Before Fiction: The Ancien Régime of the Novel

Before Fiction

BEFORE

FICTION

The Ancien Régime of the Novel

NICHOLAS D. PAIGE

UNIVERSITY OF PENNSYLVANIA PRESS

PHILADELPHIA

Copyright © 2011 University of Pennsylvania Press

All rights reserved. Except for brief quotations used for purposes of review or scholarly citation, none of this book may be reproduced in any form by any means without written permission from the publisher.

Published by

University of Pennsylvania Press

Philadelphia, Pennsylvania 19104-4112

www.upenn.edu/pennpress

Printed in the United States of America on acid-free paper

10  9  8  7  6  5  4  3  2  1

Library of Congress Cataloging-in-Publication Data

ISBN 978-0-8122-4355-0

Tout étant égal d’ailleurs, j’aime mieux l’histoire que les fictions.

—Denis Diderot

Contents

Preface

Introduction: The Three Regimes of the Novel

Chapter 1. The Impossible Princess (Lafayette)

Chapter 2. Quixote Circa 1670 (Subligny)

Chapter 3. How to Read a Mind (Crébillon)

Chapter 4. The Aesthetics of Sentiment (Rousseau)

Chapter 5. The Demon of Reality (Diderot)

Chapter 6. Beyond Belief (Cazotte)

Conclusion: On Narrators Natural and Unnatural

Notes

Bibliography

Index

Acknowledgments

Preface

This book proposes a new history of the novel in France and England in which fiction itself is the primary variable; my account then provides the ground for understanding the fictional status of a series of (mostly) canonical novels from the early French tradition—or, more to the point, for understanding why they may not in fact be fictional. Both the larger narrative and the individual readings are subtended by an approach to the evolution of literary forms that parts company with most work on the novel’s history, and this, as much as fiction, is my subject as well.

First, the big picture: I sketch out here a history of fiction. Elaborating and substantially modifying the arguments of a number of specialists of the English novel, I argue that fiction is not at all coterminous with literature or what used to be called poetry, but is a rather recent phenomenon. Saying this, I am not following common modern usage and taking fiction as a synonym for the novel; though the present study is restricted to novels, it is not about their birth. By fiction, I mean something better though more awkwardly captured by the substantive fictionality, which is to say the peculiar yet for us intuitive way that novels refer to the world: via invented characters and plots, they purport to tell us how people and institutions and abstractions like money or power work. This is peculiar logically: how can writers possibly persuade readers of their view of the world if they are just making up their evidence? More important, it is historically peculiar. For one thing, the type of invention commonly practiced by novelists starting in the nineteenth century has few analogues in earlier times, which accorded little respect to writers dabbling in subject matter entirely of their own creation, and which largely understood the term fiction to designate a form of lying as deplorable as any other. Moreover, openly invented characters were a rarity for a good chunk of the novel’s development in France and England: in the late seventeenth century and for almost all the eighteenth, novelists presented themselves as mere editors, and their inventions as real documents or reports. Modern readers have often looked back on such pretense of literal truth with a certain degree of bafflement, but our present reflex, according to which the real-world existence of the characters we read about matters not a bit, would have proved just as baffling to readers throughout the two preceding millennia.

No doubt there are many valid and useful definitions of fiction and fictionality according to which the above distinctions seem but split hairs: isn’t all literary imagining a part of what philosopher Kendall Walton has called the human propensity to make believe? And more seriously, perhaps: doesn’t Aristotle, in the West’s founding document of literary criticism, place the distinction between poetry and history front and center? Such are two main obstacles between us and a history of fiction, but they are far from insurmountable. As we will see, the principal hurdle of the Poetics is simply that we read it through our knowledge of what is to come, which is to say, fiction. And though my definition of fiction is undeniably only one of many possible definitions, it has the advantage of enabling us to distinguish between three historical regimes of literary invention in a way we cannot if we just make some consciousness of fiction the bedrock of all literary endeavor. The three regimes, which succeed one another in their dominance, are the following. Most of the Western literary tradition since Homer can be understood through the lens of Aristotle’s Poetics, which described and sanctioned an enduring articulation (not opposition) of poetry and history; according to this model, the poet adds his inventions to the renowned heroes and events of history so as to make a good plot. The second regime starts around 1670 and lasts until roughly the turn of the nineteenth century. During this time, novelists cease posing as Aristotelian poets and instead pretend to offer their readers real documents ripped straight from history—found manuscripts, entrusted correspondence, true stories, and all the rest. Following Barbara Foley, I will be calling this type of novel pseudofactual, in that it masquerades as a serious utterance. That the masquerade is almost always patent should not tempt us to confuse it with what happens under the third, properly fictional regime: the pseudofactual pact demanded that readers pretend to regard novels as true, whereas later novelists asked for something quite different—that they accept the writer’s inventions as a kind of model of reality. This is how the nineteenth century replaces the old distinction between poetry and history with fiction as we have come to know and practice it.

This narrative provides the context for the bulk of the book, comprised of six case studies illuminating the strange interregnum between Aristotelian poetics and modern fiction, a period during which the omnipresent formal feature of the novel was pseudofactual posturing. The first two chapters probe the leading edge of the regime via one novel long accepted as a canonical milestone and one that has been completely forgotten. Because of its combination of an invented heroine and a carefully drawn historical setting, Lafayette’s La Princesse de Clèves (1678) can easily be seen as a major step on the road to modern fictionality. In fact, it simply demonstrates that individual works, however successful, need not be signs of wider transformation: the apparently fictional Princess is better understood as a one-time and idiosyncratic twist on the traditional Aristotelian understanding of the poet’s use of history. Subligny’s forgotten La Fausse Clélie (1670), a Frenchified Don Quixote, allows me to tackle what has been for many decades if not centuries a basic way of understanding the novel’s history—to wit, that the modern novel replaces archaic romance. While similar though not identical divisions were made repeatedly in the period, the novel-romance opposition is not in fact able to account for texts like Subligny’s or even Cervantes’s. Romance was not dead for either writer, it was just in need of updating; and La Fausse Clélie constitutes a signal attempt to make romance forms safe for a pseudofactual age.

The subsequent four chapters are devoted to works that stretch the conventions of the pseudofactual regime without—and this is crucial—causing those conventions to crumble in favor of modern fiction. Crébillon’s Les Égarements du coeur et de l’esprit (1736–38) is the most sustained example of a new novel of manners that imported a model of invention from comedy—comedy, which since Molière had allowed writers to comment on contemporary social types without taking aim at specific individuals. But though Crébillon put no energy into bolstering the pseudofactual pretense of his memoir novel and even narrated thoughts in a manner often associated with modern fiction, his experiment, like Lafayette’s, did not change the way novels were written. Indeed, pseudofactuality had bright days ahead. The sentimental novel’s goal of giving the genre the emotional and moral gravitas of tragedy necessitated a reinvestment in reality: all available theories of aesthetic effect made the audience’s belief in the artwork the foundation of emotional experience and moral improvement. This does not mean that writers such as Rousseau and Diderot were naïve about how novels worked. On the contrary, Julie, ou La Nouvelle Héloïse (1762), commonly advanced as the epitome of quixotic fusion between reader and book, is a laboratory for producing an emotion proper to observers who identify with protagonists even as they maintain distance from them. And Diderot’s sentimental novels and tales, though often seen as signaling through their irony a fictional consciousness to come, remain thoroughly and necessarily enmeshed in pseudofactual presuppositions. The final chapter centers on novelistic subgenres that are tailor-made for thinking about aesthetic effects that don’t require literal belief—the fantastic and the gothic. But although a work such as Cazotte’s Le Diable amoureux (1772) clearly pushes hard against the governing Horatian dictum of incredulus odi—readers reject what they cannot believe—it is no more predictive of the future than any of the other works examined.

The problem of predictiveness brings me to the methodological heart of Before Fiction. For it is the relation of the individual novels examined in the six chapters to my larger narrative—a narrative of collective behavior—that separates this book from most histories of the novel. Individual works, especially great ones, are usually the data points that permit the literary historian to plot out an evolution. Thus, given the chronological arc of the novels I’ve chosen, the obvious assumption would be that I am tracing an evolution, and more precisely, that I see these works as bringing fiction into existence. But this is emphatically not the case. Before Fiction could not have been titled The Rise or The Invention of Fiction, since Cazotte, Diderot, Rousseau, Crébillon, Subligny, and Lafayette are not all engaged in the same great project. These writers are not feeling their way through the pseudofactual night toward the bright light of the fictional day. Nor have they intuited some truth about literature—say, Coleridge’s unfortunately proverbial willing suspension of disbelief—that we all now embrace. None anticipate developments that would only become dominant later, or bears witness to a collective cultural realization. And they do not relay one another: Diderot does not learn from his one-time friend Rousseau; Rousseau in turn learns nothing from Lafayette (whom he nonetheless admired). No one stands on anyone’s shoulders to get a better look into the future. At the same time, it is not that some other change—the advent of a concept of fiction, or of modernity tout court—is readable in the works of authors who were extraordinarily sensitive to their transitional moment. Isolated literary works are not signs of anything else; if they were, they would not be isolated.

What, then, can be the broader significance of the novels under study? In a sense, none: they don’t add up to anything; they don’t register momentous change. But this doesn’t mean that they are not instructive. On the contrary, their authors’ complex engagement with the problem of novelistic reference in the wake of Aristotelian poetics brings the larger narrative into focus. Most writers of the period, even good and great ones, give no more thought to the pseudofactual posture than, say, a filmmaker in our day who ends up producing a color feature lasting between 100 and 140 minutes. It’s just how things are normally done. The authors under study, meanwhile, scrutinize the problem of how to write literature that doesn’t take as its subject matter the heroes of the past. How can novelists refer to their world without writing about people who are actually part of that world? In asking this question, they push at and play with the conventions of the time; in some cases one might even say that they destroy those conventions. But to the extent that this is the right word, their destructions are local: these individuals do not alter collective practice. I will offer, especially in the Conclusion, some thoughts on how collective practice does change—thoughts, because the type of data needed would dictate a different type of study altogether. For the moment it is enough to insist that though we like to view our favorite authors as heroes or at least as paragons of historical acuity, the truth is much more plain: writers are of their time and place, which is to say, bound by a set of practices and rationales that they do indeed transform, but in limited ways, often without wider effect, and certainly not with our unborn needs in mind. By the pressure they put on Aristotelian and pseudofactual conventions, these six novelists may appear to be gesturing in the direction of fiction, but we mustn’t give in to magical thinking: the mirage is generated simply by our coming after fiction. Lafayette, Diderot, and Rousseau are not so much agents of the transformation of Aristotelian poetics into modern fiction as they are participant-observers of processes whose momentum—and inertia—outbulks the contributions of individuals, no matter how perceptive or talented. The novel, envisioned as a history of shared practices and forms, would look much the same without the great writers customarily regarded as the motors of generic change.

Studies in the morphological history of the novel: this would have made an apt if offputting subtitle for a book concerned not with what deeper things novels reflect but with how forms evolve. My subtitle as it actually reads demands a couple of qualifications. The first relates to the term I’ve chosen to designate the three approaches to literary invention—the Aristotelian, the pseudofactual, and the fictional. I attach no particular importance to the word regime itself; it appears, say, in work by Jacques Rancière and François Hartog, but my use does not follow from theirs. (When speaking of these regimes not as time periods but as ways of writing novels, I often call them modes—a term that, likewise, is not intended to recall Northrop Frye.) For me, a regime is merely a fairly stable but not monolithic way of thinking about and writing (narrative) literature; as I will take pains to point out, in no sense should it be taken as implying a perceptual or conceptual matrix on the order of Thomas Kuhn’s paradigms or Michel Foucault’s epistemes. My subtitle, then, contains some wordplay. On a basic level, The Ancien Régime of the Novel means nothing more than The Novel in the Early Modern Period; in a more important sense, the novel’s Ancien Régime is the pseudofactual regime, the interregnum between Aristotelian poetry and modern fiction.

The second remark concerns the fact that my shorthand carries some baggage better left behind at once. Since only France had an Ancien Régime, it is easy to conclude that my overarching theory of regimes applies to France alone, and that the sociopolitical context of the country explains its novel. The pseudofactual, readers might reason, must have something to do with absolutist monarchy; by extension, perhaps Aristotelian poetics suits the politics of the Classical and feudal ages, while fiction is made possible by the French Revolution. This would be nonsense, however. At most, Ancien Régime advertises that the individual works analyzed are French, while hopefully not obscuring the fact that the larger narrative covers the English domain as well. Cultural specificity matters to the novel’s history in all sorts of ways too obvious to mention. Nevertheless, and pace the many scholars who have explained the novel as first and foremost and necessarily English, Before Fiction argues that the problem of novelistic reference was shared because it was the result of a broad breakdown in Western poetic practice.

Unless otherwise indicated, translations and ellipses are my own; italics in quoted sources are not. Without completely modernizing punctuation, I have occasionally modified it for clarity. For economy I typically use last names alone when referring to authors of the period under study; readers needing greater precision of course will find it in the Index.

Introduction:

The Three Regimes of the Novel

One peculiarity of novels when they first arrived in the eighteenth century was that they told new stories rather than recomposing old ones. Their characters were singular; each novel had to introduce its readers to a new world. This has not changed.

—John Mullan, How Novels Work

Gottlob Frege’s essay On Sense and Reference, published in 1892, stands at the beginning of modern philosophical interest in fictionality—that is, in the truth status of fictional propositions. Poetry—roughly, what we now call literature—had of course long been seen as a special kind of deceit that, at least for poetry’s many defenders, led mysteriously back to the truth. The truest poetry is the most feigning, says Shakespeare’s Touchstone; The novel establishes its birthright as a lie that is the foundation of truth, writes Carlos Fuentes much more recently; and indeed, the literary ground since the Greeks is strewn with chestnuts such as these.¹ The history of Western literary theory, as one noted theorist puts it, can be summed up as a continuous debate on the classical dictum that poets are liars.² Frege’s interest was nonetheless distinct, for he was interested in semantic questions regarding language’s capacity to refer to the world; literary language was a curious subspecies that did not, he argued, refer at all. If we read, in Homer, that Odysseus was set ashore at Ithaca while he was sound asleep, we understand the proposition as having a sense even though the proper name Odysseus has no reference in the real world, and thus no truth-value. In hearing an epic poem … apart from the euphony of language we are interested only in the sense of sentences and the images and feelings thereby aroused…. Hence it is a matter of no concern to us whether the name ‘Odysseus,’ for instance, has a reference, so long as we accept the poem as a work of art.³

A matter of no concern to us, perhaps, but would the ancient Greeks have felt the same way? To be fair, Frege’s essay is only tangentially concerned with literary reference; it focuses on the way signs in general refer, and Frege, like many early theorists, felt that sharply separating out literature from natural forms of discourse clarified the issues.⁴ It is therefore not surprising that, as a theory of fiction, Frege’s treatment of Homer leaves much to be desired. But one of its shortcomings in particular is shared by more modern and elaborate theories of fiction. That shortcoming is historical. We are welcome to our doubts about Odysseus’s reality, or for that matter about Athena’s—as were, presumably, the Greeks—but Homer certainly didn’t invent them in the manner that Balzac invented Old Goriot or Dickens invented Little Dorrit. Epic heroes and the gods were quite simply attested: they were authorized by tradition. They may or may not have had reference in Frege’s empirical sense, but they didn’t need any: they possessed a type of extratextual existence that the protagonists of the typical nineteenth-century novel did not.⁵ Which is to say that along with asking what fiction is, we might also ask if fiction always is, in the same way: mightn’t calling Odysseus fictional be to mischaracterize Greek practices of poetic invention, and to read the Odyssey as if it were a modern novel?

We might offer sympathetic support for Frege’s contention that literary protagonists have no reference by limiting it to the nineteenth-century novel—a likely source of the philosopher’s conviction in the first place. The difficulty, however, is that substituting a sentence from Balzac or Dickens for Homer’s verses leads to new complications: Old Goriot or Little Dorrit may have no reference, but their inventors refer rather insistently to the Paris and London of their day—not only to places, but also to the workings of money and class and institutions. Such reference obviously falls outside Frege’s understanding of the term, predicated as it is on the proper name. We could, then, refine Frege’s proposal, perhaps noting with John Searle and others that certain fictional genres contain nonfictional commitments, which is to say, references to known people and places.⁶ This type of accommodation does not, however, solve the problem, which I repeat is at bottom historical: unlike ancient epic, the nineteenth-century novel speaks about specific, local, empirical phenomena, but it does so using completely nonexistent characters engaged in actions that never happened. Homer, meanwhile, spoke of legendary people and events, both (we may speculate) because of their intrinsic interest (heroes, by definition, are worthy of being known) and for the moral or ethical lessons they taught (heroes, by definition, are exemplary). Not without reason are we used to thinking of Western literary history along the lines laid out by Erich Auerbach in Mimesis: literature becomes more and more real, more focused on physical reality. But Frege’s remark, by its very inadequacy, reminds us that we must also factor in the converse: modern literature at its most splendidly realist is also removed from reality in a way it had never been before. It can talk a lot about history, implicitly or explicitly, but it does not claim to treat the same people and events that historians do. The difference between Homer and the modern novelist is thus not one of degree: the way the texts work, their modes of reference, are simply incommensurable. And if we agree to call the mode of Balzac and Dickens fiction, then Homer did not write fiction.

The Odyssey is not fiction? Not a novel, granted—even the many scholars who remain divided about the novel’s origin can probably agree on that much. But surely fiction and literature as such are coextensive: All literatures, including the literature of Greece, have always designated themselves as existing in the mode of fiction, writes Paul de Man.⁷ Indeed, fiction is the innocuous term used when generic objections are feared or when genre is uncertain: The Canterbury Tales and The Faerie Queene may or may not be novels, but they are surely fiction. In fact, fiction is the most unobjectionable term of all, better even than literature, a word (and therefore, troublingly, maybe even a concept) that has come into use very recently.⁸ Of course, like literature, fiction has a lexographic history. It derives from the Latin fingere, to invent; it was long used as a synonym for lies, and sometimes for poetry, especially types of poetry that did not aspire to the dignity of epic or tragedy; around the nineteenth century it became synonymous with the novel and, as I’ve noted, with narrative literature more generally.⁹ Yet my point is not ultimately lexographic: as a handful of scholars working on the early English novel have suggested, it is the operations we associate with fiction that are historically bounded.¹⁰

Broad uses of the word fiction can of course have their own logic and utility. In many ways, humans are uniquely fiction-making animals, as Kendall Walton for one has shown, and it may be that this cognitive ability is an evolutionary adaptation.¹¹ Moreover, there is certainly nothing inherently wrong-headed about using fiction as an umbrella designation for discourses about poetry.¹² Still, a clue that all literatures have perhaps not always operated in the fictional mode can be found in one of de Man’s favorite authors: Rousseau famously refused to identify himself as the author of Julie, ou la Nouvelle Héloïse (1761), a text he presented as authentic correspondence even as he called that authenticity into doubt. Clearly, it would be inaccurate to say, with Frege, that it is of no concern to him or his readers whether his heroine Julie existed. It must be of concern; otherwise he would not have gone to the trouble of addressing the issue in not one but two prefaces. Few if any were duped by Rousseau’s posture; it was unconvincing by design, as I will show. The point for now is simply that Rousseau—like many other novelists of his age—did not relinquish the Fregian reference of the proper name. And so Julie may not really be fiction any more than the Odyssey—at least not in the sense of Balzac or Dickens, who were, true to Frege’s intuition, unconcerned with the literal reference of their protagonists’ proper names. Or rather, less even than unconcerned, if by this we mean, Maybe Goriot existed, maybe he didn’t. Goriot did not exist—no hedging necessary.

This Introduction begins with some examples of how people have spoken, quite diversely, of the relation between poetry (or literature) and history (which itself is an unstable term). In these sections, Aristotle, Richardson, and a few nineteenth-century writers help flesh out some preliminary characteristics of what I will be calling the three regimes of poetic invention—the Aristotelian regime, the pseudofactual regime, and the fictional regime. Of special concern will be explaining what we lose if—following previous critics who have tried to replace the history of the novel with a history of fiction—we consider the pseudofactual novel to be merely an early version of the fictional novel: what came before fiction, as my title implies, was not fiction. (It was not inadequate, clunky, or naïve for not being fiction; it simply consisted of practices and rationales that fiction replaces or at least supplements with others.) I will finish up with some methodological remarks designed both to clarify what I mean by the term regime and to underline how the type of literary history it subtends departs from most accounts of the novel’s rise. This all amounts, I hope, to a preliminary case for the pragmatic usefulness of a historically restricted definition of fiction; given space limitations, it cannot be a full presentation and defense of modern fiction’s legitimacy (to borrow a term from Hans Blumenberg).¹³ Readers are asked to keep in mind that my analyses of the six writers treated in the chapters that follow will help fill in many of the blanks in this initial sketch; the Conclusion too chases down some problems it would be premature to tackle at this point. If, as I sometimes fear, Before Fiction opens up more questions than it answers, I can only hope that they are at least not the same questions.

Aristotle, Poetry, History

Homer, Rousseau, Balzac: one might grant differences in the way these authors invent without going so far as to deny some of them fictional credentials entirely. Why arbitrarily brand a given cultural practice as fiction proper, excommunicating writers who don’t measure up to the conventions of the nineteenth-century novel? Why not speak, rather, of different types of fictional modes? This would allow us to say that Homer operates in one mode, Rousseau in another, and Dickens in another still, while all the while not denying that their works have something in common. After all, fiction comes from fingere, as I’ve pointed out myself, and all these writers, readers know, are making or inventing to one degree or another. Surely we can agree that Homer, Rousseau, and Dickens did not write, did not want us to think that they were writing, history. Besides, Aristotle long ago carved out for poetry the domain of the possible, and opposed it to history. Why not call the underlying something that unites their texts—that is, the quality that separates them from historical assertions—fiction?

Let’s start with Aristotle, then, whose separation of poetry and history has indeed become proverbial, the place we go for an authoritative formulation of what we already know. The famous lines run as follows: The difference between the historian and the poet is not merely that one writes verse and the other prose—one could turn Herodotus’ work into verse and it would be just as much history as before; the essential difference is that the one tells us what happened and the other the sort of thing that would happen.¹⁴ The philosopher’s words jibe nicely with modern ideas about verisimilitude, realism, and probability on the one hand, and invention on the other: realistic works don’t pretend to be history, they create something of a parallel world that behaves like the world of history. Thus modern commentators often see Aristotle’s would happen as endorsing the idea that literature creates alternate, probable, or hypothetical worlds. It is the artist’s task to convince us that this could have happened, writes one critic of the modern novel: Internal consistency and plausibility become more important than referential rectitude.¹⁵ Little Dorrit, we might say, describes what might happen to an imagined debtor imprisoned in Marshalsea, not what the prison’s real inmates did and experienced. A reader could always judge Dickens’s novel unconvincing (too many coincidences, too much melodrama), but that would just make it unsuccessful fiction, not, obviously, history. Clearly, the would happen of poetry covers Dickens’s practice quite nicely; therefore the house of fiction may have many rooms, but it’s still fiction through and through.

And there are more ways still to argue that Aristotle’s theory of poetry makes way for Little Dorrit. Elaborating on the distinction between what happened and what would happen, the philosopher continues:

That is why poetry is at once more like philosophy and more worth while than history, since poetry tends to make general statements, while those of history are particular. A general statement means one that tells us what sort of man would, probably or necessarily, say or do what sort of thing, and this is what poetry aims at, though it attaches proper names; a particular statement on the other hand tells us what Alcibiades, for instance, did or what happened to him.

This passage too can underwrite an extension of the Poetics to the modern novel. First, the novel’s nonhistorical characters can be said to embody social types, human values and experiences, lessons about this or that; it is thus general. Stephen Halliwell has warned, I think rightly, that there is little to no evidence for this understanding of Aristotle’s generality, however, and so he proposes a second, less anachronistic resemblance between poetic generality and modern fiction.¹⁶ After all, we routinely speak of successful fiction as creating a thick, internally coherent world, and a small modification of the translation can reinforce this: replacing would happen with might happen or could happen aligns poetry still more closely with hypothetical or imaginable realities.¹⁷ Halliwell’s case for the relevance of ancient theories of mimesis to enduring problems of representation therefore includes the suggestion that the author of the Poetics is feeling his way … toward a notion of the fictional or the fictive.¹⁸ History is what happens, poetry is what might happen, what can be imagined as happening. Historians are given their material, poets invent it. And so do novelists, we now add.

For at least two reasons, however, Aristotle is less firm than we are in this happy division of labor. First, as Halliwell himself notes, generality—what would happen—is much more plausibly understood as a structural feature of poetry, related to plotting, causality, motives and so on. Aristotle, who disliked nothing so much as episodic plots, defines it himself as what can happen in a strictly probable or necessary sequence. Second, generality for Aristotle can hardly mean that the poet imagines or invents people and events that could realistically have existed or happened, for the simple reason that Greek poets, while inventing motives and causes, do not usually invent their heroes. Instead, they use proper names referring to the very same people historians do. Poetry is general, writes Aristotle, and he explains that this involves probability and necessity—plotting. But he cannot keep from adding, though it attaches proper names—the though registering an obstacle to a clean opposition between poetry and history. This does not destroy the criterion of poetic generality, certainly, for we can take our interpretive cue from subsequent commentators and practitioners (say, those of the French neoclassical stage) and understand the philosopher’s words like this: the historian cannot choose among things that happened to Alcibiades, whereas the poet selects certain things and invents other things said or done, in view of constructing a unified plot. But Aristotle’s though does complicate our assimilation of that generality to the modern idea of a fictional world: the world of poetry is causally coherent, but it is not invented in quite the same way as is a novel by Dickens, whose protagonists were not Gladstone or Disraeli.

The passage moreover does not stop here, as if Aristotle himself were not quite convinced that he had explained why poetry was general. That this remains a problem is made clear by Aristotle’s swerve away from tragedy to a species of poetry that is more obviously both supportive of claims to generality and distinct from history. That poetry does aim at generality has long been obvious in the case of comedy, where the poets make up the plot from a series of probable happenings and then give the persons any names they like, instead of writing about particular people as the lampooners did. The logic here is not spelled out, but presumably we are to infer that comic playwrights—as distinct from the satiric playwrights (lampooners, or more literally, iambic poets) of what we now call Old Comedy—invent and name protagonists who embody given human character types, who are therefore walking generalities. (Aristotle itemized various character types in his Ethics, and his descriptions would be greatly elaborated by his student Theophrastus.) Comedy, then, makes for a cleaner opposition between poetry and history. Yet for a second time Aristotle is drawn back to the fact that tragedy doesn’t function this way at all: In tragedy, however, they stick to the actual names. At this point, Aristotle finally drops the idea of poetry as a general statement and advances an argument that was enthusiastically developed by his Renaissance followers: conviction is instilled in viewers by real events, hence the importance of real protagonists.

The contortions of this famous passage are good evidence of Aristotle’s efforts to square his initial hypothesis on poetry’s interest in what would happen with the irrepressible fact that tragedy uses proper names. These proper names, which keep popping back up each time the philosopher seems to have the lid on the box, effectively resist the bringing of all poetry, as distinct from history, under the banner of generality. In other words, comedy and tragedy are not general in quite the same way, just as the distinction between the comic poet and the historian is not exactly the distinction obtaining between the tragic poet and the historian. By Aristotle’s own reckoning, tragedy and comedy function differently. It is not merely that one depicts people better than they are and the other worse (as he writes elsewhere in the Poetics), nor only that one speaks of distinguished families and lofty sentiments, while the other busies itself with the low born and their mundane concerns (as later commentators would repeat). Rather, tragedy deals with real people and comedy with types.

This, at any rate, was what later European commentators would take from Aristotle, and it proved surprisingly adequate to poetic practice for about 2000 years: comic characters were invented; serious protagonists were taken from history. Thus Diderot, toeing the Aristotelian line in 1758, could still divide discourse into three types: History, where facts are given; tragedy, where the poet adds to history what he thinks likely to increase its interest; comedy, where the poet invents everything.¹⁹ Of course, Diderot leaves out other possibilities, notably fable. And one can easily come up with examples that sit uncomfortably or not at all with the preference for attested subject matter. Aristotle himself backs up after declaring that tragic poets stick to the actual names and gives a dutiful nod to Agathon’s now lost tragedy Antheus, which did feature invented characters. (Antheus leads the author of the Poetics to shrug off his hypothesis that the reality of characters imparts necessary conviction, since he freely admits that invention doesn’t in the end infringe on the audience’s pleasure.) But Antheus is to all appearances a one-off, for nowhere in the classical corpus do we find references to other such tragedies. To be sure, the Greek novels of Achilles Tatius, Xenophon, Heliodorus, and others, recount the adventures of characters who have no sanction outside the text, as do the works of Petronius and Apuleius that Mikhail Bakhtin pointed to as antecedents of the modern novel.²⁰ By the same token, however, these works were long denigrated precisely because, like mere fables, they lacked the prestige of history. Any number of famed Renaissance works, from More’s Utopia (1516) and Ariosto’s Orlando furioso (1516–32) to Rabelais’s chronicles (1532–52) and Spenser’s Faerie Queene (1590–96), are hardly Aristotelian, and indeed flaunt what we can no doubt broadly call their fictionality. Yet on inspection the invention practiced by these writers bears little relation to the fictionality of a character like Balzac’s Goriot: such works are parodically inseparable from either attested heroes of the chivalric past (Ariosto, Spenser), or the truth claims of the New World travel narrative or medieval chronicle (More, Rabelais).²¹ Rather than being fictional, they parody what someone else is purported to regard as true.

It is not to impugn the creativity of pre-nineteenth-century writers that I resist speaking of literature as a house of fiction with many, many rooms. Nor does taking fictionality for something other than a universal property of literature imply that invented heroes were, in Foucauldian parlance, unthinkable for the Greeks, Romans, or Europeans of the Renaissance. In different ways, all the writers I have just mentioned invent characters from scratch, and an alternate version of the present study could no doubt inventory at length such practices. But our modern indifference to what Frege called characters’ reference keeps us from making the simple empirical