So what is the positive evidence for a romance genre? The answer lies in the texts themselves, but we shall not find it through exhaustive itemization of lieux communs. The best place to look for generic thinking is, as I have already intimated, in those moments where the generic contract is transgressed: where the effect depends on perceptible refusal to meet readerly expectations or on contamination of different generic codes. Indeed, I would submit that it is here, at the borders, that generic identities are at once most securely determined and most open to revision. They are securely determined, on the one hand, in that acts of transgression reinforce our awareness of the very norms they transgress. Let’s return to Culler’s example of the fed-up hero of Corneille who wishes to retrain as a silversmith: this example highlights the classicizing, aristocratic conventions of action in the French tragic theater. But at the same time (this is where the revision comes in) such an instance would effectively rewrite the rules of the genre for future tragedians. Such cases of aggressively ostentatious rule breaking are relatively rare: Euripides’s Alcestis represents one example from literary history. But minor adjustments of generic codes happen all the time: this is what makes literature fresh, nimble, and inventive rather than repetitively hidebound. And as a result, generic codes are always in process. “Every literary work,” writes Fowler, “changes the genres it relates to. . . . Consequently, all genres are continuously undergoing metamorphosis.”18
This, I think, is the crucial point, and it bears emphasizing. Classicists have been far too prone to assess the validity of the romance genre synchronically, as though we should be asking the same questions of the earliest texts as of the later ones. It is (to exaggerate, but only marginally) as if we were to put together a magic lantern show, Casablanca, and Avatar and ask whether the category “Hollywood blockbuster” worked for all of them. What we need instead is an account of genre that respects the diachronic fluctuations and the way in which each new novel both projects its predecessors as paradigmatic and signals its own generic reinventions.
I cannot, in the compass of a single chapter, map out this process in its entirety, but let me make some general observations and visit some particular instances. For the remainder of this chapter, I shall consider three stages in the history of the romance: the initial phase, namely Chariton’s Callirhoe and Xenophon’s Anthia and Habrocomes; Achilles Tatius’s subversive Leucippe and Clitophon; and finally Heliodorus’s Charicleia and Theagenes, the last of the surviving romances. (Longus, for present purposes, is a marginal case.) I shall proceed not with exhaustive analyses but with a few exemplary discussions of key passages.
There are two enormous questions that confront anyone considering the earliest romances. The first is “Where did they come from?” The second is “Who came first, Xenophon or Chariton?” Let us come to the first via the second, although I do not propose a conclusive answer (which would call for fresh evidence). One recent scholar, Stefan Tilg, has assembled all of the arguments and asserted strenuously that Chariton belongs in the mid-first century.19 That may well be right, but none of the evidence is conclusive: the linguistic criteria are imprecise (and who says that Atticism spread at the same rate everywhere across the empire?), the supposed references to real people implausible, the identification of the addressee Athenagoras hypothetical, and the claimed relationships between Chariton and other first-century writers unconvincing.20 All we can say is that one papyrus from the mid-second century (P.Mich. 1) offers a terminus ante quem. The evidence for dating Xenophon, meanwhile, is even more exiguous. Far too much has been made of the apparent mention of an eirenarch (“the man in charge of peace [eirēnēs],” 2.13.3; see also 3.9.5), an office first attested epigraphically under Trajan. It should not need saying that the first inscriptional mention of such an office does not necessarily mark its first institution.21
It is, however, possible to model the implications, in generic terms, of imagining precedence. Let us consider the well-known fact that (among the many similarities between them)22 the two texts open in very similar ways, with a meeting contrived by Eros between the two beautiful young people at or near a festival. The similarities of motif and even language are so close that it is unthinkable that there is no connection23—but what is the nature of that connection? There is in fact a long and inglorious history of scholarship exploring the question, but in a desperately naïve fashion: ultimately what is at issue is simply establishing chronological priority, which critics determine according to their aesthetic preconceptions about the process of literary succession, or—to use a particularly misleading word beloved of this kind of criticism—imitation.24 According to most critics of this school (and indeed to romantic literary criticism, to which it is indebted), an imitation is inherently inferior to an original. The challenge is thus to demonstrate which text is consistently “better” in those areas of similarity and posit it as the prior one.
This model is evidently outdated, both methodologically and in its estimation of the romances’ sophistication: nowadays we speak not of (passive) imitation but of (dynamic) allusion. The effect generated by the later text depends on the reader’s ability to acknowledge the similarity and to explore the tension between generic identification and local deviation from the model. Let me take just one example, perhaps the best-known point of convergence between Xenophon and Chariton. In both texts, there is a public festival: in Xenophon the phrase is epikhōrios heortē (1.2.2), in Chariton heortē dēmotelēs (1.1.4). This in itself is not surprising: infatuation at a festival is found widely in New Comedy, in Hellenistic love poetry (e.g., in Callimachus’s story of Acontius and Cydippe: Aetia 1.67.6), and indeed already at Lysias 1.20.25 But note that Chariton gives the topos a tweak. The lovers do not meet at the festival, but they bump into each other afterward: “By chance [ek tukhēs] the two met in a passageway at a corner and fell into each other” (1.1.6). This reorientation is, indeed, significant and programmatic: as has often been noted, Chariton tends to minimize direct divine intervention, preferring instead to offer psychological motivations.26 If we interpret Chariton in this way, then the little phrase by chance takes on additional resonances. First, it is heavily ironic: the festival encounter is, of course, so far from being accidentally, instead generically predetermined. Tukhē (fortune), readers of Callirhoe will discover, is a marker of self-conscious authorial intervention in the plot.27 Alternatively, the “accidental” nature of the collision can be read as a commentary on the misfiring topos: one would expect a meeting at the festival proper, but “by chance” they meet elsewhere.
None of this proves that Xenophon is prior to Chariton (though it is of course consistent with that claim). Chariton, indeed, could be playing with the topical status of the amorous meeting at a festival in preromance texts. But if we do hypothesize Xenophon’s priority, or at least the priority of another romance featuring a festival meeting, then we can see instantly how the model of genre bending that I have been proposing may work. Chariton treats the festival encounter as characteristic of the romance au degré zéro and self-consciously marks his own innovation within that frame.28
My second example comes from Achilles Tatius, who wrote in the next generation (a second-century papyrus confirms the terminus ante quem).29 In book 5 of Leucippe and Clitophon, the hero Clitophon—now remarried to Melite—discovers by letter that Leucippe is still alive (5.18.4–5). In a parallel episode in Chariton, as commentators have noted, Callirhoe—now remarried to Dionysius—learns from a letter from Chaereas that he is still alive (4.4.7–10).30 Although it serves a similar narrative function, however, Leucippe’s letter shows no signs of intertextual engagement with Chaereas’s: stylistically speaking, it is ambitious and rhetorical where Chaereas’s letter is sparse and pared down, following (as Konstantin Doulamis has