Fishing for Pearls

The first unmetrical word of Leonidas, AP 6.4 requires emendation, not explanation. On the basis of a variant in Lucian, a new textual suggestion is made. The paper explores metrical and intertextual criteria for explaining the passage, but rejects them in favour of emendation

The main purpose of this paper is to present some reflections on the relationship between metre and text in Greek literature, by concentrating on cases where emendation, in prose and verse, interacts with metrical considerations. The main case study addressed here is an epigram of Leonidas, AP 6.4, in which most editions permit an unmetrical first word to stand.
A wider interest will become clear as the paper progresses-the old question of the relative importance of 'explaining' apparent irregularities in our texts (in this case, an unmetrical line) as opposed to emending them. Very few editors of Euripides would print the first line of the Ion as transmitted by the manuscripts in violation of 'Porson's law';1 at the same time, a recent commen-Mnemosyne 76 (2023) 560-572 tator on Virgil has repeatedly pointed out that the 'rules' of Latin syntax are treated in the Aeneid more as guidelines.2 This paper is not an edition, and thus need not choose between different modes; it therefore suggests two possible solutions. The crux at issue is Leonidas AP 6.4.1, the metrical irregularity of which has been explained, with some brilliance, as a literary device, with reference both to Hellenistic textual culture broadly conceived and the specific aesthetics in Leonidas. The paper tries to test the hypothesis at issue by making the strongest possible case for this explanation, by adding a number of additional arguments and adducing a range of parallels. But the paper suggests that an emendation is not only available, but is the better solution, being based not only on palaeographical and etymological considerations (the emender's toolkit) but also on the same range of literary allusion that had been adduced in the explanation. While the explanation has much to recommend it, therefore, and constructs a sophisticated series of links between inscriptional and literary epigram and other genres, notably comedy, it will be seen that the emendation is thought preferable.
A well-curved hook, and a long rod, a line, and the baskets that receive the fish, and this pot, contrived for the swimming fish, and the harsh trident, the Neptunian spear, and the twin oars from the boat-these the fisherman Diophantus has dedicated to the lord of his craft, as is correct, the remains of an ancient craftsmanship.
Mnemosyne 76 (2023) 560-572 The poem is a list of objects dedicated by a fisherman: hook, rod,3 line, basket, pot, trident and oars.4 The first word, εὐκαμπές, 'well-curved' , is out of place metrically. Various alterations have been proposed. Hermann's suggestions of γαμψόν and γναμπτόν (the latter with Homeric authority, see Od. 4.369, 12.332), as well as Meineke's στρεπτόν require the addition of an additional particle, τ' , to correct the metre; this in effect requires two corruptions rather than one. Another suggestion by Meineke, καμπύλον, is rather better, at least restoring an adjective of correct metrical shape; the same can be said in favour of Geist's εὐπαγές and εὐαγρές. Finally, Hermann recanted and followed Blomfield's suggestion of reordering the terms: ἄγκιστρ᾿ εὐκαμπῆ-again, with the implication of two errors rather than one.5 However, the only suggestion that recent treatments have thought worthy of recording is Salmasius' εὐκαπές, a compound with an otherwise unattested second member.6 In a study of Leonidas' language, De Stefani simply records the form as Salmasius' conjecture-"la soluzione migliore, direi", echoing Gow-Page's judgement. De Stefani terms the conjecture an "intervento", but we might have expected to be told that the form is also "inventato".7 More recently, Durbec's paper prints the text of the epigram with εὐκαπές without even indicating that there is a textual problem.8 Critical opinion thus seems, to some extent, to have solidified around the idea, though Solitario's recent study cautiously retains the cruces.9 Jacobs and Geffcken, meanwhile, took a more daring approach. Comparing the treatment in Homer of ἀνδροτῆτα (apparently scanned ⏑⏑ -⏑ at Il. 16.857, 22.363, in the estimation of many on the basis of an earlier *anr̥ tāta), they propose to scan εὐκαμπές as a dactyl.10 The fact that a different historical situation 3 Gow and Page 1965, vol. 2, 360 justify their adoption of Knaack's emendation of δούρατα (MS) to δούνακα. 4 For the reception history of this epigram, see the 'stemma' in Geffcken 1896, 113;Ypsilanti 2006;and Durbec 2012/2013. Geffcken 1896 calls the piece an "inhaltsloses und geschmackloses Gedicht", and see von Wilamowitz-Moellendorf 1903, 55; contrast the more charitable assessment by Gigante 1971, 21-22. 5 Dübner 1871 underlies both forms is, for the avoidance of doubt, immaterial here: the synchronic irregularity of ἀνδροτῆτα only needs to have been available to provide the basis for an analogy (along the lines of a rule such as 'regard preconsonantal nasal as irrelevant for scansion').11 Plainly, however, wholesale discounting of preconsonantal nasals for metrical purposes did not occur: they are far more often demanded by the metre than not. Recently, however, a version of this proposal has been defended by Cusset, who sees the use of the word as an explicit aesthetic choice by a poet intent on bringing his epigram into the everyday world of the fisherman: the unmetrical word is "l'indice d'une poétique spécifique de l'humilité".12 Cusset, rather than considering the unmetrical word a license, sees it as a literary strategy. Cusset's claim is far from being unreasonable.13 Although this paper will ultimately make an alternative suggestion, Cusset's argument is nevertheless serious enough that it is worth making the strongest possible argument in its favour. Leonidas' metrical practice is known to have peculiarities.14 The second line of this very poem (6.4.2) demonstrates an example: the use of the definite article before the caesura.15 It is also true that verses produced for inscriptions-for example, the 'real' dedications on which literary epigrams such as this one are based-not infrequently feature metrical irregularities.16 Leonidas' mimesis, therefore, may be of a real, metrically defective dedication, perhaps of the sort we see in CEG II 744 (ξυνβώμοις τε θεοῖς διδασκαλ | ίας τόδε δῶρον, l. 3, with a trochee in the third foot); 770 (ii) 1 ([τόνδε, παῖ] Ἀγρετέρα Διὸς Λητοῦς τε ἰοκόλπου, third foot trochee); 789 (i) 3 (πεζοὶ δὲ ἱππῆς τε γέρας θέσαν, οὓς προέηκεν; if the start of the line is scanned, irregularly, as --⏑ --, with 11 Should this ever prove to have been a synchronic rule, there are interesting implications for the status of the reading ἀδροτῆτα in some manuscripts of the Iliad (for details, see West's apparatus ad locc.). This reading may reflect the loss of /n/ from syllable codas in later Greek, or deliberate adjustments by metrically informed copyists (and editors). In that case, what of our εὐκαμπές? 12 Cusset 2017, 39. 13 It should be noted, however, that it can hardly be reconciled with Gutzwiller's interpretation, according to which Leonidas' ornate poems are an act of the conferral of value, not a creation of ironic distance; see Gutzwiller 1998, 94-95. 14 Nonetheless, see De Stefani 2005, 157 for a comparison of Leonidas and Posidippus, according to which the latter is much freer in his hexameters. 15 Gow and Page 1965, vol. 2, 121 (on 913) and for Leonidas in particular 336 (on 2119f.). Another peculiarity, the scansion ὁρμῑη, is paralleled in Theoc. 21.11; this is likely metrically determined and not a thematic link between the two poems, and given the existence of Greek words in -ῑα there is no need to assume that the author of one of these pieces took the form from the work of the other. 16 See Todd 1939. Mnemosyne 76 (2023) 560-572 an iamb in the second foot, the line is a close match for Leonidas AP 6.4).17 Hellenistic poets, engaged in a much more rarefied exercise, were presumably able to detect 'errors' in inscriptional material; imitation of such an unrefined style of verse can perhaps not be excluded.18 Slightly disturbingly, however, this kind of line is attested already in archaic verse: ἐτρέφετ᾿ ἀτάλλων, μέγα νήπιος, ᾧ ἔνι οἴκῳ (Hes. Op. 131) ἠδ᾿ ὁπόσα τολύπευσε σὺν αὐτῷ καὶ πάθεν ἄλγεα (Il. 24.7) φωνῆς γὰρ ἤκουσ᾿, ἀτὰρ οὐκ ἴδον ὀφθαλμοῖσιν (h. Cer. 57) On the whole, the forms ἀτάλλων, ὁπόσα and γάρ are interpreted as undergoing metrical lengthening;19 the openings of all these lines would in theory work as iambics (albeit with a good deal of resolution), but the case of the Homeric Hymn to Demeter is particularly close to our Leonidas epigram.20 In short, Leonidas had a range of models on which to base this kind of metrical 'exception'; this complicates considerably our ability to gauge exactly what aesthetic effect may be in play. One might just as well see this as a 'grand' , 'rugged' or 'archaic' feature as an expression of 'humility' .21 On the other hand, the figure of the fisherman is one attested rather widely in Hellenistic poetry. Comic tropes about fishermen seem to have inspired Theocritus 21, a conversation between two fishermen-one named Diophantus, which can hardly be a coincidence.22 The fishermen of Plautus, Rudens 290ff. and Gripus' speech at 906ff., if they are reprising material from Diphilus' original, may be a further link to Hellenistic comedy (even Gripus' name, from γριπεύς, is an index of the importance of this trope). More direct evidence from 17 Obviously the hiatus would, in classical verse, be very strange. Should one understand πεζοὶ δέ (for δή?) θ᾿ ἵππης τε? 18 Our understanding of Leonidas' metrical practice will be put on reprises a number of elements from Aristophanes' Wasps 13-53. The numerous links between comedy and these Hellenistic literary fishermen suggest a modification of Cusset's theory. Rather than removing the epigram from the sphere of 'lofty' literature in more humble territory, the epigram is refitted explicitly as a piece of comedy. This is signalled not only by the content, but by the metre: an opening --⏑ --primes the ear for a fragment in trimeters, and the poetics of comedy and the everyday. The opening line of Leonidas 6.4 thus becomes a promise of comic performance, only to be thwarted by the final syllable of ἄγκιστρον, when the hexameter reasserts itself.24 In the following section we explore this theme a little further, as it seems that this can be fitted into a still wider consideration of metrical form in the Hellenistic period. ἐρέ͜ ω γὰρ οὕτω· "Κυλλήνιε Μαιάδος Ἑρμῆ" Hipponax fr. 23 (= fr. 11 Degani, preserved in the same context by Priscian) has too corrupt a text to be reliable, but seems to have had the same feature. A similar mixture is found in the 'archilochian' (-⏑⏑ -⏑⏑ -⏑⏑ -⏑⏑ | -⏑ -⏑ -x), used by Callimachus (ep. 39 Pf. = 1137-1142 HE Gow-Page). Furthermore, literary theory and practice in the ancient world knew to exploit linguistic features of Greek in metrical 'games' , in which texts could be analysed according to different metrical schemes.26 Lines such as Il. 23.644 ἔργων τοιούτων. ἐμὲ δὲ χρὴ γήραϊ λυγρῷ, Hes. fr. 270 M.-W. πίσσης τε δνοφερῆς καὶ κέδρου νηλέι καπνῷ, Empedocles 31 B 128.6 σμύρνης ἀκράτου καὶ κέδρου νηλέι καπνῷ can be scanned as iambic trimeters as well as hexameters; the observation goes back to Andronicus, recorded in the scholia to the Iliad passage. Three distinct features of Greek are being exploited: (a) the possibility of correption of long vowels as against hiatus e.g. scanning τοιούτων ---or ⏑ --; 25 Meineke 1823, 233. 26 See Kassel 1981, 11-18 = Kassel 1991, on whom the following account closely depends, and Handley 1988. Mnemosyne 76 (2023) 560-572 (b) muta cum liquida sequences being treated in separate syllables ('making position') or as syllable onsets ('Attic correption'), as λυγρῷ, καπνῷ ⏑ -; (c) treating diphthongs as sequences of two short vowels, e.g. γήραι, νηλέι -⏑⏑ or --. Now Callimachus deployed this feature of Greek in a pentameter which quoted a line from Euripides' Bacchae in an epigram, in which a theatre mask listens to a classroom discussion (ep. 48 Pf. = HE 1165-1170 Gow-Page): ἐγὼ δ᾿ ἀνὰ τῇδε κεχηνώς κεῖμαι τοῦ Σαμίου διπλόον, ὁ τραγικός παιδαρίων Διόνυσος ἐπήκοος· οἱ δὲ λέγουσιν "ἱερὸς ὁ πλόκαμος", τοὐμὸν ὄνειαρ ἐμοί.
I am sat here yawning, the double of the Samian-the tragic Dionysus listening to the boys. They say "the hair is sacred"-old news to me.
The quotation, in its original context at Bacchae 494, must be scanned ⏑⏑⏑ | ⏑⏑⏑ | -; the different treatment of muta cum liquida sequences in hexameter, however, also allows the line to be read as -⏑⏑ | -⏑⏑ | -. Callimachus' clever game exploits a metrical ambiguity, allowing the incorporation of a quotation from one metrical context into another. How are we to imagine Callimachus reading such a poem aloud? It is not inconceivable that the original prosody was used, and that a performance of elegiac metre was disrupted by a 'surprising' (half) trimeter. This would bring the case closer to Hipponax fr. 35 cited above, as well as matching the kind of practice we have suggested for Menander fr. 852; this would seem then to provide an Hellenistic counterpart to the metrical experiment suggested for Leonidas.
Mnemosyne 76 (2023) 560-572 support; nevertheless, textual correction, rather than interpretation, can be shown to be the preferable answer. At Lucian 23.14 (Prometheus), we read the following in most editions: ὅτι δὲ καὶ χρήσιμα ταῦτα γεγένηται τοῖς θεοῖς, οὕτως ἂν μάθοις, εἰ ἐπιβλέψειας ἅπασαν τὴν γῆν οὐκέτ᾿ αὐχμηρὰν καὶ ἀκαλλῆ οὖσαν, ἀλλὰ πόλεσι καὶ γεωργίαις καὶ φυτοῖς ἡμέροις διακεκοσμημένην … You'd know that these good things come from the gods, if you saw that the whole world was no longer dry and unattractive, but adorned with cities and farms and gentle plants … In place of ἀκαλλῆ, 'unattractive' , the reading of Φ, the scholiast knows the reading ἀκαμῆ, a word the scholiast glosses as ἀντὶ τοῦ ἀκατασκεύαστον, ἀπεριποίητον, ἀκατέργαστον ('unprepared, unworked, unlaboured on'). If ἀκαμής is a possible form, and its word formation is from a linguistic point of view impeccable,27 then it could be that εὐκαμές, 'well-wrought, well-worked' , is a good emendation for our Leonidas poem. A form εὐκαμές has the advantage over Salmasius's suggestion εὐκαπές of an attested (albeit weakly attested) second member of the compound, and the advantage of palaeography over the remainder, as well as satisfying metrical criteria. Leonidas is exceptionally fond of forms in εὐ-, so a solution which retains this element (unlike Meineke's καμπύλον, for example) is in keeping with his practice.28 One notes that of the various authorities canvassed above, Geist came closest to this stylistic trait, while the solution of Blomfield which ultimately satisfied Hermann sacrificed it. Perhaps the most obvious objection to this conjecture is the following: 'Well-made' is a bland epithet for a fisherman's hook-perhaps even intolerably bland. However, this is a less impressive counter-argument than might 27 We will come to consider the form's place within Greek systems of internal derivation (i.e. synchronic word-formation rules). From the point of view of modern etymological thought, the second member of the compound would be derived ultimately from the root of the verb κάμνω, *kemh 2 -; an aorist *ekemh 2 t will have given first *ekema(t) and then by regular metathesis ἔκαμε (see Strunk 1967, 133-134). Thence, the stem καμ-could be used to form derivatives such as s-stem adjectives, reinforced by genuine inherited zero-grades like παθ-(ἔπαθον, ἀ-παθής). See Meißner 2006, 199-203, and 186-197  ἀ-κμής and ἀ-κάματος (also, e.g., εὐκάματος). The relationship of these forms to κάμνω was transparent in antiquity: the etymology is given at EM 49.28-32 Gaisford (this passage also cites the form καμής, ἀκαμής).30 As explained in footnote 27, there is no reason to suspect the form ἀκαμής in principle: it may simply have been a word seldom written into literature. Nonetheless, we might also reflect on the analogies working within Greek to produce the form; and in fact a precise proportional analogy can be stated: -θάνατος: -θανής :: -κάματος : Χ, Χ = -καμής In this case, the analogy works by considering the relationship between different second members of compounds. Since competing formations in -θανής and -θάνατος (e.g. δισθανής and ἀθάνατος) existed, forms such as εὐκάματος (first at E. Ba. 66) and ἀκάματος (in Homer and Hesiod, cf. e.g. Il. 5.4) can be used as the basis for creating a form in -καμής. This is also additional support for the notion that an ἀκαμής in Lucian can be the basis for conjecturing εὐκαμής, since the first members of these compounds likewise agree with those of the assumed models in -κάματος. In any case, εὐκαμές is no less trivial than the offering of the paradosis, εὐκαμπές, 'well-curved' , itself hardly a brilliant jeu d'esprit as an epithet for a hook;31 εὐκαμπές is, however, a natural enough slip, or indeed deliberate correction, for the relatively unfamiliar εὐκαμές in an exemplar.32 It is this consideration that, on balance, makes this proposal more likely than the metrical game proposed 29 The bibliography is vast, and a complete enumeration will not be attempted: by Cusset and developed further in the opening of this article. Even though Leonidas' metrical practice shows peculiarities, it must be admitted that a line opening with a deliberately unmetrical combination of syllables is unparalleled in literary epigram; and if deliberate, it was an experiment Leonidas never repeated, nor was it imitated by other literary epigrammatists-surely significant in the case of the poet "more imitated by later epigrammatists than any other Hellenistic author".33 Furthermore, unusual word-formation, by contrast, is positively a calling-card of the poet, and has been extensively analysed and illustrated in this very poem: "leonideisch" is the laconic remark of Geffcken on ἰχθυδόκος (6.4.2) and τεχνοσύνα (6.4.8); Gow and Page comment on the semantic distinction between τέχνη and τεχνοσύνα, accepted reluctantly by Solitario; and Gigante draws attention to the further play in τεχνασθέντα (6.4.3).34 This seems at odds with our earlier conclusion, that Leonidas' use of metre in 6.4 was paralleled in other texts. Yet the fact is that none of these parallels was quite exact. They relied either on quotation or on some prosodical ambiguity in the Greek language. These prosodical ambiguities-different treatments of long vowels in hiatus, different treatments of mute plus liquid sequences-are of a different order to the metrical license alleged to permit ἀ(ν)δροτῆτα, which cannot have been sprachecht: the 'rule' that made preconsonantal nasals transparent for scansion purposes can only have been a literary license. As a result, none of the parallels was in fact as close as it appeared: instead, different phenomena conspired to produce similar looking results. It follows that the emendation of εὐκαμπές to εὐκαμές is the best solution.

4
To return to our opening reflections. This paper has expended a deal of energy to mount as strong as possible a case to justify an unmetrical text in terms of literary expression. It is important that this sort of explanation is not dismissed. It was possible to mount up parallels which served to confirm the possibility of an explanation, and furthermore these parallels suggested that we have not yet exhausted the exploration of ancient literary technique. If we then turned to emendation after all, it was because none of the parallels turned out to be precise. But emendation itself was shown to depend on literary and linguistic case of Leonidas 6.4, however, the metre remains a problem; and the prevalence of the term in exactly this sort of context may have been a contributing factor in the corruption. 33 Gow andPage 1965, vol. 2, 307. 34 Geffcken 1896, 114;Gow and Page 1965, vol. 2, 361;Solitario 2015, 59;Gigante 1971, 63. analysis of texts in quite another department of literature. Emendation, then, is part of literary appreciation, not an adjunct to it-and not simply a means of getting one's own way before the game has begun. Traditional philology draws on as sophisticated an approach to work on textuality as more modern methods do.