¹ Kannicht, R. and Snell, B. (edd.), Tragicorum Graecorum Fragmenta, Vol. II Fragmenta adespota (Göttingen, 1981).Google Scholar

² Nauck, A. (ed.), Tragicorum Graecorum Fragmenta² (Leipzig, 1889; repr. with a Supplement by B. Snell, Hildesheim, 1964).Google Scholar

³ TrGF 2.617–24 ‘ex fabrica falsarii ludaei (saec. I^a?)’. In the apparatus to fr. 624, Kannicht and Snell refer to Denis, A.-M., Introduction aux pseudépigraphes grecs d'Ancien Testament (Leiden, 1970), p. 226Google Scholar with n. 15, but this reference is not very helpful since Denis, loc. cit., only gives a summary of Ps.-Justin's De monarchia.

⁴ It is rather difficult to give an exact date for the treatise. It is commonly attributed to the second century a.d. (cf. Eiter, A., De gnomologiorum Graecorum historia atque origine [Universitätsprogramm Bonn, 1893–1897], p. 203Google Scholar; von Harnack, A., Die Überlieferung der griechischen Apologeten des 2. Jahrhunderts in der alten Kirche und im Mittelalter. [TU 1.1] [Leipzig, 1882], pp. 154f.Google Scholar; von Harnack, A., Geschichte der altchristlichen Literatur² Bd. II.1 [Leipzig, 1958 = 1897¹], p. 512Google Scholar; Harris, R., Justin Martyr and Menander [Cambridge, 1932], p. 25Google Scholar ‘earlier than Justin’; Walter, N., Der Thoraausleger Aristobulos [TU 86] [Berlin, 1964], p. 211Google Scholar; Hengel, M., ‘Anonymität, Pseudepigraphie und “Literarische Fälschung” in der jüdischhellenistischen Literatur’, Entretiens de la Fond. Hardt 18 [Genf, 1972], p. 294 n. 2Google Scholar ‘Ende 2. oder Anfang des 3. Jahrhunderts n.Chr.’; Vermander, J.-M., ‘La parution de l'ouvrage de Celse et la datation de quelques apologies’, Rev. des études aug. 18 (1972), 32fGoogle Scholar; cf. also Bardenhewer, O., Geschichte der altkirchlichen Literatur² Bd. 1 [Freiburg i.Br., 1913], pp. 236fGoogle Scholar; Speyer, W., Die literarische Fälschung im heidnischen und christlichen Altertum [München, 1971], p. 161Google Scholar; Vorst, N. Zeegers-Vander, Les citations des poètes grecs chez les apologistes chrétiens du II^e siècle [Louvain, 1972], pp. 8–9 and 18Google Scholar). But it seems to me quite certain – for reasons I cannot expand in detail in this article – that De mon. belongs to the first century b.c. or a.d. Scholars used to oscillate between whether the treatise had been compiled by a Jew or a Christian. Yet nothing within the six chapters presupposes Christian doctrine, whereas on the other hand we find many thoughts and concepts which are well known from the late writings of the LXX and Philo.

⁵ The whole passage of Clement is cited by Eusebius in his Praeparatio evangelica 13.13 (§47 for TrGF 2.624). Since Clement's Stromata are transmitted in only one manuscript, Eusebius serves as an important independent witness for the constitution of Clement's text.

⁶ (a) Ps.-Justin and Clement use the same source: Harris, , op. cit. (n. 4), pp. 17f.Google Scholar; Harnack, , loc. cit. (n. 4)Google Scholar; Walter, , op. cit. (n. 4), p. 186Google Scholar (in connection with the so-called Testament of Orpheus); Hengel, , op. cit. (n. 4), p. 295Google Scholar; Schürer, E., The History of the Jewish People in the Age of Jesus Christ², Vol. III. 1 (Edinburgh, 1986), p. 657Google Scholar; (b) Clement is dependent on De monarchia: Elter, A., op. cit. (n. 4), pp. 188, 198, 201 ff.Google Scholar

⁷ Valckenaer, L. C., Diatribe de Aristobulo Judaeo, philosopha peripatetico Alexandrino (published posthumously by J. Luzac) (Leiden, 1806), pp. 1–5.Google Scholar References to his discussion will be given with name and page only.

⁸ Euripides seems to have written two tragedies called Phrixus, see below part IV.

⁹ The lines are quoted again in Math. 1.287, but without mentioning the name of the play: Κα⋯ μ⋯ν ὅσον ⋯π⋯ τ ὑπ᾽ Εὐριπ⋯δου λεχθ⋯ντι περ⋯ θεν, τ⋯ν αὐτ⋯ν κα⋯ οἱ ἰδιται δ⋯αν ἔχουσιν. ἴσον γ⋯ρ ⋯στι τ ῞Οστις δ⋯.… Δ⋯κη, κα⋯ τ⋯ οὅτω παρ⋯ τος πολλος λεγ⋯μενον, ᾽Οψ⋯ θεν ⋯λ⋯ουσι μ⋯λοι, ⋯λ⋯ουσι δ⋯ λεπτ⋯. Stobaeus, Ecl. 1.3.15 alone adds a fifth line, wrongly printed among the ‘adespota’ in TrGF 2.490, as will be pointed out later (cf. n. 80).

¹⁰ cf. Stob. loc. cit. Εὐριπ⋯δης Φρ⋯ῳ.

¹¹ The manuscripts of Sextus give in this passage the reading καθ᾽, but τοὐφ᾽ in 1.287 (cf. n. 9) which is supported by Ps.-Justin, Clement (and Eusebius) and Stobaeus.

¹² Bothe, F. H., Euripidis fabularum fragmenta (Leipzig, 1844), p. 300Google Scholar, adopted the reading of the Stobean codex F κ⋯δικν. But κα⋯ δοκν is generally attested and can be regarded as ‘lectio difficilior’. Cf. van Looy, H., Zes verloren tragedies van Euripides (Brussels, 1964), p. 153.Google Scholar

¹³ For the probable context of the fragment, see below part IV. One could, of course, always play the devil's advocate in suggesting that the passage may be a sententious interpolation. But since Euripides is generally quite fond of gnomic sentences, the ‘onus probandi’ lies, I think, on the side of people who question the testimony of our sources. There is, at any rate, nothing whatsoever in either language or thought of fr. 835 which speaks against a Euripidean origin of the lines. For ὅστις δ⋯ θνητν οἴεται … | … λεληθ⋯ναι, | δοκε πονηρ⋯, cf. Oinomaos, fr. 575 N² ὅστις δ⋯ θνητν βο⋯λεται δυσώνμον | εἰς γρας ⋯λθεν, οὐ λογ⋯ζεται καλς κτλ. and Tro. 1203 θνητν δ⋯ μρος ὅστις ε πρ⋯σσειν δοκν | β⋯βαια χα⋯ρει κτλ.; for τοὐφ᾽ ⋯μ⋯ραν, see Cycl. 336Google Scholar ὡς τοὐμπιεν γε κα⋯ φαγεν τοὐφ᾽ ⋯μ⋯ραν; ⋯λ⋯σκεται (line 3) occurs five times elsewhere in Euripides at the end of an iambic trimeter (Med. 84Google Scholar, Hipp. 913Google Scholar, IT 1038 and 1419, fr. 811 N²)Google Scholar; for ὅταν σχολ⋯ν ἄγουσαν … cf. Med. 1238 κα⋯ μ⋯ σχολ⋯ν ἄγουσαν … For the concept of Dike, who sometimes may be inconspicuous and tardy but will always exert her power in the end, cf. Antiope, fr. 223Google Scholar N² (see below part II), fr. 979 N² (οὔτοι προσελθοσ᾽ ⋯ Δ⋯κη σε, μ⋯ τρ⋯σῃς, | πα⋯σει πρ⋯ς παρ οὐδ⋯ τν ἄλλων βροτν | τ⋯ν ἄδικον, ⋯λλ⋯ σγα κα⋯ βραδε ποδ⋯ | στε⋯χουσα μ⋯ρψει τοὺς κακο⋯ς, ὅταν τ⋯χῃ); El. 771, Archelaos, fr. 255 N² etc.; see in general Lloyd-Jones, H., The Justice of Zeus² (Berkeley and Los Angeles, 1983), pp. 151 ff.Google Scholar

¹⁴ cf. Marcovich, M., ‘Patristic Textual Criticism’, ICS 13 (1988), 143f.Google Scholar; Mette, H. J., ‘Euripides (insbesondere für die Jahre 1939–1968), Erster Hauptteil: Die Bruchstücke’, Lustrum 12 (1967), 282Google Scholar (‘beide Wörter versehentlich um 2 Zeilen nach oben versetzt’).-Walter, N., Gefälschte Verse auf Namen griechischer Dichter. In: Jüdische Schriften aus hellenistischrömischer Zeit (hg. Kümmel, von W. G.) Bd. IV (Gütersloh, 1983), p. 246Google Scholar hazards a guess that κα⋯ Εὐριπ⋯δης might be explained as a ‘sekundäre “Verbesserung”’, but this seems to me to be very unlikely. If κα⋯ Εὐριπ⋯δης was already misplaced in Clement's time, it would be easy to understand why he has disregarded the phrase.

¹⁵ The two most prominent examples are: (1) chapter 2.4 where the so-called Testament of Orpheus is quoted without the first two lines (this has commonly been regarded as a result of a mechanical distortion, cf. Elter, A., op. cit. (n. 4), p. 157Google Scholar; Erbse, H., Fragmente griechischer Theosophien (Hamburg, 1941), p. 17Google Scholar n. 37; Walter, N., op. cit. (n. 4), p. 211Google Scholar; (2) De mon. 3.1 + 2 (lines 6f. of Diphilus PCG 5.136 occur in our MSS. some six lines too early where they are totally out of place).

¹⁶ Pseudo-Iustini Cohortatio ad Graecos, De Monarchia, Oratio ad Graecos (Patristische Texte und Studien, Verlag W. de Gruyter [Berlin and New York]). I should like to thank Prof. Marcovich very much for his outstanding generosity in letting me use the typescript of his new critical edition.

¹⁷ Marcovich, , loc. cit. (n. 14).Google Scholar

¹⁸ cf. p. 3. All later discussions of the fragment are in one way or another indebted to the work of this scholar (cf. the note in the apparatus criticus TrGF 2.617–24 ‘duce Valckenaerio’; Matthiae, A., Euripidis tragoediae, Vol. IX [Leipzig, 1829], p. 298Google Scholar; Fix, Th., Euripidis fabulae [Paris, 1843], p. 822 etc.).Google Scholar

¹⁹ cf. above n. 5.

²⁰ cf. Matthiae, A., op. cit. (n. 18), p. 198Google Scholar ‘κακς facile scribi potuit propter κακ⋯ς, quod sequitur’.

²¹ I. C. Th. von Otto (ed.), Iustini philosophi et martyris opera quae feruntur omnia ³ (Corpus Apologetarum Christianorum saeculi secundi, vol. III) (Jena, 1879; 1842¹; 1849²), p. 139 n. 23. It is worth noticing that Clement and Eusebius have πρ⋯ττει for πρ⋯σσει; the latter is the correct form for Euripides (cf. fr. 835.2 N², IT 668, and elsewhere).Google Scholar

²² κερν⋯τω must be a scribe's error; cf. Stählin, O, Clemens Alexandrinus³ ii (GCS) (Berlin, 1960)Google Scholarad loc. and Kannicht-Snell TrGF 2.624 ad loc.

²³ See n. 36.

²⁴ Βροντς ⋯⋯ν ⋯κο⋯σῃς μ⋯ φ⋯γῃς vs. Ps.-Justin, De mon. 4.2 Βροντς ⋯κο⋯σας μηδαμς π⋯ρρω φ⋯γῃς; cf. ibid, line 1 (δικα⋯οις ἔργοις for the metrically correct ἔργοις δικα⋯οις); see, moreover, Strom. 5.121.1 (Diphil. PCG 5.136, quoted in Ps.-Justin, De mon. 3.2).

²⁵ cf. the parallels quoted in n. 13 (there is no εἴ τις … θνητν at the beginning of a trimeter in: Euripides).

²⁶ For the convenience of the reader, I give the text with critical notes, which entails the repetition of things mostly already mentioned above.

²⁷ Ps.-Justin: Clement (and Eusebius, cf. n. 5) δοκετε.

²⁸ Ps.-Justin and Clement (and Eusebius) θε⋯ν.

²⁹ The line is omitted by Clement (and Eusebius, cf. above and below [n. 36]). Kannicht-Snell's punctuation (semicolon) is not appropriate, see below part III.

³⁰ Ps.-Justin and Clement (and Eusebius) ἔστιν γ⋯ρ, ἔστιν.

³¹ Ps.-Justin: Clement (and Eusebius) πρ⋯ττει (cf. above n. 21).

³² Ps.-Justin: Clement (and Eusebius) κακς.

³³ Grotius, H., Excerpta ex Iragoedüs et comoediis Graecis (Paris, 1626), p. 416Google Scholar – followed by Barnes, J., Euripidis quae exstant omnia (Cambridge, 1694), p. 503Google Scholar, and Valckenaer p. 5 – reads ὑστ⋯ρῳ, contra necessitatem.

³⁴ This is, in a way, already suggested by the association with fragments of Diphilus and Euripides in both Ps.-Justin and Clement (see above).

³⁵ ‘si sententiam [sc. spectes], minus aptum nee Euripidi congruum [sc. versum]’ (p. 5).

³⁶ cf. Valckenaer, , p. 5Google Scholar ‘Noster ille posuit quo sensu Judaei dicebant et Christiani ⋯μαρτ⋯νειν εἰς θε⋯ν’. I have a feeling that this may, in fact, have been the reason why Clement avoided quoting the line; for there are other passages where one gets the impression that Clement considerately leaves out what he – sometimes rightly – thinks is suspicious in his source (cf. Strom. 5.119.2 = Men. fr. 683 Körte).

³⁷ The problem of tragic ⋯μαρτ⋯α has formed the subject of numerous books and articles; one may refer to Bremer, J. M., Hamartia. Tragic Error in the Poetics of Aristotle and in Greek Tragedy (Amsterdam, 1969)Google Scholar; Stinton, T. C. W., ‘Hamartia in Aristotle and Greek Tragedy’, CQ 25 (1975), 221–54CrossRefGoogle Scholar; and Saïd, S., La faute tragique (Paris, 1978).Google Scholar

³⁸ See Bremer, , op. cit. (n. 37), pp. 31ff.Google Scholar, and Saïd, , op. cit. (n. 37), p. 44.Google Scholar

³⁹ A careful discussion of the history of the word-group and the changes of meaning from one author to the other is given in Saïd, , op. cit. (n. 37), pp. 41 ff.Google Scholar

⁴⁰ See Saïd, , op. cit. (n. 37), p. 411Google Scholar; cf. also Aphrodite's words in the Prologue of Hipp. 21Google Scholar ἃ δ᾽ εἰς ἔμ᾽ ⋯μ⋯ρτηκε τιμωρ⋯σομαι | Ίππ⋯λυτον ⋯ν τδ᾽ ⋯μ⋯ρᾳ with the comments of Stinton, , art. cit. (n. 37), 247fGoogle Scholar, and Saïd, , op. cit. (n. 37), p. 413.Google Scholar

⁴¹ cf. the comments ad loc. by Saïd, , op. cit. (n. 37), pp. 319ff.Google Scholar One may, moreover, refer to Soph. Ant. 743Google Scholar Αι. οὐ γ⋯ρ σ⋯και⋯ σ᾽ ⋯αμαρτ⋯νονθ᾽ ⋯ρ. | Κρ.⋯μαρτ⋯νω γ⋯ρ τ⋯ς ⋯μ⋯ς ⋯ρχ⋯ς σ⋯βων; | Αι. οὐ γ⋯ρ σ⋯βεις, τιμ⋯ς γε τ⋯ς θεν πατν, and Eur. Hipp. 1433Google Scholar ⋯νθρώποισι δ⋯ | θεν διδ⋯ντων εἰκ⋯ς ⋯αμαρτ⋯νειν.

⁴² This is indeed the reason why Ps.-Justin has quoted the fragment, cf. the introduction to chapter 3 of De monarchia, Κα⋯ περ⋯ τοδε, ὅτι μ⋯νος δυνατ⋯ς ⋯στι κα⋯ τν ⋯ν τ β⋯ῳ συντελουμ⋯νων πρ⋯εων κα⋯ τς περ⋯ τ⋯ θεον ⋯γνωσ⋯ας κρ⋯σιν ⋯νστ⋯σασθαι, οἰκε⋯ους μ⋯ρτυρας παραστσαι ἕχω.

⁴³ cf. lines 25ff. τοια⋯τη Ζην⋯ς π⋯λεται τ⋯σις. οὐδ᾽ ⋯φ᾽ ⋯κ⋯στῳ | ὥσπερ θνητ⋯ς ⋯ν⋯ρ γ⋯γνεται ⋯⋯χολος, | ⋯ε⋯ δ᾽ οὔ ⋯ λ⋯ληθε διαμπερ⋯ς, ὅστις ⋯λιτρ⋯ν | θυμ⋯ν ἔχει, π⋯ντως δ᾽ ⋯ς τ⋯λος⋯εφ⋯νη. | ⋯λλ⋯᾽ ⋯ μ⋯ν αὐτ⋯κ᾽ ⋯ δ᾽ κτλ.

⁴⁴ Both fr. 835 and TrGF 2.624 similarly express this idea, see above n. 13 and below part IV.

⁴⁵ cf., moreover, Bellerophontes, fr. 303.3–5 N², Alexandr., fr. 60 N², and Alope, fr. 112 N² (three passages which stress the importance of time as bringing everything to light). In El. 953, Euripides says we should not allow ourselves to be blinded by the temporal well-being of the wicked, for they cannot overcome Dike in the long run (δ⋯ τις κακοργος ⋯ν | μ⋯ μοι τ⋯ πρτον βμ᾽ ⋯⋯ν δρ⋯μῃ καλς, | νικâν δοκε⋯τω τ⋯ν Δ⋯κην, πρ⋯ν ἂν π⋯λας | γραμμς ἵκηται κα⋯ τ⋯λος κ⋯μψῃ β⋯ου).

⁴⁶ See above part I.

⁴⁷ cf. Supp. 731 νν τνδ᾽ ἄελπτον ⋯μ⋯ραν ἰδοσ᾽ ⋯γὼ | θεοὺς νομ⋯ζω … | … τνδε τεισ⋯ντων δ⋯κην, perhaps also fr. 913 N² τ⋯ς τ⋯δε λε⋯σσων θε⋯ν οὐχ⋯ νοε …; see already Hom. Od. 24.351 Ζε π⋯τερ, ῥ᾽ ἔτ᾽ ⋯στ⋯ θεο⋯ κατ⋯ Ὄλυμπον, | εἰ ⋯τε⋯ν μνηστρες ⋯τ⋯σθαλον ὕβριν ἔτεισαν; see also G. W. Bond's commentary on Euripides' Hercules furens (Oxford, 1981), lines 772f.Google Scholar

⁴⁸ cf. Herc. fur. 841Google Scholar … ἤ θεο⋯ μ⋯ν οὐδαμο, | τ⋯ θνητ⋯ δ᾽ ἔσται μεγ⋯λα, μ⋯ δ⋯ντος δ⋯κην. See also TrGF 2.99 πς ον τ⋯δ᾽ εἰσορντες ἢ θεν γ⋯νος | εναι λ⋯γωμεν ἢ ν⋯μοισι χρώμεθα; TrGF 2.465 τολμ κατειπενν, μ⋯ποτ᾽ οὐκ εἰσ⋯ν θεο⋯. | κακο⋯ γ⋯ρ εὐτυχοντες ⋯κπλ⋯σσουσ⋯ με.

⁴⁹ Sophocles develops the same idea in Electro 245–50Google Scholar but is much more cautious in his wording: εἰ γ⋯ρ ⋯ μ⋯ν (sc. Agamemnon) θανὼν γâ τε κα⋯ οὐδ⋯ν ὢν | κε⋯σεται τ⋯λας, | οἵ δ⋯ μ⋯ π⋯λιν | δώσουσ᾽ ⋯ντιφ⋯νους δ⋯κας, | ⋯π⋯ντων τ᾽ εὐσ⋯βεια θνατν (cf. also OT 892ff.). I cannot think of any passage in either Aeschylus or Sophocles where a character is driven to deny explicitly the existence of the gods. Aeschylus only touches upon the periphery of the problem when he mentions in Ag. 369–72 atheists who dare to say that the gods disdain to care about wrongdoers: … οὐκ ἔφα | θεοὺς βροτν ⋯ιοσθαι μ⋯λειν | ὅσοις ⋯θ⋯ κτων χ⋯ρις | πατοθ᾽ ⋯ δ᾽ οὐκ εὐδεβ⋯ς

⁵⁰ cf. Drachmann, A. B., Atheism in Pagan Antiquity (Copenhagen, 1922), p. 53.Google Scholar I have argued in an article on ‘The “Atheistic” Fragment from Euripides’ Bellerophontes (286 N²) (forthcoming, ICS 15.1 [1990]) that it may have been precisely this outspoken Euripidean passage which provoked Aristophanes to impute atheism to Euripides (Thesm. 450 νν δ᾽ οτος ⋯ν τασιν τραγῳδ⋯αις ποιν | το⋯ς ἄνδρας ⋯ναπ⋯πεικεν οὐκ εναι θεο⋯ς). M. Lefkowitz, in a recent article on ‘“Impiety” and “Atheism” in Euripides' Dramas’, CQ 39 (1989), 70–82Google Scholar, has convincingly shown that such a charge against Euripides is unfounded; cf., moreover, her article ‘Was Euripides an Atheist?’, SIFC {Ser. III} 5 (1987), 149–66.Google Scholar

⁵¹ The fragment is only transmitted by Ps.-Justin, De monarchia 5.6. In lines 2–3,1 have given the reading of the Argentoratensis gr. 9 which Nauck and Marcovich (cf. above n. 16) rightly adopt, whereas the previous editor of De monarchia, Otto (n. 21), followed Matthiae's edition of Euripides (n. 18) and gave preference to the reading of the Parisinus gr. 450 (εἴ τις ⋯νθρώπων λ⋯γει, | μ⋯ τ παλαι μρος ὢν χρ⋯σθω λ⋯γῳ) – a codex which is often more reliable than the Argentoratensis, but not in this particular case.

⁵² cf. Lloyd-Jones, H., op. cit. (n. 13), p. 146Google Scholar ‘The case for women is sometimes argued with all the resources of the poet's eloquence, but so is the case against them; the same is true of almost every other regularly controversial subject’.

⁵³ See n. 50. Such an interpretation of Bellerophontes confirms the results which Lefkowitz obtained looking at other Euripidean tragedies (see CQ 39 (1989), 70ffGoogle Scholar.; it is rather surprising that she never mentions the particularly relevant fr. 286 in her fine article).

⁵⁴ cf. fr. 991 N² ⋯λλ᾽ κεἰ τις ⋯γγελ⋯λλ᾽ 〈τὠμ⋯λλ᾽ 〉 (coni. Nauck) λ⋯γῳ, | Ζεὺς κα⋯ θεο⋯, βρ⋯τεια λε⋯σσοντες π⋯θη; Herc. fur. 757Google Scholar τ⋯ς ⋯ θεοὺς ⋯νομ⋯ᾳ χρα⋯νων, θνατ⋯ς ⋯ν, | ἄφρονα λ⋯γον | †οὐραν⋯ων μακ⋯ρων† κατ⋯βαλ᾽ ὡς; ἄρ᾽ οὐ | σθ⋯νουσιν θεο⋯; 772 θεο⋯ θεο⋯ | τν ⋯δ⋯κων μ⋯λουσι κα⋯ τν ⋯σ⋯ων ⋯πᾴειν.

⁵⁵ I do not think that Willink, C. W., ‘Sleep after Labour’, CQ 38 (1988), 97CrossRefGoogle Scholar, is right in objecting to Bond's translation ‘look out’, ‘be careful’ of Herc. fur. 1072, for such a translation adequately brings out the warning overtones of ⋯ρâτε in both this passage and our fragment. Cf. also Ar. Plu.. 215 ⋯ρâτε, translated by van Daele with ‘Prenez garde’ (V. Coulon-H. van Daele, Aristophane K [Paris, 1954], p. 99).

⁵⁶ See generally Fahr, W., ΘΕΟΥΣ ΝΟΜΙΖΕΙΝ. Zum Problem der Anfänge des Atheismus bei den Griechen (Spudasmata 26) (Hildesheim and New York, 1969).Google Scholar

⁵⁷ See above, and compare the probable echo in Aristoph. Thesm. 450 (n. 50).Google Scholar

⁵⁸ See below part IV.

⁵⁹ It has strongly bothered Valckenaer (p. 5 ‘Sed cur, obsecro, dicuntur Δ⋯ς, bis, stalte peccantes gravissime?’) and, in his wake, Matthiae, loc. cit. (n. 18).

⁶⁰ Another possible explanation would be that one of the ‘mistakes’ concerns the human realm (because those people really behave wickedly towards others, cf. κακ⋯ς πεφυκςώ), and the other ⋯μαρτ⋯α consists of the neglect, or rather the denial, of a divine power; for whereas in Aeschylus every ⋯μαρτ⋯α was at the same time an offence against men and gods, the two aspects are clearly separated in Euripides, as Saïd, , op. cit. (n. 37), p. 429, has pointed out.Google Scholar

⁶¹ To have an ‘upright and good sense’ is, according to Eur. Hipp. 426Google Scholar, absolutely crucial for life: μ⋯νον δ⋯ τοτ⋯ φασ᾽ ⋯μιλλâσθαι β⋯ῳ, | γνώμην δ⋯καιαν κ⋯γαθ⋯ν, ὅτῳ παρ; cf. Hipp. 240 πο παρεπλ⋯γχθην γνώμης ⋯γαθς; cf. Soph. OT 687Google Scholar ⋯γαθ⋯ς, ὢν γνώμην ⋯ν⋯ρ etc.

⁶² I owe this important observation to C. W. Willink.

⁶³ cf. Aj. 457f.Google Scholar; El. 1485f.Google Scholar may be interpolated, as Dindorf and others assumed (cf. Dawe, R. D., Studies on the Text of Sophocles, Vol. I: The Manuscripts and the Text [Leiden, 1973], pp. 202f.Google Scholar), but are still probably by a tragedian or, at least, an actor of classical times. (N. G. Wilson drew my attention to this problem.)

⁶⁴ cf. for the thought, de Romilly, J., Time in Greek Tragedy (Ithaca and New York, 1968), p. 134Google Scholar, with regard to Euripides, ‘time oners an opportunity for reflection and rectification [sc. of passions]’.

⁶⁵ Wecklein arbitrarily separated this line from fr. 835 (see below n. 80), followed by Nauck and Kannicht-Snell who record it in TrGF 2 under No. 490.

⁶⁶ cf. E. G. Turner's comment on line 221 of P.Oxy. 2455, Oxyrhynchus Papyri, Vol. 27 (London, 1962), p. 64Google Scholar; van Looy, H., op. cit. (n. 12), p. 132.Google Scholar The hypotheses of Phrixus A and B are printed also in Austin, C. (ed.), Nova fragmenta Euripidea in papyris reperta (Berlin and New York, 1968), pp. 101 f.Google Scholar Wilamowitz's verdict ‘somniasse eos qui duplicem Phrixum extitisse dixerunt’ (Analecta Euripidea [Berlin, 1875], p. 158Google Scholar) has for a long time prohibited scholars from giving due credit to the remark of the Scholion in Aristoph. Ran. 1225 which correctly attributed fr. 819 N² to the second Phrixus.

⁶⁷ Phrixus A: fr. 821 N²; Phrixus B: fr. 819 N², 827 N², and – most likely (cf. n. 84) – fr. pap. 154 Austin.

⁶⁸ Webster, T. B. L., The Tragedies of Euripides (London, 1967), pp. 131f.Google Scholar His suggestion was based on the fact that Tzetzes argued that the prologue of Phrixus B was in fact that of Phrixus A (cf. N² fr. 819), a confusion which Webster tried to explain by the chronological priority of plot B over plot A.

⁶⁹ See Cockle, H. M., The Oxyrhynchus Papyri, Vol. 52 (London, 1984), pp. 22f.Google Scholar; cf. moreover Luppe, W., ‘ΠΑΤΡΙΚΟΣ ΚΕΡΑΥΝΟΣ im “Phrixos” A des Euripides?’, ZPE 51 (1983), 25–8Google Scholar; Luppe, W., ‘Hypothesis zum ersten “Phrixos” des Euripides’, APF 32 (1986), 5–13.Google Scholar One may refer to the two Hippolytus-plnys with nearly the same plot.

⁷⁰ The gaps which still remain can be bridged by Apollodor. 1.9.1 and Hygin. Fab. 2 and 3Google Scholar. Both show such striking parallels with the two hypotheses that one has to assume that they either rely directly on Euripides or on the hypotheses. Cf. Webster, , op. cit. (n. 68), pp. 132–6Google Scholar; van Looy, , op.cit. (n. 12), pp. 165–84Google Scholar; Cockle, , op. cit. (n. 69), p. 22Google Scholar; Luppe, , op. cit. (1983) (n. 69), 6–7Google Scholar; see generally also Mette, H. J., ‘Euripides (insbesondere für die Jahre 1968–1981), Erster Hauptteil: Die Bruchstücke’, Lustrum 23–24 (1981–1982), 299–306.Google Scholar

⁷¹ cf. Hypothesis Phrixus A 9–12Google Scholar Luppe συγκα—|λ⋯σασα γ⋯ρ τν [Θεττ⋯λων γ[να]κας ὅρ—|κοις κατησφα[⋯σατο πυρ⋯ν πεφρυγμ⋯ —| νον ⋯π⋯ η⋯ν κ [τασπορ⋯ν διδ⋯ναι.

⁷² This is one point where the two plays obviously differed from each other. In Phrixus A, Athamas rules over Thessaly, (P.Oxy. 2455. 224f.Google Scholar and P.Oxy. 3652 col. II. 19)Google Scholar, in Phrixus B over Orchomenos, (P. Oxy 2455.270).Google Scholar

⁷³ Hypothesis Phrixus A 12–17Google Scholar Luppe τς δ⋯|⋯καρπ⋯ας [ἔπεισε γεν⋯σεσθαι ἔκ—|λυσιν, εἰ Φρος [σøαγε⋯η Δι⋯ τ⋯ν γ⋯ρ|εἱς Δελøοὺς ⋯π[οσταλησ⋯μενον ἄγγε—|λον ἔπεισε ὡς [⋯γοι τοτο κεχρησμ⋯—|νον εναι.

⁷⁴ Hygin. Fab. 2Google Scholar ‘quod cum Athamas se facturum abnuisset, Phrixus ultro ac libens pollicetur se unum civitatem aerumna liberaturum’.

⁷⁵ Hygin. loc. cit. ‘itaque cum ad aram cum infulis esset adductus, satelles, misericordia adulescentis Inus Athamanti consilium patefecit’. Cf. the ⋯γών between Ino, Athamas and the slave in the Florence papyrus (Page, D. L., Select Papyri, Vol. III. Literary Papyri [London, 1950], No. 32)Google Scholar, discussed by Schadewaldt, W., ‘Zum “Phrixos” des Euripides’, Hermes 63 (1928), 1–14Google Scholar (= Hellas und Hesperien ² Bd. I [Zürich and Stuttgart, 1970], pp. 505–15).Google Scholar

⁷⁶ Hygin. loc. cit. ‘rex facinore cognito, uxorem suam Ino et filium eius Melicerten Phrixo dedidit necandos’.

⁷⁷ cf. the end of the Hypothesis Phrixus B, P. Oxy. 2455.280] τι ⋯πιβο[υλ]ευομ⋯ωι|⋯ δ⋯ τ⋯ν Δι⋯ν[υσον] ⋯πικαλε[σασ]α ⋯ναρ[γ | θ⋯νατον διώλισθεν ⋯μμανες γ⋯ρ π[οι–|⋯σας [⋯] Δ[ι]⋯νυσος Φρ⋯ν τε [κ]α⋯ [τ]ῂṿ | ⋯δε[λ]ø[⋯]ν προηγ⋯γετο εἰς τ⋯ν ⋯ρ⋯[μη]ν χώ[ραν | (285) ὥ[σ] ⋯ν⋯λωμα ⋯ν⋯λωμα μαιν⋯δω[ν ποι–|⋯σων Νεø⋯λη δ⋯ καταπτâσα κα⋯ διαρ— |π⋯σασα τοὺ[ς] ⋯αν[τ]ς κρι⋯ν ἔδωκε[ν] αὐ—|τος ⋯δηγῃ [‥]. [‥] ‥–; for the end of Hypothesis Phrixus A see Luppe, , op. cit. (1986) (n. 69), 12f.Google Scholar

⁷⁸ 4th epeisodion in van Looy, 's reconstruction (op. cit. [n. 12], p. 177).Google Scholar

⁷⁹ cf. van Looy, , op. cit. (n. 12), p. 153Google Scholar ‘Waarschijnlijk dient Athamas als de spreker beschouwd, hoewel ook Phrixus of de koorleider niet uitgesloten zijn’; Webster, , op. cit. (n. 68), p. 135Google Scholar ‘835N²…may well belong to the context when Athamas hands Ino over to Phrixos’.

⁸⁰ ρε〈ν〉 Grotius; ἔτ〈ε〉ισεν Nauck. As already mentioned, this fifth line is only transmitted by Stob. Ecl. 1.3.15. Wecklein, N., ‘Studien zu Euripides’, Jahrb. f. cl. Philol. Suppl. 7 (Leipzig, 1874), 355–6Google Scholar reads ἔρεν instead of ρεν and wanted to separate the line from fr. 835 N² (‘Vielmehr gehört der lästige Vers nicht hieher, und eine äussere Bestätigung liegt darin, dass Sext. Emp. etc. gleichfalls nur V. 1–4 in Verbindung mit anderen bringen’). This was a reaction against earlier scholars who tried to link the fifth line closer to the previous four; they either wrote τιμωρ⋯αν 〈τ᾽〉 ἔτιοεν (Musgrave, S., Euripidis fragmenta [Leipzig, 1779], p. 467Google Scholar – he suggested as another possibility τιμωρ⋯ᾳ ἔτισεν), or they printed a semicolon after ⋯λ⋯σκεται line 3 and took lines 4 and 5 together (e.g. Matthiae, A., op. cit. [n. 18], p. 297Google Scholar; a similar proposal was already made by Grotius, H., op. cit. [n. 33], pp. 415 and 959Google Scholar, who, however, felt forced to change the conjunction ὅταν line 4 to ὅτε δ᾽ ἅν; Grotius was followed by Barnes, J., loc. cit. [n. 33])Google Scholar. But all the problems disappear if we assume that the compiler of the extract has omitted a few lines between 4 and 5 (cf. below). – The phrase τιμωρ⋯αν ἕτ〈ε〉ισεν is impeccable (pace Musgrave, loc. cit.), cf. Plat. Leg. 905a6–7Google Scholar τε⋯σεις…τ⋯ν προσ⋯κοσαν τιμωρ⋯αν; for ν ρε〈ν〉 κακν, cf. Eur. Hel. 425Google Scholar κρ⋯ψας γυνακα τ⋯ν κακν π⋯ντων ⋯μο⋯|ἅρασαν…, Βα. 1114 etc.

⁸¹ cf. Eur. Oinomaos, fr. 577Google Scholar N², quoted above (part II).

⁸² With the publication of P.Oxy. 3652 (see above), it has become almost impossible to ascribe the fragments, for which the sources do not give a precise designation, to one or other of the plays. The attempts of van Looy, , op. cit. (n. 12), pp. 176–83Google Scholar, and Webster, , op. cit. (n. 68), pp. 132–6Google Scholar, who both concentrated on Phrixus B and considered fr. 835 as part of this second play, have therefore ceased to be likely, although the general considerations of both scholars remain useful.

⁸³ See Bartoletti, V., ‘Frammenti di un florilegio gnomologico in un papiro fiorentino’, Atti dell' XI Congresso Internazionale di Papirologia (Milano, 1966), pp. 1–14Google Scholar. The Euripidean fragments are printed in Austin's collection (cf. n. 66) under Nos. 152–6.

⁸⁴ Wilamowitz's attribution to the Hypsipyle was based on an inadequate interpretation of Job. Lydus, , De mens. 4.7 p. 72Google Scholar Wünsch, cf. Di Benedetto, V., ‘Giovanni Lido e due frammenti Euripidei’, Maia 17 (1965), 388ff.Google Scholar

⁸⁵ It is very interesting that Joh. Lyd., who, loc. cit., quotes four out of these five lines, changes the plural θεν into the singular θεο. As in Ps.-Justin's quotation of Eur. fr. 835 N² (see above part I), it is not absolutely certain whether this change is due to the author himself or to a scribe.

* I should like to express my sincere thanks to Nigel G. Wilson, Fellow of Lincoln College, Oxford, who kindly read an earlier version of the article, and whose scholarly approach was a source of inspiration to me, and to John H. Sykes who sacrificed a good deal of his valuable time to make my English intelligible. I am also indebted to C. W. Willink for helpful references and many critical suggestions. And, last but not least, I am grateful to the Swiss National Foundation (Schweizerischer Nationalfonds zur Forderung der wissenschaftlichen Forschung) for the generous scholarship which enabled me to spend a stimulating year in Oxford.