PHOTIUS, ΑΝΑΛΦΑΒHΤΟΣ AND ATTICIST LEXICA

Abstract Photius’ lexicon contains an entry on the rare adjective ἀναλφάβητος (‘illiterate, ignorant’) that cites Phrynichus Atticista. Based on this testimony, the whole passage has been edited as fr. 19 of Phrynichus’ Praeparatio sophistica. This article demonstrates that in this lemma Photius conflates material which comes from Phrynichus and one other source, hypothetically identified with the anonymous Antiatticist lexicon, which preserves an abridged entry on ἀναλφάβητος and which Photius employed in the compilation of his lexicon. The article also explores the possibility that the work in which Phrynichus dealt with ἀναλφάβητος was not the Praeparatio sophistica but the Eclogue. This hypothesis requires challenging some assumptions concerning the transmission of Atticist lexicography in the Byzantine era, chiefly the assumption that material from the Eclogue did not circulate at Byzantium before the fourteenth century.

This article argues that only the first sentence of Phot. α 1552 comes from Phrynichus, that it probably is not a direct quotation, and that Photius here combines two sources. The second source may be the anonymous Antiatticist lexicon, which preserves an abridged lemma on ἀ ναλφάβητος (Section 2), and which Photius employed in the compilation of other entries of his lexicon. I defend this second thesis through detailed analysis of the structure and language of Phot. α 1552 (Sections 3 and 4). In addition, the article explores the possibility that the work in which Phrynichus dealt with the suitability of ἀ ναλφάβητος was not the PS, as is usually assumed on the grounds that we know Photius consulted it (Section 3), but rather Phrynichus' more polemical work, the Eclogue. Testing this hypothesis requires challenging some common assumptions about the transmission of Atticist lexicography at Byzantium. The first such assumption is as follows: since there is no trace of the Eclogue at Byzantium before the fourteenth century, it must follow that Photius had no access to this work. This inference disregards the possibility that Photius-and other medieval lexicographers-accessed material from the Eclogue through intermediary sources (Section 5). I explore the possibility that one of the intermediaries between Phrynichus' Eclogue and Photius was the Antiatticist, in a fuller version than that which survives in the epitome contained in its sole surviving manuscript, Parisinus Coislinianus 345 (tenth century). 2 I argue that the initial portion of Phot. α 1552 is more compatible with the style and terminology of the Eclogue, and make the case on these grounds for a reappraisal of de Borries's description of the 'typical' style of the PS (Section 5).
Valente's text marks a significant improvement on Bekker's edition and proves that the pedigree of ἀ ναλφάβητος was a matter of debate for the Atticists. As Photius reports, Phrynichus deemed ἀ ναλφάβητος a 'cheap' (εὐ τελές) word, typical of laymen. The reasons behind this judgement are not stated, but it is possible that Phrynichus' dislike concerned the second member of the compound adjective (ἀ λφάβητος). 7 ἀ λφάβητος is not the standard name of the alphabet, for which Greek uses γράμματα, documented from Herodotus (5.58) onwards. While the first attestations of the letter-names ἄλφα and βῆτα occur in Xenophon and Plato, ἀ λφάβητος is first documented in Irenaeus' Aduersus haereses (second century A.D.) and in the apocryphal Gospel of Thomas of roughly the same date. 8 If ἀ λφάβητος arose in Christian-Jewish contexts, the Atticist dislike of its compound ἀ ναλφάβητος could concern both its formation pattern and its rarity in comparison to ἀγράμματος, the standard word for 'illiterate'. 9 The Antiatticist and Photius, with their references to Nicochares, provide the only evidence that ἀ ναλφάβητος was already in use in classical Greek. It crops up subsequently in Athenaeus (4.176e-f), where it describes an ἰδιώτης. In Athenaeus, as in Photius, being ἀναλφάβητος is paired with being εὐτελής, confirming the impression that the adjective was associated with the lower classes. 10 The next testimony, Hesychius (α 4426 Latte-Cunningham = EM α 1277 Lasserre-Livadaras), is also the first to provide a synonymic explanation of ἀναλφάβητος as 'uneducated' 6 For the interpretation of this line, perhaps describing Polyphemus, see Orth (n. 5), 55-6. In fr. 4, from the same play, Polyphemus is described as ἀπαιδεύτερος … Φιλωνίδου τοῦ Μελιτέως: see Orth (n. 5), 53. 7 It is unclear whether ἀ ναλφάβητος derives from ἀ λφάβητος (a Dvandva, i.e. a copulative compound whose constituents are logically linked by 'and': 'alpha-and-beta'), because this word is attested much later, or constitutes a parasynthetic compound based upon the common sequence ἄλφα-βῆτα (e.g. Pl. Cra. 431e10, Arist. Metaph. 1087a8-9). On this latter category, see E. Risch, 'Griechische Komposita vom Typus μεσο-νύκτιος und ὁμο-γάστριος', MH 2 (1945), 15-27, at 15-16 = Kleine Schriften (Berlin and New York, 1981), 112-24, at 112-13. ἀ λφάβητος is usually masculine, but a rare feminine usage appears in the Greek version of the mid sixth-century A 10 Athenaeus discusses the Alexandrians' musical competence vis-à-vis that of his contemporaries: καὶ οὐ λέγω περὶ κιθαρῳδίαν μόνην, ἧς καὶ ὁ εὐτελέστατος παρ' ἡμῖν ἰδιώτης προσέτι τε καὶ ἀναλφάβητος οὕτως ἐστὶ συνήθης ὡς τάχιστα ἐλέγχειν τὰ παρὰ τὰς κρούσεις ἁμαρτήματα γινόμενα ('I am not speaking only of competence on the kithara, with which even the most worthless among us, indeed any uneducated plebeian, is so familiar that he could instantly expose any mistakes when the notes are struck').
(ἀπαίδευτος) rather than 'illiterate'. 11 The two meanings, in fact, are often found side by side. Procopius of Caesarea uses ἀναλφάβητος to describe Emperor Justin I, who was so ignorant as to be truly illiterate (Anecdota 6.11.6 ἀμάθητος δὲ γραμμάτων ἁπάντων καὶ τὸ δὴ λεγόμενον ἀναλφάβητος ὤν; cf. Suda α 1470 and 1952), while in the ninth-century Life of Cosmas 'the Melode' and John of Damascus, dubiously attributed to Michael Syncellus, ἀναλφάβητος refers to a man who had to be taught how to join letters into syllables. 12 Attestations increase in Byzantine literature, including in major authors such as Eustathius, Nicetas Choniates and Maximus Planudes. In his Exegesis in canonem iambicum pentecostalem (1.82.23-6 Cesaretti-Ronchey), Eustathius employs ἀναλφάβητος to paraphrase a line of the Canon pentecostalis which alludes to the common topos of the illiterate (ἀγράμματοι) Apostles, whom the Holy Spirit has enlightened (ἀγραμμάτους δὲ ἢ τοὺς μὴ εἰδότας ὅλως γράμματα λέγει, οὓς ἀναλφαβήτους ἔφη τις, ἢ τοὺς ὀλιγογραμμάτους, κτλ.: '[the canon] calls ἀγράμματοι those who are completely illiterate-whom some called ἀναλφάβητοιor those who are but poorly educated'). 13 Eustathius's recourse to both a paraphrasis and the synonym ἀναλφαβήτους to explain ἀγραμμάτους is noteworthy: the parenthetic οὓς ἀναλφαβήτους ἔφη τις seems to be Eustathius's way to avoid ἀναλφάβητοι, an attribution which might have sounded offensive if referred to the Apostles, probably because it was a low-prestige term. This would explain the rarity of this word in Greek, Phrynichus' dislike, and also why the author of the Antiatticist-who had a different attitude towards contemporary language-attempted to defend it by showing that it was employed by an author from the fifth century B.C. 14 Photius or his sources may have been seeking a compromise between earlier Atticist condemnation of the term and the linguistic practice of later times.
While this interpretation of the lexicographical sources is plausible, the precise implications of Phot. α 1552 are more difficult to unravel. At first sight, what Photius seems to be saying is that Phrynichus, initially persuaded that ἀναλφάβητος was a 'cheap' word, changed his mind because Nicochares 'turned it into an ancient word' (Νικοχάρης δὲαὐ τὸἐ ποίησεν ἀ ρχαι̑ον). 15 It is unlikely, on the other hand, that Phrynichus was defending the permissibility of such a low-prestige term against another scholar's overt criticism, since he was more prone to criticize other scholars for their 11 Latte (followed by Hansen and Cunningham in their edition) includes this lemma among those that Hesychius derived from Diogenianus' lexicon (first half of the second century A.D.  laxness. 16 The first scenario is more plausible because we know from Photius (Bibl. cod. 158) that in elaborating his long work Phrynichus discussed its lexical material with friends, who responded with corrections and references that he had missed. However, the rehabilitation of a rare, low-register word through a minor comic poet is not Phrynichus' habit, whereas it is routine for the Antiatticist.  18 What has survived indicates that Phrynichus avoided polemical expressions and prescriptive language in this work: most of the extant entries address semantics and issues of style and register rather than linguistic correctness. This is consistent with the information provided by Photius (Bibl. cod. 158), who describes the PS as a 'collection of words and short expressions' (λέξεών τε συναγωγὴ καὶ λόγων κομματικῶν), of which 'some form short phrases, expressed and formulated in elegant and innovative ways' (ἐνίων δὲ καὶ εἰς κῶλα παρατεινομένων τῶν χαριέντως τε καὶ καινοπρεπῶς εἰρημένων τε καὶ συντεταγμένων). Using 'the standards of unadulterated and pure Attic speech' (εἰλικρινοῦς δὲ καὶ καθαροῦ καὶ Ἀττικοῦ λόγου κανόνας), Phrynichus divided these phrases into categories 'appropriate to oratory, written composition and conversation, some utilized for derisive or contemptuous speech or delivered within amatory modes' (τὰς μὲν γὰρ αὐτῶν ῥήτορσιν ἀποδεδόσθαι, τὰς δὲ τοῖς συγγράφουσι, τὰς δὲ συνουσίαις ἐφαρμόζειν, ἐνίας δὲ καὶ εἰς τὰς σκωπτικὰς ὑπάγεσθαι λαλιάς, ἢ καὶ εἰς τοὺς ἐρωτικοὺς ἐκφέρεσθαι τρόπους).
The information preserved in Photius' Lexicon concerning the 'cheap' status of ἀναλφάβητος would therefore be an anomaly in the context of the PS-and not on these grounds alone. Another suspicious element is the rehabilitation of a low-register term through Nicochares, whom Phrynichus is not known to have quoted elsewhere. Although the quotation of a minor poet of Old Comedy may seem to conform to Phrynichus' usual practice, a careful study of the comic references in the PS reveals no other recourse to Nicochares.
The fragments of Nicochares are all transmitted in Greek works of scholarship, chiefly Byzantine lexicography. 21 Pollux cites Nicochares eight times, usually in reference to words for tools, some of which are, in Pollux's own admission, alternatives to purer Attic expressions. 22 Apart from the defective lemma on ἀναλφάβητος, the Antiatticist quotes Nicochares in δ 18 (δαρθείς⋅ ἀ ντὶτου̑δαρείς. Νικοχάρης Κενταύρω ͅ ) to defend δαρθείς, the aorist passive participle of δέρω ('to skin, to thrash'), an analogical formation that arose beside δαρείς and that is still attested in late Byzantine texts. 23 Since there is no evidence for Phrynichus' recourse to Nicochares, who is quoted in defence of post-classical usages by less severe Atticists, we must consider whether the part of Photius' lemma quoting Nicochares might come from a source other than Phrynichus. This proposal is supported by a survey of the nine entries in Photius' Lexicon that preserve Nicochares' name (α 430, 443, 898, 1200, 1643, 1798, 3411, 3467, 3479), none of which can be securely linked to Phrynichus. 24 24 The fact that all the lemmas begin with α could be explained either by assuming that Photius directly consulted only the book(s) of the PS containing words in α or, more probably, that he did not use the PS directly but via intermediary sources that had already selected the material. The most important of these intermediaries was an expansion of the Synagoge: Alpers (n. 16), 74 and n. 71 below. For the relationship between these texts, see Reitzenstein (n. 5), xxix-xlii; Cunningham (n. 1), 13 and 38-41.
(α 430, 443, 898, 1798) derive from the expanded Synagoge. 25 Phot. α 1200 (ἀμιθρεῖν), which has no parallels in the Synagoge, is thought to derive from Diogenianus. 26 The sources of α 1643 and 3479 are unidentified. 27 None of these eight lemmas mentions Phrynichus, nor is he known to have treated the words with which these entries are concerned.
The ninth lemma, α 3411, which is about the expression ἄφυκτον λαβήν, is more ambiguous. This entry surfaced in the Supplementum Zavordense (S z ), unavailable to Reitzenstein for his 1907 edition of Photius' Lexicon, used by de Borries in his edition of the PS. 28 Theodoridis marks Phot. α 3411 as deriving from the PS because of the expression ὁ λόγος ἐναργής, which de Borries identified as typical of Phrynichus' style. 29 However, the only secure parallel for this expression is in the epitome (PS 12.9). The other three possible parallels are found in fragments quoted by Photius, of which only fr. 23 (= Phot. α 2058) mentions Phrynichus' name; fr. 91 (= α 414) and fr. 185 (= α 1784) are attributed to the PS by de Borries because of their use of ἐναργής. This reasoning is circular: ἐναργής is a frequent adjective in Greek rhetorical and grammatical criticism, as well as in Photius' Bibliotheca, and hence may well be Photius' own rendering of his source.
Let us now move on from the specific case of Nicochares to Phrynichus' general attitude towards Attic comedy as a source of approved linguistic usages. In both the Eclogue and the PS, Phrynichus mostly turns to the major authors of Old Comedy, with only a few concessions to minor Old Comedy playwrights, as well as to the major authors of Middle and New Comedy: in general, when he quotes later Attic authors, these are never poets. 30  This interpretation can be confirmed by looking at Old Comedy quotations in the epitome of the PS. It preserves 26 direct references to Aristophanes and 31 to the other major poets (12 to Cratinus, 8 to Plato Comicus, 7 to Eupolis, and 2 each to Hermippus and Pherecrates); other poets are limited to one quotation each (Aristomenes, Crates, Cantharus and Strattis). There is also a gradation in the use of Old Comedy playwrights other than Aristophanes. While Eupolis and Pherecrates, two authors whom Phrynichus admired, can be used to illustrate correct Attic usages, 33 most quotations of minor poets-such as Nicochares-appear to be aimed at providing alternative meanings for the main lemma and never at defending a certain expression or prescribing a rule. 34 It may be objected that the epitome of the PS is missing too many of its original references for this analysis to adequately represent Phrynichus' range of citations in this work. It is worth, therefore, considering the unattributed entries for which de Borries suggests a comic quotation, often based on parallel entries in other lexica. Old Comedy poets other than Aristophanes provide some 60 references, with Cratinus (23 references) and Eupolis (15) as the most quoted authors, followed by Plato Comicus (8) and Phrynichus Comicus (8). As in the case of those entries which preserve the loci classici, these restored references mostly quote the minor Old Comedy playwrights (Alcaeus Comicus, Ameipsias, Aristomenes, Cephisodorus, Hermippus, Philonides, Lysippus, Strattis, Sannyrio) to illustrate the exact meanings of certain expressions, some of which are quite rare. 35 Only two entries have a clear prescriptive tone, with a third being more ambiguous. In the entry on ἁλίπαστα (46.8-9), the simple meaning 'pickled food' (τὰ ταρίχη) is contrasted with the correct Attic meaning 'food preserved in salt'; the reference restored by de Borries (followed by Kassel and Austin) is Aristomen. fr. 12, based on Ath. Deipn. 14.658a. PS 58.8-11 prescribes γλωττοκομεῖον as the correct term for 'a case that contains the reeds or tongues of musical instruments' against γλωσσόκομον, employed by the ἀμαθεῖς (the same prescription can be found in Phrynichus, Ecl. 33 See the lemmas ἀνωφέλητος ἄνθρωπος (4.11-13), quoting Eupolis (fr. 409), and θυμέλη (74.9-12, cf. Ecl. 135), quoting Pherecrates (fr. 28). The lemmas ἀπιστίαν βλέπει (5.15), which quotes Eupolis (fr. 332), and ἀρχαῖος (21.11), which quotes Pherecrates (fr. 228), merely provide semantic explanations and are neutral. 34 For such 'descriptive' entries, see e.g. 23.13-24.2 on ἀλλόκοτον (Crates, fr. 49); 37.14-16 on ἀγωγεύς (Strattis, fr. 55); 39.15-40.10 on ἀκρατίσασθαι (Aristomenes, fr. 14); 86.5-6 on λέκιθος (Cantharus, fr. 13). Phrynichus makes similar use of the occasional New Comedy reference: 28.4-8 on αὐθέκαστα <λέγειν> (Men. fr. 592); and 60.14-18 on γρυμεῖα (Diphilus, fr. 128). 35  On the whole, this survey confirms that Phrynichus considered it appropriate to resort to minor Old Comedy playwrights for the purposes of linguistic elucidation and mild prescription. However, in none of these lemmas do we find an argumentative structure similar to that of fr. 19 (= Phot. α 1552), in which a minor Old Comedy poet is used to defend the use of a term deemed εὐτελές. In both the PS and the Eclogue, Phrynichus admits quotations from less-approved playwrights only for those words with an uncontroversial pedigree. One example is Ecl. 175, where the correct noun for 'incense' (λιβανωτόν, rather than λίβανον, which is used by Sophocles and judged a uox poetica) is defended through a reference to Menander's Samia. Similar conclusions can be reached by turning to the nine lemmas of the PS that may have contained quotations from Middle Comedy (6) and New Comedy (2). 37 Owing to epitomization, most of these entries are ambiguous or, at best, neutral; none of them preserves clear Atticist prescriptions.
The evidence reviewed in this section strongly suggests that Phrynichus did not quote Nicochares in the PS to defend the admissibility of ἀ ναλφάβητος, and confirms the unique character of fr. 19. Elsewhere in his two lexicographical works Phrynichus censors expressions popular in later periods, even those that have a classical attestation. Consider, for example, the criticism of Menander's compound καταφαγᾶς in Ecl. 402, even though its classical pedigree is clear from a fragment of the Old Comedy playwright Myrtilos (fr. 5), which Phrynichus in fact quotes in full.

AN ALTERNATIVE PROPOSAL: PHOT. α 1552 COMBINES TWO SOURCES, AND THE SECOND IS THE ANTIATTICIST
In this section we consider an alternative scenario for Phot. α 1552, namely that it conflates two different sources: the first condemning ἀ ναλφάβητος, the other citing Nicochares to redeem it. I argue that this second source was the Antiatticist, owing to its different approach to linguistic correctness as well as its frequent recourse to minor comic poets to defend koine expressions.
In this lemma, Phrynichus judges the adverb προαλῶς 'rashly' as being 'typical of women' through a construction in which δοκεῖ μοι is followed by the disparaging description γυναικῶν εἶναι τοὔνομα. He then expresses surprise at its use by Favorinus. The fact that a 'worthy man' uses this word is not sufficient to redeem it for Phrynichus: Favorinus was Phrynichus' contemporary and not a classical authority.
ῥύμη: the Athenians use this word to refer to an onrush, but our ignorant contemporaries for a narrow street. This word too seems to me to be of Macedonian origin. It is necessary to use στενωπός for a narrow street and to call an onrush ῥύμη.
We know from Pollux's Onomasticon (9.38) that ῥύμη was used with the meaning 'alley' by the New Comedy poet Philippides in two different plays (frr. 22 and 14). 39 It is likely that Phrynichus had this poet in mind when he characterized ῥύμη as a 'Macedonian' word: Philippides was a friend of Lysimachus (Plut. Dem. 12.8), the Macedonian officer who succeeded Alexander to the throne of Macedonia. 40 It is consistent with the different agendas of the two lexicographers that Phrynichus condemns a word even though it was used by a fourth-century Athenian poet, while Pollux is more open to admitting the model of New Comedy. A parallel lemma in the Antiatticist (ρ 2) responds to Atticist condemnation by showing the admissibility of the word through a classical author (probably Philippides himself) that has now been lost. The sense of ῥύμην⋅ οὔ φασι δεῖ<ν> λέγειν, ἀλλὰ στενωπόν would then be: 'strict Atticists say that one should not use ῥύμη but στενωπός; however, ῥύμη is attested in X'. In both of these entries, Phrynichus does not defend the word that he perceives to be incorrect. This strengthens the impression that Phot. α 1552 may be combining two different authorities, as perhaps implied by the contrastive δέ (that is, this entry could perhaps be a compressed version of lexicographic structures such as οἱ δὲ λέγουσι 'but other authorities say'). There are two reasons for suggesting that the Antiatticist comment that Photius (or his source) found in a text concerned with Θιβρώνειον νόμισμα 'counterfeit coin'. The adjective Θιβρώνειον is otherwise only transmitted by one manuscript (cod. Par. graec. 2646) at Poll. Onom. 3.86 in a list of words qualifying coins. 39 This meaning is also found in Aeneas Tacticus. Pollux's passage is analysed in S. Valente, 'Osservazioni su συνήθεια e χρῆσις nell'Onomastico di Polluce', in C. Mauduit (ed.), L'Onomasticon de Pollux: Aspects culturels, rhétoriques et lexicographiques (Lyon, 2013), 147-63, at 154-5, which addresses Pollux's more tolerant attitude towards foreign words used by Attic authors. 40 The καί 'also' in the text may refer to Ecl. 354, where παρεμβολή is defined as 'Macedonian'. may be the second authority. First, the Antiatticist is the only other Atticist lexicon that preserves an entry (albeit abbreviated) on ἀ ναλφάβητος. Second, the Eclogue and the Antiatticist share similar material to a degree that points in all likelihood to some kind of direct relationship between the two lexica, although the exact dynamics of this relationship are a matter of debate. The next section gives full attention to this point because it is central for our interpretation of Phot. α 1552. For the time being, suffice it to say that most of the parallel entries are found in Eclogue Book 1, with a few in Book 2, suggesting that Eclogue Book 1 was written before the Antiatticist, which seems to respond to it. The Eclogue shares twenty-six entries with the Antiatticist. 41 In twenty-four of these, the Antiatticist expresses an alternative view to that of Phrynichus. 42 In six, this alternative view is supported by a reference to an Attic author, as in Antiatt. β 37 (βούδια⋅ οὐ μόνον βοίδια. Ἕρμιππος Κέρκωψι), which seems to reply to Ecl. 61 (νοίδιον καὶ βοίδιον ἀρχαῖα καὶ δόκιμα, οὐχὶ νούδιον καὶ βούδιον) by showing that these forms were used by Hermippus (for the other five entries, see Table 1).
In ten other entries, the Antiatticist adopts a strategy more characteristic of its approach to Greek linguistic history, namely the recourse to classical authors who are not admitted into Phrynichus' canon (Herodotus, Epicharmus, Solon, the poets of Middle and New Comedy, etc.). In all these entries, the Antiatticist retrieves information aimed at showing that the condemned expression is in fact ancient or (in modern terms) classical, and hence that it is acceptable in contemporary Greek. Consider, for instance, the pair Ecl. 93 ἀκμὴν ἀντὶ τοῦ ἔτι⋅ Ξενοφῶντα μέν φασιν ἅπαξ αὐτῷ κεχρῆσθαι, σὺ δὲ φυλαττόμενος ἔτι λέγε and Antiatt. α 21 ἀκμήν⋅ ἀντὶ τοῦ ἔτι. Ὑπερείδης Ὑπὲρ Κρατίνου. Here, the Antiatticist challenges Phrynichus' statement that the adverb ἀκμήν is attested only once in Xenophon by drawing readers' attention to its use by Hyperides, an author quoted fifteen times in the lexicon. The other nine parallels (Table 2) adopt an identical strategy.   42 In ι 2 and μ 1, the Antiatticist seems to preserve the same opinion as Phrynichus (Ecl. 3 and 122).
The same structure is displayed in four entries in the Antiatticist where the cited author is lost but can be restored with fair certainty; by Phrynichus' standards, the restored authority is always an 'uncanonical' author.
These parallels should inform our approach to the remaining three lemmas of the Antiatticist, the contrastive nature of which is signalled only by the use of οὐ or ἀντί. The structure of the longer entries clarifies that οὐ or ἀντί introduces an alternative form that the Antiatticist is defending and that Phrynichus condemns in the Eclogue. 44 As an illustration, consider Antiatt. γ 28, γαγγαλίζειν⋅ οὐ γαργαλίζειν, which is a reply to Ecl. 68: γαργαλίζειν λέγε διὰ τοῦ ρ, ἀλλὰ μὴ διὰ τῶν δύο γγ, γαγγαλίζειν. The compressed entry in the epitome of the Antiatticist can be interpreted, 'it is admissible to use γαγγαλίζειν and not (only) γαργαλίζειν'. Ecl. 62 = Antiatt. ο 13 (on ὀσμή vs ὀδμή) can be interpreted in the same way; for Ecl. 144 and Antiatt. σ 2, see below.
Another element that supports the hypothesis that the second part of Photius' entry comes from the Antiatticist is the latter's well-known practice of employing the full range of Attic comic authors (278 references, Aristophanes excluded) to defend post-classical usages. 45 The excellent indexes in Valente (2015) offer a comprehensive picture of direct references and possible lost loci classici (see Table 4).
The variety and the frequency of Old Comedy minor playwrights in the Antiatticist provide a fitting context for the use of Nicochares as a model to redeem a koine feature. We will consider two examples here.  43 The epitome wrongly identifies the author as Aristophanes; the title of the play, Metrophon, clarifies that this is a mistake for Antiphanes. See Valente (n. 1), 143. 44  At κ 37 (κλιβανίτης ἄρτος⋅ Ἀμειψίας Ἀποκοτταβίζουσιν), concerning the name of a type of bread cooked in a pan, the Antiatticist implicitly defends the variant κλιβανίτης against κριβανίτης, which was considered to be the true Attic form (see Kassel and Austin on Ameipsias, fr. 5 and Ath. Deipn. 3.110c). 46 At σ 2 (συμπατριώτης⋅ Ἄρχιππος. τὸ μέντοι πατριώτης Ἄλεξις) the nouns συμπατριώτης (used by Archippus) and πατριώτης (used by Alexis) are implicitly presented as admissible, against Pollux's statement that the barbarians employ them in place of πολίτης (3.54.1-2 οἱ δὲ βάρβαροι ἀλλήλους οὐ πολίτας ἀλλὰ πατριώτας λέγουσιν). See too Phrynichus' condemnation of συμπολίτης in Ecl. 144, although it is not clear to what extent this entry is related to Antiatt. σ 2.
The evidence reviewed reflects a small proportion of the Antiatticist lemmas that employ comedy to defy Atticist prescriptivism. 53 It is sufficient, however, to indicate a systematic practice and to corroborate the hypothesis that the Antiatticist responded in α 143 to the Atticist condemnation of ἀναλφάβητος by referring the reader to Nicochares.

THE FIRST SOURCE QUOTED BY PHOTIUS COULD BE THE ECLOGUE RATHER THAN THE PS
We have operated up to this point within the parameters of the traditional assumption that all references to Phrynichus in Photius' lexicon come from the PS. Building on our hypothesis that Photius accessed information on the classical pedigree of ἀναλφάβητος through the Antiatticist, we shall now consider the more speculative possibility that the work in which Phrynichus dealt with ἀναλφάβητος was not the PS but the Eclogue. The absence of an entry on ἀναλφάβητος in the extant Eclogue is not an obstacle, since the extant version has probably undergone modifications ( pace Fischer: see below). The obscurities surrounding the composition and transmission of the Eclogue and the PS are legion. First, which was composed first? While Naechster thought that the PS preceded the Eclogue, de Borries and others have proposed the reverse. 54 The sixty-six lemmas which the two works have in common cannot be used to confirm the precedence of either, and the possibility of later interpolation in the epitome of the PS further complicates the issue. 55  one of the sections of his critical edition 'Die Ekloge ist nicht verkürzt', assuming that whatever traces one finds of epitomization belong to the manuscript tradition and not to the archetype on which it depends. 56 The lack of ancient information about the original Eclogue makes it impossible to say whether Phrynichus himself composed it as a series of short annotations. But the selection of lemmas in the extant Eclogue is unlikely to reflect the original without omissions or modifications, as it would then be a unique case in Greek lexicography. Third, the title Ἐκλογή is not attested before the fourteenth century; the Suda seems to have known the text as Ἀττικιστής. 57 Ἐκλογή is suspiciously suggestive of an abridgement, and attempts to defend it as Phrynichus' own selection of older material are not persuasive. 58 In the light of all this, it is not impossible that Photius was drawing on the Eclogue, perhaps through an intermediary source. Let us look at some supporting evidence for this hypothesis. The first element in Phot. α 1552 that points in the direction of the Eclogue rather than the PS is the terminology of the first sentence, where ἀναλφάβητος is deemed to be εὐτελές, an adjective not found elsewhere in the PS. 59 In its metaphorical sense εὐτελής qualifies worthless individuals and, by extension, those who do not belong to the class of the σεμνότεροι-the plebeians to whom Phrynichus otherwise refers with the expressions οἱ πολλοί, οἱ ἰδιῶται and οἱ σύρφακες. 60 Disparaging terminology such as this is a distinctive feature of the Eclogue, where undesirable words are marked with adjectives such as ἀμαθής, ἀπαίδευτος, ἀμελής, μανείς, βάρβαρος, σόλοικος, ἔκφυλος, ἀνενήλλιστος, by ethical and aesthetic terminology (ὀρθός, καλός, κακός, κάκιστος, μιαρός, νόθος, δεινός, αἰσχρός, διεφθαρμένος) and by metaphors taken from the sphere of numismatics and commerce (ἀδόκιμος, κίβδηλος, παρασεσημασμένος, ἀγοραῖος). Those who use 'bad' words are accused of κατακηλιδοῦν 'defiling', συσσύρειν 'spoiling', or μολύνειν 'staining' the language. The PS, in contrast, makes infrequent use of evaluative terminology: disapproval is most often conveyed by the all-purpose adjective ἀδόκιμος. 61 When explicit, prescriptions in the PS are occasionally marked by δεῖ (3 occurrences), χρὴ λέγειν (4 occurrences) and χρὴ φεύγειν (1 occurrence), whereas overt prohibitions (for example φυλάσσω) are not attested in the epitome. 62 If Phrynichus did address ἀναλφάβητος in the Eclogue, this might better explain the Antiatticist's adoption of the opposite stance, given that-as discussed in the previous section-the two lexica have much material in common. Following Latte's authoritative discussion, there is a scholarly consensus that in many entries of Eclogue Book 2 (lemmas 230-411 in Fischer's edition), Phrynichus took a polemical stance against the Antiatticist. 63 Latte also noted that in Book 1 (lemmas 1-229), Phrynichus does not polemicize against the Antiatticist. This led him to hypothesize that the Antiatticist was composed after the publication of Eclogue Book 1, and that Phrynichus replied by composing Book 2. 64 Latte hesitated, however, to draw the further conclusion that the Antiatticist, written after the publication of Eclogue Book 1, also directly criticized Phrynichus' ideas, since only Ecl. 134 (= Antiatt. ε 83, on ἐλλύχνιον) seemed to him to demonstrate a reply to Phrynichus, rather than simple dependency on common sources. 65 Latte's caution was not shared by Fischer, who proposed a bolder hypothesis: Phrynichus used the Antiatticist in Book 1 also, and in this scenario the Antiatticist would precede the whole Eclogue. 66 Valente regards Fischer's proposal with scepticism and inclines toward Latte's view. 67 The use by both lexica of the same sources complicates the task of determining their relative chronology; 68 however, Latte's hypothesis of the precedence of Eclogue Book 1 over the Antiatticist receives some support from structural features.
The lemmas of Book 1 paralleled in the Antiatticist do not show traces of alphabetical ordering, whereas those of Book 2 often do. 69 Perhaps the two lexica used the same sources but in different ways: the Antiatticist by arranging the lemmas in an alphabetical order, Eclogue Book 1 by incorporating them in no particular order. However, a better explanation would be that the difference in alphabetical organization between Eclogue Book 1 and Eclogue Book 2 depends on the fact that, in compiling the latter, Phrynichus used the Antiatticist, which was not available to him when he composed Book 1. Moreover, the polemical style of some of the Antiatticist lemmas that have parallels in the Eclogue suggests that the Antiatticist is directing its criticism against a prominent Atticist work; given the overlap between the two lexica, it is more economical to think that its target was the Eclogue, part of which must therefore have already been published when the Antiatticist was composed. The absence of a clear polemical structure in the Antiatticist for some of the parallel lemmas-an issue that induced Latte to view the hypothesis that Eclogue Book 1 preceded it with some caution-is hardly conclusive evidence: as noted by Latte himself, we read the Antiatticist in a heavily abbreviated epitome that is missing many of the extended explanations and critical comments that characterized the original (see Section 4 above for some examples). 70 The different hypotheses and problems reviewed so far also bear on the specific question of whether it is conceivable that Phrynichus devoted an entry in the Eclogue to ἀ ναλφάβητος. The only certainty is that an abbreviated lemma on ἀ ναλφάβητος is preserved in the Antiatticist. Judging from this lexicon's usual structure, it would be logical to think that in defending this word the Antiatticist is responding to some other scholar's criticism. Photius provides us with another fact: Phrynichus dealt with ἀ ναλφάβητος, although in which work he does not say. The structure of Photius' entry also suggests that-if two sources are implied-Phrynichus' critical judgement preceded another source's rehabilitation of ἀναλφάβητος. All of this leads to the possibility that a note on ἀ ναλφάβητος may have been included in Book 1 of the Eclogue, to which the Antiatticist later replied.
This tentative scenario must remain hypothetical owing to the insurmountable gaps in our knowledge of the transmission of Atticist lexicography. While there is evidence that Photius knew the PS and used its material in his lexicon (perhaps exclusively through the Synagoge tradition), there is no evidence that he ever came across the Eclogue. 71 Fischer touches upon the issue, closing with a lapidary statement: 'Das attizistische Material in der Συναγωγὴ λέξεων χρησίμων, im Photios-und im Suida-lexicon bietet zwar zahlreiche Parallelen zu Gl<ossen> der Ecl.; es gibt aber kein Indiz dafür, daß die Ecl. selbst in diese Kompilationen eingearbeitet wurde' (my emphasis) . 72 Yet lexicographical entries which imply use of the Eclogue are not lacking in Byzantine lexicography; and it is splitting hairs to ask whether Byzantine scholars accessed the Eclogue directly, viz. in the form we read it, or through an intermediary source, which still implies that the Eclogue circulated in one form or another at this stage. 73 Our perception of Photius' use of the PS is profoundly influenced by de Borries's edition, which attributed to this lexicon all of the lemmas in Photius that mention Phrynichus, as well as other passages that, according to him, were reminiscent of Phrynichus' style (see above), although the latter can hardly be reconstructed in a comprehensive way from the short snippets preserved in the epitome of the PS. The fundamental need to revise de Borries's methodology is one of the reasons why a new edition of the PS is required. 74 In reality, we know little about how much of the PS Photius read, or in what form. This leaves open the question of whether some of the references to Phrynichus in Photius may instead point in the direction of the Eclogue. 70 See the examples discussed in Latte (n. 44), 378. 71 One should not infer from the description of the PS in Bibl. cod. 158, however, that Photius had already read the PS when he compiled his lexicon. The latter is a juvenile work, some twenty years older than the Bibliotheca. It may well be that at this stage Photius based his entire knowledge of Phrynichus on other sources. He may have consulted the PS only subsequently for the composition of the Bibliotheca. Further, while Photius seems acquainted with the general organization and aim of the PS, his main interest revolves around the prefatory letters and their dedicatees. We may wonder how many of the individual lemmas of the PS Photius actually consulted. Regarding Photius' direct knowledge of older lexica in general, see Alpers (n. 16), 72-4; also n. 25 above. 72  The Eclogue seems to have seen very limited circulation before the early Palaeologan era, when it suddenly resurfaced in manuscripts and indirect testimonies, such as Thomas Magistros's lexicon. 75 It is almost certain that Photius did not access the Eclogue directly but rather found its material quoted in other sources under Phrynichus' name. Since a lemma on ἀναλφάβητος is preserved in the epitome of the Antiatticist, the logical inference would be that Photius found the information he provides in α 1552 in the Antiatticist. Although much is uncertain in this area also, there are some facts we can pin down.
Whatever version of the Antiatticist Photius or his source(s) consulted, this was a more complete work than what survives in cod. Par. Coisl. 345, since there are several instances in which Photius preserves interpretamenta and loci classici missing from the extant Antiatticist. 76 Photius also accessed much of the Antiatticist through the Synagoge tradition: the expanded Σ b (preserved in cod. Par. Coisl. 345) alone contains about one hundred and ten Antiatticist lemmas. 77 However, no lemma on ἀ ναλφάβητος is preserved in either Σ (the epitome transmitted in cod. Par. Coisl. 347) or Σ b . This is not an obstacle to our hypothesis: Photius' lexicon contains other Antiatticist lemmas not preserved in the Synagoge. Faced with these cases, some scholars assume that Photius derived these lemmas from a different version of the Antiatticist. 78 Other scholars prefer to posit his use of a lost intermediary version of the Synagoge. 79 In α 1552 Photius probably transmits information on Nicochares' use of ἀ ναλφάβητος that he gathered in a version of the Antiatticist containing the author's name, the title of the play and a more complete discussion of the Antiatticist's stance on this word and on Phrynichus' condemnation of it. Advancing this hypothesis has highlighted the need to reconsider Photius' references to Phrynichus in a critical light to assess whether he drew from both of Phrynichus' works rather than only from the PS. One potentially fruitful approach, as employed in this paper, would involve a systematic assessment of the style and terminology of Phrynichus' fragments in Photius, with the aim of identifying elements more coherent with the character of the Eclogue than with that of the PS, such as we know it.