Chapter 6

Beyond Belief (Cazotte)

In his famous essay on the uncanny, Freud does not attempt to provide a literary history of this effect of “dread and creeping horror,” but he gives enough examples to allow us to sketch one out.¹ He finds the uncanny in the stories of Hoffmann and Schnitzler, and even in a Strand magazine piece in which carved crocodiles come to life. He does not find it in “Homer’s jovial world of gods,” nor in Dante and Shakespeare, whose works nonetheless abound in spiritual entities. Fairy tales, meanwhile, are “crammed with instantaneous wish-fulfillments”; “and yet,” writes Freud, “I cannot think of any genuine fairy-story which has anything uncanny about it.” Sheer literary talent cannot be the explanation for the uncanny, since the “thoroughly silly” story in the Strand easily beats Shakespeare and Homer. So even if Freud doesn’t actually draw the conclusion, it would seem from his examples that the uncanny is above all a historical phenomenon: the literature of many periods has supernatural or irrational content, but modern literature is the uncanny’s only home. Such a periodization has been confirmed by subsequent research into the literary genre that best overlaps with the uncanny, the fantastic: both are defined by an intrusion of the irrational into the seemingly rational world, and both appear to come into their own in the nineteenth century.²

One popular explanation for the uncanny’s historical intrusion has been to stress an evolution in mentality—in people’s beliefs about the supernatural, in their faith in reason. Freud’s essay points firmly in this direction. Since the uncanny is generated by the shocking eruption of the irrational into the rational world, it cannot very well exist before the world itself is viewed rationally; indeed, the uncanny is precisely the return of beliefs “surmounted” by the process of rationalization, but which “still exist within us ready to seize upon any confirmation.”³ Freud himself provides no indication of when that surmounting occurred, but others have singled out the Enlightenment as the moment when ghosts exit the material world and enter literature. For E. J. Clery, the early 1760s seem emblematic: in 1762, the so-called Cock Lane ghost attracts much attention, both skeptical and credulous; two years later, one of the skeptics, Walpole, converts the polemic into entertainment by penning the gothic harbinger The Castle of Otranto.⁴ In her analysis of gothic-era ghost stories, Terry Castle sets the shift a bit later, at the turn of the nineteenth century, “precisely [the] moment when traditional credulity had begun to give way, more or less definitively, to the arguments of scientific rationality.”⁵ Either way, our uncanny literature allows us to savor at nostalgic remove a realm of experience otherwise lost to Enlightenment rationality. “Aesthetically induced demonic dread of the sort Freud describes,” writes Victoria Nelson, “is finally all that we superstition-free rationalists possess of the numinous.”⁶

Fantastic literature may well provide life support for beliefs that have otherwise become embarrassing, but throughout this chapter I will sidestep appeals to a sea-change in Western rationality as the genre’s source or explanation. First, it would be a tall order indeed to decide just when we surmounted the “traditional credulity” of our ancestors: when Castle tells us that the literary uncanny arises “precisely” at the point that scientific rationality “more or less” “had begun” to win out, she unwittingly highlights the elasticity of a transition that can be positioned pretty much wherever the argument requires.⁷ Second, and much more important, the proper place of the supernatural, the irrational, and the unbelievable in literature had long been a subject of debate—in Enlightenment England and France, certainly, but also in the neoclassical French seventeenth century, and before that in the Italian Renaissance; all these writers and thinkers, meanwhile, were turning over some basic if ambiguous precepts found already in Aristotle and Horace. Chief among these was the latter’s dictum incredulus odi, typically taken to mean that unbelievable subject matter could give no pleasure.⁸ No doubt real beliefs, both popular and elite, did change over such a long period. But the advent of the uncanny or fantastic genre, as well as its break from previous accepted uses of the supernatural, is a problem whose solution lies less in the process of rationalization than in aesthetic and literary history.

The reason we can date Freud’s uncanny to the beginning of the nineteenth century has to do, then, with beliefs about literature, beliefs that change at least in part due to the experiments of writers. Only when novels start to prove, through their own aesthetic efficacy, that we can indeed take things we read seriously without crediting them as having really happened—only then is the door open to a genre that mingles the supernatural and the realistic. The world of such a type of novel is no longer the ersatz real world of the pseudofactual novel. Instead, it becomes a fictional world soliciting a type of belief that can be quite vivid but that remains qualitatively distinct from the “historical faith” Richardson had supposed necessary to the success of Clarissa. The change was both sudden and a long time coming, and the present chapter approaches it through a short novel that has often been advanced as the original fantastic narrative, Cazotte’s Le Diable amoureux (1772). While the fantastic designation seems to me doubtful, Le Diable amoureux is actually more interesting, historically speaking, than it would be if it were a beginning: it is a self-conscious effort to overcome the limits traditionally imposed on the marvelous that does not, however, produce a formula that will be copied.

My argument proceeds in three stages. First, I delve into the tradition of thought about the poetic marvelous that Cazotte confronted. Writers—even neoclassical writers like Corneille and Dryden—were not so much hostile to the supernatural as convinced that there should be two classes of literature that did not communicate. I call this doctrine—though it is really more like a set of interlocking assumptions—the Great Divide. On the one hand, there was a literature of illusion and emotion that functioned because its audience believed it literally (if incompletely); on the other, there was a literature that appealed to the sensory imagination alone and produced a distinct sort of pleasure. It was this divide that Cazotte, steeped in earlier uses of the marvelous ranging from Ariosto to fairy tales, sought to overcome, and so my second section examines the work whose very subject is the protagonist’s love for a devilish creature who herself (apparently) longs to become human, that is, subject to the laws and pleasures of the flesh. The chapter’s last part reckons with the demonstrable fact that Le Diable amoureux was part of no movement or school: even if the novel can legitimately be called fictional, it did not initiate any broad change in literary practice; in France, only under the influence of the English gothic in the 1790s did the supernatural start to intrude into the world of the novel. And the same can be said for the apparently pioneering work of the English gothic: Walpole’s Castle of Otranto (1764) sought to bridge the Great Divide, but it was essentially a one-off. Though both novels can arguably considered fictional, they were so in isolation; neither hit upon the formula that powers modern supernatural narrative.

The Great Divide

The story of Cazotte’s supernatural obsessions leads back to what many think of as a most ununcanny place—to the neoclassical seventeenth century, ruled by the sensible, sing-song demands of vraisemblance and bienséance. Critics often point to Corneille’s early tragedy Médée (1635) as a last gasp of the literary irrational. Mythological, supernatural, and violent, it was the Other that Corneille himself had to repress in order to found a political and psychological tragedy, a tragedy that was “civilized” in Norbert Elias’s sense of the word, “disenchanted” in Max Weber’s.⁹ After the infanticidal sorceress of Médée exits the stage in a flying chariot pulled by dragons, Corneille distances himself from mythological subjects, and as goes Corneille goes neoclassical tragedy more generally: no flying chariots are to be found in Racine’s tragedies, certainly; only Voltaire, under the influence of Shakespeare, will make an anemic attempt to bring a ghost back on stage in Sémiramis (1748).

This, however, is only part of the picture—half of it, precisely. Corneille himself was not hostile to myth, and in 1650 offered audiences Andromède—a machine play that challenged recently imported Italian operas through the use of music, elaborate sets, and special effects. And so when Medea pops up again at the end of the century, she does so in an opera, with music by Charpentier and a libretto by Corneille’s younger brother Thomas.¹⁰ The gods of old were not put to death by reason, after all. With Medea and her chariot, they simply migrated across the Great Divide that separated a literature of belief from a literature of the imagination.

This migration marked a decisive modification of longstanding thought about the poetic marvelous—a thought that was both complicated and downright paradoxical. Complicated, because the marvelous was a grab-bag category comprising anything out of the ordinary that could stimulate the audience’s sense of wonder. The category included the pagan gods of Greece and Rome, and the Christian God as well; but the marvelous was also to be found in other types of content—purely human actions that were extreme enough, for example. (Medea was just as “marvelous” for killing her children as for driving her dragon chariot.) Moreover, the passions elicited by marvels—wonder and admiration—could be provoked not only by what was represented but also by the manner of representation itself. Ingenious plot reversals, unexpected recognitions, episodic digressions that impressed by their variety, even poetic language itself—all of these were evoked in discussions of the marvelous. Muddling matters still further were generic divergences: epic narration was widely held to be more tolerant of the marvelous than tragic representation. And beneath all this lay a kind of paradox. Since belief in the poem was essential to the stimulation of the audience’s passions, the very marvelous that grabbed our interest and provoked our admiration worked against our absorption into the artwork. “Whatever I cannot for a moment believe, I cannot for a moment behold with interest or anxiety”: such was Johnson’s paraphrase of Horace’s famous incredulus odi line.¹¹ Like the God of Job, then, the marvelous both giveth and taketh away.

Though the fine line between stimulating interest and prompting rejection is arguably still with us, the French seventeenth century witnessed a generic reorganization that mitigated substantially the paradox of the marvelous. This was not primarily a theoretical achievement: disagreement about the marvelous was no less prevalent among seventeenth-century authorities such as Chapelain, Rapin, or Le Bossu than in the writings of Castelvetro and Tasso that inspired them. It is rather in the period’s poetic production that a generic division of labor became obvious. Marvels related to plotting, for example, were reserved for the romance novel and for tragicomedy. Tragedy proper focused on two kinds of marvelous emphasized by Castelvetro: men doing horrible deeds either deliberately or against their will. The marvelous gods, meanwhile, were exiled to opera.¹²

If the new generic topography was not justified by a full-blown theoretical program, it was not arbitrary either: its logic may be gleaned by crossing isolated or occasional remarks. Audience emotion, people reasoned, was of essentially two different sorts. On the one hand the literary work could stimulate the passions of the heart; according to this model, belief was indeed a prerequisite, for without it, we would not feel—with Johnson’s “interest or anxiety”—the contagious passions of the characters.¹³ On the other hand, an art form like opera produced a different sort of emotion—a pleasure that was visual and auditory. Corneille, in his remarks on his machine play Andromède, is quite explicit: “My main goal here,” he writes, “has been to satisfy the gaze by the spectacle’s brilliance [éclat] and richness [diversité], and not to touch the mind by the force of argument, or the heart by the delicacy of the passions…. I might as well admit that this play is for the eyes alone.”¹⁴ Whereas theorists from Chapelain to d’Aubignac routinely predicated the effects of tragedy on the carefully cultivated belief of the audience, both defenders and critics of the new opera referred instead to a different kind of illusion, a sensory enthrallment that was not belief—ravissement, charme, and enchantement are some of the terms used by writers such as Saint-Évremond and La Bruyère. “There are therefore poetical reasons why at the opera I can judge excellent things that I would find detestable at the theater,” wrote Perrault in 1674, and he couldn’t have put it better.¹⁵

Perrault’s recognition of opera’s specific pleasures is moreover appropriate, for the writer was involved with another fin-de-siècle genre that was likewise supposed to enchant by recycling a bygone mythology that demanded no credence. This was the fairy tale, of which Perrault published one of the first collections, the 1697 Histoires ou contes du temps passé; though the vogue’s first wave subsided by about 1710, its effects reverberated throughout eighteenth-century literature. It is easy to assume that fairy tales were merely transcriptions, more or less embellished, of oral narratives, and that they were destined for children. In fact, only about half of the eighty tales published between 1696 and 1705 have folkloric antecedents, after which a full ninety percent of the production is invented.¹⁶ And they were not invented to be read to children. On the contrary, they were often embedded within longer novels in a manner that figures their own intended mode of consumption: in novels like Histoire d’Hypolite, comte de Duglas (d’Aulnoy, 1690—apparently the first published example), Inès de Cordoue (Bernard, 1696), or Les Petits Soupers de l’été de l’année 1699 (Durand, 1702), the tales are exchanged by aristocrats wanting to divert, impress, or seduce one another. The title of d’Aulnoy’s 1698 collection Les Illustres Fées, contes galants dédiés aux dames [the illustrious fairies, gallant tales dedicated to ladies]—speaks to the role these works played within the context of elite sociability. And they encoded as well a literary program. Because they owed nothing to the erudition of classical learning, they could serve as a kind of “manifesto of modernity,” especially for women writers.¹⁷ Unlike laborious poetic imitatio, which arranged traditionally sanctioned subject matter in new way, what Addison (citing Dryden) would call the “fairie way of writing” came straight from the unbounded imagination of its creators.¹⁸ One embedded narrator refers to her tale as containing “the marvels born of imaginations unrestrained by the trappings of truth”; another celebrates the possibility of ex nihilo creation: This tale, she says, “is absolutely my own.”¹⁹ So fairies were not simply in the tales: through their inventions, women authors actually became “modern fairies,” according to the title of Murat’s Histoires sublimes et allégoriques, dédiées aux fées modernes.

Yet, as Perrault’s title—“stories and tales of times past”—attests, fairy tales were as self-consciously old as they were new. A tale like “La Belle au bois dormant,” Jean-Paul Sermain points out, is strewn with references to chivalric romance; Sermain goes on to list some of the forgotten titles that demonstrate the extent to which fairies, though on the one hand associated with aristocratic oral exchange, were also emissaries from a dusty (and written) medieval past—as in Lhéritier de Villandon’s La Tour ténébreuse et les jours lumineuses, contes anglais … tirés d’une ancienne chronique composée par Richard, surnommé Coeur de lion (1705), Nodot’s Histoire de Geoffroy, surnommé la Grand’Dent, sixième fils de Mélusine (1700), or Gueulette’s Soirées bretonnes (1710), which presents itself as an eighth-century manuscript.²⁰ If opera provided a way of savoring unbelievable classical mythology, texts like these recycled indigenous beliefs, transforming harmless superstitions into poetry. Addison says as much: the fairy way of writing demands on the part of the poet “a particular cast of fancy, and an imagination naturally fruitful and superstitious”: “he ought to be very well versed in legends and fables, antiquated romances, and the traditions of nurses and old women, that he may fall in with our natural prejudices, and humour those notions which we have imbibed in our infancy.”²¹ Fairy tales are thus what Susan Stewart has called a “distressed” genre, in the sense that they mark a deliberate return to bygone beliefs, beliefs “pried from a context of function and placed within a context of self-referentiality.”²²

To fairy tales, old in their inspiration, new in the type of literature and authorship they represented, one could certainly add still other manifestations of the distressed sensibility, most obviously Galland’s Les Mille et Une Nuits (1704–1717), the translation of a series of Arabic manuscripts the Orientalist scholar had come across during his travels in the Islamic world.²³ It is difficult to overstate the importance of fairy and oriental tales to the literary landscape of the eighteenth century: along with endless reprints and continuations, there were numerous appropriations of these fanciful forms—by writers like Diderot, Crébillon, and Voltaire—for satiric “philosophical” purposes. By the same token, however, it is difficult now to understand this popularity. I am not making a judgment of quality, here, but merely remarking that whatever the merits or present readability of these works, the genre was useful to writers for reasons relating to the literary field as it was then contested and configured, reasons we can scare now comprehend. As the title of a book like Les Mille et Un Quart-d’heure (Gueulette, 1715) suggests, the valuation of fancy led to the deliberate cultivation of a frivolity that was the measure of the writer’s inventiveness, freedom, and ability to entertain. Small wonder, then, that a book as vital to literary history as Galland’s translation of the Arabian Nights has become, in the words of one critic, an “invisible masterpiece.”²⁴ To be sure, we can still appreciate fanciful literature, but our aesthetic landscape is no longer that of the early eighteenth century.

For the landscape I have been describing—and it was the landscape of England as much as that of France—was cleaved in two. In a text from 1762 that has become important to the history of the English gothic—another distressed and marvelous genre—Richard Hurd sums up well the opposite but complementary rationales attached to the literatures of illusion on the one hand and fancy on the other.

In those species that address themselves to the heart and would obtain their end, not through the imagination, but through the passions, there the liberty of transgressing nature, I mean the real powers and properties of human nature, is infinitely restrained; and poetical truth is, under these circumstances, almost as severe a thing as historical. The reason is, we must first believe, before we can be affected. But the case is different with the more sublime and creative poetry. This species, addressing itself solely or principally to the imagination—a young and credulous faculty, which loves to admire and to be deceived—has no need to observe those cautious rules of credibility so necessary to be followed by him who would touch the affections and interest the heart.²⁵

Hurd’s Letters on Chivalry and Romance, which make a case for the interest of a literature that operates by its own special rules, are sometimes said to lay the groundwork for Walpole’s Castle of Otranto, published just two years later.²⁶ This is probably a bit misleading, as we will see, but for the moment it is enough to remark that conceptually speaking, Hurd’s argument is standard-issue: he merely reprises a division that had been theorized on and off, and indeed practiced, at least since Corneille made his remarks about Andromède being for the eyes alone. In 1712, Addison was already claiming the territory of the imagination for England, whose people are “naturally fanciful, and very often disposed by that gloominess and melancholy of temper … to many wild notions and visions”²⁷; and since then it has been easy to view the country of Shakespeare as constitutionally opposed to the country of Voltaire, and the natural home of the gothic sensibility. In fact, the Great Divide between neoclassical illusion and distressed literatures of the imagination ran right through both countries. English proponents of the gothic merely appropriated for their own purposes the aesthetic rationale worked out by the people promoting the possibly more “French” genres of opera, fairy tales, and oriental tales.²⁸

If we want to understand why Freud could think of no uncanny fairy tales, we need look no further than the Great Divide. On one side there was the literature of real human feelings and moral instruction that founded itself on history and illusion and literal truth—tragedy, epic, the historical romance, and the pseudofactual novel. On the other side was a literature that appealed to the senses, to the imagination—opera, fairy tales, and oriental tales, which is to say genres that like the later gothic trafficked in the beliefs of other times or other places and that demanded from the present audience only an imaginative investment, ironic or distanced. In practice these logics were not always mutually exclusive: by the mid-eighteenth century, for instance, there were calls for a type of nonmythological, purely human opera that could serve as a vehicle for sentimental sympathy.²⁹ But allowing for such inevitable complications, the literature of truth and the literature of fancy did not meet; the supernatural could not erupt within everyday reality, producing Freud’s slowly creeping sense of dread, for these were different practices, each with its own conventions and traditions. Until Cazotte, in Le Diable amoureux, tried to mix them together.

Incredible Love

The oeuvre of Cazotte, a navy administrator who was not a writer by profession, is mostly forgotten today. This is not surprising: his energies were almost entirely devoted to the literature of fancy whose contours were drawn around the turn of the eighteenth century. He participated in polemics on opera, notably opposing Rousseau’s attempts to abolish its marvelous subject matter; he composed Ollivier (1763), a chivalric prose poem in the vein of Ariosto’s Orlando furioso, and the distressed “medieval” ballad “La Veillée de la bonne femme” (1752, pub. 1788); working from an Arabic manuscript we now know was forged, he penned a continuation to Galland’s Les Mille et Une Nuits (1788); and he cut his teeth on a number of fairy tales, or more accurately self-conscious fairy tales that incorporate characters who know they are trapped in the genre, people who pretend to be fairies, and so on. In some respects, he was distinctly of his time: many other writers rode Galland’s coattails by adding to his endlessly popular collection; Crébillon and Voltaire too used oriental tales for satirical or philosophical purposes; Hamilton wrote self-conscious fairy tales in the spirit of salon banter, or badinage. But in one respect Cazotte surprises, for a few of his works try hard to give readers a marvelous they can believe in even though it is unbelievable. First among these is Le Diable amoureux, in which fancy, in the form of a horrible camel, sticks its head into the sensible and verisimilar world of the pseudofactual novel.

Le Diable amoureux is retrospectively narrated by the Spaniard Alvare, who in his twenties served in the guards of the king of Naples (presumably Charles, also duke of Parma, who would become Charles III of Spain in 1759). When, after a smoky evening of idle chatter about the esoteric or “cabalistic” sciences, Alvare is approached and interrogated by a mysterious participant who claims to be an initiate, he quickly accepts his interlocutor’s proposal to introduce him to the occult. Led down to a secret chamber beneath the recently discovered ruins of Herculaneum, in the town of Portici, Alvare is shown a pentacle scratched onto the floor and given the formula needed to summon Beelzebub. His hair on end, a chill in his veins, he steps into the pentacle and pronounces the magic words. A window high in the underground vault flies open, and in a blaze of light a hideous camel’s head appears. “Che vuoi?” it asks.³⁰ Immediately, Alvare orders the apparition to take the form of a spaniel, and the camel vomits the specified dog at the narrator’s feet. Alvare presses on, demanding that a feast be served up for his cabalist friends, and that the dog appear there as a page dressed in his new master’s livery.

All this occurs—to the amazement of Alvare’s group, but also, we may presume, to that of the contemporary reader. Cazotte begins in a manner that recalls a popular book like Villars’s Le Comte de Gabalis (1670), in which the first-person narrator describes with ironic detachment the superstitious doctrine of a renowned German “sage.”³¹ Yet before we know it, the initiate proves to be just that, the doorkeeper to another reality. And Alvare’s own character blurs our expectations. On the one hand, he seems the very type of the rationalist skeptic, responding to the initiate’s openings by asserting the Cartesian tabula rasa of his mind—“Je ne connais rien des esprits, à commencer par le mien, sinon que je suis sûr de son existence” (8). Alvare’s play on the word “esprit” makes his assertion difficult to translate: he knows nothing, he says, about (supernatural) spirits, no more than he knows about his own spirit, or mind; except that, like Descartes, he is convinced of the latter’s existence. And indeed, there in the ruins below Herculaneum, he displays the intrepid bravery associated with the skeptical esprits forts that we have seen in Subligny, those who have no truck with superstition.³² At the same time, however, Alvare is characterized by the insatiable “curiosity” of the gullible soul: the desire for something beyond “ordinary knowledge” possesses him (8).

The reader familiar with earlier works with supernatural content must wonder: is this a satire of Alvare? or is Alvare our skeptical surrogate? Neither. Alvare’s credo is something more like: I don’t believe but I want to believe. And so the tone here is not debunking, as if we were but temporarily unsure of the all-too-natural causes we know to be behind the observed phenomena; nor is the initiate’s discourse displayed for us to savor ironically as the ravings of a madman. And it goes without saying that we are far from a fairy world where anything is possible. Rather, a supernatural that no one can believe in is nevertheless treated as a reality. When Alvare demands the presence of an opera star to entertain his feasting friends, she materializes and sings; his guests are “astounded by the truth of the scene, to the point of rubbing their eyes” (17).³³

The conceit of Le Diable amoureux is indeed bracing, generically speaking. The work takes precisely the opposite route from Cazotte’s earliest ironic fairy tales, undertaken much in the spirit of Hamilton or Caylus. Les Mille et Une Fadaises (1742), for example, is told to an insomniac Marquise in the hopes that the clichés of the genre will finally enable her to get some rest.³⁴ By contrast, in Le Diable amoureux, Cazotte injects the supernatural into a pseudofactual frame and finds that it is able to produce effects at the other end of the emotional spectrum from boredom-induced sleep. Fright, first: at the apparition of the camel’s head, Alvare describes a complete capsizing of his being, as “a multitude of feelings, ideas, and reflections touch my heart, pass into my mind, and leave their mark on me all at once” (13). And then, soon, passions of a more suave sort, as Alvare is moved “to the bottom of [his] heart” (18) by the singing of the diva, who, he notices, is a further transformation of the graceful pageboy Biondetto. Biondetto, who looks good dressed as a woman, Alvare notes. Biondetto, who reveals that he is in fact a she, Bionsdetta, a sylph who has betrayed the devil out of admiration for Alvare’s bravery, and who in the process discovers love: “I realized,” she tells Alvare, “that I had a heart” (49). Alvare, of course, feels much the same way.

Le Diable amoureux is, precisely, a kind of fairy tale that has lost its distance from our world; beings that normally frolic in their own domain ask us to believe that they have become real. Le Diable amoureux’s plot revolves entirely around the inflamed protagonist’s uncertainty in this respect. What is the exact relationship between Biondetta, the spaniel, and the camel’s head? Can it really be possible that a creature of air has become flesh and blood? Naturally, Alvare is wary; at the same time, Biondetta seems so very alive. After she is seriously wounded by a knife-wielding, jealous rival, Alvare at last contemplates her “beautiful bleeding body, cut by two enormous gashes that seemed as if they must both be attacking the wellspring of life” (45); shortly thereafter, during the bandaging of her wounds, her sex is “confirmed” (47). “Man was a mixture of a bit of mud and water,” Alvare muses. “Why wouldn’t a woman be made of dew, terrestrial vapors, and rays of light, the condensed debris of a rainbow? Where is the possible …? Where is the impossible …?” (93–94, ellipses in original). Periodic memories and dreams of his pious mother bolster Alvare’s continued delays and resistance; he tells Biondetta that they will be married once his mother blesses their union. During the trip back to Spain he succumbs to temptation. Carnal satisfaction is quickly followed by Biondetta’s revelation of her true identity. Alvare cowers under the bed until he is pulled from his hiding place some time later and finds his tormentor gone. Soon reunited with his mother, he is told by a theologian that his true remorse and good intentions have preserved him.³⁵

Reading Cazotte nowadays can be a familiar operation, and not only because of his archetypal exaltation of maternal purity at the expense of carnal woman. For Le Diable amoureux is generically familiar: we cannot help but read it through what we know follows, which is the romantic fantastic. Enrolling Cazotte under the banner of the fantastic has a long literary-historical sanction: Pierre-Georges Castex, for example, called Cazotte “the true creator of the French fantastic tale,” and practitioners such as Tieck and Hoffmann, Nodier and Mérimée, had already invoked Le Diable amoureux as an inspiration.³⁶

Unsurprisingly, then, Tzvetan Todorov’s now-classic The Fantastic pointed to Cazotte’s novel as the genre’s true beginning. But Todorov’s structuralism made for an account that was not a typical generic history: the fantastic, he claimed, occupied a specific spot on a generic grid. Todorov divided supernatural narratives into three sorts—the uncanny [l’étrange], in which apparently irrational events are explained rationally; the marvelous, in which the laws of rationality are suspended; and finally the fantastic, in which characters and readers hesitate over “whether or not what they perceive derives from ‘reality’ as it exists in the common opinion.”³⁷ A complication in this generic grid was necessitated by the temporal dimension of the reading process, since the characters and readers can well hesitate for a while, only to resolve their doubts by the book’s end. Hence, Todorov ended up with a total of five categories—the uncanny, the fantastic-uncanny (in which the supernatural was shown to have a rational explanation), the fantastic-marvelous (hesitation was followed by the confirmation of supernatural agency), the marvelous, and finally the “pure” fantastic, in which hesitation is maintained until the very end.³⁸ Todorov’s comments on Le Diable amoureux are occasional and ambiguous, but the work does occupy a strategic spot in The Fantastic: Alvare’s hesitations provide Todorov with his opening example of the genre’s sine qua non condition.³⁹ Historically, moreover, the work’s importance is underlined: the fantastic “appeared in a systematic way around the end of the eighteenth century with Cazotte.”⁴⁰

But there are a number of ways in which Cazotte does not fit very well into Todorov’s structuralist grid. The first thing to notice is that the critic’s characterization of Alvare’s hesitation is inconsistent, shifting. In his initial presentation, Todorov implies that the main hesitation is on the order of a hoax: is Biondetta “simply a woman,” in spite of the fact that “the way this being first appeared clearly suggests that she is a representative of the other world”?⁴¹ In virtually the same breath, the critic then shifts the hesitation a bit: Alvare is said to wonder whether sylphs exist, or if he is not dreaming the whole thing. And at another point, later in the study, the critic locates Alvare’s search for a rational explanation in a short episode at the book’s beginning when the initiate, Sobanero, calls out to an invisible servant to take away his pipe and bring it back lighted; when this happens, “we seek, with [the narrator], a rational explanation for those bizarre phenomena.”⁴² As a result, it is difficult to know just where on his grid Todorov places Cazotte’s novel. As an example of the “pure” fantastic, one would expect, given that he opens his account with Alvare: “The ambiguity is sustained to the very end of the adventure: reality or dream? truth or illusion?”⁴³ Several pages later, however, he leaves Cazotte behind on the grounds that the hesitation of the protagonist is only “momentary.”⁴⁴

There is good reason for these difficulties. Cazotte does thematize hesitation, but it is not the hesitation that will characterize the later fantastic (or, as we will see, the English gothic). For all Todorov’s textual examples rewrite Cazotte so as to bring him in line with the critic’s definition. Alvare spends no time thinking up rational explanations for the pipe lighting, but immediately becomes Sobanero’s acolyte, pestering him for access to the other world; the question is never whether Biondetta is just a woman after all, but whether the sylph has become human. And in the passage to which Todorov alludes, Alvare does say, “All this seemed to be a dream,” but this is the beginning of a reasoning process designed to convince himself of the dreamlike nature of all reality: “but is human life ever anything else?” he asks (50). In addition, many critics have pointed out that Todorov passes over the most ambiguous part of the tale—another moment at the very end when Alvare cries out, “Is everything false in the dreadful dream I’ve been dreaming?” (84). This possibility creates numerous interpretive ambiguities; one critic has gone so far as to call it “one of the most underdetermined texts in narrative literature,” arguing that the dream hypothesis destroys any hope of producing a “literal reading” of Le Diable amoureux.⁴⁵

These ambiguities are real, and if I do not go into them here, it is simply because our difficulty in determining just what has happened to Alvare has nothing to do with fantastic hesitation regarding the possible reality of apparently supernatural phenomena. It is hard to tell whether Alvare has been dreaming, and if he has, where that dream may have started; some have called into doubt whether, dreaming or not, Alvare has actually slept with Biondetta, or even whether we can consider him “saved” at the novel’s end.⁴⁶ This may be intentional on Cazotte’s part, or merely the result of sloppiness. But given that all critics agree that there is no textual evidence for the possibility that Alvare is dreaming the entire narrative, none of this changes the fact that the devil actually appears to him, and to us, in the ruins of Herculaneum.⁴⁷

For strangely—or, strangely for the reader of the later fantastic—Alvare immediately grants the apparitions at Portici a certain type of reality: yes, there is an order of beings with which man can be in contact, and which have power over the natural world, even though they are not of that world. What Alvare himself must be brought to consider is the possibility that the two orders of reality can become confused. The hideous camel’s head, the spaniel, the pageboy, the opera diva—all these exist for Alvare (and for the reader) as “real” apparitions. The illusion does not concern the existence of the supernatural, but lies in Biondetta’s purportedly human nature. Alvare asks: can it be true that this beautiful, embodied creature is not—is no longer—the devil? And Cazotte answers: she was the devil all along. Whatever hesitation it presents to us, it is not that of the fantastic as Todorov defines it, which is said to start “from a perfectly natural situation [only] to reach its climax in the supernatural.”⁴⁸ On the contrary, Le Diable amoureux is structurally almost the inversion of the fantastic, since it starts with the supernatural and then asks whether the supernatural has become part of the natural world.⁴⁹

This is not to exalt the novel as genre-bending. It is hardly surprising that Cazotte does not obey conventions that did not exist when he was writing, or that Alvare’s nonfantastic hesitation makes more sense in the context of what comes before it than what comes after. And what comes before is the literature of fancy, whose imaginative pleasures were of a qualitatively different sort from those experienced by audiences who were assumed to believe—literally yet not completely—in the reality of spectacle or narrative. Contemporary reactions are not especially plentiful, nor are they unambiguous, but they all situate Cazotte against this literature of the marvelous. The most extensive of these emphasize Cazotte’s break with this tradition. Hence, the anonymous reviewer for Le Mercure de France opens with a meditation on generic resemblance: today’s tragedies, novels, tales, comedies usually have the same “physiognomy” as their predecessors, but every now and then certain writers are able to “cast in a new mold” the works created by their “imagination.” Such is Le Diable amoureux, whose surface of “frivolity” and “banter” hides a moral lesson about resisting our weaknesses.⁵⁰ Élie-Catherine Fréron, in his Année littéraire, is even more categorical. France, where “shameless plagiarism” is the normal recipe for success, finally has something “original and new” to rival Shakespeare, the operas of Quinault, and the Italian “enchanters,” presumably Ariosto and Tasso. For aside from Quinault, Fréron writes, all the French have had until now are fairies—and “their magic wand is short.” Casual observers may think that the marvelous genres are easy, but this is far from the case: supernatural subject matter makes it hard to produce involvement, or intérêt, on the part of the reader. Le Diable amoureux is a “model” in that respect. It creates “limits” for the marvelous; and the way events are presented “gives an appearance of verisimilitude [un air de vraisemblance] to what is furthest from nature.”⁵¹ These two reactions were not the only ones—a couple more short notices simply saw here a badinage indistinguishable from the familiar French marvelous.⁵² But they do credit Cazotte with an original use of the imagination; Fréron states specifically that originality lay in the way Cazotte treated the supernatural as if it were natural.

Le Diable amoureux’s sharp departure from normal practices of the marvelous was not dulled by time: when the French Romantics looked back at Cazotte a half-century later, he was still the only French example of a non-frivolous use of the supernatural. In Jean-Jacques Ampère’s appreciation of Hoffmann, published the year before the posthumous appearance of the German writer’s Contes fantastiques (1829) in France, Cazotte is adduced as the exception that proves the rule: unlike the Germans, the French have been uninterested in using the marvelous as anything but “a ridiculous phantasmagoria or a purely satirical frame.” Of the abundant French production in the vein of the new marvelous, only Le Diable amoureux, “a masterwork of imagination and grace, … gives some idea of the pleasure provoked by the proper use of [the marvelous], and should reconcile to it all those who have been put off by the nonsense of certain novels and melodramas.”⁵³ Prosper Duvergier de Hauranne’s review of Les Contes fantastiques reserves the same spot for Le Diable amoureux as the only French text capable of giving any idea of Hoffmann’s genius. Eighteenth-century France, he writes, was a “unpoetical” time “of skepticism and cold reason” that could only admit the marvelous as patent allegory; there were fairy tales, but their invention was too obvious; the tales in Les Mille et Une Nuits provoke “ravishment,” yet remain incapable of “mov[ing] us as profoundly” because they are so “foreign to our own beliefs.”⁵⁴ In Nerval’s essay on Cazotte, Le Diable amoureux is again placed against the backdrop of the marvelous—fairy tales, operas, allegorical fables—in order to bring out the originality of the novel’s serious marvelous. Even when writers of the caliber of Montesquieu, Voltaire, and Diderot dabbled in fables, they were still hampered by the limits assigned to the marvelous: “all this was invention, wit, and nothing more, albeit of the finest and most charming kind.” Only Cazotte broke from this: “But the poet who believes in his fable, the narrator who believes his legend, the inventor who takes seriously the budding dream of his thought, this is what one hardly expected to find in the middle of the eighteenth century.”⁵⁵

Such critical consensus may help us understand better Cazotte’s somewhat peculiar use of hesitation in his protagonist—peculiar because it does not ask: “might the supernatural exist?” but rather: “has the supernatural (which exists) become human?” In Hoffmann and the later fantastic, but also in much English gothic, readers and protagonists share assumptions about the disenchanted material world that are then troubled by the possibility of immaterial agency. In Cazotte, such assumptions are immediately sloughed off and supernatural agency acknowledged; yet the dividing line between the real and the supernatural, drawn on the very body of Biondetta, remains the focal point of the narrative. And indeed, unlike in fairy tales, where in Duvergier’s words “fiction makes itself felt from the first to the last lines,”⁵⁶ the question of reality—of the difference between the natural and the supernatural—does not disappear, but is posed in a way specific to the literary world of Cazotte. Diderot, echoing the common relegation of the supernatural marvelous to an inferior form of aesthetic response, wrote that “the enchanted world can amuse children, [but] only the real world is pleasing to reason.”⁵⁷ Cazotte, meanwhile, features an adult who is very much pleased by supernatural enchantments—pleased, not amused in the normal ironic fashion. What might it mean, Cazotte’s plot asks, for a “fantastic being” to leave its allotted space and “to borrow the form [les traits] of truth and of nature” with such perfection (43)?

The title of Le Diable amoureux is in itself a program: gone is the conceit of Lesage’s Le Diable boiteux (1707), where the devil was a (remarkably successful) narrative device, what we’ve seen Ampère calling a “purely satirical frame”; in its place Cazotte has installed a real devil and, moreover, a love story, which is to say, the very stuff of the serious novel. Hence, Le Diable amoureux is about the enchanted world’s claim on our reason and heart, its ability to seduce us, to get us to rub our eyes. The novel starts to tell the tale of its own powers of persuasion. “I was moved to the bottom of my heart,” writes Alvare, listening to the aria of the opera diva he had conjured, “and I almost forgot that I myself was the creator of the charm that held me in its spell” (18). It is no doubt significant that in a departure from the Faust myth, Biondetta doesn’t as much promise power, or even pleasure, as represent a longing—Cazotte’s longing—for a literature that incarnates the unreal, that makes the unbelievable possible. In that, the self-reflexivity of the book’s plot is patent. The writer can summon creatures whose incredible nature does not keep them from exercising real seduction over us. Given the bifurcated generic field that had given verisimilar drama and narrative on the one hand and the new marvelous on the other, Cazotte’s experiment has radical implications: supernatural subject matter is not foreign to a poetics of illusion; a different and more potent use of marvels is possible.

Le Diable amoureux, then, is new.⁵⁸ But to what does this newness really amount? The typical way of pursuing the problem is simply to agree to see in this one novel a sign that some deep cultural and conceptual substrate is shifting: fiction is coming into existence, people are realizing that their novels can seem true even when they are literally unbelievable.⁵⁹ It’s quick and convenient, because the substrate is invisible and can contain whatever we need it to. But if we actually look at the evidence—subsequent novels that make use of the supernatural—can we really say that Cazotte changes something? After all, as we’ve already seen, Alvare’s hesitation is not really very fantastic at all. Cazotte is new, but surely not all inventions change the future by catching on and altering behavior. Maybe his was not exactly forgotten, but not exactly useful, either. Maybe we’ve simply misremembered him as a precursor so that we do not need to confront the vagaries of literary history.

Weird Devices

Todorov asserts that the fantastic “appeared in a systematic way around the end of the eighteenth century with Cazotte.”⁶⁰ One might wonder what the critic considers a system. A half-century later, French Romantics remember Cazotte as the only one to bridge the Great Divide, and their memories were better than Todorov’s: there are precious few French texts from the period that try to make a serious use of the supernatural, and all are as idiosyncratic as Le Diable amoureux. There is Potocki’s Manuscrit trouvé à Saragosse, which Todorov trumpets for its elaborate use of supernatural hesitation. But it is hard to see how it fits into a “system,” either: its partial, sporadic, and largely ignored publication starts thirty-odd years after Cazotte. Two other texts, not mentioned by Todorov, don’t get us much farther. Beckford’s Vathek, published in its original French the year following its 1786 appearance in English translation, suits the chronology better, and it is much more serious in tone than most riffs on Arabian Nights-type oriental fancy. Nevertheless, it does not even try to orchestrate a meeting between real life and the irrational.

The closest example of the type of mixing of the fanciful and the real that Cazotte practices is a book that also references Villars’s Comte de Gabalis: Bibiena’s La Poupée presents us with a doll come to life as a sylph. But here too the date—an early 1748—dilutes the notion of a big literary “change” or “realization”; and in spite of some wonderfully uncanny pages where the doll first becomes animated, Bibiena’s sylph is essentially a satirical device enabling the display of the period’s ridiculous characters. (As such, it remains firmly in the tradition of the devil-voyeur of Lesage’s Le Diable boiteux and the metempsychosis of Crébillon’s Le Sopha [1742].) Cazotte’s contemporary admirers explictly noted verisimilar treatment of the marvelous as the noteworthy element of Le Diable amoureux. Its difference, then, was clear, for it showed that a writer could bring the supernatural into the real, sentimental world of the novel—that readers could fall for things they did not believe. Yet it never became a template for later French writers. Gargoyle-like, Cazotte remained perched on a lonely outcropping overlooking the gay and reasonable eighteenth century.

But perhaps we need to take into account gargoyles, precisely: if Cazotte’s tale has no French emulators, the place to look may be the English gothic. (Todorov, prizing only “pure” fantastic hesitations, marginalizes a movement known for its “supernatural explained.”) It has been suggested that a few features of Biondetta flit across the beautiful face of the temptress Mathilda in Lewis’s The Monk (1796).⁶¹ Even if this is right (and it is probably not), borrowings of this sort leave us far from proof that the 1772 work was in any way responsible for the English gothic novel that exploded around 1790. Indeed, it is hard to see what gothic writers might have got from Cazotte that they didn’t already have at their disposal—notably, the precedence provided by Smollett’s Ferdinand Count Fathom (1753), Walpole’s Otranto, and some other titles I’ll mention in a moment. The distressed gothic atmosphere, be it supernatural (Walpole) or not (Smollett), owes nothing to Alvare’s satanic dabbling. No gifts of divination are required to indulge in the counterfactual prediction that in the absence of Cazotte the main lines of the gothic vogue would not change a bit.⁶²

Perhaps, then, things operate in the other direction: was Cazotte, in 1772, working in his own way the gothic vein opened by Walpole a few years before? A French translation of The Castle of Otranto appeared in 1767, and was reissued in 1774; conceivably Cazotte, whose extant correspondence gives little sense of his literary interests, might have read either. But beyond the fact that Walpole seems to have had little market penetration in France, Le Diable amoureux, hardly Walpolian in the first place, merely extends the interest for the marvelous that Cazotte had already shown in earlier texts like the ballad “La Veillée de la bonne femme” and Ollivier.⁶³ Cazotte, in other words, had no more need of the early English gothic than the later English gothic needed him.

If direct influence appears dubious, and if we persist in our refusal of those ghostly “conceptual” mutations that can always be trusted to resolve difficulties like these, there remains a sense in which Cazotte does indeed deserve to be considered in the context of the English gothic supernatural. For what Cazotte shared with early practitioners of the gothic novel was quite simply an articulated set of aesthetic commonplaces regarding the Great Divide. In the second edition of The Castle of Otranto, in which the author revealed the hoax of the first (there was no discovered manuscript after all), Walpole was explicit in laying claim to a “new route” for poetry that would manage to blend the discourse of “common life” and “probability” with “the great resources of fancy.”⁶⁴ Reeve found Walpole’s blending ineffective—too much fancy (gigantic swords and helmets, walking pictures), not enough probability—which is why in her Old English Baron (1777) she tried again. The goal was to blend “ancient romance” and the “modern novel” by providing “a sufficient degree of the marvelous to excite the attention; enough of the manners of real life, to give an air of probability to the work; and enough of the pathetic, to engage the heart in its behalf.”⁶⁵ Cazotte himself left us no such explicit statement, but we’ve seen that Fréron’s review of Le Diable amoureux made very similar points, giving the tale the same coordinates mentioned time and again by the English commentators who in the face of the neoclassicism of Dryden and Pope strove to rehabilitate the “gothic” sensibility—Shakespeare and Spenser, but also Ariosto and Tasso.⁶⁶ And of course the point was made by the tale itself, in which it becomes possible to fall in love with an illusion because the illusion is made so real. This is why the English and French works possess a structural resemblance even as on other levels—setting, plot, character, framing devices—they look nothing like one another. That resemblance is due neither to direct influence nor to a common conceptual substrate on which they all stood. It is because Walpole, Cazotte, and Reeve confronted the same inherited problem that their solutions do much the same thing. The problem was not an eternal one; it was inscribed in the aesthetic history they shared. What might a modern literature of the marvelous look like?

Walpole and Reeve shared something else with Cazotte: they too innovated in isolation. Such an assertion will probably be viewed with doubt by scholars of the English gothic, who can point to a rich tradition of mid-eighteenth-century medievalism that was largely absent in France. Of course, it is true that Walpole and Reeve do not look nearly as lonely as Cazotte, stranded as he is in the country of Voltaire, and any genealogy of the English gothic has an abundance of different texts to rope in. There are, first, the antiquarian—Hurd’s Letters of Chivalry and Romance (1762), Wharton’s Observations on the Faerie Queene (1754, revised 1762), Percy’s Reliques of Ancient English Poetry (1765). Then there is contemporary production in the medieval vein, most obviously the forgeries of Macpherson (1760) and Chatterton (pub. 1777). Finally, there are the novels that, while not overtly supernatural, contain gothic accents or settings—Smollett’s Ferdinand Count Fathom, perhaps Leland’s Longsword, Earl of Salisbury (1762), certainly Lee’s The Recess (1783–1785). All of which add up, admittedly, to a serious vogue, while the French side of the scale is relatively weightless—Caylus’s “translation” of a Spanish chivalric romance, Tiran le blanc (1740), Tressan’s many adaptations of chivalric legend, a passing enthusiasm for the Ossian poems.⁶⁷

Rather than conclude, however, that the reasonable French were simply resistant to the charms of gothic gloom, we might recognize the extent to which English writers did not find it any easier than their French counterparts to marry the novel and the supernatural. Before the explosion of gothic novels in the late 1780s and 1790s—at which point one can reasonably speak of a new literary genre for reasons I will explain—England boasts, after all, only two practitioners to France’s one.⁶⁸ Perhaps on account of the centrality to the national canon of Shakespeare and Spenser, enchanted writers both, English contemporaries do return repeatedly to the Middle Ages in a way the French generally do not. Nevertheless, their novels are hardly more receptive to the supernatural, as the case of Walpole and his avowed imitator Reeve attests.

If we simply draw a line between Walpole’s interest in a middle way or “new route” and the explosion of gothic novels in the 1790s, we paper over this difficulty: why is there only one supernatural gothic novel—Reeve’s—in the rough quarter century between Walpole and the earliest uses of the “supernatural explained” in the work of Fuller (Alan Fitz-Osbourne, 1787) and Smith (Emmeline, the Orphan of the Castle, 1788)? There was, it turns out, a big difference between appreciating the supernatural in works of the past and making it work in novels. It was one thing to argue, as Hurd did, that medieval romance had historical reasons for existing, possessed its own beauties, and should not be judged by neoclassical standards. But even Hurd did not advocate bringing back this “world of fine fabling”: “I would advise no modern poet to revive these faery tales in an epic poem.”⁶⁹ Hence the overwhelmingly nostalgic tone of the texts lauding the poetic receptivity of the medieval mind: Shakespeare and Spenser represented that last happy moment before Don Quixote appeared and “chivalry vanished, as snow melts before the sun.”⁷⁰

This is why immediate critical reaction to The Castle of Otranto shifted significantly once Walpole’s second edition gave up the pseudofactual pretense of the first. Two sets of reviews, from the Critical Review and the Monthly Review, register the problem quite clearly. Reacting to Otranto’s first edition—the one that presented the text as the translation of a thirteenth-century manuscript—the anonymous account in the Critical Review raises the issue of the “translator”’s seriousness, and hints at fraud (“we doubt strongly whether pictures were fixed in panels before the year 1243”). Still, the reviewer, less the dupe of Walpole’s pretense than its relay, plays along, praising the characters and the narrative. But when six months later the same reviewer takes into account Walpole’s new preface, the tone changes. Walpole’s aim of blending “the two kinds of romance” is a bad idea: “we cannot but think that if Shakespeare had possessed the critical knowledge of modern times, he would have kept these two kinds of writing distinct.” If we pretend the text is medieval, all is well. If the text is given as modern, it must respect the Great Divide.⁷¹

John Langhorne’s two contributions to the Monthly Review do a similar about-face, one that is nearly as perplexing to the modern reader as Richardson’s protests to Warburton over admitting his authorship of Clarissa.⁷² Here too “the absurdities of Gothic fiction,” its “machinery of ghosts and goblins,” are tolerated because Otranto is presented as the work of a translator, bound to an original which he himself criticizes as unbelievable. In a footnote to the word “translator,” Langhorne expands on his attitude toward Walpole’s pretense: “This is said on the supposition that the work really is a translation, as pretended.” What does the reviewer mean by “supposition”? Langhorne, like the contributor to the Critical Review, changes his tune after Walpole’s second-edition revelations, but I do not think that he does so because he “is indignant about Walpole’s subterfuge,” as one modern scholar has claimed.⁷³ Rather, he is indignant about Walpole’s choosing to do away with a subterfuge that was as necessary as it was transparent. Langhorne’s reasoning, couched again in a language of supposition and not of empirical truth, must be quoted at length.

When this book was published as a translation from an old Italian manuscript, we had the pleasure of distinguishing in it the marks of genius …; we were dubious, however, concerning the antiquity of the work upon several considerations, but being willing to find some excuse for the absurd and monstrous fictions it contained, we wished to acquiesce in the declaration from the title-page, that it was really a translation from an ancient writer. While we considered it as such, we could readily excuse its preposterous phenomena, and consider them as a sacrifice to a gross and unenlightened age. —But when, as in this edition, the Castle of Otranto is declared to be a modern performance, that indulgence we afforded to the foibles of a supposed antiquity we can by no means extend the singularity of a false taste in a cultivated period of learning.

Righteous indignation of the duped? Retroactive revision of his original attitude, now said to be “dubious”? Not at all. Rather, we have here the strange contortions made so often when writers and readers of the period pressed hard on pseudofactual conventions that were meant not to be believed literally so much as treated as if we believed them literally. Walpole’s original presentation was the excuse, in Langhorne’s terms, for not stamping the work with the Horatian condemnation—incredulus odi—he goes on to invoke. Langhorne, in his first review, obligingly followed through on the pseudofactual posture Walpole had initiated; but if Walpole was no longer prepared to play the game, then the critic would not play it either. The author wants to call this a modern work? Fine, he must take the consequences, and submit it to modern aesthetic categories—those, again, of the Great Divide.

Of course, not everyone agreed with bean-counting reviewers: the philosophe Grimm reports his hair standing on end; Warburton paid the book the high praise of “effect[ing] the full purpose of the ancient tragedy, that is, to purge the passions by pity and terror.”⁷⁴ Some, then, thought the experiment a success, and ultimately Walpole would indeed “win,” in that a literature in which one could not believe and that affected us nonetheless was possible. But not quite yet, and not with the devices Walpole was using. The repeated association of the essence of poetry with the medieval period and popular superstitions did not translate directly into a new novelistic practice. Walpole remained alone in trying to inject fancy directly into the genre’s truthful illusions. At home and abroad, some liked it, some didn’t, but Walpole had to wait more than a decade for an imitator, Reeve, and he wasn’t satisfied: her modifications involved carefully reducing the amount of supernatural agency in the plot to some mysterious thumps and groans, and a door swinging open of its own accord.⁷⁵ Only at the end of the 1780s were there a series of novels in which the supernatural appears to intervene—by Fuller and Smith, as I’ve mentioned, and then of course Radcliffe. At that point, writers and readers found themselves able to take the supernatural as seriously as writers of more unenlightened ages did. As one French reviewer remarked of Lewis’s The Monk,

many passages cannot be read without an involuntary shudder…. The author appears to have the goal of frightening, like Shakespeare, by supernatural means, by dreams, spells, communication between the living and the dead, between man and fantastic beings; it is all quite impossible, quite absurd, and [yet] it all produces an effect, so great is the effect of the imagination’s power on us, no matter what our age. Try as we may to say that it’s only a story, the story makes us quiver and captures our interest. After, one can shrug it off with condescension, but only once the book has been closed.⁷⁶

In a way, little had changed from thirty years before, when Grimm said that being a philosophe was no guarantee against the fear provoked by Otranto, “so much are the sources of the marvelous the same for all men.”⁷⁷ What was new was simply that writers had invented a recognizable and reproducible form that could bridge the divide between fancy and belief.

But why, precisely, around 1790? Since the interest in the supernatural had long been in place, what kept writers from bridging the Great Divide and exploiting that interest in the immediate wake, say, of Addison? Why was Reeve Walpole’s only imitator for a quarter century? One set of explanations—the most common—read through the gothic novel. They take this maligned genre and discover in it the workings of serious spiritual, cultural, or subjective forces, often using what we already know about the historical moment to motivate its appearance in the 1790s. Such are the Weberian hypotheses I alluded to at the outset: the gothic marked the last refuge of the numinous, finally expelled from the disenchanted Enlightenment and exiled to novels.

Also reading through the gothic are cultural explanations, for instance, that of Ronald Paulson, who has proposed that the gothic tapped into deep anxieties prompted by the French Revolution.⁷⁸ Marshall Brown sees the genre as part of the history of modern subjectivity, occurring at the moment when Cartesian self-examination is replaced by Kantian self-consciousness.⁷⁹ To these possibilities one could add the more materialist explanation of Clery, who appeals to the politics of circulating libraries as the reason Walpole did not have more immediate imitators: “supernatural fictions along the lines of Otranto should have been the ideal commodity for the libraries…. Instead, in the period up to the 1790s, pure commercial interest was counterbalanced and constrained by the representation of the library as a civic institution with moral responsibilities.”⁸⁰ The idea is that the commercial power the gothic novel will demonstrate in the 1790s must have been blunted by something, or else it would have shown itself earlier.

Variously illuminating as these explanations may be, they are perhaps needlessly clever. Why weren’t ghosts, demons, and the like “from the first unproblematically available as a resource for writers”?⁸¹ Because people had to invent forms. Yes, they needed to be interested enough in the supernatural to bother to try to invent forms, but interest alone only takes writers so far. Consider, for example, the detective novel—another commercial powerhouse, and one that didn’t always exist. Certainly, the detective novel seems unimaginable without at the very least an interest in crime-solving that is scarce before the advent of urban policing in the nineteenth century. It needs something else, however: the decodable clue. In Chapter 1, I alluded to Franco Moretti’s work on the spreading of the device of the clue; what he finds is that clues did not spread in a predictable, linear manner. Conan Doyle may have invented them, but his rivals did not immediately borrow the device; instead, a generation was necessary before clues really caught hold and came to define the genre.⁸² However, Lafayette’s seemingly prescient use of a completely invented heroine in La Princesse de Clèves did not, I’ve shown, behave in the same way: some “innovations” go nowhere.

At first glance, supernatural narrative appears to show a pattern closer to the one Moretti observes in the detective novel—two isolated early practitioners, Walpole and Reeve, followed by a later generation for whom the supernatural had become a successful formula. But if we look more closely, the “generational” hypothesis, derived from an analogy with natural selection, fails. For Walpole and Reeve do not possess the device that will form the backbone of the gothic novel as it starts to take off in the late 1780s. Walpole, true to his plan, tries to feature characters who react plausibly to the supernatural, where plausibility concerns the construction of a subjective space; we are indeed brought inside the characters’ heads.⁸³ But one thing that we don’t find in their heads is hesitation over the reality of the supernatural. This, I propose, is the device that gives the gothic novel (and many other subgenres since) its form and attraction. Hesitation’s advantages over the brute supernatural are clear: not only does it create suspense, it avoids demanding that we accept the supernatural by persuading us to accept only the latter’s possibility. Once we narrow our sights on this signature device, Walpole’s want of imitators does not seem so strange. Some people liked his brute supernatural, but no one cared to reproduce it (especially because Walpole’s second preface had burned the bridge that was the sanctioning pretense of medieval authorship). Only with the first examples of the “supernatural explained” in the novels of Fuller, Smith, the anonymous author of The Spectre (1789), and of course Radcliffe (1790) do writers find a convincing way to introduce the supernatural. Unlike the clue, hesitation spreads immediately. Like the clue, it is still with us today.⁸⁴

So my narrative, which I choose to start in the mid-seventeenth century, goes like this. The supernatural had long been a point of interest for writers, who saw very well that literature did not have to be rational to provoke pleasure. Their solution was to carve out genres that appealed to the imagination as opposed to the faculties of reason and the passions—opera, and fairy and oriental tales. A few isolated figures (Walpole, Cazotte) attempt to bridge what becomes an accepted divide between the literature of fancy and the literature of sentiment and psychology; this program is made explicit either by them (in Walpole’s case) or by their contemporary critics (in Cazotte’s). But for whatever reason—general satisfaction with the Great Divide, lack of a good formula for introducing the supernatural into the novel—nothing happens. Around 1790 some writers in England start using the device of hesitation: protagonists are confronted by irrational phenomena that they register as such; they continually ask themselves, “Can this be?” Initially, the device is used in the English gothic (and by French “copycats” such as Ducray-Duminil). Another generation, starting with Hoffmann and then Nodier and Mérimée, will fit the device to a new context, bringing it into a more immediate present (no more far-away castles), eliminating romance plotting (no more faithful lovers of mysterious birth), and substituting new explanations—especially madness—for the creaky supernatural explained. (It is probably at this stage that Freud would start to recognize the uncanny, for it is only here that the supernatural can intrude into modern life; Mérimée said that “to do the fantastic right, you have to begin by dressing your hero in a flannel vest.”⁸⁵) From this point, there has essentially been no end to the device’s re-deployment, occasionally in “high” literature (Nerval’s Aurélia, James’s Turn of the Screw), very often in twentieth-century pulp. Todorov is right to single out the importance of hesitation, but wrong to make the perfectly ambiguous work the cornerstone of the fantastic. Just as tragedy is defined by its subject matter and not its resolution—there need not be blood—the fantastic lies in the hesitation itself. And historically speaking, hesitation was the only bridge over the Great Divide that could support any traffic.

Cazotte has become a bit lost in the gothic mists, and it is time to bring the discussion back to him and to the question of fiction. Of all the works I’ve been discussing, which are “fictional”? At what moment does fiction “exist”?

Cazotte’s Le Diable amoureux has a good claim to being fictional and not pseudofactual for two main reasons. First, it unfolds in a shared contemporary world while recounting events that are impossible. So instead of pretending to believe the story literally, which is what the pseudofactual asks of its readers, we just find the story “believable,” which is to say, internally vivid, psychologically plausible, and so on. Second, though Le Diable amoureux takes the familiar form of the memoir, we are never asked—however ironically—to credit it as a “found manuscript.”⁸⁶ One clue that the pseudofactual memoir is being used to different effect is the “nouvelle” designation in the work’s subtitle. Nouvelle normally implies a true story, in contradistinction to conte, or tale, the preferred designation for narratives that take place on the other side of the Great Divide.⁸⁷ In that, the subtitle redoubles the claim to truth put forth by the memoir. But the curious thing is that I know of no other cases where a memoir novel is called a nouvelle: the former is in the first person, the latter in the third; they are incompatible forms. The supernatural subject matter of Le Diable amoureux, it would seem, short-circuits accepted classificatory systems. Calling a memoir a nouvelle entails revising what the terms can possibly mean; the subtitle here begins to signify something like “novel,” behaving as a marker of the fictional status of what otherwise looks to be a memoir. Given all this, it would be churlish to deny the label “fiction” to Cazotte’s work, and we can probably extend it, mutatis mutandis, to Walpole and Reeve as well.⁸⁸

In the end, however, discerning fictional “firsts” keeps us on the level of the individual, whereas fiction must be understood on the level of the group: one fictional work (or three, spread over a dozen years and two countries) does not mean that fiction “exists.” For where could fiction possibly reside, apart from in shared practices and rationales? Unquestionably, Cazotte and Walpole and Reeve share something—a desire to blend fancy and illusion, two categories bequeathed to them by an entire poetic tradition. And each in his or her manner does blend them. But only later, after the invention of hesitation, do writers have a systematic way to forestall the charge of incredulus odi: they incorporate within their works the suspense-laden question of whether the impossible can be real. Belief becomes the novel’s very pivot. Not belief about the real world, of course, but a belief that is cultivated in the space of reading itself. And for that to occur, a subtle shift is necessary, a shift that is more technical than conceptual: belief is not something readers are asked to (pretend to) bring to their reading; rather, belief is introjected into the space of the novel. And so authors can start from a premise shared by readers—the supernatural does not exist—and yet end up persuading those readers of the possibility of something that both parties know is untrue. Reading allows us to entertain possibilities that we need not entertain in real life; emotion and edification, it turns out, do not depend on our literal belief after all. In the end, the assumptions of old aesthetic theory were just wrong. The gothic novel gave the lie to Richardson, who like so many others thought of “historical faith” as a fragile thing, destroyed by the least intimation of impossibility. Radcliffe and Lewis’s works were not “more fictional” than Cazotte and Walpole’s solitary efforts. They were, however, part of a shared practice, and thus a much more credible reason to proclaim, if proclaim we absolutely must, fiction’s birth.