SOURCE: "The Kinds of Historical Fiction: An Essay in Definition and Methodology," in Genre, Vol. XII, No. 3, Fall, 1979, pp. 333-55.

[In the following essay, Turner identifies three categories of historical fiction—documented, invented, and disguised historical novels—and discusses the boundaries between history and fiction.]

"Everyone knows what a historical novel is; perhaps that is why few have volunteered to define it in print" [Avrom Fleishman, The English Historical Novel: Walter Scott to Virginia Woolf 1971]. So begins Avrom Fleishman's foray into the morass of defining historical fiction. Now there is, without doubt, a significant portion of truth in this disarming gambit: except for an occasional skirmish over how far back a novel must be set to count as history, critics have assumed that the term "historical novel" effectively explains itself. But the remainder of Fleishman's theoretical discussion—not to mention the ensuing chapters on individual novels—attests to a fundamental problem which this appeal to common sense only serves to mask: namely, that neither history nor fiction is itself a stable, universally agreed upon, concept. Small wonder, then, that as soon as Fleishman moves beyond the conventional definition and tries to distinguish between history and fiction, he creates as many problems as he solves. Not only does the whole discussion threaten to founder, but it usually circumvents that danger by presuming greater unanimity about what history actually is than exists among historians. Thus we find Fleishman asserting, for example, that "most philosophers of history" subscribe to the "covering law" theory of Popper and Hempel, when in fact there is anything but agreement on the question of the authority of historical explanations. Worse yet is the way Fleishman selectively appeals to contending schools of historiography when trying to argue his thesis that history and fiction can be differentiated according to "their similarities of aim and differences of means." Using Collingwood, Gallie, and Dilthey to assert the affinity between history and fiction, only to bring in Popper and Hempel to insist on the points of divergence, creates a tortuous, even if not altogether loaded, argument. We might just as easily juggle the contending schools differently and arrive at the contrary thesis: that history and fiction share the same means but have quite different ends. There is, for that matter, no reason to rule out the other two possible combinations in between.

My point, I should perhaps hastily add, is not that Professor Fleishman has nothing to tell us about historical fiction. His chapters on individual novels are often quite good. But his attempt at definition confuses as much as it clarifies. The problem with the opening chapter, moreover, is not simply one of definition, although it proves inadequate even for Fleishman's own purposes. Rather, the confusion inheres in his basic method, and in this regard he is by no means alone. Despite their divergent conclusions, in fact, nearly every commentator on the genre has opted for the same procedure: starting out with a comparison of history and fiction, they arrive at (depending on their preferences or presuppositions) a conception of historical fiction that revolves around either what fiction shares with history or what fiction claims for itself. Thus Georg Lukics and Harry Henderson build their critical approaches, although quite different, on the observation that history and fiction possess certain common traits. Conversely, Lion Feuchtwanger and Floyd C. Watkins erect their conceptions of the genre, although again quite different, on the argument that history and fiction are first and foremost distinct.

Now much has been accomplished in this manner, to be sure; but I question the advisability of adopting any method that depends for its success on the assumption that we can all agree on precisely what history is, much less how fiction relates to it. Either one of these questions is enough to detain discussion for a very long time; take them together and we shall never get out of the starting blocks. For with two such essentially contested concepts as history and fiction, comparison only compounds the contention.

Despite what common sense would appear to tell us,.… definition is a problem with the historical novel. Despite what the composite nature of the genre would seem to require, moreover, comparison is not the solution. But if not by distinguishing between history and fiction, you might ask, how are we to characterize the genre? Not, I think, by setting out formal elements shared by its members. First, because there is an inevitable circularity, if not a certain pseudo-empiricism, in such attempts at generic description: isolating common denominators restricts us, by and large, to recapitulating an a priori definition. Secondly, because formal properties may not be the genre's distinguishing characteristic: it is the content more than the form, after all, that sets historical novels off from other fiction. Finally, and perhaps most importantly, because the very diversity of the genre frustrates our rage to generalize, and condemns our results—should we persist—to triviality: for there are at least three distinct kinds of historical novels (those that invent a past, those that disguise a documented past, and those that re-create a documented past), and therefore no single description can blanket all three without losing most of its significance. We can succeed, then, if we wish, in forcing the genre to conform to a single conception of fiction or of history, but only at considerable costs. For the impulse to create historical fiction comes from incredibly disparate sources; it finds expression, moreover, in an equal variety of fictional forms. Similarly, the motives for reading historical fiction are manifold, and the satisfactions it provides are just as varied. Aside from the tautology of common sense (all historical novels are novels about history), in short, all we can say in general about the genre is that it resists generalization.

Rather than thinking in terms of necessary formal features, therefore, we need to begin with a question: exactly what do readers decide when they place a novel within the category of historical fiction? Notice that their decision entails two choices: whether they will consider the text a historical novel, and whether they will consider it a historical novel. There are, in other words, two boundaries to be set if we are to define the genre: one between historical novels and other fiction, one between historical novels and narrative history. Instead of trying to establish them simultaneously, however—which is precisely what the comparison of history and fiction in previous discussions was meant to do—we would do well to take up the boundaries one at a time. Instead of trying to establish them for all historical novels, moreover, the boundaries must be redrawn (and in different ways) for each of the three types of historical fiction. With these two strictures in mind, then, return for a moment to Avrom Fleishman's formulation of the commonly accepted definition. He mentions the need for a past setting ("beyond an arbitrary number of years" as he puts it) and for a number of historical events ("particularly those in the public sphere," he adds), but he insists that to be genuinely historical a novel must contain "a real person among the fictitious ones." Now it is clear, as many reviewers have pointed out, that this last requirement is simply arbitrary: it excludes from the genre, among other works, Middlemarch and Absalom, Absalom!, which is more than enough to raise a few hackles. Even Fleishman chafes under the restrictions of his own definition, as time goes on, arguing for example that Nostromo "is more solidly historical than many novels that more neatly fit the definition." But the point to be drawn from this contradiction is not, as Murray Baumgarten would have it [in "Novel: Some Postulates," Clio, 4 (1975)], that Fleishman has thereby "reduced his own, elsewhere insisted upon, generic definition" to shambles. Rather, confusion results only because of the failure to differentiate between kinds of historical novels. Thus the definition can be put to good use—and not just salvaged—with but the slightest of changes: each member of the genre need only meet some, and not all, of the requirements. The presence of a real person in a novel may not be a necessary condition, for example, but it comes very close to being a sufficient one. When we read a novel about actual people from the past, we automatically assume that we have a historical novel. And though it may well be that in rare instances we might finally reject that initial assumption, even that possibility in no way disturbs our definition, for we have not posited any necessary condition. Generally speaking, then, novels with an actual historical character can be considered historical fiction. Novels of this type I should like to call documented historical novels, to emphasize their direct links with recorded history.

The justification for this distinction, as Fleishman points out, is that having actual people in a novel raises the problem of the ontological status of "real" as opposed to "invented" characters. I shall return to this issue shortly; all I wish to stress here is that its very presence sets off documented historical novels from other fiction. Not only do these actual characters set the stage for much of the quarreling over historical "accuracy" (as in William Styron's The Confessions of Nat Turner, say), but their non-fictive status creates a series of narrative problems that do not arise when the novelist is working entirely with his own inventions. Thus documented historical novels require separate consideration. Then, too, this distinction has the further advantage of helping to explain some of the inconsistencies that crop up in categorizing historical fiction. Novels set in the author's own lifetime—Robert Coover's The Public Burning, say—or even novels that are more nearly autobiographical—Norman Mailer's Armies of the Night, for example—are frequently thought of as historical novels, even though they violate one or more of the criteria in the commonly accepted definition. The explanation, of course, is that both these works partake of many of the same qualities, and raise many of the same difficulties, as documented historical novels; they both have enough specific links to actual persons and events—particularly those in the public sphere—that we tend to think of them as providing historical interpretations. Now we could, to be sure, deplore the inconsistency and cling tenaciously to the definition; but that would be to ignore the way certain novels align themselves with obvious members of the genre, thereby engaging historical as well as generic expectations.

In any case, on the fiction side of the boundary between narrative histories and historical fiction stands the documented historical novel. Delaying for the moment the problem of deciding how to make that demarcation, let us move away from narrative histories along a theoretical continuum in the direction of novels generally. At some point, by no means imperceptible, we pass from documented historical novels to what I shall call disguised historical novels. Robert Penn Warren's All the King's Men will serve as an example. In contrast to something like The Confessions of Nat Turner, Warren's novel includes no actual characters or events. Nevertheless the work stands somewhere between documented history and conventional fiction: Willie Stark may be Warren's invention, but the parallels between the careers of Huey Long and Willie Stark are so close that we find ourselves reading the novel as a disguised account of a documented past. The critical issue here, therefore, entails defining the novel's relationship to recorded history and determining what aesthetic significance it might have. The very fact that Warren has chosen to transform the figure of Huey Long, after all, might warrant the conclusion that the whole question is irrelevant: by changing the name of the historical character, you might say, Warren circumvents the reader's impulse to judge the novel as a historical re-creation. Thus you could argue, and with some conviction, that within the terms that the novel sets for itself All the King's Men has nothing to do with Huey Long.

Despite this formalist argument and the fictional disguise on which it depends, historical questions persist. Although the name has been changed to protect the fiction, the striking similarities between the novel and recorded history continually tease the reader beyond the novel itself. Perhaps the closest, although not altogether satisfactory, analogy would be to the dynamics of allegory, for in both cases it is up to the reader to supply the identifications that lie outside the text. In contrast to the way allegory normally works, however, All the King's Men does not insist on a dual interest. Warren is perfectly willing—indeed, would almost prefer—to leave the Huey Long question aside. And for good reason: to insist on the historical re-creation brings the fictional disguise under attack. Thus once having decided to translate history into his own terms, Warren is better off not calling attention to that fact. If he is to maintain the autonomy of his fiction, he must assert its independence from history by presenting Willie Stark as if he were a completely fictitious character. Still, the parallels exist to qualify the fictional claim: even with the disguise, we can read All the King's Men as an interpretation of Huey Long, though it is not exclusively (nor perhaps even primarily) that. The link to past actuality may be attenuated, but it is nevertheless there. Neither drawn directly from recorded history nor composed as an outright fiction, Warren's disguised historical novel situates itself half-way between documented and invented history. Now it is possible, certainly, to imagine a disguised historical novel in which the camouflage is so complete or the historical antecedents so obscure that the reader never recognizes the disguise. And it is equally possible to imagine a disguised historical novel in which the historical parallels are so extensive and precise that the reader wonders why there should be any disguise at all. But between these two extremes, there is a good deal of historical fiction that operates in the manner of All the King's Men, simultaneously striking the reader as being completely new creation and disguised recreation. Because of this peculiarly double focus (and the variety of ways that a novelist might engage it), moreover, the disguised historical novel exhibits certain distinctive characteristics, which here again merit separate consideration.

Thus only in the final stage of the continuum—which I shall call the invented historical novel—do we face any real problem establishing the boundary between historical novels and other fiction. Here, since the principal characters and events are all invented, any insistence on a generic distinction must contend with the fact that most novels are presented as if they were history. More often than not, as Henry James points out, the realistic novelist aspires to the authority of history: "It is impossible to imagine what a novelist takes himself to be unless he regards himself as an historian and his narrative as a history.… As a narrator of fictitious events he is nowhere; to insert into his attempt a back-bone of logic, he must relate events that are assumed to be real" [Partial Portraits, 1888]. Thus it would seem that all novels, or at least all realistic ones, are "historical." Interestingly enough, in fact, Georg Lukács—in his seminal work on the genre—argues that historical novels are no different from any others:

If then we look at the problem of genre seriously, our question might be: which facts of life underlie the historical novel and how do they differ from those which give rise to the genre of the novel in general? I believe that when the question is put in this way, there can only be one answer—none. An analysis of the work of the important realists will show that there is not a single, fundamental problem of structure, characterization, etc. in their historical novels which is lacking in their other novels, and vice versa.… The ultimate principles are in either case the same. [The Historical Novel, 1937]

Such a conflation of realistic and historical novels, though true enough perhaps in the sense that Lukács formulates it, nevertheless confuses the distinction between genres and kinds: the "ultimate principles" may well be the same, for they are indeed both novels, but finer discriminations at a lower level of generality (and with a less organic conception of literary forms) can still be made. In this instance, the novel's setting and the author's relationship to it become the determining factors. The farther back in time the fiction is set, that is, the more likely we are to treat it as an invented historical novel. Here, then, we must almost insist on a temporal requirement: either Fleishman's suggestion of an admittedly arbitrary "two generations" or Harry Henderson's less-restricted notion of "the world that existed before the author was born" [Versions of the Past: The Historical Imagination in American Fiction, 1974]. Removed thus from the author's experience, the characters and events—though invented—begin to require the exercise of a distinctively historical imagination.

On these grounds alone, then, we might differentiate between this last type and fiction generally. In addition, however, invented historical novels are frequently structured to highlight the problems of historical interpretation. William Faulkner's Absalom, Absaloml, for example, is not only set in the unexperienced past but also examines the nature of historical inquiry. Admittedly, Quentin's attempt to learn the truth about Thomas Sutpen involves something that is historical only within the artifice of Faulkner's fiction. But with all the conflicting interpretations that stand between Quentin and the "real" Sutpen, his narrative stance takes on a historical significance that it does not have in realistic fiction. Much more than the illusion of verisimilitude is involved; indeed, the meaning of history and the possibility of recovering it become the primary concerns in Faulkner's novel. Determinations of fact, assessments of documents and oral reports, interpretations of motive—these are all methodological issues for the historian, as they are for Quentin. Thus by predicating a gap between the narrator and his story, Faulkner raises the issue of how it is to be closed. Herein lies the difference between many invented historical novels and other fiction: in realistic novels, though the narrator may pose as a historian, attention is usually diverted from the problem of how the narrator can know his story; in invented historical novels, by contrast, that very possibility is often brought into question, turning the novel into a reflection on the way we know history. Generally speaking, then, unless they have such a complex narrative structure, invented historical novels remain distinct from other fiction only by placing the action far enough into the past as to claim for themselves the status of a historical reconstruction.

To summarize, then, historical novels fall into three different categories, which represent distinct stages along the continuum (although individual novels may combine elements of all three). Each of the three kinds, moreover, can be distinguished from other fiction (albeit with decreasing ease as we move from documented, through disguised, to invented historical novels) in terms of the kind of characters and events that predominate. In setting the other boundary, quite naturally, the problematic is reversed. Neither invented nor disguised historical novels seem to run any risk of being confused with the works of narrative historians. Admittedly, these two kinds pretend to be referential; we can, to be sure, even judge the adequacy of their historical re-creation. But we do not, with invented historical novels anyway, expect to find these fictional creations mentioned in any historical documents. We are being presented with fictional, not historical, events—events that never were observable outside the text, nor ever can be, since they do not exist before or apart from the words that create them. Then, too, with disguised historical novels—if we accept the translation on its own terms—the same ontological difference obtains: the very act of the disguise asserts a fundamental discontinuity between the novel and recorded history, amounting to an insistence on the fictional status of the characters and events. With documented historical novels, however, the conventional distinction between history and fiction threatens to collapse, for the novelist and the historian share the same event. By drawing on recorded history rather than disguising it or inventing his own, moreover, the novelist appears (by virtue of eschewing these generic alternatives) to be making the historian's claim to recreate an extra-textual reality.

Certainly, the precise nature of this apparent claim to historical truth has been the subject of most of the debate surrounding the documented historical novel. On the one hand, it is argued that the novelist is not a historian and that no one expects him to be: thus so long as he does not violate our sense of what Robert Penn Warren has called "the spirit of his history," he is free to change facts and invent whatever he needs to tell his story. Whether or not his novel is "good history" does not matter; the important requirement, as Warren and others have insisted, is that it should make "historical sense" in a thematic or symbolic way. On the other hand, it has been argued that once the novelist moves into the arena of history he must comport himself as a historian: thus he must restrain his inventiveness, work within the existing evidence. The fact that he is a novelist is not relevant; he has pretended to be a historian, and therefore owes allegiance to the documents. Now neither of these positions in isolation, nor the very way of posing the issue, illuminates very much, except that there is a tension inherent in the genre (expressed in the oxymoron of the name itself). Readers conventionally expect, in other words, that a novelist will adhere fairly closely to his sources—why else call it "historical"? At the same time, however, they also assume that the novelist can take greater liberties than the historian—why else call it "fiction"? Where the problem lies, therefore, is that these expectations represent conflicting commitments, which can never be simply (nor ever finally) resolved. For so long as we look for a single answer, we are forced to privilege one of the two conventional expectations; and no matter which we choose to emphasize, neither the dependence on or independence of recorded history, there are documented historical novels aplenty to contradict the choice. Almost inevitably so: not only does a tension exist within the terms of the genre, but there is a further complication built into the way a novelist can work either within or against these conventional expectations. Thus to argue that documented historical novels are or are not valuable because they do or do not stick to the "facts" amounts to something of a false problem.

Instead of trying to decide once and for all about the question of fidelity to history, then, we need to consider the status of historical events in fiction. This issue, which has occupied the attention of theorists since Aristotle, has frequently been resolved by arguing that historical events are transformed the very moment they take their place in the poetic fiction. Thus, as Aristotle formulates it, the poet can take his subject from history, but in doing so he "is none the less a poet for that; since some historic occurrences may very well be in the probable and possible order of things; and it is in that aspect of them [rather than in their 'actual' aspect] that he is their poet." Aristotle insists here on the autonomy of all fictions, even historical ones. Murray Krieger, in commenting on the Poetics, explains the point this way:

Every element taken from [historical] reality—an incident, a character, an idea, even words and their normal meanings—must be newly justified by the role it must play in that closed teleological pattern. As a result, that element must change its meaning, indeed its very nature and ontological status, by virtue of these interlocked functions. ["Fiction, History, and Empirical Reality," Critical Inquiry, I (1974)]

Thus the responsibility of the documented historical novelist, according to this view, would be not to history but to the teleology of his poetic fiction. It would follow, then, that even documented historical novels are discontinuous from history and thus no different than disguised or invented historical novels.

The argument for autonomy, in short, provides a way of securing the distinction between narrative histories and documented historical novels, but at the cost of collapsing the one between documented and other types of historical fiction. Far more debilitating, however, is the way that Krieger's argument minimizes the dynamics of aesthetic response: while it may well be that every historical character and event must be "newly justified by the role it must play in that closed teleological pattern," there is still a resistance to that very process built into the distinct ontological status of the historical material. Consider in this light a slight variation on L.C. Knights' argument for aesthetic autonomy, the proverbial "How many children had Lady Macbeth?" Now in an invented or disguised historical novel, Knights' premise would hold: Faulkner, for example, can tell us that Rosa Coldfield never bore children, and there is the end to the question. But if a novelist were to write about Queen Elizabeth, the reader would expect from the very first that Elizabeth bore no legitimate children and that the crown was passed on to James I. It is precisely this sort of historical expectation that the argument for autonomy, with its synchronic formal analysis, does not take into account; and it is precisely along these lines that we can describe the differences between documented historical novels and other fiction, even other historical fiction. Faulkner could just as easily have decided to arrange for Rosa to marry Sutpen and bear his coveted male heir. That would be quite a different novel from Absalom, Absalom!, to be sure, but no one would know what they were missing. In a novel about Queen Elizabeth, by contrast, the rhetorical situation is not the same: the reader expects certain things to happen and will no doubt be surprised—perhaps even disappointed—if they do not. Now that is not to rule out the possibility—it should be stressed—that a novelist might create, and convince his readers to accept, a story about Queen Elizabeth's hitherto unknown illegitimate son. Rather, it suggests only there may well be resistance on the part of some readers to tampering with recorded history. Mary Renault explains why: "One can at least desire the truth; and it is inconceivable to me how anyone can decide deliberately to betray it; to alter some fact which was central to the life of a real human being, however long it is since he ceased to live, in order to make a smoother story, or to exploit him as propaganda for some cause." Whereas Renault is committed to historical accuracy as a fundamental requirement of good historical fiction, however, I wish only to insist on it as a conventional expectation. The point, then, is not to try to legislate how readers should react to a novelist's divergence from recorded history; it is, rather, simply to assert that the reader will in all probability react, and that this reaction—whether it leads to insight or disappointment—is different from any provoked in invented or even disguised historical novels. For only by dealing directly with historical events does the novelist create the expectation that he will follow what actually happened; and only when this expectation exists does compliance with (or divergence from) recorded history take on internal significance.

Insofar as he generates historical expectations, then, the documented historical novelist is in a position similar to the historian's; but because he is writing fiction, and not history, the documented historical novelist operates within a different set of generic conventions. Crudely put, the difference is that the historian is supposed to restrict himself to historical events, while the documented historical novelist need not. The novelist is free, more so than the historian, to fill in with imagined details the gaps in recorded history. Admittedly, this distinction is basically a matter of degree: historians have never been prohibited from speculating, nor have they even been prevented from inventing characters (usually to stand for representative types). Nevertheless, the historian conventionally must identify his speculations and inventions, refer them to what documents exist, and defend them against previous interpretations. The documented historical novelist, by contrast, usually slips in his inventions and speculations unannounced. The reason for this, as we shall see, is that the novelist conventionally does not wish to call attention to himself or to the fact that his "history" is part fiction. For the moment, though, the important point is that the boundary between documented historical novels and narrative history needs to be drawn along the lines of generic conventions, and not in terms of Aristotle's universality and particularity (which so much of the previous criticism on the genre has invoked).

Once having recognized the importance of conventions and expectations in documented historical novels, more-over, we find that a critical approach to all of historical fiction immediately presents itself: namely, reader-response criticism. In contrast to the synchronic formalist model, a theory of aesthetic response obviates the need to define in advance what history is or how a novel should relate to it; instead, it allows us to follow the way the implied reader adjusts his conception of history to the one the novelist is creating. Its focus on the actions involved in responding to a literary text is thus ideally suited to the process of expectation and revision that is engaged when novelists write about the past. Furthermore, it should also be clear that the number and kinds of historical expectations increase as we move along the theoretical continuum from invented, through disguised, to documented historical novels. Aside from the idea that each of the three types generates historical expectations differently, all that remains to be observed is that it is possible, even within a single kind, for a novelist to bring generic expectations to bear on historical ones in a great many ways.

Before proceeding further, then, we need a general sketch of the respective generic conventions. Invented historical novels exhibit very little that is unique. Generally, they rely on novelistic forms that can be loosely termed realistic—avoiding, that is, anything that might draw attention to the fact that they are not history but fiction. Thus invented historical novels avoid any self-conscious reflections on their ontological status; instead, the narrator comports himself as a historian, pretending that his statements refer to an extra-textual reality. Thus whatever historical expectations the reader brings to such texts are fairly generalized; since they can only be applied indirectly to these fictitious characters and events, moveover, the invented historical novel retains much, if not all, of the conventional autonomy of fiction.

With disguised historical novels, the same preference for realistic forms presents itself. The same conventions hold therefore, with one addition: the reader is expected to grant the novelist's right to disguise history and then do with it what he will. The critical difficulty in analyzing this kind of historical novel, however, has less to do with defining its conventions than with explaining the double articulation of the text—its autonomy and its historical referents—and the resulting double role of the reader. In principle, at least, the effect of the generic conventions would be to inhibit historical expectations. In a situation where anything is possible reader expectations tend to be minimal or non-existent; and since the disguise signals the author's privilege to change history in any way he should choose, it becomes rather difficult for the reader to know in advance what will happen. As time goes on, of course, the reader may begin to discern certain patterns or tendencies in the disguise; and from this interpretive process, he may even begin to develop assurances about what to expect. But the reader can never be certain if, or when, the pattern might be broken. Moreover, he really has no grounds on which to object to any of the changes, no reason for being disappointed in the way that one might be disappointed by a documented historical novel that recounted Custer's victory at Little Bighorn. Perhaps, then, we should posit a certain passivity on the part of readers of disguised historical novels, less an anticipating of the disguise than a waiting to see how it is developed. We might say, in other words, that in conventional examples of this type of historical fiction the reader interprets the referential component of the novel retrospectively. Having followed the story of Willie Stark, as it were, we can superimpose it on what we know of Huey Long; but we almost have to wait until we have gotten Willie's story before we can make the comparison. Granted, some of this goes on all along, and the reader does have expectations generated within the terms of the fiction itself. So we might, from this perspective, conceive of the reader being engaged in a process of constantly measuring these internal and forward-looking glances against what he knows to have actually happened. But even in this more complicated model, the reader is operating on two distinct levels—the historical and the fictional; he is in fact reinforcing his sense of the separation. Thus it is not until the end of the book that the two can ever quite be brought together with any assurity. All the while, moreover, the awareness of the disguise is focusing the reader's attention on the design of the novel itself. Thus we might say that the disguise points more to the closed teleological pattern of the novel than it does to history. And to this extent, it could be argued, disguised historical fiction maintains a conventional autonomy.

Beyond these tentative hypotheses, however, we would do well to stick to individual cases, noting only that a position of this kind on the theoretical continuum makes one of two distinct rhetorical strategies likely. There are, in other words, some disguised historical novels—All the King's Men is a good example—that are written in much the same way as invented historical novels. Here, although the disguise may be obvious, it is never insisted upon. The novel proceeds as if its characters and events were entirely fictitious. In direct contrast, there are some disguised historical novels—E. L. Doctorow's The Book of Daniel is a good example—that self-consciously flaunt the fact of their double status, that force the reader constantly to engage the disguise. A brief comparison of the two novels should clarify this distinction.

Doctorow's disguised account of the Rosenberg trial and its aftermath presumes the same generic conventions than operate in Warren's novel. In The Book of Daniel Susan Isaacson loses her parents when they are electrocuted for conspiring against the United States. She eventually responds to this childhood trauma by attempting suicide and then willing her death. Now the generic conventions prevent us from objecting to this narrative premise on the grounds that it does not comport with what actually happened. To this extent, and it is no small matter, The Book of Daniel shares the same autonomy as All the King's Men. The way that Susan's story sheds light on the experience of the Rosenbergs, therefore, is oblique and retrospective: having been convinced by Doctorow's fiction that Susan's suicide is inevitable, we are left to wonder about the fact that the Rosenberg children did not respond in the same way.

Despite this important resemblance, however, the generic status of The Book of Daniel is far more problematical than that of All the King's Men. At the same time as the disguise preserves the autonomy of the novel, Doctorow tries to generate the kinds of historical expectations that are created when a novelist deals directly with recorded history. Rather than invented historical fiction, which is what Warren's novel resembles, Doctorow's novel keeps moving in the direction of documented historical fiction, as evidenced by the way Doctorow manipulates his narrator, Susan's brother Daniel. Notice, for example, the elaborate game of hide-and-seek that Daniel plays with his implied reader at the beginning of the novel. At first he uses his adopted name, Daniel Lewin, and for the first twenty pages he enjoys his anonymity, even going so far as to entertain himself with ironies that only he can appreciate. In this process of toying with the reader, of course, Daniel is dropping clues to his public identity, which he reveals before too long by recounting his appearance at a rally in New York to free the Isaacsons. Daniel insures that his reader makes the identification by self-consciously referring to the game that he has been playing:

Oh, baby, you know it now. We done played enough games for you, ain't we. You a smart lil fucker. You know where it's at now, don' you big daddy. You got the picture. This the story of a fucking, right? You pullin' out yo lit-er-ary map, mutha? You know where we goin', right muthafuck? (My emphasis)

The shift in tone and diction that occurs here points to an emotional involvement in his story that has been controlled up to this point in the novel. For Daniel to display his aggression and outrage is itself striking; that he should turn it against his audience, by mocking the reader's sense of discovery, is altogether disconcerting.

Even more importantly, however, Daniel's self-conscious reference to his public identity brings the fictional disguise into play in a radical fashion. Within the terms of Daniel's story, Paul and Rochelle Isaacson are actual historical characters. Thus when Daniel finally identifies his parents, he can be sure that his imagined readers will "know where we goin'." But for the actual reader of The Book of Daniel, as opposed to the one Daniel pretends to address, this passage elicits a more complicated response. Since the Isaacsons never really existed before the novel was written, the reader must first make the connections to the Rosenbergs before he can make any sense out of Daniel's remarks. Without the identification there can be no expectations. Thus to the extent that Daniel believes his reader already knows his story, the actual reader must interpret the novel on two levels: for if he is to occupy the same position as the audience that Daniel addresses, he must continually translate, even as he willingly accepts, the fictional disguise that stands between him and the history that Daniel purports to be recounting.

Therein lies Doctorow's unconventional strategy. Unlike All the King's Men, which never depends on the existence of Huey Long for its meaning or its effect, The Book of Daniel cannot be read without taking the Rosenbergs into account. Doctorow insists, in short, that we do and do not accept the disguise. Clearly, the risks are considerable, for he is threatening the very nature of the generic contract. No matter how hard he pushes the disguised historical novel in the direction of documented history, in fact, there is a resistance to that transformation built into the conventions of this fictional kind. And besides, there are places where he needs our acceptance of the disguise. Thus Doctorow all but faces the same rhetorical problem as the little boy who cried "wolf" one too many times: if he expects the reader to accept the disguise in some places, then he had better not subvert it at others, and vice versa. That is not to say, however, that the novel fails; the self-consciousness with which Doctorow points to the problem anticipates, and partially defuses, the reader's objections. But it does suggest the conventional affinities between disguised and invented historical novels, as well as account for much of the confusion expressed by reviewers of the novel. In addition, it highlights the potential impact of playing historical expectations off against generic conventions, since whatever its flaws The Book of Daniel is a powerful novel and at least part of its success comes from the rhetorical risks that Doctorow takes.

Turning now to the conventions of documented historical fiction, we find the same basic principles operating. Most of the generic conventions are designed to preserve the autonomy of the fiction. Thus, as we have said, the novelist is traditionally allowed greater freedom than the historian to speculate in order to create what history has failed to provide, an unbroken record of the past. At the same time, the novelist is not required to disturb his illusion by identifying his inventions. More specifically, however, we find a number of strategies designed to help circumvent the problems created by the distinct ontological status of historical and fictional events. More than relying on realistic forms to lend their fictions the illusion of historicity, for example, documented historical novelists frequently center the narrative on their own invented characters. Observing this pattern in the novels of Sir Walter Scott, Georg Lukács even raises it to the level of a critical precept, arguing that world-historical figures are not an appropriate subject for the novelist and so should be kept in the background of the fiction. Now Lukács' reasons for insisting on this "classical form of the historical novel" have as much to do with his Marxist interpretation of history as they do with any inherently formal limitations or requirements of the genre. But even without broaching the issue of Marxist historiography, we might quarrel with Lukács' restrictive definition on the grounds that there are, quite simply, many good historical novels that take major historical figures as their prime subject (George Garrett's Death of the Fox, for example). Nevertheless, Lukács has drawn our attention to an important point: that the subordination of history to fiction is a frequent strategy, since it has the advantage of freeing the novelist to do what he presumably does best—invent. This movement toward invented historical fiction permits the novelist, in effect, to recover much of the autonomy that he has sacrificed to history. But even with a preponderance of invention, the problem of handling the interaction of history and fiction persists. Indeed, to some extent it is actually exacerbated by the compartmentalizing strategy. The farther you keep actual and fictitious characters apart, that is, the more you remind the reader of the difference, and the more you deprive yourself of one of the great advantages of documented historical fiction—the tendency for the novelist's inventions to accrue historicity from their very proximity to historical events. Once history and fiction are rigidly separated, moreover, there are limitations imposed on the way they can interact: generally speaking, at least, history must invade the fiction rather than the reverse if the illusion of historicity is to be preserved.

This elaboration of the basic conventions and strategies could be extended, but the underlying principle is clear: if his fiction is to carry the weight of history, the novelist does well to distract any attention from himself or the artifice he has created, to gesture through his text to the past he seeks to recapture. In recent years, however, various writers of historical fiction—such as John Barth, Thomas Berger, and Ishmael Reed—have taken the opposite tack, creating what might be called comic historical fiction (by which I mean, less that they engage in humorous treatment of certain historical materials, although they do, than that they create their comic perspectives on history by poking fun at these generic conventions, playing delightful variations on the interaction of their fictions with history, and generally flaunting the inescapable artifice of their creations). This trend toward self-reflexive historical fiction merits careful attention: its connections with both conventional historical novels and other recent self-conscious fiction are historically and artistically significant. But rather than broaching that enormous subject here, I should like to take a brief look at one of the more conventional examples of recent historical fiction, William Styron's The Confessions of Nat Turner, if only because any satisfactory approach to the genre should be able to bring some coherence, even if not consensus, to the Nat Turner controversy.

Certainly, a great many issues are brought into focus by Styron's novel, ranging anywhere from disputations of "fact" to the ideological content of historical fiction. Just as certainly, the controversy itself merits critical and historical analysis, here again ranging anywhere from the details of a particular interpretation to reflections on the political dynamics of culture. Not only are all these matters worth pursuing, moreover, but they suggest something of the complexities of literary, historical, and cultural analysis required by many documented historical novels. I shall have to restrict myself here, however, to the question of artistic value, particularly as it relates to Styron's handling of generic conventions. Let us begin, then, with Styron's own conception of historical fiction, which he boldly sets forth in his prefatory note (and which he was later to defend with appeals to the authority of Lukács). As Styron conceives it, the virtue of historical fiction resides in the novelist's conventional freedom to speculate and invent:

During the narrative that follows I have rarely departed from the known facts about Nat Turner and the revolt of which he was the leader. However, in those areas where there is little knowledge in regard to Nat, his early life, and the motivations for the revolt (and such knowledge is lacking most of the time), I have allowed myself the utmost freedom of imagination in reconstructing events—yet I trust remaining within the bounds of what meager enlightenment history has left us about the institution of slavery.[The Confessions of Nat Turner, 1967]

Though we might well question whether the enlightenment that history affords is as "meager" as Styron believes it to be, I shall leave that argument to the professional historian. I only wish to pursue the artistic consequences of Styron's decision to speculate "in those areas where there is little knowledge."

In itself, Styron's strategy is neither unusual nor problematical; the conventions of the documented historical novel clearly countenance it, and the practice of a great many historical novelists makes it even predictable. But we might ask whether Styron has taken that conventional freedom too far. The entire credibility of the novel, after all, hinges on the illusion that it is the actual Nat Turner who speaks to us. Putting aside the stylistic problems this creates, there are still extraordinary difficulties built into the attempt to know the mind of a historical character. Much has been said about the daring—some would say foolhardiness, or worse—of a white author attempting to render the thoughts of a black revolutionary. Perhaps more needs to be said about the existentialist perspective that Styron brings to Nat Turner's religious convictions. But even after the opinions from both sides are in (and notice how much depends on the critic's own conception of what a "true" account would be), we still need to examine Styron's formal solutions to the rhetorical difficulties presented by his subject.

Styron thinks, to be sure, of The Confessions of Nat Turner as an unconventional historical novel; he calls it a "meditation on history" as opposed, presumably, to a simple chronicle of events. But what he has given us is the height of conventionality, a documented historical novel that not only presumes almost unlimited freedom to speculate, but one that asks the reader to accept without question the pseudo-historicity of its autobiographical form. The critical issue, therefore, is not whether Styron is entitled to speculate and invent, but whether the form he has chosen for the novel allows him to reflect on his own historical reconstruction. And it must be said, I would contend, that it does not. While the novel is clearly a meditation—in the sense that it is Styron's imaginative re-creation—it is not presented in the form of a meditation. In fact, any time the reader becomes aware that it is Styron's meditation the novel's necessary illusion has slipped, for it asks us to believe that Nat is doing the telling. Now precisely how well Styron succeeds in creating or maintaining that illusion has been, and will no doubt continue to be, endlessly disputed. Whether it convinces or not, however, the strategy of the first-person narrative builds into the novel an unresolvable contradiction: for if it is Nat Turner's "Confessions," then it cannot be Styron's "meditation," and vice versa. The difficulty, therefore, goes beyond the inevitable problem of convincing the reader that Nat Turner is speaking, to a deeper irony: that to succeed, Styron must efface himself completely, forfeit any attempt at "meditation."

My point, in short, is not that Styron should have fettered his imagination, given up any hope of penetrating the veil of mystery that stands between him and this particular past. On the contrary, the historical novelist is obligated to exercise his imagination to the full. But the more one's historical subject stands shrouded in mystery, the greater one's responsibility to create a formal correlative for the necessary expense of imagination. There is, to be sure, a little of this in Styron's Confessions; and I do not wish to underestimate his achievement. Clearly, he has created a striking, albeit controversial, image of the conditions of slavery. Nevertheless, Styron's own essay, "This Quiet Dust," is a far superior "meditation" on that same history. In that autobiographical sketch he wrestles with the difficulty, yet necessity, of recovering this past; captures the relevance, as well as many of the ambivalences and complications, of his own involvement in the historical recreation; and explores the significance of both. Yet such selfconscious reflections run directly counter to the heavily conventional strategies and design of his own novel. It is almost as if Styron has been seduced by the very lack of historical documentation, betrayed by the very freedom conventionally allowed the novelist. Whatever the cause, however, what is missing from The Confessions of Nat Turner is any formal representation of the limits of historical knowledge, any acknowledgment of his own wrestlings with the recalcitrance of history.

To judge Styron's Confessions lacking in this respect and therefore seriously flawed does not, I think, impose an irrelevant criterion on the novel, nor does it amount to a reversion to the idea that the novelist is bound by existing historical evidence. Rather, it implies only that historical novelists—particularly those who re-create an actual past—invite historical judgments; and further that one of the primary concerns in such evaluations revolves around the novelist's conception of his own relationship to the past. In much the same way, therefore, as Hegel distinguished among three classes of historical consciousness (Original, Reflective, and Philosophical), we need to draw similar divisions among historical novelists: those who write in the Original mode, where the principal concern is to create a compelling picture of the past—history primarily in itself; those who write in the Reflective mode, where the chasm between past and present is recognized only to be bridged—history in and for itself; and those who write in the Philosophical mode, where the primary concern becomes how, or if, history itself is possible—history in and for, but primarily about itself

This new set of categories attributes value, to be sure; there is no avoiding it, nor would we want to. The best historical fiction, in my view, is ultimately about itself, about the meaning and making of history, about man's fate to live in history and his attempt to live in awareness of it. While I would be willing to defend that view, I have sought to avoid imposing it on the theoretical problems of definition and methodology. There are many good historical novels (and Styron's is only one of them) that either do not achieve, or more often never even aspire to, the Philosophical mode. There are even—lest it seem that I have unwittingly privileged self-consciousness—a good many self-reflexive historical novels (Thomas Berger's Little Big Man, for example) that remain almost entirely within the Reflective mode. There are, for that matter, novels in the Philosophical mode (George Garrett's Death of the Fox, for instance) that share much of Styron's commitment to imaginative invention, while resisting most of the comedy of Barth's self-reflection in The SotWeed Factor. It would be a mistake, then, to define the genre in such a way as to privilege the Philosophical mode, and I would stress that the initial categories of analysis (invented, disguised, and documented historical novels) are valueneutral: whatever sense there may be of a progression along the continuum relates only to the increasingly distinctive interpretive problems that arise as we move closer to the boundary with narrative history. We should be wary (as much as it is possible), moreover, about confusing the value of a novel with the amount of analytical criticism that it requires or the specifically theoretical issues that it raises. It should be observed, then, that the Hegelian categories of historical consciousness just introduced do not align themselves with the purely descriptive generic categories that I proposed at the start. Faulkner's Absalom, Absalom! is as much in the Philosophical mode as John Barth's The Sot-Weed Factor, though the first is invented and the second documented. Nor is any absolute hierarchy of value intended: there is no reason why we cannot appreciate novels in the Original or Reflective modes for what they have to offer.

Finally, it should be noted that these two sets of categories, and my discussion, have touched only briefly, if at all, on other important issues. There are, for example, historical problems to be faced when dealing with nineteenth-century historical fiction. I have side-stepped the issue here by drawing my examples from modern, predominantly contemporary, fiction. But there is more than enough flexibility within an aesthetics of reader response to accommodate the historically conditioned shifts in conceptions of history and the novelist's relationship to it. Still other questions remain—the novelist's ideological commitments, his affinity for a particular type of historical explanation, his ideas on the nature or course of history, his conception of the relevance of his work. Depending on our interpretive concerns, therefore, other discriminations, along somewhat different lines, would need to be made. In particular, if we wished to pursue primarily historiographical issues (such as the role of narrative in historical explanations, for example, as in The Sot-Weed Factor), we would find not only that thematic categories are far more appropriate, but that they cut across the original differentiations into kinds. The necessity for the categories of invented, disguised, and documented historical novels, therefore, is exclusively interpretive, a matter of differing expectations and conventions involved in our reading. Similarly, the real advantage of the categories derived from Hegel is evaluative: that they point to what I take to be the central issue, the meaning and making of history. And while metahistory is by no means all there is to history, nor all that there is to historical fiction, it remains the subject that has engaged all the great historians. A historical novelist—even a literary critic in search of definition and methodology—could do much worse.