γάρος, γάρον
([Hdn.] Philet. 5)

D. General commentary

An entry in the Philetaerus (A.1) prescribes the masculine gender for the noun γάρος ‘fish sauce’, quoting passages from Aeschylus (C.1) and Sophocles (C.3). Although the entry as transmitted does not specify what the rejected form was, other erudite sources clarify that the question as to whether the word was masculine or neuter was the subject of debate. There existed, in turn, two separate neuter variants, an o-stem γάρον and an s-stem γάρος, γάρους, both of which must be regarded as innovations from the linguistic perspective.

The noun γάρος, of unknown etymology (possibly a loan from Akkadian garûm ‘cream’, itself from Sumerian; see Watson 2013, 186), is scarcely attested in the classical period. Early attestations are all in the form of fragmentary quotations by later authors, taken from comedy (C.4, C.5, C.6) and satyr play (C.1, possibly C.2: see F.1); as Olson and Seaberg (2018, 51) observe, fish sauce was ‘far too undignified a substance to be mentioned onstage in tragedy’. In prose, the only occurrence from the classical period is found in Clearch. fr. 57 Wehrli ap. Ath. 1.5f–6a. The variant τὸ γάρον is first attested with certainty in Strabo (C.7) and subsequently occurs in papyriPapyri (P.Flor. 3.334.5 [= TM 19373], Apollonopolites, 2nd century CE), although, of course, most forms in the paradigm are indistinguishable from the masculine ones; the s-stem τὸ γάρος occurs relatively frequently in papyri from the 2nd century BCE onwards (P.Iand. 8.146.3.3 [= TM 5229], Arsinoites, ca. 180 BCE; see Gignac 1981, 98–9; DGE s.v.) but remains rare in literary sources until late antiquity (first in Gal. De comp. med. sec. loc. 12.622.17 Kühn). Meanwhile, it is unlikely that the spelling γάρρον, which occurs in a textually problematic entry of Orion’s lexicon (B.4), recovers a linguistically authentic variant. The creation of τὸ γάρος is comparable to other instances in which o-stems are occasionally reinterpreted as s-stems, a phenomenon that is older for some words, such as ὁ/τὸ σκότος ‘darkness’, but is mostly documented in Roman and Byzantine times (see Meissner 2006, 95 and the entry σκότος). As far as τὸ γάρον is concerned, it is also worth noting that all attested compoundsCompounds (ἐλαιόγαρον ‘fish preserved in oil’, ὀξύγαρον ‘sauce of vinegar and fish’, οἰνόγαρον ‘fish sauce mixed with wine’, πεπερόγαρον ‘peppered fish sauce’, ὑδρόγαρον ‘fish sauce prepared with water’) are neuter, and may have influenced the gender of the simple noun. Of greatest significance in this respect is ὀξύγαρονὀξύγαρον, which is attested from the 2nd century CE onwards (Arr. Epict. 2.20.30), given that the other compounds are only attested since late antiquity; in fact, the masculine ὀξύγαρος is only transmitted in Ath. Epit. 2.2.3 and in [Zonar.] 1456.5 (= Orus fr. A 67), where the editors correct it to ὀξύγαρον, comparing Phryn. PS 97.10Phryn. PS 97.10 and Ath. 2.67e (in the latter three sources, the spelling with <υ> is prescribed against ὀξόγαρον).

Latin borrowed the Greek noun as the neuter o-stem garum, -ī, which, unlike its Greek counterpart, exhibits no oscillation in gender (the anonymous reviewer kindly directed our attention to an instance of garos in Beda gramm. 7.279.10 muria id est garos, which may be either a borrowing from the o-stem γάρος or a simple transliteration of the Greek word: Bede, living in the 7th–8th centuries CE, had access to now-lost Latin grammatical sources with extensive Greek glossing, and garos may have been added to his source at some point in late antiquity to assist native Greek speakers). It is possible that Latin simply borrowed the innovative variant τὸ γάρον, a hypothesis that can neither be ruled out nor made inevitable by the relative chronology of attestations (garum is attested since Varro, but as Ernout [1954, 72] observed, the word was surely in use since considerably earlier). Alternatively, the change in gender may be due to several Latin-internal reasons. André (1968) discusses parallel instances of borrowing from Greek into Latin accompanied by a change in gender, including several in which the motivation lies in the gender of the corresponding Latin word: note that the native Latin equivalent of γάρος, liquamen, is a neuter noun. Grainger (2018; 2021, 13–43) has recently confronted the thorny issue of terminology, arguing that, in imperial times, there existed two different sauces: liquamen, the more common of the two, was made from whole fish like the Greek γάρος, while garum (sometimes called γάρος αἱμάτιον or γάρον μέλαν in Greek sources) denoted a different sauce that was made from fish blood and viscera and restricted to elite consumption. However, prior to the latter’s invention close to the end of the Republic, garum could, in fact, be used to translate γάρος, and as Grainger herself acknowledges, instances in which garum means ‘primary fish sauce, liquamen’ in imperial-period authors (e.g. Celsus) persist, so that the distinction cannot be taken too rigidly. Latin did also possess a masculine noun garus (-os), -ī, mentioned by Pliny the Elder (C.8) as the name of a small fish that was originally used to prepare garum, which would have been named after it. As Thompson (1947, 43) remarked, γάρος does not occur as a fish-name in Classical Greek, although it occurs in the 6th-century-CE Latin translation of Rufus of Ephesus’ Podagra (10), and an apparent diminutive γαρίσκος is used by Marc. Sid. 33. Moreover, if the hypothesis of a Semitic loanword meaning ‘cream’ is correct, the semantic shift from ‘fish’ to ‘sauce’ likely would not have occurred within Greek. Thus, if the Latin sources are not simply mistaken (as suspected by Thompson and by de Saint-Denis 1947, 40), it is possible that the sauce’s name was secondarily applied to a species of fish that was commonly used as an ingredient (a rare but not unparalleled occurrence: see Strömberg 1943, 88). At any rate, the Romance outcomes attest that garus did exist in spoken Latin as a fish name (see Meyer-Lübke 1923 s.v. garus). One may speculate that the diffusion of garum in Roman times (when the attestations are also most numerous also for the Greek form), together with the word’s stable neuter gender in Latin, may have served as an additional reason for the Philetaerus’ prescription. Indeed, this prescription would have been useful to (even if not expressly intended for) Greek-learners whose first language was Latin, or who hailed from a part of the Empire in which Latin was the dominant language and who might have superimposed the gender of the Latin garum onto the Greek noun (see Adams 2003, 519–20 for examples of contact-induced changes in the gender of nouns). Some influence from the Latin on the Greek within this lexical family is suggested by the noun γαράριον ‘jar for fish sauce’, attested since the 1st or 2nd century CE (BGU 3.781.3.7 [= TM 25637, provenance unknown), which may be derived from γάρος with the Greek suffix -άριον but may also be a borrowing from an unattested Latin *garārium < garum (see Dickey 2023, 101). In Byzantine times, liquamen itself was borrowed as τὸ λικουάμεν (see E.).

Herodian may have been the source of the Philetaerus’ entry, having discussed the gender of γάρος at least in the Περὶ μονήρους λέξεως (B.2), his only work to survive intact, adducing the same Aeschylean quote found in the Philetaerus (on the presence of authentic Herodianic material in the Philetaerus, see the entry [Herodian], Φιλέταιρος (Philetaerus)). The existence of a scholarly debate surrounding the word during the 2nd century CE is confirmed by Athenaeus (B.3), who also prescribes the masculine, and by Pollux (B.1), who mentions the word in a list of condiments with no explicit grammatical prescriptions but clearly once again using it in its masculine form (indeed, Olson and Seaberg [2018, 50] suggest that Pollux and Athenaeus, both of whom cite C.2 as well as C.4, share a common source, despite some minor textual differences). None of the above sources state whether the proscribed form is τὸ γάρον, τὸ γάρος, or both. Much later, the Suda (B.5) continues to prescribe the masculine, whereas Eustathius (B.6), in a passage dependent on Athenaeus but supplemented by Eustathius’ own readings (see Van der Valk ad loc.), reports the existence of τὸ γάρον alongside ὁ γάρος as an example of gender metaplasm, albeit without expressing a preference for either form.

Possibly due to its meaning, γάρος, while frequent in medical and technical works, is not frequently employed by high-register authors of the imperial age. An understandable exception is Alciphron, an author who is both linguistically and thematically indebted to Attic comedy and who employs the noun three times, one of which is unambiguously in the masculine (1.21.1.6: τὸν χρηστὸν καὶ ἡδὺν γάρον, ‘the useful and pleasant fish sauce’). The single occurrence in Aelius Aristides (49.35 Lenz–Behr = 25.497.7 Dindorf) was also thought by Schmid (Atticismus vol. 4, 670) to be drawn from the comic vocabulary, but this assumption is unnecessary, since the context – unlike in Alciphron – is contemporary, and the word had never fallen out of use in spoken Greek, in which it has, in fact, survived to the present day (see E.). Around the time of the Second Sophistic, the noun also occurs as a masculine form in Artemidorus of Daldis (C.9).

E. Byzantine and Modern Greek commentary

All three variants – ὁ γάρος, τὸ γάρον, and τὸ γάρος – continue to be attested in the medieval and early modern periods (Kriaras, LME s.vv.). Almost all attestations in Byzantine literature derive from medical and technical texts, in which a mixture of forms may be found, although the o-stem predominates (often in oblique cases or in the acc. sing. without the article, in which cases it is impossible to disambiguate the gender). The 10th-century farming manual Geoponica (20.46), for instance, contains several recipes for the production of fish sauce (see Grainger 2021, 18–21) that use τὸ γάρος (4x) in addition to the o-stem forms γάρων and τῷ γάρῳ and even the Latin borrowing τὸ λικούαμεν (2x). Standard Modern Greek has retained the masculine ο γάρος, meaning ‘fish sauce’ or ‘brine for pickling fish’. However, Modern Greek dialects attest both masculine (e.g. Epirote γκάρους) and neuter forms, such as the Peloponnesian το γάρος and the Pontic το γάρον (see ILNE s.v.; Shipp 1979, 189). Although masculine nouns in -ος did develop neuter variants in -ο(ν) and -ος during the Byzantine period (see CGMEMG vol. 2, 298), it seems more economical to consider the Medieval and Modern Greek neuter forms as the continuation of those already attested in late antiquity and proscribed by the purists.

F. Commentary on individual texts and occurrences

(1)    [Hdn.] Philet. 5 (A.1)

The Philetaerus adduces two quotations from Athenian playwrights to demonstrate the masculine gender of γάρος. The first omits the source (possibly lost in the course of transmission), but we know from Herodian (B.2) that it is a fragment of Aeschylus’ satyr play Proteus (C.1). The second quote (C.3) is attributed to Sophocles, and its similarity with the fuller quote from Sophocles’ Triptolemus (C.2, see F.2) transmitted by Athenaeus himself and by Pollux has led several scholars to identify in fr. 799a a doublet of fr. 606. Radt (1966)    opposed this view, observing that, since the Philetaerus’ aim is to clarify the noun’s gender, he (or his source) must have selected this quotation precisely because it shows the name in the accusative singular accompanied by the article τόν: thus, it cannot be a confused reminiscence of fr. 606, where the noun occurs in the genitive.

(2)    Soph. fr. 606 (C.2)

Radt’s critical apparatus (TGrF vol. 4, 450) records several attempts to restore the iambic metre in this fragment of Sophocles’ Triptolemus, which is transmitted in slightly different form by Pollux (B.1) and Athenaeus (B.3). It remains uncertain whether the Triptolemus was a tragedy or a satyr play: Olson and Seaberg (2018, 51) argue that the mention of fish sauce and the parallelism with the Aeschylus’ Proteus, which is known to have been a satyr play, weigh in favour of the latter hypothesis, first defended by Brunck (1788, 451). Indeed, ancient lexicographers were interested in satyr dramaSatyr drama because its vocabulary, like that of comedy but unlike that of tragedy, attested to words relating to Realien and everyday life; in particular, it was useful for the Atticist purists who wished to trace the earliest classical attestations of a given word (see Cipolla 2017; 2021). Therefore, given the lack of attestations in Attic prose (see D.), it is unsurprising that the lexicographers turned to this genre, in addition to comedy, to prove the Attic pedigree of γάρος.