Favi ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 Article copyright held by the author(s) and made available under the Creative Commons Attribution License CC-BY https://creativecommons.org/licenses/by/4.0/ A Contribution to the Study of P.Oxy. 1803 (Atticist Lexicon) Federico Favi SHEET OF A PAPYRUS CODEX from the sixth century CE, written in a sloping ogival majuscule, contains twelve entries of an Atticist lexicon, some only partially pre- served from damage, all of which begin with sigma. The editio princeps is by Hunt.1 Physical and palaeographic features are discussed by him and, more recently, by Esposito,2 who also comments on the mise en page and the use of paragraphematic signs.3 As Hunt notes (163), the purpose of the lexicon is “rather scholastic than scientific.” Unlike general lexica (like Dio- genianus, Cyrillus, Hesychius, etc.), the specific aim of P.Oxy. 1803 is to teach its readers how to use Attic correctly.4 To this end, most of its entries are provided with references to fifth- and fourth-century BCE writers who exemplify the approved Attic 1 A. S. Hunt, “1803. Glossary,” The Oxyrhynchus Papyri XV (1922) 163–166 (TM 65081 = LDAB 6322 = MP3 2124.300). 2 E. Esposito, “Fragments of Greek Lexicography in the Papyri,” Trends in Classics 1 (2009) 251–297, at 294–295, and “Aristophanes 8. P.Oxy. XV 1803, fol. 1r, 9–16,” Commentaria et lexica Graeca in papyris reperta2 I.1.4 (Berlin 2012) 68–69, at 68 n.1. 3 Previous bibliography is collected by E. Esposito, Trends in Classics 1 (2009) 294. The only contributions which have appeared after 2009 are his in Com- mentaria et lexica Graeca in papyris reperta2 I.1.4 (2012) 68–69 and 75–76 (“Aristophanes 10. P.Oxy. XV 1803, fol. 1v, 1–4”; first edition in 2006). 4 For a distinction in the typologies of lexica depending on their scope see further R. Tosi, “Typology of Lexicographical Works,” in F. Montanari et al. (eds.), Brill’s Companion to Ancient Greek Scholarship (Leiden 2015) 622–636, at 627–628. A 310 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 usage. Some of these authors, like Aristophanes, Eupolis, Thu- cydides, and Demosthenes, are uncontroversial models of good Attic, and they are regularly included in the Atticist canon by all lexicographers. However, this is not the case with two other quoted authors, Menander and Xenophon, whose inclusion in the Atticist canon was often regarded as controversial in an- tiquity. No clear evidence allows us to say when the lexicon contained in P.Oxy. 1803 was originally compiled, nor do we know any- thing about its earlier textual history. Some of the linguistic phenomena discussed in the entries of this lexicon are paralleled in late texts, which suggests that the lexicon too is late in date, but one must handle this evidence with the necessary caution.5 The aim of this paper is twofold. In the first section I discuss the high number of quotations from Menander in P.Oxy. 1803 and outline the profile of the compiler against the background of Menander’s reception in Atticist lexicography. In the second section I examine two entries that have received little or no attention after the editio princeps and suggest new interpretations based on linguistic and lexicographical analysis. 1. Menander’s reception in Greek lexicography and the position of P.Oxy. 1803 Menander is quoted in five of the twelve extant entries in P.Oxy. 1803. This is a surprisingly high figure. By way of com- parison, Aristophanes and Xenophon are quoted twice, Demos- thenes, Eupolis, and Thucydides once. Although the surviving evidence is limited, it seems reasonable to infer that Menander received sustained attention by the compiler.6 This fact is worthy of detailed examination.7 5 See in §2 below the discussion of P.Oxy. 1803 fol. 1, recto, 35. 6 We shall see that in several of the entries that quote Menander other more canonical fifth-century BCE authors could readily have served the compiler. 7 Cf. A. Körte, “Literarische Texte mit Ausschluss der christlichen,” ArchPF 7 (1924) 225–258, at 247: “daß Menander in einem attizistischen Lexikon für FEDERICO FAVI 311 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 Menander’s inclusion in the Atticist canon was a notorious matter of contention among Atticists. Even though he was widely read in antiquity, or perhaps precisely because of his popularity, his language attracted the criticism of purists. Par- ticularly famous are Phrynichus’ violent attacks, especially in the Eclogê.8 Several modern scholars have assumed that Phrynichus’ views are representative of the general attitude towards Menan- der in Atticist lexicography; these same scholars have concluded that because of this Atticist condemnation his comedies were no longer copied after late antiquity.9 This view, though still widely held, has not met with unanimous approval.10 Some have convincingly argued that Menander’s comedies were no longer transmitted after late antiquity for more complex and multi-layered reasons;11 the controversial status of his Attic language might be just one of den Schulgebrauch des VI. Jahr. noch so stark herangezogen wird, ist be- merkenswert.” 8 The most famous case is Ecl. 394 Fischer. See too Ecl. 170, 304, 341, 367, 390, 391, 392, 393, 397, 402, 408, 410, and 411. 9 See A. Blanchard, “Destins de Ménandre,” Ktèma 22 (1997) 213–225, at 222–224; W. G. Arnott, Menander2 I (London 1997) xxiii–xxiv; H. D. Blume, Menander (Darmstadt 1998) 22; M. Lamagna, “Il lessico di Menandro nella disputa sull’atticismo,” in J. A. López Férez (ed.), La lengua científica griega III (Madrid 2004) 195–208, at 198; F. Montana, “Menandro (e Aristofane) ad Alessandria: qualche riflessione,” in E. Dettori et al. (eds.), La cultura letteraria ellenistica. Persistenza, innovazione, trasmissione (Rome 2007) 257–269, at 261– 262; and E. W. Handley, “The Rediscovery of Menander,” in D. Obbink et al. (eds.), Culture in Pieces. Essays on Ancient Texts in Honour of Peter Parsons (Oxford 2011) 138–159, at 146. 10 For an early refutation see G. Zuntz, “Die Aristophanes-scholien der Papyri. Teil III. Schlussfolgerungen,” Byzantion 14 (1939) 545–614 and 704, at 547 n.1. 11 See especially P. E. Easterling, “Menander: Loss and Survival. ζώεις εἰς αἰῶνα (AP 9.187),” in A. Griffiths (ed.), Stage Directions. Essays in Ancient Drama in Honour of Eric W. Handley (London 1995) 153–160. 312 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 them, certainly not the only one.12 Furthermore, Menander’s reception in Atticist lexicography is more nuanced than is gen- erally thought. The attitudes represented range from the open rejection of Menander from the approved canon (Phrynichus) to his full inclusion (Aelius Dionysius, the Antiatticist, Orus). The middle ground was occupied by lexicographers who accepted Menander, not always wholeheartedly, and judged his linguistic choices on a case-by-case basis (Pollux).13 How does P.Oxy. 1803 fit into this picture? The relatively high number of quotations from Menander suggests that his comedies were still valued in late-antique education, not just for their 12 A brief, though nuanced and far-reaching, discussion is provided by G. Cavallo, “Conservazione e perdita dei testi greci: fattori materiali, sociali, culturali,” in A. Giardina (ed.), Società romana e impero tardoantic IV Tradizione dei classici. Trasformazioni della cultura (Rome 1986) 83–172 and 246–271, at 119 (repr. G. Cavallo, Dalla parte del libro. Storie di trasmissione dei classici [Urbino 2002], at 102): “il suo [i.e. Menander’s] naufragio è dovuto al fatto che o non vi furono mai interventi che ne confortassero la conservazione o se qualche intervento vi fu, questo non riuscì a proiettarsi fino alle soglie della rinascenza macedone.” L. Del Corso, “Aristofane in Egitto. Osservazioni sulla docu- mentazione papirologica (e non),” in G. Mastromarco et al. (eds.), La commedia attica antica. Forme e contenuti (Bari 2017) 231–279, provides a valuable and far- reaching discussion of the circulation of Aristophanes in Egypt, which is relevant to Menander’s case. Cf. also F. Montana, “Aristophanes,” Com- mentaria et lexica Graeca in papyris reperta2 I.1.4 (2012) 3–12, and E. Esposito, Tra filologia e grammatica. Ricerche di papirologia e lessicografia greca (Bologna 2017) 6– 7. A thorough investigation of Menander’s manuscript tradition and the processes which determined its fate is provided by R. Carlesimo, Il testo di Menandro: verso un riesame critico delle fonti (diss. Univ. Basilicata 2019). 13 The most up-to-date and thorough discussion of Menander’s presence in Greek lexicography is O. Tribulato, “ ‘Not even Menander would use this word!’: Perceptions of Menander’s Language in Greek Lexicography,” in A. H. Sommerstein (ed.), Menander in Contexts (New York 2014) 199–214 (from which I adopt the tripartite view of Menander’s reception in lexicography). Tribulato emphasizes that against the general background of Atticist lexicography Phrynichus appears extreme and comparatively marginal. S. Nervegna, Menander in Antiquity. The Contexts of Reception (Cambridge 2013) 257–258, offers a multi-focal approach to the loss of Menander that de- emphasizes the role of Atticist lexicography and Phrynichus. FEDERICO FAVI 313 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 moral content, but also as models of Attic diction. Even if the original compilation of the lexicon transmitted in P.Oxy. 1803 were older than the date of the papyrus,14 its canon of approved authors must still have been acceptable when the papyrus was copied. Menander’s name, moreover, is left out in two of the entries and only the title of the play is provided.15 Given that authorship is a crucial piece of information in Atticist lexicogra- phy, P.Oxy. 1803 presupposes familiarity with the comedian’s output. The name’s absence may also be the result of earlier epitomizing, although one would expect the omission of the play’s title sooner than the author’s name. The plays quoted without authorship are Ἐγχειρίδιον and Φιλάδελφοι. Mosaics at Mytilene that include scenes from the former attest to its fame as late as the fourth century CE.16 This is consistent with what we know about the circulation of Menander’s text, whether in full or excerpted, in late antiquity and early Byzantine times, as well as in the context of education at various stages of the curriculum.17 I am not claiming that the Menandrian quotations in P.Oxy. 1803 presuppose direct access to the actual text of the corresponding plays. It is likely that the compiler of the lexicon drew some of his material from previous lexicographical (ar- guably Atticist) compilations, and that his readers relied on the lexicon and not on the quoted sources. But the fact that diction 14 See in §2 below the discussion of P.Oxy. 1803 fol. 1, recto, l. 35. 15 See the quotations from Ἐγχειρίδιον (“The Dagger”), fol. i, verso, 8–10, and Φιλάδελφοι (“Siblings in Love”), fol. i, recto, 20–26. Aristophanes’ name is left out as well in a quotation from Knights (fol. i, recto, 27–34). Τhe failure to note the author’s name may betray a lack of accuracy on the copyist’s part. 16 See E. G. Csapo, Actors and Icons of the Ancient Theatre (Chichester 2010) 140–167; and Nervegna, Menander in Antiquity 136–169 and 264. 17 Fourteen papyri of Menander’s plays date between the fifth and seventh century (see the list in Nervegna, Menander in Antiquity 271–279). K. Alpers, Das attizistische Lexikon des Oros (Berlin 1981) 104–105, and Arnott, Menander I xxiii-xxiv, collect passages that attest to the circulation of Menander’s plays in late antiquity and early Byzantine times. See Nervegna 201–251 on the circulation and use of Menander in schools. 314 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 from the comedies of Menander is considered in the context of linguistic education shows that in the sixth century, when the papyrus was copied, his plays were famous and he was well regarded at least by some who aspired to good Attic.18 I will now take a closer look at the ways in which Menander is used in our lexicon. In all cases he illustrates uncontroversial and uncontested classical Attic diction:19 1. Fol. i, verso, 1–7: στιφρόν· ὃ οἱ πολλοὶ στριφνό(ν), / ὡς Ἀριστοφάνης Γήραι· “καὶ / µὴ⟨ν⟩ ὑπόστιφρόν γε τὴ(ν) / φωνὴν ἔχεις” (Ar. fr.134 PCG). καὶ Μένα/νδρος ἐν Συναριστώσαις· “ὡς / ἀεὶ στιφρὰς20 ἐσοµένας / καὶ νέας, ταλάντατος” (Men. fr.343 PCG).21 The entry indicates that στιφρός is good Attic, as proven by Ari- 18 Whether people could read Menander’s plays in full is not relevant to his inclusion in the canon. Eupolis and Cratinus, who are regularly included, did not survive late antiquity, but they were revered in late-antique and Byzantine lexicography. 19 Throughout this paper I use a slash / to indicate line end in the papyri; and two vertical lines || to indicate verse end in poetic quotations. 20 The transmitted reading is στεφρας, which from a linguistic point of view is an interesting slip. The copyist is already influenced by the later vocalism with [e] in the middle syllable, attested in adverbial στερπνά and στρεφνά (“tightly,” a formulaic use for people embracing one another) in the Digenis Akritis (στερπνά in the Grottaferrata version, ms. Crypt. Z.α.XLIV (444), at 2.280; στρεφνά in the Escorial version, MS. Escorial.gr. 496 (Ψ.IV.22), at 481, 915, and 1594); and by the variant reading στεφρός in place of στιφρός at Xen. Cyn. 9.13 (see F. Ruehl, Xenophontis scripta minora [Leipzig 1912] 174). This change in the vocalism of unstressed [i] is most likely due to [r] in the neighboring syllables (see D. Holton et al., Cambridge Grammar of Medieval and Early Modern Greek I [Cambridge 2019] 68–71, who trace this development back to the late koine and describe /i/ > /e/ as one of the most consistent and widespread phonetic changes of medieval Greek). 21 “στιφρόν (“firm/solid”), which common people spell στριφνόν: so Aristophanes in Old Age: ‘Indeed, you have a really strong voice’. So also Menander in The Women who have Breakfast Together: ‘(Women) who will always have a firm body and who will always be young, poor man’.” The adjective στιφρός also occurs in Ar. fr.148.3. FEDERICO FAVI 315 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 stophanes and Menander, whereas στριφνός is the common post- classical form, the one used by οἱ πολλοί.22 The same doctrine is in Moer. σ 10 Hansen στιφρόν Ἀττικοί· στριφνόν Ἕλληνες.23 Another instance of στιφρός in Menander is in a passage of Epitrepontes.24 The deprecated form στριφνός is attested in early treatises of the Corpus Hippocraticum (VM 15.11, 15, 16 Littré; De carnibus 3.29),25 but it is not attested in Attic before Theo- phrastus.26 22 On the category οἱ πολλοί in Pollux see S. Matthaios, “Pollux’ Ono- mastikon im Kontext der attizistischen Lexikographie. Gruppen ‘anonymer Sprecher’ und ihre Stellung in der Sprachgeschichte und Stilistik,” in C. Maudit (ed.), L’Onomasticon de Pollux. Aspects culturels, rhétoriques et lexicographiques (Lyon 2013) 67–140, at 95–105, who points out that in the Onomastikon this category of speakers must be taken primarily in a quantitative sense. 23 “Those who speak Attic Greek (say) στιφρόν, (while) those who speak common Greek (say) στριφνόν.” 24 Men. Epit. 384–385 [οὐκ]οῦν οὑτοσὶ µὲν φαίνεται || ἀλεκτρυών τις καὶ µάλα στιφρός (“This one here looks like a cock, a really meager/solid one”; on the meaning of στιφρός in this passage see W. Furley, Menander. Epitrepontes [London 2009] 162). The reading στιφ[ ]ς in P.Oxy. LX 4022 fr.2 should be preferred over στριφνός in the Cairo codex (and in fact, the last editors print στιφρός: see Furley 52; A. Blanchard, Ménandre II [Paris 2013] 91; and R. Kassel and S. Schröder, PCG VI.1 Menander (Berlin 2022) 151; Furley 162 also comments on the metathesis of /r/). Notice that A. W. Gomme and F. H. Sandbach, Menander. A Commentary (Oxford 1973) 321, though they did not yet know the reading in P.Oxy. 4022 (published in 1994), already suspected that the reading of the Cairo codex may be a trivialisation. On στιφρός and στριφνός in later comedy see also W. G. Arnott, “Some Orthographical Vari- ants in the Papyri of Later Greek Comedy,” in A. Willi (ed.), The Language of Greek Comedy (Oxford 2002) 191–218, at 214. 25 Both treatises are usually dated to the late fifth century BCE: M. J. Schiefsky, Hippocrates. On Ancient Medicine (Leiden 2005) 63–64; E. M. Craik, The ‘Hippocratic’ Corpus. Content and Context (London 2015) 48. But στιφρός and στριφνός occasionally alternate in the manuscripts, as in De salubri diaeta 2 and 3; cf. J. Jouanna, Hippocratis de natura hominis (Berlin 1975) 208 and 210. 26 Theophrastus not only uses στριφνός (Hist.pl. 3.11.4, unless it is a cor- ruption, see n.24 above) but also στιφρός (Sens. 44 and 50) and its comparative στιφρότερος (Hist.pl. 3.12.5 and 5.1.11). 316 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 2. Fol. i, verso, 8–10: Σάραπιν διὰ τοῦ ᾱ, ὡς ἐν Ἐγ/χιριδίωι· “ὡς σεµνὸς ὁ Σάρα/πις θεός” (Men. fr.4 PCG).27 Menander’s fragment is one of the earliest attested mentions of Sarapis in Greek. The aim of this entry was to defend Σάραπις against forms of the theonym with a different vocalism in the first syllable (see §2). 3. Fol. i, recto, 20–26: συναγαγεῖν τὸ συναθροῖ/σαι. καὶ συλλέξαι δὲ τὸ αὐτὸ / τοῦτο, ὡς ἐν Φιλαδέλφοις· / “(A) χωρίδιον πρίω συνα/γαγὼν πάνθ’ ὅσα || ἔχεις, / τὸ δ’ ἐγὼ δώσω. (B) σχολῇ / µοι σύλλεγε” (Men. fr.394 PCG).28 This entry is concerned with the synonyms συνάγω, συναθροίζω, and συλλέγω (“to bring together/collect/assemble/gather to- gether”), which are offered as equivalents. Menander’s quo- tation, where συνάγω and συλλέγω appear in the same context, provides direct evidence for this semantic proximity. The com- piler might have also quoted Ar. Lys. 584–585 κᾆτ’ ἀπὸ τούτων πάντων τὸ κάταγµα λαβόντας || δεῦρο ξυνάγειν καὶ ξυναθροίζειν εἰς ἕν29 (note here too the coppia contigua). Poll. 4.29 offers a com- parable use of these verbs to indicate gathering or collecting money or goods (as in Menander): δι’ ὧν ἔστι πόρους ἐξευρεῖν, ἀργυρολογῆσαι, φορολογῆσαι, χρήµατα συναθροίσασθαι, συλλέξαι, συστῆσαι συστήσασθαι, συναγαγεῖν.30 There are many more 27 “(One must say) Σάραπις with alpha, as in The Dagger: ‘How venerable (is) the god Sarapis’.” 28 “συναγαγεῖν (means) συναθροῖσαι (‘to collect’). συλλέξαι too (means) this same thing, as in Siblings in Love: ‘(A) Buy yourself a small piece of land, gathering all the things you have. I will give it. (B) Gather (them) for me slowly’.” 29 “And so, taking the wool from all these cities, (it is necessary to) gather them here and assemble them in one mass.” 30 “With these, one can πόρους ἐξευρεῖν (‘to procure financial provisions’), ἀργυρολογῆσαι (‘to collect money’), φορολογῆσαι (‘collect the tribute’), χρή- µατα συναθροίσασθαι, συλλέξαι, συστῆσαι συστήσασθαι, συναγαγεῖν (all FEDERICO FAVI 317 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 parallels in erudite literature where συνάγω, συναθροίζω, and συλλέγω appear in the same context or gloss one another. 4. Fol. ii, recto, 36–37: ]ωργῶι (lege e.g. Μένανδρος ἐν Γε]ωργῶι)· “ὡς σχολῇ πορεύ/εθ’ οὑτοσί” (Men. fr. 6 PCG).31 Regarding this entry, Hunt writes (165): “this is no doubt part of a note on σχολῇ with the sense of βραδέως or οὐδαµῶς,” for which he compares Suda σ 1803 σχολῇ γ’ ἄν· ἀντὶ τοῦ οὐδ’ ὅλως, βραδέως, οὐδαµῶς. Σοφοκλῆς (OT 433–434)· “ἐπεὶ || σχολῇ σ’ ἂν οἴκους τοὺς ἐµοὺς ἐστειλάµην.”32 In the context of Atticist lexicography, one may also compare Poll. 3.93 σχολῇ περιπατεῖ.33 There is abun- dant fifth-century Attic evidence for this use of σχολῇ.34 5. Fol. ii, verso, 60–69: σιωπήσοµαι ἀντὶ τοῦ σιω/πήσω καὶ σιωπήσει καὶ / σιωπήσεται ὡς ἐν τῷ πε/ρὶ τοῦ στεφάνου· “κἀγὼ στέρ/ξω καὶ σιωπήσοµαι” (Dem. De cor. 112). καὶ / Μένανδρος ἐν Φανίῳ· / “σιωπήσ⟨ε⟩ι πάλιν || ἐν τῷ µέ/ρει” (fr.392). κατὰ τ[α]ῦτα δὲ καὶ ἀ/κούσοµαι καὶ ἀκούσει καὶ / ἀκούσεται καὶ πηδήσοµαι.35 The form σιωπήσει, like ἀκούσει in the last sentence of the entry, must be the 2nd person singular future indicative.36 This entry synonyms for ‘to collect money’).” Pollux is referring to diplomatic missions, which are discussed at the end of his preceding paragraph. 31 “(In) The Farmer: ‘How slowly this man walks’.” 32 “σχολῇ γ’ ἄν: meaning ‘not wholly’, ‘slowly’, ‘not at all’. So Sophocles: ‘I would not have summoned you to my house at all’.” 33 “(He/she) walks about slowly.” 34 See LSJ9 s.v. σχολή B. 35 “(One should use) σιωπήσοµαι instead of σιωπήσω, and also σιωπήσει and σιωπήσεται, as in On the Crown: ‘And I will be content and remain silent’. So also Menander in Phanion: ‘You will remain silent again in your turn’. Like these (verbs), (one should use) ἀκούσοµαι and ἀκούσει and ἀκούσεται and πηδήσοµαι.” 36 The evidence for this spelling is discussed by W. G. Arnott, “Some Orthographical Problems in the Papyri of Later Greek Comedy II: -ει or -η(ι) 318 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 points out that, in the future tense, verbs like σιωπάω, ἀκούω, and πηδάω are always middle, never active.37 This topic is often the object of interest by Atticist lexicography (particularly in Moeris’ lexicon).38 The occurrences of ἀκούσοµαι, σιωπήσοµαι, etc. in fifth-century BCE Attic are legion.39 The evidence reviewed proves that Menander is a central figure in the compiler’s canon of approved authors.40 This is an important point: Menander was not just one more author, he was the object of special consideration. To illustrate proper Attic usage, the compiler readily turned to Menander.41 And what- ever the date of the original compilation of this lexicon, it seems fair to conclude that when P.Oxy. 1803 was copied and circulated as the Ending of the Second Person Singular Middle and Passive in the Present and Other Tenses of Verbs in -ω,” ZPE 135 (2001) 36–40. 37 For a list of Greek verbs with a middle future see R. Kühner and F. Blass, Ausführliche Grammatik der griechischen Sprache I.2 (Hannover 1892) 244–245. 38 Cf. Moer. α 81, α 83, β 33, β 36, θ 7, ο 8, π 2, and π 3; and George Lacapenus Epistula 8 p.67.14–17, discussing p.63.26–27 (ed. S. Lindstam, Georgii Lacapeni et Andronici Zaridae Epistulae XXXII [Göteborg 1924]). 39 For these forms in Ptolemaic papyri see E. Mayser, Grammatik der griechischen Papyri aus der Ptolemäerzeit I.2 (Berlin 1970) 130. 40 There would have been some (for no. 3) or many other occurrences (for nos. 4 and 5) in fifth-century Attic authors to choose from. For the other two cases (in particular, for no. 1), the existence of quotable fifth- and fourth- century alternatives seems plausible, even if we know of none. 41 Cf. Pollux’ criteria for quoting evidence in support of common and less- than-common usages as formulated in his prefatory letter to Book 3 of the Onomastikon: “In selecting the words which were used by the authors whose language is approved, I deemed it unnecessary to provide an indication of those who employed such a word, if those who used it are many; but when those who used a certain form are few, I selected only one, the one who uses the most beautiful language, just as in court cases, where a single trustworthy witness is enough in place of many witnesses” (οἷς µὴν τῶν ὀνοµάτων οἱ δόκιµοι τὴν γλῶτταν κέχρηνται, ταῦτα παρ’αὐτῶν λαβών, εἰ µὲν πλείους ἦσαν οἱ χρησάµενοι, τὸ µηδὲν ἐπισηµήνασθαι περὶ τῶν εἰπόντων ᾠήθην ἀπαρκεῖν, ἐλαττόνων δ’ ὄντων ἕνα τὸν καλλιφωνότατον αὐτῶν ἐπελεξάµην, ὥσπερ ἐν ταῖς δίκαις εἷς ἀξιόχρεως ἀντὶ πολλῶν µαρτύρων ἀρκεῖ). FEDERICO FAVI 319 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 as a pedagogical tool, Menander was still familiar and valued for language instruction. If it is true that Menander eventually fell out of favor for its language, this process was hardly far along, much less completed, by the sixth century CE. The approach in P.Oxy. 1803 may be compared with other forms of Atticist lexicography. That the compiler often quotes Menander as a source aligns him with Atticists like Aelius Dio- nysius, the Antiatticist, and Orus, who are invariably in favor of including Menander in the canon and who often quote from his plays.42 Furthermore, he pursues a comparatively mild Atticism. Besides his tolerating γί(γ)νοµαι and γι(γ)νώσκω,43 the appearance 42 For Aelius Dionysius see Tribulato, in Menander in Contexts 212 nn.25–27. According to S. Valente, The Antiatticist. Introduction and Critical Edition (Berlin 2015), ten entries in the Antiatticist explicitly go back to Menander, and in many more cases the lexicon comments on forms and constructions used by Menander. See further Tribulato 205–211, who envisages the possibility that the Antiatticist programmatically aims to rehabilitate Menander as a suitable model of good Attic. On Orus see Alpers, Das attizistische Lexikon des Oros 100: “In seiner attizistischen Schrift löst er sich von dem allmächtigen Dogma des Analogismus und vertritt einen anomalistischen Standpunkt. Sein Kanon mustergültiger Autoren, die ihm die Maßstäbe des Ἑλληνισµός verbürgen, ist durchaus unabhängig von dem des Phrynichos, und zwar teils weiter (Oros läßt z.B. Menander, Lysias und Xenophon gelten), teils aber auch enger: Aischylos und Sophokles fehlen!” 43 See fol. i, verso, 11–15: συγγίγνεσθαι λέγεται κατ’ ἐ/πένθεσιν τοῦ τ̅ (lege τοῦ γ̅ ) καὶ γιγνώ/σκειν. µάλι⟦ον⟧στα οἱ παλαιοί. ἀξιοῦσι δὲ καὶ χωρὶς αὐτοῦ (“With the addition of gamma, one says συγγίγνεσθαι and γιγνώσκειν. The ancients especially (say this). But (scholars) also approve of the form without it (i.e. gamma)”). As evidence of a stricter position, one may compare Moer. γ 3: γίγνεται Ἀττικοί· γίνεται Ἕλληνες (“Those who speak Attic Greek (say) γίγνεται, (while) those who speak common Greek (say) γίνεται”). Imperial Atticizing writers show the full range of different practices, in that some use only γιγν- (Aristides), some alternate γιγν- and γιν- (Lucian), some use only γιν- (Aelian); see R. J. Deferrari, Lucian’s Atticism. The Morphology of the Verb (diss. Princeton 1916) 36–39. The spelling γιγν- rarely appears even in those papyri of the Imperial age which otherwise display a more pretentious language, and in most cases γιν- was clearly regarded as an acceptable spelling even in openly Atticizing texts: see R. Luiselli, A Study of High Level Greek in the Non- 320 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 of Xenophon twice (and in cases for which he cannot have been the only available evidence)44 marks a clear difference from the canon of approved authors adopted by strict Atticists. One need only think of Phrynichus (in the Eclogê, but not equally so in the Praeparatio sophistica) and Moeris.45 There is, however, some con- tinuity between P.Oxy. 1803 and Pollux. Although he too in- Literary Papyri from Roman and Byzantine Egypt (diss. UC London 1999) 157, 160– 162. One may also compare the absence of γιγν- in the Petra papyri, which otherwise adopt classicizing spellings: M. Vierros, “The Greek of the Petra Papyri,” in A. Arjava et al. (eds.), The Petra Papyri V (Amman 2018) 8–34, at 13. While γιν- is the normal spelling in papyri of Roman times, γιγν- enjoys a revival in papyri dating to the Byzantine period: F. T. Gignac, A Grammar of the Greek Papyri of the Roman and Byzantine Periods I (Milan 1976) 176. 44 For σῖτος indicating an army’s provisions, P.Oxy. 1803 fol. 2, recto, 39–44 quotes Xen. An. 2.1.6, but there are innumerable other examples in a variety of authors for this generic use of σῖτος in military and non-military contexts. See, e.g., Thuc. 2.6.4: καὶ µετὰ ταῦτα οἱ Ἀθηναῖοι στρατεύσαντες ἐς Πλάταιαν σῖτόν τε ἐσήγαγον καὶ φρουροὺς ἐγκατέλιπον (note that Thucydides is quoted in P.Oxy. 1803 at fol. 2, verso, 53–55). 45 Phrynichus quotes Xenophon only twice in the Eclogê, and both times he criticizes his language choices (Ecl. 62 and 93). Moeris often quotes Xenophon and frequently points out that he is an isolated source for rare forms: α 149: Xenophon is the only Attic author who uses ἀκµήν with the meaning ἔτι “still,” as the Ἕλληνες do; γ 25: Xenophon (alone) uses γνωστήρ “surety,” cf. Poll. 9.151; δ 45: Xenophon uses dative δένδροις from τὸ δένδρον “tree,” a neuter noun of the thematic declension, rather than δένδρεσι from τὸ δένδρος, a neuter s-stem (cf. Aelius Dionysius δ 6 Erbse); ε 39: only Plato and Xenophon use the rare form εὐµορφία “beauty of form”; ε 43: Xenophon alone uses ἐθέλεχθρος “bearing one a grudge”; ε 67: the adjective ἐπιτήδειος “convenient/necessary/friendly” should not be used only for the relatives, as Xenophon does, but also for the friends, as in Plato; µ 31: Xenophon used µεριµνάω “to care for” with the meaning φροντίζω “to consider/reflect”; υ 6: Xenophon alone used ὑπνοµαχέω “to withstand sleep.” For Xenophon’s re- ception in ancient and Byzantine lexicography see L. Gautier, La Langue de Xénophon (Paris 1911) 17–18; A. Sgobbi, “Lingua e stile di Senofonte nel giudizio degli antichi,” in G. Daverio Rocchi et al. (eds.), Il Peloponneso di Senofonte (Milan 2004) 219–255; I. Pérez Martín, “The Reception of Xeno- phon in Byzantium: The Macedonian Period,” GRBS 53 (2013) 812–855, at 849–851. The most complete general treatment remains that of K. Mün- scher, Xenophon in der griechisch-römischen Literatur (Philologus Suppl. 13.2 [1920]). FEDERICO FAVI 321 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 cludes Menander in his larger canon and quotes him a few times (but far less often than the poets of Old Comedy),46 Pollux does not always regard Menander a model of good Attic. More than once he openly reproaches him when he makes use of post- classical forms.47 That P.Oxy. 1803 quotes Menander only as evidence for uncontested Attic usage, while it condemns fourth- century, and more generally post-classical, usage (as happens with στιφρός/στριφνός) shows that the compiler was disinclined to approve post-classical forms and did not espouse any ‘anti- puristic’ agenda. Finally, I draw attention to the fact that there is no mention of tragedy in P.Oxy. 1803. This may be just a matter of chance, but it may also reflect the compiler’s canon. The exclusion of tragedy from the canon is common in Atticist lexicography, the more eminent examples being the scarce attention given to, and general criticism of, tragic language in Phrynichus’ Eclogê 48 and the exclusion of Aeschylus and Sophocles from Moeris’ and Orus’ canon. But the fragmentary nature of the papyrus pre- cludes a definitive conclusion. 2. A reconsideration of two neglected entries A small number of entries in P.Oxy. 1803 have received atten- tion after Hunt’s editio princeps. Here I discuss two understudied 46 For general data on quotations of Old and New Comedy in Atticist lexicography see M. Sonnino, “I frammenti della commedia greca citati da Prisciano e la fonte del lessico sintattico del libro XVIII dell’Ars,” in L. Martorelli (ed.), Greco antico nell’Occidente carolingio. Frammenti di testi attici nell’Ars di Prisciano (Zurich 2014) 163–204, at 191–192. 47 Poll. 1.79, 2.82, 3.29, 6.26, 6.38, 6.161, 9.139, 10.98. Although Menan- der is not openly criticized in several other cases, the fact that forms he uses are contrasted with those of fifth-century BCE comic poets or treated as examples of new Attic betrays a more cautious attitude towards his language and the language of Middle and New Comedy as a whole. 48 See especially Ecl. 157, 200, 318, and 401. The Praeparatio sophistica is a different case: tragic materials abound, perhaps because of this lexicon’s more pronounced focus on matters of style. 322 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 cases for which I suggest a new interpretation. The entries are cited in the order they appear in the papyrus. The first, cited above, is fol. i, verso, 8–10 (Men. fr.4 PCG): Σάραπιν διὰ τοῦ ᾱ, ὡς ἐν Ἐγ/χιριδίωι· “ὡς σεµνὸς ὁ Σάρα/πις θεός” Hunt argued that this is an orthoepic prescription indicating that [a] in the middle syllable is long.49 However, not only could Hunt provide only Latin evidence for the sanctioned prosody with a short middle syllable, but the available evidence is limited to a single occurrence, Prudentius C. Symm. 2.531: nil potuit Serapis deus et latrator Anubis (hex.).50 There are further reasons beside the lack of parallels to doubt Hunt’s interpretation. For a start, though it may seem familiar to us, a formulation like διὰ τοῦ ᾱ is an odd way to indicate [a:]. Ancient lexica normally spell out the orthoepic prescriptions with verbs like ἐκτείνω/µηκύνω (“to lengthen”) or βραχύνω/συστέλλω (“to shorten”) or by indicating that a vowel is µακρόν or βραχύ.51 Menander’s fragment, more- over, would hardly provide evidence in support of the allegedly prescribed Σάρᾱπις, given that the middle syllable occurs in an anceps position. A more suitable interpretation makes this entry proscriptive, not prescriptive: it rejects alternative, mostly (though not always) later, forms of the theonym, like Σέραπις/Σείραπις, Σόραπις/ Σώραπις, and Ὀσάραπις/Ὀσέραπις/Ὀσόραπις.52 Given that Σέραπις is by far the most widely attested of these alternatives (Serapis is almost the standard form in Latin), it seems likely that the entry 49 Hunt 165. This explanation is accepted without comment by Arnott, Menander I 373, and the notion, based on Hunt’s explanation, that P.Oxy. 1803 is concerned with prosody is repeated in later bibliography on this papyrus. 50 “The god Sarapis and barking Anubis could do nothing.” 51 A thorough exemplification of this terminology is found in C. Vessella, Sophisticated Speakers. Atticistic Pronunciation in the Atticist Lexica (Berlin 2018) 269– 277. 52 On the etymology of Sarapis and its variant forms see the discussion and bibliography in G. Renberg, Where Dreams May Come. Incubation Sanctuaries in the Graeco-Roman World (Leiden 2017) 404–405 n.29. FEDERICO FAVI 323 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 was specifically aimed against it.53 If one accepts this inter- pretation, διὰ τοῦ ᾱ would then mean “with the letter α,” a very usual formulation which is also paralleled at fol. i, recto, 35 συνεθίζεσθαι διὰ τοῦ ῑ (i.e. συνειθίζεσθαι).54 For the σύρµα above a letter one may compare fol. i, verso, 11–12 κατ’ ἐπένθεσιν τοῦ τ̅ (lege γ̅).55 The second entry appears at fol. 1, recto, 35: συνεθίζεσθαι διὰ τοῦ ῑ καὶ56 – – – – – – – – – This entry is interrupted where the papyrus breaks off and its aim remains elusive. Various interpretations have been ad- vanced. Hunt (165) argued that διὰ τοῦ ῑ “presumably refers to the spelling εἰθίζειν.” This proposal is likely, even though διὰ τοῦ ῑ is not how one would have expected the spelling ει- to be indicated: διὰ τοῦ ει would have been the usual formulation. However, evidence that supports Hunt is provided by cases like Ecl. 22 (πιοῦµαι· σὺν τῷ υ λέγων οὐκ ὀρθῶς ἐρεῖς· πίοµαι γάρ ἐστι τὸ ἀρχαῖον καὶ πιόµενος ἄνευ τοῦ υ. Δίων δὲ ὁ φιλόσοφος σὺν τῷ υ λέγων ἁµαρτάνει)57 and 28 (ποῖ ἄπει· οὕτω συντάσσεται διὰ τοῦ ι· ποῦ δὲ 53 A hint in this direction was already made by K. F. W. Schmidt, review of P.Oxy XV, GGA 186 (1924) 1–17, at 15 n.5. For some examples where the later spelling Σέραπ- is corrected into the older one Σάραπ- on Hellenistic inscriptions see P. Bruneau, Le Sanctuaire et le culte des divinités égyptiennes à Érétrie (Leiden 1975) 78–79. 54 “συνεθίζεσθαι with iota (i.e. συνειθίζεσθαι).” 55 “With the addition of gamma.” On this use of the horizontal bar above the letters in grammatical and lexicographical papyri see E. Dickey, Ancient Greek Scholarship (Oxford 2007) 122–123. For the analysis of a concrete example one may also compare Esposito, Tra filologia e grammatica 135–136. 56 “συνεθίζεσθαι with the addition of iota and ….” 57 “πιοῦµαι: if you say (it) with upsilon, you will not say (it) correctly. For πίοµαι is the ancient (form), as is πιόµενος (too), without upsilon. The phi- losopher Dion, who uses (the form) with upsilon, is wrong.” 324 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 ἄπει, διὰ τοῦ υ, ἁµάρτηµα. εἰ δὲ ἐν τῷ υ, ποῦ διατρίβεις).58 Hunt then compared this spelling εἰθιζ- with the metri causa use of εἰθίζου in the Pythagorean Carmina aurea (35 εἰθίζου δὲ δίαιταν ἔχειν καθάρειον, ἄθρυπτον; but see also εἰθίζεο in Carmina aurea 9 ταῦτα µὲν οὕτως ἴσθι. κρατεῖν δ’ εἰθίζεο τῶνδε), and he pointed out that συνειθίζω is unattested (this is not so, see below). Finally, comparing fol. i, verso, 14–15 ἀξιοῦσι δὲ καὶ χωρὶς αὐτοῦ (i.e. the second γ in γί(γ)νοµαι and γι(γ)νώσκω),59 Hunt cautiously sug- gested that this entry could have gone on to say something like καὶ / [χωρὶς αὐτοῦ] (“and (without it)”). In sum, he believed that the compiler considered συνεθίζεσθαι and συνειθίζεσθαι equally legitimate spellings. This proposal is implausible. The alternative forms εἰθίζω/-ειθίζω and εἰθίζοµαι/-ειθίζοµαι are late variants at- tested in imperial and especially late-antique and Byzantine texts (see below). One would not expect them to be treated on an equal footing with ἐθίζω/-εθίζω and ἐθίζοµαι/-εθίζοµαι and to be recommended to an aspiring Atticist. An alternative interpretation by Schmidt60 suggests the sup- plement συνεθίζεσθαι διὰ τοῦ ῑ καὶ <παρατατικῷ καὶ ἀορίστῳ καὶ συντελεστικῷ χρόνῳ λέγεται>.61 The sense would then be that while ἐθίζω/-εθίζω and ἐθίζοµαι/-εθίζοµαι are the correct spell- ings in the present tense, augmented and reduplicated forms with initial εἰ-/-ει- are correct in the imperfect, aorist, and per- fect. There are two difficulties with this interpretation. First, the presence of καί before the list of tenses with initial spelling εἰ-/-ει- is otiose and odd. Second, it is puzzling that the compiler would make a list of all the other tenses in which εἰ-/-ει- is the correct 58 “ ‘Whither are you going away?’: it is construed like this, (using the form) with iota (i.e. ποῖ). But ‘Where are you going away’, with upsilon (i.e. ποῦ), is a mistake. If (one uses the form) with upsilon (i.e. ποῦ), (the correct usage would be) ‘Where are you spending time?’ ” 59 “But (scholars) also approve of the form without it (i.e. gamma).” 60 Schmidt, GGA 186 (1924) 15. 61 “One says συνεθίζεσθαι with iota in the imperfect, the aorist, and the perfect.” FEDERICO FAVI 325 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 spelling, instead of stating that ἐθίζω/-εθίζω and ἐθίζοµαι/-εθίζοµαι are the correct spellings in the present. Finally, consideration of the spelling of augmented and reduplicated verbs beginning with epsilon is hardly connected specifically with ἐθίζω/-εθίζω and ἐθίζοµαι/-εθίζοµαι. It is there- fore unclear why it should be applied to συνεθίζοµαι. Once again, I believe that a more convincing interpretation emerges if the aim of the entry is proscriptive: the compiler deprecates συνειθίζεσθαι as a variant form of συνεθίζεσθαι. Besides the evidence for εἰθίζου/εἰθίζεο in the late Pythagorean Carmina aurea,62 one should note that συνειθίζεσθαι is attested only once in Galen (unless it is a copying mistake)63 and then more frequently in Byzantine texts written in medieval and early-modern Greek.64 Hence, the available evidence supports the notion that συνειθίζοµαι was an emerging variant of συνεθίζοµαι in late Greek, and it is plausible that it may have been condemned as non-Attic by the compiler of our lexicon.65 62 The dating of the Pythagorean Carmina aurea is a thorny issue that exceeds the scope of this article. For a fourth-century BCE date see J. C. Thom, The Pythagorean Golden Verses (Leiden 1995) 35–58. For a critique of this new dating see, e.g., the reviews by P. W. van der Horst, Mnemosyne 49 (1996) 351–352, and J. Bussanich, BMCR 1995.12.01. Following LSJ9 s.v. ἐθίζω (where one reads that εἰθίζω is poetic for ἐθίζω), Thom notes (45): “the lengthened form […] is used metri causa.” This view is unpersuasive given that ἐθίζω is a prosaic verb and that these two would be the sole instances of εἰθιζ- metri causa. The evidence collected above for the spellings εἰθίζω/-ειθίζω and εἰθίζοµαι/-ειθίζοµαι suggest that the Carmina aurea date to late Imperial times, in line with the traditional view of earlier scholarship. 63 De humero iis modis prolapso, XVIIIa 407.1 Kühn (συνειθιζοµένοις). 64 See especially the thirteenth-century Assisiae regni Hierosolymitani (ed. Sathas 40 and 41 συνειθίζει, 245 συνειθίζουν). For what it is worth, συνειθίζω is also used in early modern texts such as Nicodemus the Hagiorite and Procopius of Nazianzus (eighteenth century). 65 The form ἐνειθίζεσθαι also occurs in Suda κ 1254: κεκονιµένος· σπουδάζων πάνυ, καὶ σχεδὸν κόνεως πεπληρωµένος. […] “πρὸς γὰρ τὴν τοιαύτην δίαιταν ἔλεγον οἱ ἅγιοι τρεῖς παῖδες κεκονῖσθαι τὴν τῶν ὀσπρίων.” ἀντὶ τοῦ 326 P.OXY 1803 (ATTICIST LEXICON) ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 If one accepts this interpretation, it is reasonable to infer that after συνειθίζεσθαι the entry went on to condemn the spelling συνηθίζεσθαι. Indeed, ἠθίζοµαι/-ηθίζοµαι too are later, minor alternative forms of ἐθίζοµαι/-εθίζοµαι. Not only does a form συνηθίζω occur in the sixth-century Platonic and Aristotelian commentator Olympiodorus,66 but it is also attested in Byzan- tine texts written in medieval Greek;67 the simple verb ἠθίζω too is equally documented in early Byzantine texts.68 To conclude, one may envisage that the text, after the papyrus breaks off at line 35, was originally along the lines of καὶ / [διὰ τοῦ η̅] (“and [with heta]”), followed by a caveat indicating that συνειθίζεσθαι and συνηθίζεσθαι are incorrect and should be avoided by the aspiring Atticist (e.g. ἀδόκιµα “unapproved (forms),” φυλάττου “abstain from (these forms),” χρὴ φεύγειν “you should avoid (these forms),” vel sim.).69 Such attention to developments in late Greek suits the chronological placement of P.Oxy. 1803 at the threshold of the Byzantine age. If my view of this entry is correct, this interest in ἐνειθίζεσθαι (“κεκονιµένος: one who is in great haste and almost covered in dust. […] ‘Regarding such a way of life, the three holy youths said that they were dusted with a (diet) of pulses’, meaning that they were accustomed (to it)”). 66 In Platonis Phaed. 8.2 (συνηθίζεσθαι). Cf. ἀποσυνηθίζειν in the sixth- century medical writer Aëtius, Iatricorum liber iv 28.7–8. 67 Historia Alexandri Magni (recensio φ) 23.9 Veloudis (ἐσυνήθιζαν); because the augment is placed before the preverb, the corresponding present stem is συνηθίζω). 68 The first occurrence is in Stephanus of Athens, the sixth/seventh-cent. commentator on Hippocrates: In Hippocratis aphorismos commentaria III–IV, comm. 4.60.89 (ἠθίζουσιν). 69 If I am right, the entry would feature first the proscribed, not the pre- scribed, usage. One may compare, e.g., Phryn. Ecl. 21: ἀνειλεῖν βιβλίον διὰ τοῦ ἑτέρου λ κάκιστον, ἀλλὰ διὰ τῶν δύο, ἀνείλλειν (“ἀνειλεῖν βιβλίον (“to unroll a book”) with just one lambda is very bad, but rather (there should be) two lambdas, i.e. ἀνείλλειν”); Ecl. 22 πιοῦµαι· σὺν τῷ υ λέγων οὐκ ὀρθῶς ἐρεῖς (“πιοῦµαι: if you say it with upsilon, you will not say it correctly”). Cf. further Ecl. 26, 36, 110, 153. FEDERICO FAVI 327 ————— Greek, Roman, and Byzantine Studies 62 (2022) 309–327 the late spellings εἰθίζοµαι/-ειθίζοµαι and (possibly) ἠθίζοµαι/- -ηθίζοµαι suggests that the lexicon need not be many centuries earlier than the date of P.Oxy. 1803. Galen’s συνειθίζεσθαι may well be the earliest evidence of εἰθίζοµαι/-ειθίζοµαι, unless it is a Byzantine copying mistake.70 As observed above, the spelling ἠθίζοµαι/-ηθίζοµαι is not attested before the sixth century. Yet one must be cautious when inferring a date for the lexicon from this kind of evidence, for it is possible that εἰθίζοµαι/-ειθίζοµαι and ἠθίζοµαι/-ηθίζοµαι were already in use before they surface in extant texts. And we cannot be sure that modern critical editions faithfully report this kind of variant readings, which are easily (and often tacitly) amended and relevant only for linguistic studies. Finally, we cannot rule out the possibility that the entry on the spellings εἰθίζοµαι/-ειθίζοµαι and (possibly) ἠθίζοµαι/- -ηθίζοµαι was only added to the lexicon at a later stage, when these spellings had made their appearance.71 October, 2022 Dipartimento di Studi Umanistici Università degli Studi del Piemonte Orientale “Amedeo Avogadro” federico.favi@uniupo.it 70 For the late date of the Pythagorean Carmina aurea see n.62 above. 71 I wish to thank Elena Esposito, Lucia Prauscello, Olga Tribulato, and Giuseppe Ucciardello for discussing earlier drafts of this article with me. I am grateful to the anonymous reader of GRBS for useful suggestions. I also wish to thank Tom Nelson and Theodore Hill, who have kindly improved the English of this piece. This article is part of a project that has received funding from the European Research Council under the European Union’s Horizon 2020 research and innovation program (Grant agreement No. 865817).