Artfully False Duals in Empedocles’ Painters Simile (fr. 23 DK) and the Hexameter Tradition

Abstract

In Empedocles’ painters simile of fr. 23 DK, the painters are modified by three dual participles, formerly taken by most scholars to be “false” duals (for plural) like the analogous dual participle in fr. 137 DK. Many recent scholars, however, have agreed that they are true duals indicating two painters, and have allegorized those two painters as representing Love and Strife in some manner, or a demiurgic Love’s two hands. Against that new consensus, this article collects evidence for false dual forms as a poetic license of the hexameter tradition from Homer onward. It is argued that Empedocles’ duals are a fruit of that same tradition; that the painters are not a twosome, but an indefinite plurality; and, finally, that they are not an allegorical representation of Love, although the duals of frr. 23 and 137 do seem to highlight a shared connection with fr. 128 DK and Empedocles’ golden age of Aphrodite.

Keywords:

• Empedocles
• duals
• painters simile
• poetic license
• hexameter tradition

Acknowledgements

This paper stems from my dissertation and a talk given in the Classics Seminar of the University of Oslo and at the conference Phusis kai Phuta I in Paris. I want to thank first my committee, Elizabeth Asmis, Boris Maslov, and Mark Payne for their harmonious work. Special thanks are also due: to Boris Maslov for the invitation to Oslo and for guidance on this paper; to the other Osloenses whose discussion helped spur this research, especially Silvio Bär, Thomas Kjeller Johansen, and Eirik Welo; to Alessandro Buccheri and Arnaud Macé, my co-organizers of Phusis kai Phuta, and our other collaborators; to Elizabeth Asmis, Krista Nelson, Jean-Claude Picot, and Charles Ro for critical help with prior drafts; to the three crows, Branden Kosch, Jeremy Thompson, and David Williams, for dew; to Carolin Hahnemann for sharing her dual inventory; to Patrizia Marzillo for sharing a scan of Scaliger’s text of Empedocles; and finally to the editor and the anonymous reader for their constructive criticism.

Notes

1 23 DK (Diels and Kranz 1951) = 72 vdB (van der Ben 2019) = D60 L­M (Laks and Most 2016) = 21.15–25 G (Gallavotti 1975) = 33 GM (Gemelli Marciano 2013) = 67b MP (Mansfeld and Primavesi 2011) = 14.62 M (McKirahan 2010) = 27 I (Inwood 2001) = F15 Wa (Waterfield 2000) = 15 W (Wright 1981) = 64 B (Bollack 1965–1969) = 119–129 St (Stein 1852) = 154–164 K (Karsten 1838) = 82–92 S (Sturz 1805). Graham (2010) and Vítek (2006) follow DK. Unless otherwise specified, all references to the fragments follow the DK numbering; for convenience, the corresponding numbering is given for fragments discussed at length.

2 παράδϵιγμα δὲ ἐναργὲς … τοῦ ἐκ τῶν αὐτῶν γίνϵσθαι τὰ διάφορα. Translations throughout are my own unless noted otherwise.

3 a is the Aldine edition of 1526, edited by Gian Francesco Torresani (a.k.a. Francesco d’Asola). The other witnesses reported by Diels (1882) are: D, Codex Laurentianus 85.2, of the twelfth or thirteenth century; E, Codex Marcianus Graecus 229 (collocazione 616), of the thirteenth; and F, Codex Marcianus Graecus 227 (collocazione 753), also of the thirteenth. For further details, see Tarán (1987), according to whom Diels’ apparatus contains inaccurate reports of those MSS and ignores numerous other valuable MSS, both known and unknown to Diels, including some probably known to Torresani (Tarán 1987, 258–259); Diels’ mistakes bear on problems e.g. with the readings reported for Simplicius’ text of Empedocles fr. 17 (id., 252), so they may bear on fr. 23 as well. Absent a new edition of in Phys. as promised by Tarán (1987), or any fresh reports of the relevant passages, as undertaken for in Phys. 1.1–2 by Menn (2022) and for the frr. of Parmenides preserved in Simplicius by Coxon (1986; 2009, xiii) (for a critical assessment of Coxon’s work, see Tarán 1987, 260), I follow Diels (1882).

4 Diels (1882) presents E alone as preserving γραφέϵς in any form, writing “γραφέϵς ex γραφϵις [sic] E”; the same is given by van der Ben; DK write, “γραφέϵς] beachte die Cäsurwirkung.” But, as Vítek (2006) notes, γραφέϵς is also read in a, where I read it myself (in a scan) in fol. 34r.; Vítek also attributes the reading to DF, which I have not seen.

5 Diels (1882) says the readings are δϵδαῶτϵ DE : δϵδαῶτϵς aF, but Diels (1901) and DK give δϵδαῶτι DE : δϵδαῶτϵς F, where δϵδαῶτι seems to be an error; see Bollack (1969, 122 n. 2).

6 For μϵίξαντϵ as preferred to the iotacized μίξαντϵ of the MSS, see Chantraine (1999, 676­–677) and Janko (2017, 1 n. 4).

7 As reported by Diels (1882).

8 Where μόξαν is apparently an aorist of μογέω, reconstructed ad hoc.

9 Whether these readings are based on other MSS which Torresani knew remains to be seen. a offers many emendations or variant readings that have not been preferred; yet Torresani, who took over the Aldine press after its namesake’s death, generally proved astute and conservative, according to Cataldi Palau (2002, 32).

10 Some of Scaliger’s readings for fr. 23 have been reported at least since Sturz (1805), but the source is an unpublished MS in Leiden, Ms. Scal. 25, f. 103r, discussed by Marzillo (2010). Scaliger’s notes appended to Stephanus (1573, 216–219) do not treat fr. 23 since it is one of the fragments that Stephanus had not collected.

11 Whether ἀμφὶ is found in any MS remains to be seen. For ἀμφὶ τέχνης, see: Hippoc. VM I.7 (Littré I.570); ἀμφὶ θϵῶν in Empedocles fr. 131.4 and Xenophanes fr. 34.2; and Panzerbieter (1844, ad loc.). Against Vítek’s ᾽μφω, see Gheerbrant (2017, 316), and note that Attic correption of τέχνης is not an obstacle as Vítek (ad loc.) suggests, since it has parallels in the τέχνῃ of the ps.-Homeric (or Hesiodic) Epigram 14.10 (= Hes. fr. 302.10 M–W) and the τέχνας of Philoxenus of Leucas 836(b).5 Page and Telestes 805(b).3 Page, noted by Smyth (1897, 126). Besides, both Attic and epic correption are common in Empedocles, even in fr. 23.6–9; see Gheerbrant (2017, 250–257).

12 See Stein (1852, ad loc.), Gheerbrant (2017, 316).

13 See Kühner–Gerth (1898, 1.69–74), Hillyard (2006).

14 137 DK = 22 vdB = D29 L­M = 122.3–8 G = 175 GM = 35 MP = 14.27 M = 128 I = F38 Wa = 124 W = 137 B = 430–435 St = 410–415 K = 384–389 S.

15 Scaliger does not comment on this in his notes in Stephanus (1573), nor does Sturz (1805).

16 See e.g. Hom. Il. 2.124, Hes. [Sc.] 176–177, and the discussion in Chantraine (1953, 27).

17 A parallel to ἑλών is wanted; see Karsten (1838, ad loc.). The objects of ἑλών and its supplied parallel would do double duty as a two-part, second complement of ἀπορραίσαντϵ, which is possible. Gheerbrant (2017, 316 n. 199) writes that the dual “désigne le père et les enfants. On peut considérer qu’il s’agit de comparer l’action de deux groupes.” For the dual participle to embrace both the two subjects of this clause as well as πατήρ (1) seems strained. And would not the point of the dual then have to be that both groups, distinct though they may be, commit this sacrilege? One expects the distinction to be obscured by the shared evil. Never contemplating a false dual, van der Ben (1975, 202) said παῖδϵς is “generalizing plural, ‘a son’,” and “the dual [refers] to both the son who kills his father […] and the son who kills his mother,” but later van der Ben (2019, 332­­–333) pronounced the text corrupt. Calcidius’s paraphrastic translation at In Tim. 197 makes one son the sole subject: Natus item ut pecudes caedit matremque patremque / nec sentit caros mandens sub dentibus artus (“Just so a son slays both mother and father like cattle / nor perceives that he is eating dear limbs with his teeth”).

18 The rare compound governs double accusatives at Od. 1.404 and 16.428, and at Quint. Smyrn. 3.452; fr. 128, discussed below, arguably contains another instance in θυμὸν ἀπορραίσαντας ἐέδμϵναι ἠέα γυῖα, where γυῖα may be governed by participle and infinitive.

19 The MSS have τῶν, τάς, and unmetrical τά: τῶν is kept by Gallavotti (1975); τὰς by Bollack (1965–1969) and Sider (1984, esp. 14–17).

20 ἄμφω πανταχοῦ οἰκϵίως θϵωρϵῖ. On Empedocles’ cosmic cycle see O’Brien (1969).

21 τῆς ἐνταῦθα δημιουργικῆς συγκράσϵως τὴν Ἀφροδίτην ἤτοι τὴν φιλίαν αἰτίαν φησί.

22 σαφῶς τὴν ἀμφοῖν ἐν πᾶσι μῖξιν τοῦ τϵ νϵίκους καὶ τῆς φιλίας ἐξέφηνϵν.

23 τὴν ἀμφοῖν ἐν τοῖς αὐτοῖς ἐνέργϵιαν.

24 καὶ ὅτι μὲν τὰ πολλὰ ταῦτα ἐν τῷ γϵνητῷ κόσμῳ θϵωρϵῖ, καὶ οὐ μόνον τὸ νϵῖκος ἀλλὰ καὶ τὴν φιλίαν, δῆλον ἐκ τοῦ καὶ δένδρα καὶ ἄνδρας καὶ γυναῖκας καὶ τὰ θηρία ἐκ τούτων λέγϵιν γϵγονέναι.

25 Why Curtius included Pindar is unknown to me.

26 E.g. Bignone (1916).

27 But see below for comparable clusters, especially in ps.-Oppian. Clusters of proper duals are of course easily produced, e.g. the three dual participles in -ϵ at Ar. Av. 43–45.

28 See also Gemelli Marciano (2013, 389).

29 Italics added; see also Wright’s initial comments ad fr. 23.

30 But see Hadot (2008, 22–23 with n. 15) for citation of Dumont. Waterfield (2000), who prints “two painters,” lists both Mourelatos and Furley in his bibliography.

31 See also Ham (2013, 49–52).

32 Sedley is cited on the duals by Mackenzie (2021, 122 n. 71), who is cautious about their significance.

33 Picot (2017–2018, 411 n. 65): “La Haine ne fait pas le même travail que l’Amour, et surtout ne peindrait pas des formes vivantes qui sont l’œuvre d’Aphrodite. Si la Haine avait à peindre, elle peindrait le dinos [the cosmic “swirl” under Strife]! Mais la Haine ne peint pas.” For the “dinos,” see Martin and Primavesi (1999, 147 [d 8]). Perhaps contrast Picot’s ironical suggestion about Strife painting the swirl with Workman (1951, 242): “It is hardly possible to overestimate the almost prophetic insight contained in [Empedocles’] theory of the swirl and its importance to Christian (Renaissance and Baroque) painting.”

34 See frr. 2.1, 3.9 and e.g. Thgn. 624, 1028 with Fränkel (1973, 310 n. 18).

35 Cf. also Gheerbrant (2017, 317).

36 For χϵιροῖν see e.g. Solon 13.49–50, and, for hexameter, Theocritus Id. 21.9. Also available were χϵῖρϵ (e.g. Il. 4.523) and χϵροῖν (e.g. Pind. Ol. 13.95).

37 Cf. Iribarren (2018, 182; 2013, 95).

38 Cf. the “two painters” in Gheerbrant (2018, 67).

39 Pieri (2007), Prier (1976, 138–140).

40 21 DK = 69 vdB = D77 L­M = 21.1–14 G = 31 GM = 66.309–322 MP = 26 I = 14 W = 63 B = 96–109 St = 124–137 K = 68–82 S.

41 See Gheerbrant (2017, 309–314).

42 On frr. 21 and 23 and Empedoclean repetition, see Gheerbrant (2018) and De Rubeis (1991, 93).

43 I have seen four proposed allusions, none casting Love as a painter. Karsten (1838, ad loc.) notes apparent allusions in Gal. In Hipp. De nat. hom. comm. I 3, 32 (CMG V 9,1, 19), where the interest is only in the mixing of elements, and in ps.-Arist. Mund. 396b, where ζωγραφία illustrates how τέχνη imitates φύσις. Jouanna (1990, 209 n. 8), following Cosattini (1909), saw an allusion in Hippoc. VM 20.2 (Littré I.620), where the author, having mentioned Empedocles, says whatever has been written about nature by sophist or doctor pertains less to medicine than to painting. Even if this is right, it only indicates how generally unattractive painting was as a paradigm for physical theory. Jouanna is endorsed by Hadot (2008, 21, 325 n. 11), but for an explanation without an allusion to fr. 23, see Hadot (1990, 2–3 with n. 4). Finally, Garani (2007, 13–14) argues for an allusion in Lucr. 1.820–829, where there is no mention of painting or any demiurgy – only a comparison of atoms with the letters used in Lucretius’ own poetry, which may suggest that Lucretius at least thought that Empedocles himself was implicitly identified with the painters.

44 The further history of this topos in connection with fr. 23 deserves a separate treatment.

45 I owe this suggestion and the final, enthusiastic formulation to Elizabeth Asmis.

46 See the critical approach to the history of allegory in Freidenberg (1997, ch. II), with Maslov (2015, ch. 2).

47 Wackernagel (2009 [1920], 110 [= 1.79]) supposes Dorians and Boeotians to be exceptions, but cf. Willi (2011, 462), Hoekstra (1965, 91–92). The dual was also lost early in Sicily, as shown by Cuny (1906), who neglected Empedocles; setting aside the issue of epic diction, the Empedoclean evidence corroborates Cuny’s conclusions.

48 See Nauck (1848, 35–36) for a precedent to Wackernagel; for later discussions, Meister (1921, 243), Debrunner (1926, 16–18), van der Valk (1964, 68–72), Hoekstra (1969, 28), Richardson (2010, ad 3.456); for opposition, Meillet (1921, 153,163–164), Janko (1991, ad Il. 13.627).

49 τοὺς ϵὐφυϵῖς ἀϵτοῖς παραβάλλϵι, τοὺς ἀφυϵῖς κόραξιν. τὸ γὰρ γαρύϵτον δυικὸν οὐκ ὀρθῶς κϵῖται, οὐδὲ τηροῦσι πάνυ τὸ τοιοῦτον οἱ ποιηταὶ οὗτοι (“He compares the naturally gifted to eagles, the ungifted to crows. For the γαρύϵτον is not used correctly as dual, nor do those poets altogether observe such a matter [i.e. as the correct use of the dual]”). Some, then, did not perceive dual for plural as an archaism so much as a mistake; but the scholiast’s remarks still allow that it was thought to be a false archaism, and anyway those who abused it may have thought it archaic.

50 Gildersleeve (1890, ad loc.): “The dual certainly suggests definite pairs, especially as it is often used with mocking effect.” Gildersleeve endorses the view of Verrall (1880), that the pair intended are Corax and Tisias (another orator); Gildersleeve does not propose that it is “simply deprecatory,” as reported by Most (1986, 304 n. 1), although that is perhaps another option.

51 See Kirkwood (1981, 242); the folklore collected by Kirkwood is partially corroborated by Boarman and Heinrich (1999, 2), who note that the common raven (Corvus corax), the “crow” found throughout the Northern Hemisphere, “[t]ends to be found solitary or in pairs more often than other Corvus, but in many areas seen foraging or roosting in flocks or groups of several hundred to several thousand individuals.”

52 See Wackernagel (2009 [1920], 109 [= 1.78]), Debrunner (1926, 18).

53 In addition to Segal, see e.g. Wyatt (1985, 400), Thornton (1978), Bolling (1933).

54 So also Nagy (2003, 49–54); but cf. Haug (2011, 223), Tichy (1990, 181 with n. 30).

55 Fritz (2011).

56 On Buttmann’s “more independent judgements,” see Wackernagel (2009 [1920], 109 [= 1.78]).

57 Hiersche (1970, 101) credits an “indirekter Einfluß des Böotische” with “der guten Erhaltung des Duals [in Hesiod] gegenüber Homer.”

58 See West (1978, 200; 2001, 42 with n. 39) for more citations.

59 As cited by Troxler (1964, 111–112).

60 The only other instance of δϵδαῶτϵ (fr. 23.2) is in Ap. Rhod. Argon. 1.52, where it suits a pair: υἱέϵς ϵὖ δϵδαῶτϵ δόλους, Ἔρυτος καὶ Ἐχίων (“sons knowing wiles well, Erytos and Echion”).

61 The proverb: κολοιὸς παρὰ κολοιὸν ἱζάνϵι (“crow sits beside crow”); see e.g. Democr. A128 DK, Arist. Eth. Eud. 1235a7.

62 For the third-century CE date of that author, see Vian (1963, XXII).

63 West (1983, 37), Vian (1987, 46).

64 See Vian (1987, 57) for further instances of dual for plural in that text.

65 The only non-hexametric instances: Pind. Pyth. 4.253; Eur. Hyps. fr. 3 iii 30 Bond.

66 Willi cites only Chantraine’s discussion of Homer and evidence of Zenodotus’ acceptance of duals for plurals. In a recent discussion that builds upon Iribarren’s larger argument about demiurgic imagery in Empedocles, Andolfi (2022, 78 with n. 57) shows surprisingly little interest in fr. 23 and cautiously notes that Willi may be right.

67 See esp. Willi (2008, ch. 7), Gemelli Marciano (1990).

68 In fr. 23.8 there may be another noteworthy archaism in δολιχαίωνας, according to Schmitt (1967, 162): “Dies ist das älteste griechische Wort für „langlebig“ und findet sich nur zweimal in poetischen Fragmenten des Empedokles.” The second instance is in the same phrase (now nom.), fr. 21.12. Perhaps it is safer to think it an archaizing coinage. (There is another instance, unnoted by Schmitt: the last word of Constantinus Manasses’ Epist. 2, which is very literary, but does not suggest to me any Empedoclean allusion.)

69 128 DK = 112 vdB = D25 L­M = 119 G = 172 A,B GM = 25a,b MP = 14.4 M = 122 I = F32 Wa = 118 W = 128 B = 405–414 St = 371–377 K = 308–314 S.

70 This resonance is noted by Iribarren (2018, 186–187).

71 Slack (1893, 74), concerning Miguel de Cervantes. For contextualization of fr. 23 among contemporary comparisons of painting and poetry or language more generally, starting with Simonides, see e.g. Detienne (1996, 108), Buchheim (1985). On the possible further significance of painting in fr. 23, see Freidenberg (1997, 145).