¹ I would like to thank Drs James Diggle, Stephen Harrison and Donald Mastronarde, and Sir Charles Willink for their acute criticisms.

² See especially Diggle, J., CQ n.s. 34 (1984), 50–1CrossRefGoogle Scholar, who argues for Sakorraphos' αὐτι and also for the reading πολίταις, without, however, being satisfied that the problems of ϕυγι are solved; and Harrison, S. J., CQ n.s. 36 (1986), 260CrossRefGoogle Scholar, who solves the latter by reviving Pierson's ϕυγς.

³ See Appendix below. Our century has not been inclined to see deep corruption here and has convinced itself that all is basically sound. Lenting's attitude, by contrast, though conservative in its result, was more gloomily realistic: ‘Hanc omnium et MSS. et edd. lectionem vitio laborare, nemo dubitat.’ Then, after citing evidence that the text was basically the same in late antiquity, he says ‘Corruptam igitur lectionem retinere malui, quam incertam alicuius conjecturam in textum recipere’ before going on to put forward a tentative suggestion.

⁴ E.g. by Wilamowitz, p. 821.

⁵ This point was made by Ritschl, pp. 148–9 (rpt., pp. 749–50): ‘da doch das οὐκ ν κατὠικει den Zeitbegriff κατοικεῖ gibt, wie kann von der Gegenwart πντα ϲυμϕρουϲ’ ’Ισονι ausgesagtwerden, wovon ja gerade das Gegentheil wahr ist?’ and by Hübner, p. 21: ‘Das Motiv κατώικει… | ξὺν νδρί – anders 32f. ϕκετο | μετ' νδρóc – führt vorübergehend sogar in die Irre.’ See also the similar argumentation of Schröer: ‘Dem οὐδ' ἂν κατώικει können also die Participia praesentis …nicht zu– und untergeordnet sein da beide abgeschlossen hinter Medea liegen: ausserdem kann νδνουϲα ( – günstiges für Medea – ) gedanklich nicht dem οὐδ' ἂν κατώικει ( – Verwiinschung – ) zugeordnet werden’. The impossibility of the conjectures whichfollow these sensible remarks in no way reduces their accuracy.

⁶ See the scholars cited in nn. 5 and 10.

⁷ Willink's ‘νûν δ in 16 does double duty’ puts the matter altogether too kindly.

⁸ Cf. Willink, n. 21: ‘we cannot be expected to “understand” anything adverse about Jason before we have heard 16ff.’.

⁹ The following are the only prima facie examples I could find of things other than main clauses paired with main clauses: Alc. 353 (‘accusative in apposition with the sentence’ with main clause), 591 (prepositional phrase: text corrupt), Med. 500 (parenthetic, with elliptical őμως δ᾽), Held. 997 (probably μν solitarium rather than paired with δ᾽ in 1000), Su. 873 (adjective phrase), El. 35 (but μν solitarium in Diggle's punctuation), Hel. 261 (prepositional phrase), Ph. 1421 (adverb), Ba. 224 (πρóϕασιν μν set against main clause), I.A. 569 (prepositional phrase). (Med. 1316 rightly diagnosed as unsound.) This allows us to say that the pairing of a phrase with a clause is not impossible (as we knew in any case from Denniston's Sophoclean examples) but that there are no other examples in Euripides of the sequence ‘main clause, participial phrase with μν, main clause with δ’. See, e.g. Sup. 34 and 204 and Alc. 338 for examples of the expected and logical use.

¹⁰ The same is true of the acute suggestion of Stephen Harrison, per litieras, to attack κατώικει, replacing it with a verb in the aorist meaning ‘come to’, thus allowing the participles to refer exclusively to past time. This line of attack has never been proposed before to my knowledge, and I pass it on in case some future critic can make use of, e.g.‘κατώικει] μετστη post Harrison Kovacs cl. Med. 551’ as part of a larger solution. Until such larger solution is put forward, however, its attractions are limited.

¹¹ Schneidewin suggested transposing 11 to follow 12 and writing καὐτ δ in 13, but with 12 following on 10 ϕἱκετο must refer to her arrival in Iolcus, ineptly introduced into a discussion of her departure; while νδνονϲα μν καὐτ δ πντα ξνμϕρονϲ’ ’ Ιϲονι, referring to Jason's love for Medea and her cooperation with him, seems oddly restricted to their arrival in Iolcus. Usener likewise transposed 11 and 12 but wanted to mark a lacuna of two or three lines after 10, a highly uneconomical procedure.

¹² Cf. the sequence in the prologue to the Andromache. There the first blow was the loss of city and husband. Then even in slavery Andromache formed hopes for the survival of her son by Neoptolemus. But subsequently, even these hopes are dashed by Hermione, Neoptolemus' new wife.

¹³ I am especially grateful to Charles Willink for criticizing an earlier and longer version of the supplement. The second supplementary line is substantially his.

¹⁴ See West, M. L., BICS 25 (1978), 118Google Scholar, on the statistical probability that there are lacunae in tragedy not so far detected.

¹⁵ I mention last a way of construing transmitted ϕνγ⋯ι πολιτ⋯ν that has been put forward at least since Lenting and has recently been championed by Roux, ‘by her avoidance of the citizens’. Roux argues that what the Nurse must have said was that Medea pleased Jason by living a retiring and unsociable life. For Medea herself, when she is trying to win the Chorus over to her side, remarks that some people win a reputation for slackness from their quiet manner of life, and she seems to be asking them not to hold her retiring ways against her. Yet, attractive as it is to retain a MS. reading seemingly confirmed by the rest of the play, it runs afoul of its context: if we supply ʼΙϲονι with νδνουϲα, then αὐτ⋯ι makes no sense, while transmitted αὐτ is just as difficult as before. If we understood νδρ κα τκνοιϲιν with νδνονϲα, then αὐτ⋯ι would make a certain amount of sense, separating the more important Jason off from the children. But it makes little sense to say that Medea's quiet life pleased husband and children, for though one could see why a Greek husband might be happy that his wife remained indoors, why would the children be pleased? If there is really a reference here to Medea's retiring way of life, we must be more courageous and adopt Nauck's λανθνουϲα. But at present, I favour Pierson's reading.