PURA. Purism In Antiquity: Theories Of Language in Greek Atticist Lexica and their Legacy

Lexicographic entries

(Antiatt. α 137, [Hdn.] Philet. 7)

A. Main sources

(1) Antiatt. α 137: ἀνθέων· ἀντὶ τοῦ ἀνθῶν, ἀπὸ <τοῦ> τὰ ἄνθη.

ἀπὸ <τοῦ> τὰ ἄνθη Alpers (1981, 154), Valente : ἀπὸ τῆς ἄνθης cod.

ἀνθέων: In place of ἀνθῶν (‘flowers’), [genitive plural] of τὰ ἄνθη (‘flowers’).

(2) [Hdn.] Philet. 7: ἀνθέων γραπτέον καὶ μὴ ἀνθῶν, ὡς βελῶν καὶ γενῶν, ἵνα μὴ συνεμπέσῃ τῷ ἀνθ’ ὧν ἔγραψας, ἀνθ’ ὧν ἔδωκας.

You should write ἀνθέων, and not ἀνθῶν, like βελῶν (‘arrows’, gen. pl.) and γενῶν (‘races’, gen. pl.), lest it coincide with ἀνθ’ ὧν ἔγραψας (‘wherefore you wrote’), ἀνθ’ ὧν ἔδωκας (‘wherefore you gave’).

B. Other erudite sources

(1) Orus fr. A 9a (= [Zonar.] 192.18–9): ἀνθέων δεῖ λέγειν, οὐκ ἀνθῶν, ἵνα μὴ συνεμπέσῃ τῷ ὧν ἄρθρῳ, οἷον τὸ ἀνθ’ ὧν καὶ ὧν τινων.

Cf. Orus fr. A 9b = Choerob. Orth. GG 4,1.LXXXI in Et.Gen. λ 169 ~ [Zonar.] 1295.3–7.

One should say ἀνθέων, not ἀνθῶν, lest it coincide with the pronoun ὧν, as [in] ἀνθ’ ὧν and ὧν τινων.

(2) [Hdn.] Περὶ τῶν ζητουμένων 249.8–13: ὁμοίως πλημμελοῦσιν οἱ λέγοντες ἀπὸ τοῦ ἄνθη τὴν γενικὴν τῶν πληθυντικῶν τῶν ἀνθῶν· ὡς βέλη, τῶν βελῶν· καὶ τὰ γένη, τῶν γενῶν· ἀλλὰ τὰ μὲν ἄλλα περισπᾶται ὁμολογουμένως, τὰ δὲ ἄνθη οὐκέτι τῶν ἀνθῶν, ἐπειδὴ ἄλλο τι ἐμφαίνεται οἷον ἀνθ’ ὧν ἔδωκας, ἀνθ’ ὧν ἔγραψας· ὅθεν τοῦτο μόνον διαιρεῖται ‘τῶν ἀνθέων’, ὡς καὶ οἱ παλαιοὶ κέχρηνται.

Cf. Et.Gen. α 890.

In the same way, those who say that the genitive plural of ἄνθη is ἀνθῶν, like τῶν βελῶν [is the gen. pl. of] βέλη, (‘arrows’) and τῶν γενῶν [of] τὰ γένη (‘kins’), are wrong: while the others have the circumflex accent on the final syllable conformably with [the norm], τὰ ἄνθη does not give τῶν ἀνθῶν [in the genitive], because it would indicate another thing, as [in] ἀνθ’ ὧν ἔγραψας ‘against whom/which you wrote’, ἀνθ’ ὧν ἔδωκας ‘against which you gave’. Consequently, this form only, ‘τῶν ἀνθέων’, is uncontracted, as the ancients also use [it].

(3) Σb α 1416 (ex Σʹ) (~ Phot. α 1956, Su. α 2494, ex Σʹʹ): ἀνθέων· τὴν γενικὴν ὁμοίως τοῖς Ἴωσιν οἱ Ἀττικοί. Ἕρμιππος Ἀθηνᾶς γοναῖς· ‘καιροσπάθητον ἀνθέων ὕφασμα καινὸν Ὡρῶν’. καὶ ἐν τοῖς ἑξῆς· ‘λεπτοὺς διαψαίρουσα πέπλους ἀνθέων γέμοντας’. καὶ Ἀρισταγόρας Μαμμακύθῳ. οὕτως δὲ καὶ βελέων καὶ ὀρέων καὶ ἕτερα πλείω Ξενοφῶν διαιρεῖ.

On the transmission of this gloss within the Synagoge’s tradition, see F.1. | καιροσπάθητον (‘close-woven’) Toup (1790, 482) : κηροσπάθητον (‘trowelled in wax’) codd. | καὶ Ἀρισταγόρας Μαμμακύθῳ omitted in Phot. | οὕτως δὲ καὶ βελέων καὶ ὀρέων καὶ ἕτερα πλείω Ξενοφῶν διαιρεῖ is only in Phot. α 1956 (= Su. α 2494, ex Σʹʹ).

ἀνθέων: Attic speakers [inflect] the genitive in the same way as Ionic speakers. Hermippus in Birth of Athena (fr. 5 = C.1): ‘a new garment of the Hours, close-woven with flowers’ and, in the following lines: ‘[a breeze] brushing away the fine woven cloths loaded with flowers’. Aristagoras too [inflects the genitive as ἀνθέων] in the Mammakythos (fr. 1 = C.2). Xenophon uses the uncontracted forms in this way: βελέων, ὀρέων and many others.

(4) Thom.Mag. 1.9–10: ἀνθέων, οὐκ ἀνθῶν. Φιλόστρατος ἐν Εἰκόσιν· ‘ἡ γὰρ τῶν ἀνθέων ἐλευθερία’.

ἀνθέων, not ἀνθῶν. Philostratus in Images (1.2.5 = C.4): ‘the freedom of flowers’.

C. Loci classici, other relevant texts

(1) Hermipp. fr. 5:
καιροσπάθητον ἀνθέων ὕφασμα καινὸν Ὡρῶν.
λεπτοὺς διαψαίρουσα πέπλους ἀνθέων γέμοντας.

The two verses are highly unlikely to be consecutive, although printed together (with a spacing) by Kassel and Austin. The expression ἐν τοῖς ἑξῆς (literally, ‘in the following lines’, B.3) does not necessarily imply consecutiveness, nor a short distance between the lines (see Comentale 2017, 58). | καιροσπάθητον Toup (1790, 482) : κηροσπάθητον codd. | λεπτοὺς διαψαίρουσα πέπλους codd. : πέπλους διαψαίρουσα λεπτοὺς Meineke (FGC vol. 1, 139, for πέπλος usually exhibits correptio Attica outside paratragic and paraepic passages) | ἄνθος can have a double meaning: ‘flower’ or ‘dye (obtained from flowers)’, ‘colour’, i.e. ‘dyed yarn’; see Comentale (2017, 61–2); Gow (1955, 296); C.3 in apparatus.

A new garment of the Hours, close-woven with flowers. [A breeze] brushing away the fine woven cloths loaded with flowers.

(2) Aristagor. fr. 1 = Σb α 1416 (~ Phot. α 1956, Su. α 2494) re. ἀνθέων (B.3).

(3) Pherecr. fr. 51:
ταχὺ τῶν ἐρίων καὶ τῶν ἀνθῶν τῶν παντοδαπῶν κατάγωμεν.

Cf. Σb α 1415 (= Phot. α 1965, ex Σʹʹʹ) | κατάγωμεν Meineke (1814, 11) : κατάγoμεν Σb Phot. cod. b | The translation understands ‘τῶν ἐρίων καὶ τῶν ἀνθῶν ἄνθος’ as a hendiadys for dyed wool (Pellettieri 2024, 56).

Quickly, let us spin wool of all sorts of colours.

(4) Philostr. Im. 1.2.5: ἡ γὰρ τῶν ἀνθέων ἐλευθερία παραιτεῖται τὴν χεῖρα ὡς μαραίνουσαν αὐτὰ πρὸ τοῦ χρόνου.

The freedom of flowers avoids the hand, as though it might wither them before their time.

D. General commentary

Atticist lexicographers compare two forms – ἀνθέων (uncontracted) and ἀνθῶν (contracted) – as the gen. pl. of ἄνθος (‘flower’) and are unanimous in approving ἀνθέων and stigmatising ἀνθῶν. ἄνθος is a neuter s-stem noun, inflecting with ablaut alternation between the suffix -es-/-os- (type γένος). Greek s-stem neuters continue the late PIE pattern (Meissner 2006, 54). They show the o-grade of the suffix outside the nom. sg., acc. sg. and voc. sg., while all other forms take the e-grade. Cases that take the e-grade – and thus the suffix -es- between stem and ending – regularly loose the intervocalic σ (e.g. gen. sg. *γεν-εσ-ος > γέν-εος, from the PIE *génh1-es-os > génh1-os; on the inflection of these nouns see Meissner 2006, 56–60). In Attic and koine Greek, this gives rise to vowel contraction (e.g. gen. sg. γένεος > γένους, nom. acc. pl. γένεα > γένη). For the gen. pl. of ἄνθος specifically, the pattern may be reconstructed as follows: *ἀνθ-εσ-ων > ἀνθέων > ἀνθῶν.

Uncontracted and contracted forms of neuter s-stem nouns coexist in Greek, their distribution depending on dialect and genre. The uncontracted forms are preserved in Homer, in poetry more generally (ἀνθέων occurs, among other places, in Sapph. fr. 94.17; Pi. N. 5.54; B. 11.18; the uncontracted -εων of the gen. pl., disyllabic, is presented as an epic trait preserved in elegy by Passa 2008, 269), and in Herodotus’ Ionic, whereas the contracted inflection is typical of Attic (Chantraine 1961, 69). Attic inscriptions, both in the Old Attic and in the Ionic alphabet, show that the contraction of neuter s-stems was the rule in Attic, whereas uncontracted forms, being a Ionic feature, only occur in poetic texts, particularly during the Roman period (Threatte 1996, 134–5), when they are perceived as a refined trait (see Cassio 1996, 167). Nevertheless, uncontracted forms are attested in tragedy (ἀνθέων occurs, among others, in Eur. Med. 841; Soph. El. 896) and comedy (C.1, C.2, on the latter see Bagordo 2014, 15–6). Xenophon also uses them occasionally, with a different treatment of the gen. pl. of s-stems: in the Anabasis, for instance, the gen. pl. of βέλος is systematically βελῶν, but the gen. pl. of ὄρος (‘mountain’) is more often ὀρέων (11x, while ὀρῶν οccurs only 3x, see below); on Xenophon’s language see Gautier (1911, 79–80, on gen. pl. of s-stems), Cavenaile (1975, 241, on the form ὀρέων anticipating Hellenistic tendency), Huitink, Rodd (2019, 23–9). Both ἀνθέων (C.1, C.2) and ἀνθῶν (C.3, on which see Pellettieri 2024, 52–7) are used in Old Comedy: the decision to use either of the two is based on metrical requirement. For Hermippus’ passage (C.1), a paratragicParody context has been proposed on the grounds that ὕφασμα (‘garment’) and σπαθητός (‘compactly woven’) and compounds are characteristic of tragic language (see Cobet 1858, 29; Comentale 2017, 58–9): the presence of the uncontracted form may also point toward an echo of tragedy in this passage.

Although the contract forms of neuter s-stems may be expected in Attic, ancient doctrines concerning ἀνθέων/ἀνθῶν indicate that the contracted gen. pl. was a special case for Atticist lexicographers, perhaps because the contracted form was also typical of koine Greek. The Antiatticist (A.1) gives ἀνθέων as the expected genitive plural of ἄνθη (the ‘regularly’ contracted nom. pl.): given that this entry is likely to have undergone epitomisation, it remains uncertain whether the penchant for the uncontracted form originally concerned all cases, particularly because the manuscript reading of ἀπὸ <τοῦ> τὰ ἄνθη represents a corrupted ‘τῆς ἄνθης’. The item of the Synagoge’s tradition (B.3) argues that Attic speakers inflect the gen. pl. as Ionic speakers do (that is, in its uncontracted form, as in Herodotus) and recommends the uncontracted form ἀνθέων, alongside βελέων and ὀρέων, for which Xenophon’s authority is invoked (here, διαιρεῖνδιαιρέω, literally ‘to divide’, has the technical meaning of ‘using an uncontracted form’ or ‘resolving a contracted form’, see Dickey 2007, 231). By contrast, other ancient scholars condemn the individual form ἀνθῶν, while admitting the contracted gen. pl. forms of other s-stem neuters. According to the pseudo-Herodianean Περὶ τῶν ζητουμένων (B.2), whereas gen. pl. forms such as βελῶν and γενῶν are acceptable (περισπᾶται ὁμολογουμένως, ‘they have the circumflex accent on the final syllable conformably with [the norm]’), ἀνθῶν must be rejected on the basis that it is confusingly similar to ἀνθ’ ὧν (‘against whom/which’). This doctrine may also be expressed in the Philetaerus (A.2), where examples of admitted contracted gen. pl. forms coincide. This precept appears to reflect the actual use: whereas βελῶν and γενῶν are indeed regularly used in high-register texts and their occurrences outnumber the corresponding uncontracted forms, ἀνθέων is preferred to ἀνθῶν and scores a higher number of occurrences in literary texts. Aside from ἄνθος, ὄρος and χεῖλος (‘lip’) also receive special treatment. Such uncontracted forms are ubiquitous in the koine and are attributed to the influence of Ionic, which favoured uncontracted variants (see e.g. Horrocks 2010, 138). The SeptuagintSeptuagint is an exemplary case, as it systematically opts for the uncontracted genitives ὀρέων (77x), χειλέων (46x) and also τειχέων (12x), although it retains contracted forms elsewhere (see Thackeray 1909, 151; Meecham 1935, 80; on ὀρέων in medieval texts see E.; cf. also Mayser 1906, 17). According to Moulton (1901, 435), the diffusion of such forms was also encouraged based on their being analogous with other gen. pl. forms, such as πόλεων (of πόλις, ‘city’) and βασιλέων (of βασιλεύς, ‘king’). Some forms thus appear to have established themselves as the default option in Greek, likely as a result of their use in certain literary registers or dialects. The use of uncontracted forms in Post-classical Greek is not always determined by metrical requirements, as suggested by the presence of forms such as πληθέων (‘of the multitudes’, unattested elsewhere) in the so-called Problema bovinum attributed to Archimedes (SH 201.43).

While the Synagoge item (B.3) derives from Atticist material, as both the mention of Attic speakers and the use of comic texts as authorities suggest, its source remains uncertain. Latte (1915, 377) identified it with the Antiatticist (A.1), but Valente (2015, 17–8) refutes this attribution, alleging that many Synagoge glosses that share the Antiatticist’s approach are more likely to depend on another source than on an extended version of the Antiatticist. He suggests Orus (Valente 2015, 18), while remaining open to Reitzenstein’s (1907, 139) suggestion, that Phrynichus may be the source of this doctrine. Ritschl (1832, LXXVII) also attempted to trace the doctrine on ἀνθέων to Phrynichus, suggesting that Thomas Magister’s entry (B.4) may derive from the Eclogue. Alpers (1981, 155) rightly dismissed this hypothesis on the grounds that no trace of the gloss is found in family q of the Eclogue; the derivation from Phrynichus is moreover an unnecessary speculation, since, when Thomas Magister (B.4) prescribes the uncontracted form, he depends both on a long tradition of doctrines that recommended it as correct Attic and also on the established use of ἀνθέων in high-register texts (see F.). In invoking the authority of a 2nd–3rd century CE Atticist, such as Philostratus (C.4), Thomas exhibits his originality in enriching of the canon with post-classical authors.

E. Byzantine and Modern Greek commentary

Neuter nouns in -ος remain a productive category in koine and Medieval Greek. Aside from the inherited paroxytone nouns, this nominal category is enriched with newly-formed words and even with forms that resulted from metaplasmsDeclension metaplasm into the s-stems of neuter and masculine nouns of the thematic declension (see CGMEMG vol. 2, 661–3; on masculine nouns developing a later neuter paradigm, see also the entry σκότος). For the gen. pl. of paroxytone nouns, the contracted perispomenon ending in -ῶν – already predominant in koine Greek – continues to prevail, whereas paroxytone gen. pl. forms in -ων are rare in Medieval Greek. -έων survives as an ‘archaic poetic ending’, and its use is restricted to high-registerRegister texts (although ὀρέων survives in mixed-register texts of Medieval Greek too, see CGMEMG vol. 2, 672–3). The uncontracted gen. pl. is particularly helpful in poetic texts, given that forms in -έων serve the needs of dodecasyllables, which require a tonic accent on the penultimate syllable; Manuel Philes, for instance, uses the uncontracted ἀνθέων twelve times, nine of which occur at the end of the verse.

Tracing the occurrences of ἀνθέων and ἀνθῶν in literature (with a plausible margin of error in light of manuscript transmission and the editors’ normalisation) reveals the influence of the Atticising norm that prescribed the uncontracted form. ἀνθέων prevails in high-register texts of late antique (e.g. Libanius, Gregory of Nazianzus) and Byzantine authors (Nicephorus Basilaces, Michael and Nicetas Choniates, Manuel Philes). Several learned authors also use the contracted form (e.g. Photius: ἀνθῶν 7x) but less frequently than the uncontracted form (e.g. Psellus: ἀνθέων 9x, ἀνθῶν 3x, Leo the Wise: 9x, 2x, Theodore II Lascaris: 4x, 0, Nicephorus Gregoras: 9x, 0). In Medieval Greek, ἄνθος is replaced by το λουλούδι (‘flower’), attested from the 15th century (in the Digenis Akritis cod. E; on the etymology of λουλούδι, either from Albanian, or Latin, or Greek λειρίον ‘lily’, see LKN, s.v.), or becomes a masculine noun of the thematic declension (i.e., ο ανθός (attested from the 14th–15th century), with an accentual shift believed to be attributable to the influence of καρπός (‘fruit’, see ILNE and LKN, s.v.).

In Modern Greek, neuter s-stems take the contract endings (e.g., nom. acc. sg. το λάθος, ‘error’, gen. sg. του λάθους, nom. acc. pl. τα λάθη, gen. pl. των λαθών). Note that in Modern Greek, the genitive plural witnesses a ‘full revival, in contrast to its previously more marginal status in earlier demotic’ (Horrocks 2010, 463): it is one of the learned traits that the modern language retrieves. The word used to denote a flower is το λουλούδι, while το άνθος, alongside ο ανθός, primarily features in idiomatic expressions (e.g., ‘το άνθος της ηλικίας’, ‘in [his/her] prime’) and as a rare learned word for which the uncontracted gen. pl. ανθέων, although irregular, is admissible (see LKN, s.v. άνθος). In dialects the noun survives in different variant forms (e.g. ἄθθος in Euboea, Cyprus, the Dodecanese, etc.: see ILNE s.v.) because the sequence νθ is a learned cluster in Modern Greek (since [nθ] yields [θθ], which in some dialects goes to [θ]: see CGMEMG vol. 1, 159).

F. Commentary on individual texts and occurrences

(1)    Σb α 1416 (ex Σʹ) (~ Phot. α 1956, Su. α 2494, ex Σʹʹ) (B.3)

The agreement between Photius, the Suda, and expansion B of the Synagoge verified that the doctrine on ἀνθέων entered the Synagoge’s tradition from an unidentified Atticist source during the production of version Σʹ; at this stage, many Atticist materials were indeed added to Σ (Cunningham 2003, 13). Thereafter, Σʹʹ expanded the gloss of Σʹ, adding information on Xenophon’s use of the uncontracted genitive: this further stage is witnessed by the agreement between Photius and the Suda, while such information is missing from Σb.


Alpers, K. (1981). Das attizistische Lexicon des Oros. Untersuchung und kritische Ausgabe. Berlin, New York.

Bagordo, A. (2014). Alkimenes – Kantharos. Einleitung, Übersetzung, Kommentar. Freiburg.

Cassio, A. C. (1996). ‘La prose ionienne postclassique et la culture de l’Asie mineure à l’époque hellénistique’. Brixhe, C. (ed.). La koinè grecque antique II. La concurrence. Paris, Nancy, 147–70.

Cavenaile, R. (1975). ‘Aperçu sur la langue et le style de Xénophon’. LEC 43.3, 238–52.

Chantraine, P. (1961). Morphologie historique du grec. 2nd ed. Paris.

Cobet, C. G. (1858). Novae lectiones quibus continentur observationes criticae in scriptores Graecos. Leiden.

Comentale, N. (2017). Ermippo. Introduzione, traduzione e commento. Mainz.

Cunningham, I. C. (2003). Synagoge. Συναγωγὴ λέξεων χρησίμων. Texts of the Original Version and of MS. B. Berlin, New York.

Dickey, E. (2007). Ancient Greek Scholarship. A Guide to Finding, Reading, and Understanding Scholia, Commentaries, Lexica, and Grammatical Treatises, from their Beginnings to the Byzantine Period. New York, Oxford.

Gautier, L. (1911). La Langue de Xénophon. Geneva.

Gow, A. S. F. (1955). Theocritus. Edited with a Translation and Commentary. Vol. 2: Commentary, Appendix, Indexes, and Plates. Cambridge.

Horrocks, G. (2010). Greek. A History of the Language and its Speakers. 2nd edition. Chichester.

Huitink, L.; Rood, T. (2019). Xenophon. Anabasis. Book III. Cambridge.

Latte, K. (1915). ‘Zur Zeitbestimmung des Antiatticista’. Hermes 50, 373–94.

Mayser, E. (1906). Grammatik der griechischen Papyri aus der Ptolemäezeit. Mit Einschluss der Gleichzeitigen Ostraka und der in Ägypten Verfassten Inschriften. Vol. 1: Laut-und Wortlehre. Leipzig.

Meecham, H. G. (1935). The Letter of Aristeas. A Linguistic Study with Special Reference to the Greek Bible. Manchester.

Moulton, J. H. (1901). ‘Grammatical Notes from the Papyri (Continued)’. CR 15.9, 434–42.

Pellettieri, A. (2024). Pherekrates. Doulodidaskalos – Korianno (frr. 43–84). Heidelberg.

Reitzenstein, R. (1907). Der Anfang des Lexicons des Photios, herausgegeben von R. Reitzenstein. Leipzig, Berlin.

Ritschl, F. (1832). Thomae Magistri sive Theoduli monachi Ecloga vocum Atticarum. Halle.

Thackeray, H. (1909). A Grammar of the Old Testament in Greek According to the Septuagint. Vol. 1: Introduction, Orthography and Accidence. Cambridge.

Threatte, L. (1996). The Grammar of Attic Inscriptions. Vol. 2: Morphology. Berlin, New York.

Toup, J. (1790). Emendationes in Suidam et Hesychium et alios lexicographos Graecos. Oxford.

Valente, S. (2015). The Antiatticist. Introduction and Critical Edition. Berlin, Boston.


Giulia Gerbi, 'ἀνθέων (Antiatt. α 137, [Hdn.] Philet. 7)', in Olga Tribulato (ed.), Digital Encyclopedia of Atticism. With the assistance of E. N. Merisio.
DOI: https://doi.org/10.30687/DEA/2974-8240/2022/01/005

This article provides a philological and linguistic commentary on the form ἀνθέων, discussed in the Atticist lexica Antiatt. α 137, [Hdn.] Philet. 7.