γρᾶες, γραῦς, γραῖαι
(Philemo [Laur.] 357, Moer. γ 9, Thom.Mag. 72.2–3)

D. General commentary

Atticist erudition deals with the noun γραῦς (‘old woman’) with respect to both its inflection, in light of the rise of an analogical form of the nominative plural (γραῦς in place of γρᾶες, A.1, A.2, A.3), and its coexistence with the synonym γραῖα (A.2, A.3, B.1, B.2). Both interests are often intertwined in the sources.

Although its inflection is assimilated to that of themes in diphthong (e.g. βοῦς, ‘ox’, Ζεύς, ‘Zeus’), in γραῦς the diphthongDiphthongs is not original, but results from the presence of a suffixSuffixes. Indeed, γραῦς goes back to the zero grade of the IE root *ǵerh₂-, ‘to grow old’; from the same root, with the e-grade, one also has γέρων (‘old man’) and γέρας (‘honour’). The reconstruction of the origin of γραῦς requires facing an array of problems and has been the subject of much scholarly debate whose most recent account is provided by Höfler (2017, 345–9); in that which follows, I am much indebted to the rich and valuable comments of this entry’s anonymous reviewer. Setting aside the now-outdated theory that γραῦς is a root noun (cfr. Chantraine 1933, 97), the discussion centres on two reconstruction proposals offered by Peters (1980, 252 n. 210) and Nussbaum (Nussbaum’s proposal was presented orally but never published; it is summarised by Nikolaev 2003, 179–80). According to Peters, the Ionic and Homeric form γρηῦς (or, in its disyllabic form, γρηΰς) derives from PIE *ĝréh₂-i̯u‑s, with the suffix -i̯u‑, an element that is also found in υἱύς/υἱός, ‘son’ (< *suH-i̯u, see DELG, EDG s.v. According to Szemerényi 1960, 29 n. 2, instead, ‑i̯u‑ goes back to *h₂oi̯u‑ ‘age’, but this solution is hardly convincing). The presence of the suffix ‑i̯u‑ is accepted by Nikolaev (2003, 194–5) – according to whom the element ‑i̯u‑ may be explained as a secondary u-stem derivative of an i-stem whose traces survive exclusively in onomastics; one might thus postulate an original acrostatic stem *ĝórh₂-i‑/*ĝérh₂-i‑ ‘old age’ – and Steer (cf. Höfler 2017, 347). Nussbaum proposes instead that γραῦς derives from the amphikinetic PIE *gérh₂s-ou̯- (later levelled into *ǵr̥h₂su̯-), with gen. sing. *ǵr̥h₂su̯-és, a possessive adjective derived from the s-stem *ǵerh₂-(e)s, ‘old age’. Höfler (2017, 348–9) reaches similar conclusions and suggests that the noun *ǵerh₂-s, ‘old age’ (cf. γέρας, ‘honour’, γῆρας, ‘old age’) is the base for the creation of the possessive adjective *ǵr̥h₂s-o- (lit. ‘who has old age’), which eventually became substantivised into the *gérh₂s-ou̯- proposed by Nussbaum. Given that it envisages a known morphological process and since forms that would otherwise require an analogical explanation are more easily accounted for under his analysis, Nussbaum’s proposal is preferable. Further evidence in support of his theory may be found in the Callimachean (fr. 513) γραῦις /grau̯u̯is/, which may reflect the Aeolic outcome of *-hu̯- < *-su̯-, with the first compensatory lengthening. Note also that γραῦς is comparable to the Mycenaean form ka‑ra‑we, a feminine appellative in KN Ap 694, interpreted as /grāwes/ by Chadwick, Baumbach (1963, 254) and perhaps better viewed as /graiwes/ or /grā(h)wes/. The word γραιβία/γραιτία in Hsch. γ 880, instead, is not connected to γραῦς, since it most likely stands for the ethnic *Γραιϝία, ‘Greek’ (cf. Restelli 2000).

In Attic, the noun inflects as follows: nom. sing. γραῦς, gen. sing. γραός; dat. sing. γραί; acc. sing. γραῦν; nom. pl. γρᾶες, gen. pl. γραῶν, dat. pl. γραυσί(ν), acc. pl. γραῦς). The diphthong ᾱυ shortens to αυ before σ in the nom. sing., acc. pl., and dat. pl., whereas the other cases retain ᾱ. Aside from its standard meaning of ‘old woman’, γραῦς developed two secondary meanings, denoting (1) milk skin, possibly with a play on its wrinkled surface, which would resemble that of an old woman’s skin (this meaning, first attested in Ar. Pl. 1206, is later extended to the film that forms atop a boiling liquid and is commented on by multiple erudite sources, see e.g. Phryn. PS 60.9–10Phryn. PS 60.9–10; Hsch. γ 907, ζ 223; Su. γ 428, B.3, schol. Ar. Pl. 1206); (2) the ‘sea crab’ (Arist. HA 601a.18). Sporadically, γραῦς is also used as an adjective (cf. e.g. Eur. Tr. 490: δούλη γυνὴ γραῦς, ‘an old enslaved woman’).

γραῖα derives from γραῦς with the suffix *‑i̯ă (IE feminine suffix *‑ih₂, see DELG, DGE s.v. γραῦς). It is used both as a noun (‘old woman’) and as an adjective (besides C.1, cf. e.g. Aesch. Eum. 150: γραίας δαίμονας, ‘ancient deities’). In the plural, it is also the proper name of the deities Graiae (cf. Hes. Th. 270). γραῖα takes on the entire semantic spectrum that γραῦς encompasses, including its secondary meanings ‘foam’ and ‘sea crab’ (LSJ s.v.).

In late Attic and Post-classical Greek, u-stems (see entry ἰχθῦς, δρῦς) and stems in diphthong (see entry νῆες, ναῦς, νῆας) develop alternative endings for the nominative and accusative plural – ‑ῦς and ‑ας, respectively. Both innovative forms are analogical: the nom. pl. ending ‑ῦς is shaped in analogy with the inherited acc. pl. (Schwyzer 1939, 564, 578), while the new acc. pl. ending ‑ας is, in turn, modelled on the inherited nom. pl. in ‑ες in analogy with the ending in -ας of thematic nouns of the first declension. An example of this inflectional re-organisation is ναῦς, for which earlier Attic has nom. pl. νῆες, acc. pl. ναῦς, while later Attic and Post-classical Greek have (also) nom. pl. ναῦς, acc. pl. νῆας (see entry νῆες, ναῦς, νῆας, and AGP vol. 2, Morphology, forthcoming, with further bibliography).

The situation of γραῦς is only partially analogous to that of ναῦς: whereas both analogical forms of ναῦς (nom. pl. ναῦς, acc. pl. νῆας) are recorded in literature (although they are attested much later than the analogical forms of u-stems, see entry νῆες, ναῦς, νῆας and AGP vol. 1, 263–4), only the analogical nom. pl. γραῦς is attested, with no trace of an analogical acc. pl. γρᾶας, whose isovocalic hiatus would have been cacophonic. Moreover, the nom. pl. γραῦς, to my knowledge, is not extant in literary texts, in which nom. pl. γρᾶες is the norm. γρᾶες is standard in Old Comedy: attested as the title of a play by Pherecrates, the form is also implied by its use as a vocative in Aristophanes (C.3). γρᾶες then occurs, among others, in Timocles (C.4), Lucian (Philops. 9.7), Plutarch (De superstitione 368d.12), and Pausanias (2.35.7). Conversely, the analogical nom. pl. γραῦς was avoided in literary texts and in formal register and, aside from erudite sources discussing it (A.1, A.2, B.2, Orio 14.13–5, Epim.Hom. in Il. 1.410.B2, and Et.Gen. 1409.3–5 = EM 170.55–7), its only trace is found, to my knowledge, only in P.Cair.Zen. 2.59215 (3rd century BCE, C.5), a private letter in which a certain Dioscurides complains about having received animals that are of little use, among which are two old female pack mules. The occurrence of the nom. pl. γραῦς in a documentary papyrus suggests that the form was in use in the spoken languageColloquial language of the Hellenistic period.

In view of the virtual absence of the analogical nom. pl. γραῦς from literary texts, one would expect Atticist lexicographers to condemn its use, as Philemon (A.1) does. However, the opposite doctrine also exists: Moeris (A.2, on which see F.1) and Thomas Magister (A.3), although within the scope of the opposition of the lexemes γραῦς and γραῖα, surprisingly promote the analogical form. Broadening the scope to the wider debate on the admissibility of the analogical nom. pl. and acc. pl. endings, of which the doctrines on γραῦς are a part, it emerges that the Atticists have a nuanced attitude towards these forms. While they unanimously criticise the innovative forms of ναῦς (see entry νῆες, ναῦς, νῆας), they sometimes adopt a more tolerant stance on the analogical nom. and acc. pl. of u-stems in accordance with the use of Atticising authors (see entry ἰχθῦς, δρῦς). One may conceive that, similarly, the openness toward the nom. pl. γραῦς shown by Moeris (A.2) is due to the fact that literary occurrences of the analogical form existed and were debated; however, the absence of such occurrences in extant literature cause this to remain speculative.

The alternative nom. pl. οἱ βοῦςβοῦς, which is accepted by Philemon (Laur. 356)Philemo (Laur. 356) and recommended by Moeris (β 13Moer. β 13, see F.1), is afforded the same tolerance. It may be possible, although unprovable, that the acceptance of the nom. pl. βοῦς influenced the Atticists’ judgement regarding γραῦς. While Thomas Magister (55.10–3)Thom.Mag. 55.10–3 discourages its use, he signals the form βοῦς as a nom. pl. in Aristophanes (fr. 798, on which see Bagordo 2017, 237) owing to metrical constraints. Although unidentified, Thomas’ source is plausibly Atticist (possibly Phrynichus, according to Kaibel). Considering Thomas’ testimony, the Atticists’ appreciation for the nom. pl. βοῦς – where the diphthong is seen as the result of a regular contraction from βόες, cf. Choerob. in Theodos. GG 4.1.238.27–35 – may be explained by the fact that they read it in Aristophanes’ text or were aware of his use of the form. In this case, the classical pedigree of the nom. pl. βοῦς, guaranteed by Aristophanes’ authority, may have prompted the Atticists to accept – and perhaps even recommend – mutatis mutandis, other alternative forms, such as the analogical nom. pl. γραῦς.

The issue of the admissibility of the innovative nom. pl. is closely connected with that of the overlap between γραῦς and γραῖα. The latter is discouraged by stricter Atticists (A.2, A.3, B.1) and is described by Eustathius as belonging to common language (B.3), while it is admitted by the Antiatticist (B.2). γραῖα does not lack a classical pedigree: it is well attested since Homer (Od. 1.438), occurs multiple times in tragedy (e.g. C.1, C.2), and is used by Aristophanes (Th. 1024) and Plato (Ly. 205d.2). The purists’ disapproval of the form may be attributable to multiple factors. The distribution of γραῖα and γραῦς in canonical authors favours the latter: γραῦς is the standard word for ‘old woman’ in comedy (and particularly in Aristophanes, more than 30x, contra γραῖα once), in Attic orators (Andocides, e.g. Myst. 1062; Demosthenes, e.g. 19.283.6), Plato, and Xenophon. In terms of register, moreover, γραῖα’s abundant occurrences in tragedy, together with its rarity in prose, may have led lexicographers to view it as a poetic wordPoetic language that was unsuited to prose composition (A.3). Moreover, γραῖα was the form that speakers commonly usedColloquial language, as Moeris, Philemon, and Eustathius remark (A.2, B.1, B.3) and as is also suggested by the fact that it (and not γραῦς) is continued in Modern Greek (see E.). In this framework, the Antiatticist (B.2) stands out for defending γραῖα: the entry, which has been handed down to us in a highly condensed form that makes it sound prescriptive, was most likely meant to state the admissibility of both γραῦς and γραῖα, plausibly on the base of a classical occurrence of γραῖα (on the different contexts in which the negation οὐ is used in the Antiatticist, see Valente 2015, 48–9).

E. Byzantine and Modern Greek commentary

Byzantine authors use both γραῦς and γραῖα. γραῦς, which is more frequently employed in literary texts than γραῖα overall, is preferred by authors writing in high register (e.g. Anna Comnene, Michael Choniates, Nicephorus Gregoras) but also occurs in mixed register. γραῖα is instead mostly found in mixed register.

γραῦς has not made it into Modern Greek – a fate that it shares with most athematic (third declension) nouns – although it is used up to the 19th century by Greek scholars supporting the use of Classical Greek in education and erudition (e.g. Neophitus Ducas, Ep. 1335.4). γραῖα is instead continued in the form γριά, resulting from synizesis and the shift of the accent on the last vowel occurring in Medieval Greek (on synizesis see CGMEMG vol. 1, 99–105; note however that γριά tends to be disyllabic, see CGMEMG vol. 2, 508). In literary texts γριά is attested from the 14th century onwards: it occurs, for instance, in the anonymous tale Alexander and Semiramis (e.g. 6.82); however, the form plausibly took ground much earlier in spoken language. In Medieval Greek, γριά has two alternative nom. pl. forms: γριές (since the nom. pl. ending -αι/-αί of the first declension tends to be replaced by -ες/-ές of the third declension, see CGMEMG vol. 2, 458) and the imparisyllabic γριάδες, with the ending -άδες typical of words belonging ‘to the semantic field of status/family relations of females’ (CGMEMG vol. 2, 516). Aside from γριά, Medieval Greek also includes the alternative form γρά, with loss of the /j/ owing to the difficulty in pronouncing the consonant cluster /γrj/ (CGMEMG vol. 2, 512).

F. Commentary on individual texts and occurrences

(1)    Moer. γ 9 (A.2)

Given the absence of the nom. pl. γραῦς from literary texts, Moeris’ precept (A.2) is rather surprising, and it would be tempting to dismantle the discrepancy by emending the entry’s text with Pierson (1759, 110), who corrects γραῖαι to γραῖας, so that the entry would refer to the acc. pl.; see also Sicking (1883, 73). However, Pierson’s attempt proves unnecessary, since the preference for the nom. pl. γραῦς is explicitly expressed by Thomas Magister (A.3), who ostensibly relies on Moeris. The reconstruction of Moeris’ stance on these analogical plurals is no easy task. Moeris appears to express his preference for the analogical nom. pl. in the case of βοῦς: see Moer. β 13, which in Hansen’s edition reads βοῦς ἑνικῶς καὶ πληθυντικῶς <Ἀττικοί>· βόες κοινόν, ‘<Users of Attic> [use] βοῦς as both [nominative] singular and [nominative] plural; βόες [is] common’. To complicate matters further, this entry occurs only in codd. E and F and is thus suspected of being a later interpolation to Moeris’ text (on the manuscript tradition of Moeris’ lexicon, see Cavarzeran, forthcoming). This position may be attributable to the alleged classical pedigree of the nom. pl. βοῦς (see above, D.), which may also be the underlying reason for Moeris’ preference for the nom. pl. γραῦς. However, Moeris rejects the analogical nom. pl. ναῦς and prescribes the old Attic νῆες: in this case, the historiographical tradition – in which νῆες is ubiquitous – may have affected his judgement. In short, Moeris plausibly adopted different stances on the alternative nom. pl. forms depending on the individual word, sometimes following the distribution of the forms in the approved authors (as in νῆες/ναῦς), but in other cases based on an alleged peculiarity of the Attic dialect (βοῦς in Aristophanes?) or analogical reasoning (as in γρᾶες/γραῦς). In any case, A.2 has likely been transmitted in a condensed form, which may justify its unusual formulation. We may plausibly assume that Moeris’ original entry contrasted γραῦς with γρᾶες (as in A.3) and further addressed the issue of the overlapping γραῦς/γραῖα (but it is unclear whether it would be hasty to assign to Moeris a judgement regarding register; cf. Thomas discouraging the use of γραῖαι on the grounds that it is poetic, A.3).