scholar.

^{page 22 note 4} Triclinius deleted ἰώ here, and editors have followed him. But ἰώ (or Wilamowitz's ἰώ ⟨ἰώ⟩) is perfectly appropriate at the head of this feignedly enthusiastic speech of welcome to the queen. Similarly ἰώ (or ἰὼ ἰώ) is prefixed to welcoming anapaests at Su. 1114, IA 590, Rh. 379.

^{page 24 note 1} For monometer rather than dimeter following ἰώ see A. Ag. 1455 , where Fraenkel comments, ‘Editors who note here: “ἰώ semel codd.: bis Blomfield” do less than justice to Blomfield's circumspection. He says: “Notent tirones ἰὼ extra metrum esse; nisi legendum ἰώ ἰώ’”.

^{page 24 note 2} Better than Badham's deletion of καὶ σός and much better than Matthiae's deletion of καὶ.

^{page 24 note 3} Andr. 348 MAV) Dindorf) is no instance of a vocative τλήμων (P. T. Stevens's note ad loc. (Oxford, 1971) is misleading). This is the nominative, used in an exclamation: see Kühner–Gerth II, p. 46, Schwyzer, , Griechische Grammatik II, p. 65Google Scholar, Page on Med. 61, Broadhead on A. Pe. 733.

^{page 25 note 1} On which see Richardson (supra p. 22 n. 2), pp. 326–8. It is perhaps better to write Καλλιχόροισι with a capital.

^{page 26 note 1} ‘κύπριδας mut. in (compendium ους quod super ας scriptum erat erasum itemque accentus super ι), ἀθεμίτους P’, Wecklein. Bayfield proposed ἀθεμίτος, very plausibly. As Badham had said (proposing ἀθεμίτας), ‘aliter… ipsius Veneris nuncuparentur’. Once Κύπριδος had become Κύπριδας, the other change was almost inevitable.

^{page 26 note 2} Apart from (which is invariable), tragedy yields only these analogies: A. Pe. 901, S. Ant. 339, 607, - pars codd.) A. PV 185.

^{page 26 note 3} Dale ad loc. gives an improbable analysis of these lines.

^{page 26 note 4} Dale, , Lyric metres, p. 66Google Scholar, defends ἔνοπλον with the surprising plea that a resolved dactyl is ‘a still stranger anomaly’ than a pair of tribrachs in responsion with a pair of dactyls. For resolution of dactyls in non-tragic lyrics see Dale, p. 25 n. 2.

^{page 26 note 5} For divided resolution in iambics, trochaics and dochmiacs, see Parker, L. P. E., CQ n.s. xviii (1968), 241–69CrossRefGoogle Scholar. It is instructive to examine Euripides' practice in resolving one of the longs in the choriamb of his aeolo-choriambic cola. I can find only one example (and that not a striking one) of divided resolution. Here are the instances of resolution which I have found in Murray's text (it is possible that other instances might be added to this list): Alc. 971a ∼ 981a, Hcld. 777 (Zuntz's, G. remarks on the responsion, The political plays of Euripides (Manchester, 1955), p. 120Google Scholar, are refuted by several of the following examples), Su. 978, HF 641, 794 (but probably corrupt), Ion 122 ∼ 138 (unless 122 + 3 ∼ 138 + 9 are iambic trimeters), 497 (unless iambic dimeter), 1231 (unless iambic), Tr. 1065 (unless ἱράν), El. 126 (unless iambic dimeter), 445, 458, 709 ∼ 722, [718 corrupt], 732, [IT 425 ∼ 442 would give divided resolution in 442 if telesilleans; but for a better analysis see Platnauer's edition, p. 182 n. 1], IT 1101 (unless ἱράν), Hel. 1110 ∼ 1125, 1119, 1301 ∼ 1319, [1308–9 ∼ 1326–7 are iambic: ‘uncertain whether…iambic dimeters… or… resolved glyconics’, Dale, p. 150 of her edition; since glyconics would have three divided resolutions, the choice is clear], 1459, 1489; Ph. 206, 221, 227, 234 (unless ἱράν), 237, Ba. 107 ∼ 122, 123, 865, IA 165 ∼ 186, 183 ∼ 204, 207, 222, 771, 781, 794, 1038, 1048, Hyps. fr. 1. ii. 17. Against the unanimity of nearly forty unequivocal instances of undivided resolution I can set only Hyps. fr. I. ii. 23 1. iii. 26 .

^{page 27 note 1} On p. 145 Stinton considers the possibility of scanning , with the responsion , an abnormality which requires something better to commend it than his analysis of S. Tr. 825 ∼ 835.

^{page 27 note 2} On the colon 457 ∼ 469 see also Parker, L. P. E., CQ n.s. xviii (1968), 257Google Scholar.

^{page 27 note 3} On this see Parker, , CQ n.s. xvi (1966), 1–26CrossRefGoogle Scholar.

^{page 28 note 1} Elmsley's conjecture was proposed in the Classical Journal viii (1813), 214Google Scholar. Less satisfactory is Dobree's ἔπλεξ᾽. ἥ: a stop after the second foot is not common in Euripides (Denniston, , CQ xxx (1936), 73–9Google Scholar). To keep the negative and print a question mark after 1281 produces a not very natural question. This proposal is ascribed by Wecklein, to Mekler, S. (Euripidea (Vienna, 1879), p. 53)Google Scholar, but it had already been suggested by Lenting, , Noua acta lit. soc. Rheno-Traiectinae i (1821), 106Google Scholar, and had been adopted even earlier by Matthiae (Leipzig, 1815).

^{page 28 note 2} Our surprise is not to be allayed by any such artifices as this précis by Owen: ‘noticing that Creusa is at the altar (1275–8), [Ion] says that such sanctuary will not help her; finally (1279–81) he calls the attention of the others to her position.’ Such reassuring summaries are easy enough to write, but only when the eyes are closed and the words on the page have ebbed from the mind.

^{page 29 note 1} Similarly Verrall, ‘Thy appeal for mercy yields to mine, and to my mother's’; D. W. Lucas in his translation (London, 1949), ‘The pity you ask for I keep for myself and my mother’; Burnett, Anne Pippin, Ion by Euripides (Englewood Cliffs, 1970)Google Scholar, ‘And do not plead your sufferings, for all that you have felt, / and more, have been among the agonies my mother knew, and I.’

^{page 29 note 2} Nauck (μὴ μὴ κτλ. Boissonade), L. See Barrett on Hi. 503–4.

^{page 30 note 1} Conjecture has failed to effect any improvement. To the conjectures recorded by Wecklein add Herwerden, , Mnemosyne n.s. xxvii (1899), 237Google Scholar.

^{page 30 note 2} This puts an end to the further transposition proposed in this speech by Kirchhoff and adopted by Wecklein. He wished to place 1266–8 (the instructions to the servants to seize Creusa) to follow at the end of the speech after the lines already transposed to the end by Musgrave. Unlike Musgrave's proposal, this was never a compelling one. Since Ion does not see that Creusa has fled to the altar until 1279, there is no reason why he should not tell his men to seize her at this earlier point in the scene.

^{page 30 note 3} Indeed, as Professor Page warns me, the presence of these lines at this point in the text would be more easily explained (as would also the words .) if we regarded them not as an interpolation in the ordinary sense but as a marginal ‘parallel’ erroneously inserted. ‘Whoever wrote . meant just what Carter says, and it is wholly alien to this context. The insertion of the lines in an unsuitable place (for, if they are genuine or interpolated, Musgrave's transposition must be accepted) seems to me rather more likely if they are a scribe's incorporation of something from the margin.’ I should like to agree; but I wonder how likely it is that a lost play would provide room for the remark ‘Apollo's altar and temple will not save you; greater is the pity for you felt by me and by my absent mother’.

^{page 31 note 1} This statement, as we now know, but as Badham could not know, is wrong. P's omission of σ' is typical carelessness: see the apparatus criticus at e.g. 1199, 1293, 1335, 1348, 1360, 1362.

^{page 31 note 2} Paley has this not very lucid note: ‘θεός may be a monosyllable, as Dindorf suggests; for initial anapaests should be of one word; but perhaps we should read , or even omit the unnecessary σε’. He may have thought of this independently of Badham; but he is sometimes remiss in ascribing conjectures to their authors. Wecklein accepted Badham's conjecture in both editions (without comment). The problem was unknown to Wilamowitz (‘ὁ θεός wie oft zweisilbig zu sprechen… weil dannder Versanfang normal ist’), although he had once regarded P's reading as correct (Analecta Euripldea (Berlin, 1875), p. 27Google Scholar).

^{page 32 note 1} His main conclusion is that ‘die Tragiker die Synizesen von θεός und θεά nur in der Arsis des ersten und (namentlich) in der Thesis und Arsis des dritten Fußes angewendet haben’ (p. 247). The only exception which he admits is Su. 926 (see below, p. 33).

^{page 32 note 2} See Rumpel, p. 245 n. 9.

^{page 32 note 3} Rumpel counted 157.

^{page 32 note 4} As Rumpel agrees, p. 247.

^{page 32 note 5} Cf. Descroix, J., Le trimètre iambique (Mâcon, 1931), pp. 162–3Google Scholar.

^{page 32 note 6} Animaduersiones in poetas tragicos Graecos (1874), known to me only from Bursian iii (1874/1875), 405Google Scholar.

^{page 32 note 7} A further instance in this position was introduced by Seidler's ⟨τὸ⟩ θεοῦ at 315. And another by Stiblinus's conjecture at HF 1228 (τῶν del. Stiblinus). Here either Headlam's (CR xv (1901), 105Google Scholar) or Vitelli's τὰ θεῖα (RF viii (1880), 458Google Scholar) is likely to be right. Reiske's deletion of γε is no better than Stiblinus's deletion of τῶν, since could mean only ‘falls suffered by the gods’.

^{page 33 note 1} This comes in a section of the play which is perhaps not by Euripides: see Page, , Actors' interpolations in Greek tragedy (Oxford, 1934), p. 179Google Scholar.

^{page 33 note 2} The other conjectures on this verse are: Valckenaer, Heimsoeth, Hermann (supported by Rumpel, p. 247), Hartung, Paley.

^{page 33 note 3} Op. cit. (supra p. 32 n. 6).

^{page 33 note 4} I expect that further instances will be found among conjectures not accepted by Murray. Two which I have noticed are Hermann's conjectures at Hcld. 437 and IT 387.

^{page 34 note 1} This total includes several instances in lines which are possibly or certainly spurious, but I have not thought it worthwhile to complicate the discussion by separating such passages (except that I have ignored IA 1578 to the end, and two fragments mentioned in the next note). Zieliński, T., Tragodumenon libri tres (Cracow, 1925), pp. 199–200Google Scholar, counted 327 initial anapaests in words other than proper names. Add to this total a further 29 instances from Cycl., which he does not consider, and his total of 356 matches my total of 355.

^{page 34 note 2} I ignore fr. 112. 2 (λάλος Canter, ἄλλος Stob., alii alia). The text is uncertain, and the fragment may not belong to Euripides. Equally I ignore fr. 953, which contains three initial anapaests, two of them divided. These lines (containing among other faults two violations of Porson's law) are not by Euripides.

^{page 34 note 3} He ignores Cycl. 590; and perhaps I should have ignored this play too, since its metrical practices are not those of the tragedies.

^{page 35 note 1} But possibly not at Ion 1537–8 , (‘my mind is confused ⟨whether⟩ the god…’). The ellipse of εἰ is scarcely possible, and Reiske's εἰ θεὸς restores normal syntax. Wilamowitz retains ὁ θεὸς by placing a question mark after μαντεύεται. If this solution is adopted, I should prefer to write ταράσσει⟨ς⟩ with Blaydes, , Aduersaria cnttca in Eurlpidem (Halle, 1901), p. 198Google Scholar.

^{page 35 note 2} See above, p. 31 n. 2.

^{page 35 note 3} And less often than is implied by Porson on Or. 393 [399] (‘non raro’).

^{page 36 note 1} There are ten instances of synizesis in Euripides' trochaic tetrameters. In only one does a short syllable precede: Ion ;. Baier proposed , which violates the diaeresis. Perhaps ἒμηνεν; and yet Sophocles has in trochaic tetrameters at OT 1519. In the spoken iambic trimeters of Aeschylus synizesis is found only three times: 1¹Ch. 148, 3²Pe. 404, ScT 276. In Sophocles I have counted 27 instances (Rumpel counted 28) distributed as follows: 1² 10, 3¹ 10, 3² 7. In all except one of the Aeschylean and Sophoclean instances a long syllable precedes synizesis. The exception is Ph. 1036 . Jebb introduced a further exception by conjecture at El. 1264 (Jebb, ὅτε codd.) (Reiske, ὤτρυναν codd.) μολεῖν.