² See Turyn, 333–5; for the date, Wilson, N. G., Scrittura e Civiltà 7 (1983)Google Scholar. Fairly full but unreliable reports in Elmsley, Matthiae, and Kirchhoff(on whom see next note); thereafter only very occasional readings reported by editors. Collated by me from microfilm; doubtful readings checked by autopsy.

³ See Turyn, 358. Collated (negligently and incompletely, as he admits) by Elmsley; reported by Matthiae and by Kirchhoff, in his separate edition of Medea (Berlin, 1852)Google Scholar but not in his later edition of Eur. (Berlin, 1855). (My knowledge of the 1852 ed. is derived from Benedetto, V. Di, La Tradizione Manoscritta Euripidea (1965), 14–16Google Scholar; cf. also Matthiessen, K., Studien zur Textüberlierferung der Hekabe des Euripides (1974), 47.)Google Scholar Thereafter ignored by editors. Collated in Hi. by Barrett (from whom I have adopted the siglum C). Collated by me from photographs.

⁴ See Turyn, 335–7. Fairly full but unreliable reports in Elmsley, Matthiae and Kirchhoff; thereafter only very occasional readings reported by editors. Collated by me from microfilm; doubtful readings checked by autopsy.

⁵ See Turyn, 325–9. Not previously collated, except in Hi. by Barrett (from whom I have adopted the siglum E). Collated by me from a microfilm loaned by the Institut de Recherche et d'Histoire des Textes (Paris).

⁶ Dated ‘early 14th cent.’ by Turyn, , 360Google Scholar; but see Zuntz, G., An Inquiry into the Transmission of the Plays of Euripides (1965), 162 nGoogle Scholar. ‡, Matthiessen, K., GRBS 10 (1969), 299Google Scholar and Studien (n. 3 above), 47. Reported by Prinz and Wecklein; but no inferences should be drawn from their silence, which has led Murray and Page into one serious misstatement (F has νέον not κακόν at 37). Collated by me from photographs.

⁷ See Turyn, 329–31, Prinz, R., RhM 30 (1975), 129–33Google Scholar. Fairly full but unreliable reports in Elmsley, Matthiae, and Kirchhoff; thereafter only very occasional readings reported by editors. Collated by me from photographs.

⁸ See Turyn, 348–51, Matthiessen, , Studien, 129Google Scholar. Not previously collated. Collated by me from photographs.

⁹ I report these manuscripts from my own collations, made either from published facsimiles (HBLP) or from photographs (AV). Doubtful readings in B and L have been checked by autopsy. – By the symbol Tr I refer to Triclinius, corrector of L. That P is a copy of a copy of L in Medea (and the other annotated plays), first argued by Vitelli, (Museo Italiano di Antichità Classica 3 (1890), 287–300)Google Scholar and accepted by e.g. Barrett, 73, seems to me far more probable than the alternative hypothesis (of e.g. Turyn, 264 ff.) that P is a twin of Zuntz, L. (Inquiry, 35–8)Google Scholar offers evidence from Medea which strongly supports the former hypothesis, but afterwards (175–80) argues in favour of the latter (positing an intermediate copy between P and the source of L), on grounds which I find inadequate. Like Matthiessen, (Studien, 32–3)Google Scholar, but unlike Barrett, I do not regard acceptance of the former hypothesis as justifying the exclusion of P from the apparatus criticus.

¹⁰ See Longman, G. A., CQ n.s. 9 (1959), 129–41CrossRefGoogle Scholar.

¹¹ See Matthiessen, K., Hermes 93 (1965), 148–58Google Scholar.

¹² See Matthiessen, , Hermes 94 (1966), 398–410Google Scholar.

¹³ Pack, R. A., The Greek and Latin Literary Texts from Greco-Roman Egypt (ed. 2, 1967)Google Scholar.

¹⁴ Pack says that this papyrus is in the Fitzwilliam Museum, Cambridge. Only part of it is there (lines 1156–60, 1165–77, 1191–9). The rest is still in the Central Library of the Selly Oak Colleges, Birmingham (see Powell, J. E., The Rendel Harris Papyri (1936), v)Google Scholar.

¹⁵ I am indebted to the Institut de Papyrologie, Université des Sciences Humaines, Strasbourg, for sending me a microfilm of this papyrus, from which I was able to have made a photograph of excellent quality. Some of the comparatively little that has emerged from an inspection of this photograph is reported in Part II. Very much more has recently been achieved in the parts of the papyrus which preserve Phoenissae by Mastronarde, D. J., ZPE 38 (1980), 1–42Google Scholar. The papyrus is dated c. 250 B.C. by Turner, E. G., Greek Manuscripts of the Ancient World (1971), 60Google Scholar.

¹⁶ They roughly correspond with those adopted by Barrett, 92–3.

¹⁷ In this section no inferences about the readings of Hn and Nv should be drawn from my silence. Nor do I report testimonia.

¹⁸ In Hi. (the only other play which C preserves) there is no such relationship between O and C.

¹⁹ II ¹¹ omits 725–6 and continues with 729, 727, 728 in that order. This is surely the right solution to this problematic passage.

²⁰ The evidence of gE was not available to Barrett. The most telling evidence from C was not available to Matthiessen when he published the readings of gE, since it consists of the agreement between gE and C in unique errors, and Barrett (reasonably enough) does not record C's unique errors. I cite C from my own collation, made from photographs.

²¹ DE are also closely related in Hi.: see Barrett, , 69–72Google Scholar, Turyn, 336.

²² cf. Alc. 23 (V), Hi. 907Google Scholar.

²³ Note also that with this hypothesis we should have to accept that D went back to its original model in order to copy 769–903. For these lines cannot have been copied from the B-model: see 802 (τίcει BOAV: δώcει DLP: [CE]) and possibly 875 (κοιράνοιc BOCEAVL: τυράννοιc DP). The latter corruption is not necessarily evidence of affiliation, since it is so frequent (1299, IT 1080 (P), Ph. 1643, fr. 337, A. Ag. 549, [A.] PV 958, Ar Ach. 472).

²⁴ In his ‘Annotatio’ (for which see Part II, n. 1), p. 14.

²⁵ Coniecturae in Euripidem (1822), 28–30.

²⁶ But although not attested, this (and not εὐν⋯τηc) is the form which we should expect: see Fraenkel, E., Geschichte der griechischen Nomina agentis auf—τ⋯ρ, τωρ, -τηϲ ii (1912), 126Google Scholar, Redard, G., Les noms grecs en -τηϲ, -τιϲ (1949), 6Google Scholar. On the variations in the mss. between the forms εὐνα- and εὐνη- see Björck, G., Das Alpha Impurum unddie tragische Kunstsprache (1950), 139–40Google Scholar.

²⁷ εὐν⋯τοραϲ VP unmetrically: εὐν⋯ϲ Seidler (rightly, I think), ἄνδραϲ Beyer.

²⁸ See p. 353.

²⁹ 162 is corrupt: see Diggle, , Studies 32Google Scholar. I exclude El. 700 ~ 714, where the traditional division is objectionable: see Dawe, , Studies i. 256Google Scholar.

³⁰ See Diggle, , Studies 97Google Scholar.

³¹ See Dale, A. M., The Lyric Metres of Greek Drama (ed. 2, 1968), 168Google Scholar. On S. OC 1564 ~ 1575 (adduced by Dale) see Pohlsander, H. A., Metrical Studies in the Lyrics of Sophocles (1964), 86Google Scholar. On Ale. 252 ~ 259 (adduced by Stinton) see Dale, 164.

³² cf. Wilamowitz, , Griechische Verskunst (1921), 412Google Scholar, Denniston, J. D. in Greek Poetry and Life: Essays presented to Gilbert Murray (1936), 141Google Scholar, Parker, L. P. E., CQ n.s. 18 (1968), 246 n. lCrossRefGoogle Scholar.

³³ See Dale, 102. This might be interpreted (bacchiac + dochmiac): see Part II (on 1255–7 ~ 1265–7). For resolution of the second long of the bacchiac see Dale, 74.

³⁴ See my note on the passage in ICS 6. 1 (1981), 95–8Google Scholar, Stinton, , JHS 91 (1977), 143Google Scholar.

³⁵ loc. cit. 132.

³⁶ θρηνεῖ (with Wunder's conjecture) is accepted by R. D. Dawe (Teubner, 1979). Dr Dawe reminds me that θροεῖν is a well attested variant for θρηνεῖν at S. Ai. 582.

³⁷ The fragment is not Euripidean: see Diggle, , Studies 85, 104Google Scholar.

³⁸ CR 4 (1890), 10Google Scholar = Classical Papers 116.

³⁹ Even a second subject added non-parenthetically need not affect the gender: Here. 774–6 ⋯ χρυϲ⋯ϲ ἄ τ' εὐτυχ⋯α…δ⋯ναϲιν ἄδικον ⋯φ⋯λκων.

⁴⁰ Miscellanea Critica (1745), 97 (= ed. Kidd [1827], 188–9).

⁴¹ Kritische Studien zu den griechischen Dramatikern ii (1886), 333–4Google Scholar.

⁴² SBAW 143 (1901), 8–9Google Scholar; cf. Weil, , REG 13 (1900), 417Google Scholar, Pistelli, E., SIFC 11 (1903), 446Google Scholar.

⁴³ SIFC 25 (1951), 80Google Scholar.

⁴⁴ Bemerkungen zum Text der Medea des Euripides (diss. Heidelberg, 1962), 53–7Google Scholar; an excellent dissertation, which deserves to be better known.

⁴⁵ Roberts believed that a second hand may have written α over ουϲ. Dr R. A. Coles, who kindly inspected the papyrus for me under the microscope, believes that the trace is more likely to be the accent above ι.

⁴⁶ See p. 352.

⁴⁷ For similar attempts in our manuscripts to remedy a n accidental omission see 588 καλ⋯ϲ γ' ἄν οἶμαι (Nauck: οὖν μοι DLP: οὖν ϲὺ OAV: οὖν ϲοι C: οὖν BE) τ⋯ιδ' ὑπηρ⋯τειϲ (OCDALPV²: -⋯τηϲ V: -⋯τηϲαϲ E: ⋯ξυπτηρ⋯τειϲ B) λ⋯γωι, where the variants appear to be different attempts to amend καλ⋯ϲ γ' ἂν οὖν τ⋯ιδ' ὑπηρ⋯τετιϲ; 1139 δμ⋯εϲ δι' ⋯των δ' εὐθὺϲ ἦν πολὺϲ (ἦν πολὺϲ EAVLP: ἦν BD: ⋯μ⋯ν ἦν O) λ⋯γοϲ.

⁴⁸ On these manuscripts (and Neap. II. F. 41, mentioned below) see Barrett, 76. I have discussed their relationship with the other manuscripts of Hi. in CQ n.s. 33 (1983), 34–43CrossRefGoogle Scholar.

⁴⁹ Or. 512 is rather different (see Fraenkel, 414 n. l).

⁵⁰ ‘The ⋯γὼ is here clearly emphatic: “fear not for your children; I will take care of them”’ (Paley). But to claim that the pronoun is indispensable (Matthiae) is to go too far.

⁵¹ Reiske first drew attention to the anomaly: ‘non placet π⋯ρι, non quidem ideo, quod versum proxime superiorem pariter claudat, sed magis, quod incongrua dictio sit θ⋯ϲω περ⋯ τ⋯νδε εὖ pro θ⋯ϲω τ⋯ τ⋯νδε εὖ'. At Ant. 34 (in (iii) above) supply τ⋯ δ' ἔνδον ⋯μ]εῖϲ (Arnim) or the like.

⁵² Editors of Ag. do not mention Med. 926. But Wilamowitz, (Hermes 15 [1880], 497)Google Scholar quotes the restoration of Ag. 1672–3 in support of τ⋯νδ' ⋯γὼ θ⋯ϲω π⋯ρι.

⁵³ Leo, (Hermes 15 [1880], 315 n. 1)Google Scholar proposed τ⋯νδ' ⋯γὼ θ⋯ϲω β⋯ον. This was listed by Wecklein, (Bursian 1881, 49)Google Scholar as one of Leo's ‘wenig probable und zum Theil unmethodische Vermuthungen'. In his 1891 edition Wecklein misreported Leo's conjecture as θ⋯ϲομαι β⋯ον. In his 1899 edition he reported ‘τ⋯νδ' ⋯γὼ θ⋯ϲω β⋯ον…coniciebam’.

⁵⁴ Blaydes's, π⋯τμον (Adversaria Critica in Euripidem (1901), 34)Google Scholar is a less choice word. Other emendations may be found in Wecklein.

⁵⁵ Adv. Cril. 41; wrongly attributed to Wecklein by Diehl.

⁵⁶ For the agreement of Hn with D and LP see pp. 354–5. Editors, misled by Matthiae's note, wrongly ascribe the deletion to Seidler. See Seidler's Epistola Critica appended to Lobeck's, Ajax (1809), 433Google Scholar.

⁵⁷ In his 1886 edition, known to me only from Bursian 1886, 296. I have written δ⋯ϲτηνε for δ⋯ϲτηνε, since these cannot be lyric anapaests.

⁵⁸ Herwerden's, π⋯ντηι π⋯ντωϲ (Mnem. n.s. 5 [1877], 25)Google Scholar should be accepted; but instead of π⋯ντηι we need π⋯νται or π⋯ντᾱ: cf. Schwyzer i. 550, Barrett on Hi. 563.

⁵⁹ On this manuscript see n. 48.

⁶⁰ On this manuscript see n. 48.

⁶¹ I owe my information about the mss of Hecuba to the generosity of Dr Matthiessen. For s explanation of the sigla see his Studien (above, n. 3).

⁶² This is an interesting case. All mss except O have τε in one place or another, a few have it in both places. The majority have it before θε⋯ϲ: highly abnormal word-order, though accepted by most editors (the scholia show disquiet: Σ^m ⋯ τε πλεον⋯ζει, Σ^rec το⋯το τ⋯ τε πρ⋯ϲ τ⋯ χρυϲ⋯αν ἄμπυκα ϲ⋯ναπτε). Few mss have it after χρυϲ⋯αν, which (if it is to be accepted) is its proper place. If we do not accept it there, we must omit it altogether, as Murray does (with O). But if O is right, it is hard to see why it got into the text. It is more plausible to assume that it originally stood after χρυϲ⋯αν, was accidentally omitted, was placed above the line, and was then replaced in the wrong position.

⁶³ Page's report of the reading of ‘plurimi’ in his OCT is mistaken, as Zuntz, G. points out (PCPS n.s. 27 [1981], 93)Google Scholar.

⁶⁴ See Dawe, , Studies i. 245–6Google Scholar.

⁶⁵ For the rhythm (third foot monosyllable followed by tribrach) see Diggle, , Studies 112Google Scholar.

⁶⁶ See Dawe, , Studies i. 253Google Scholar.

⁶⁷ In order to avoid repetition I have sometimes not transcribed the readings of the other mss in full but instead have referred to an earlier page where this information may be found.

⁶⁸ He was not the first to pronounce this verdict. See, for example, Wilamowitz, , Analecta Euripidea (1875), 254Google Scholar.

⁶⁹ See pp. 355, 356–7.

⁷⁰ On this passage see most recently Haslam, M. W., Arktouros: Hellenic Studies presented to Bernard M. W. Knox (1979), 95–6Google Scholar.

⁷¹ I am not convinced by the elaborate attempt of Christmann, , op. cit. (n. 44), 47–53Google Scholar, to show that the verse (with ἥλικαϲ) is indispensable.

⁷² For another trace of a connection between Nv and Σ^o see p. 355.

⁷³ Euripideische Studien i (Mém. Ak. Imp. Sc. St. Pétersbourg, , ser. vii. 1 [1859], 106–39), 133Google Scholar.

⁷⁴ Adversaria Critica, 39.