Sherlock Holmes Was Wrong

by Pierre Bayard

  * A list of Holmes’s failures in stories not narrated by Watson figures at the beginning of “The Problem of Thor Bridge.”

  * See also on this point Umberto Eco’s analyses in The Limits of Interpretation, Bloomington: Indiana University Press, 1994.

  Counter-Investigation

  I

  What is Detective Criticism?

  TO HIGHLIGHT THE SORT of problem that the Sherlock Holmes method exemplifies, I created a dozen or so years ago my own method of investigation, to which I gave the name detective criticism. The aim of this method is to be more rigorous than even the detectives in literature and the writers who create them, and thus to work out solutions that are more satisfying to the soul. Before applying this method to the most famous of Conan Doyle’s novels, I would like to give a brief presentation of it, recalling the circumstances of its creation and explaining its principles.

  The first inspiration for detective criticism was Sophocles’ Oedipus Rex, or more precisely my reading of texts by American critics—such as Sandor Goodhart and Shoshana Felman—who cast doubt on the traditional version of the murder of Laius by Oedipus. Studying the contradictions in Sophocles’ text, the two authors, who drew inspiration from Voltaire’s ironic remarks on the plausibility of the plot, came to the conclusion that it was never actually proven that Oedipus was guilty of the crime of which he finally accuses himself.*

  One of the problematic details is the number of Laius’s attackers. The only witness to the murder, a servant of the king’s, declared that his master had been killed by several people; he never changed his testimony. Yet once Oedipus has convinced himself of his own guilt, the witness is not brought back for questioning, although his testimony flagrantly contradicts the results of Oedipus’s investigation. A strange oversight, and one that encourages all sorts of suppositions—including that the accused Oedipus is innocent.

  We can imagine all the consequences that an innocent Oedipus might entail. To take just one example, one of the most important theories of our era, psychoanalysis, is largely based on this ancient myth, and upon the conviction that Oedipus killed his father. The theory elaborated by Freud would of course not collapse if the Greek hero were proved innocent, but it would not emerge entirely unscathed, either. While some specialists in mythology such as Jean-Pierre Vernant have long expressed doubts about the “Oedipal” nature of the criminal act—a term that depends on an anachronism—the reinterpretation offered by Goodhart and Felman goes farther: it calls into question the very existence of the crime itself.

  These studies were a revelation for me. My only reservation was that they had opened a path of research but did not go far enough. The American critics were content to point out the improbabilities in Sophocles’ text, and went only as far as suggesting that Oedipus did not necessarily kill Laius. So their approach was simply negative. But they didn’t undertake to go on to the obvious next step, the constructive one: to solve the mystery disclosed by their reading by unmasking the real murderer.

  So the main difference that separates detective criticism not only from other studies based on investigation, but also from the rest of literary criticism is its interventionism. While other methods are usually content to comment on texts in a passive way, whatever scandals those texts might present, detective criticism intervenes in an active way, refusing to go along. It is not content with pointing out the weaknesses in texts and casting doubt on presumptive murderers; it boldly risks any number of consequences by actually looking for the true criminals.

  The main premise of detective criticism is this: many of the murders narrated in literature were not committed by the people accused by the text. In literature as in life, the true criminals often elude the investigators and allow secondary characters to be accused and condemned. In its passion for justice, detective criticism commits itself to rediscovering the truth. If it is unable to arrest the guilty parties, it can at least clear the names of the innocent.

  Having arrived at this theoretical premise, it made sense to flesh it out. Agatha Christie offered a favorable field for research, both because of her eminence in the realm of the detective story and also because of the literary quality of her work. It would have been all too easy to show that unpunished criminals lurk inside books not originally conceived of as detective novels.

  To make the demonstration even more convincing, I decided to work on Christie’s book The Murder of Roger Ackroyd, regarded as a masterpiece of precision.* This novel draws its celebrity from the fact that the murderer is the narrator. Through entries in his journal, the narrator,Dr. Sheppard, tells how he gets involved with the investigation that the detective Hercule Poirot is conducting into the murder of the village squire, Roger Ackroyd. But, according to Poirot, the doctor leaves out of his story the fact that he himself is the criminal—that it was Sheppard who killed Ackroyd, to keep him from publicly accusing Sheppard of being a blackmailer. In the final pages of the book, the triumphant Poirot turns to Sheppard, accuses him of committing the murder, and encourages him to commit suicide.

  The nice idea of having an investigation narrated by the murderer himself ensured the book’s fame, but at the same time ignored a whole series of concrete problems. I won’t go over all the contradictions in the text here, but no serious investigator today can blithely accept Hercule Poirot’s conclusions. In short, the ingenuity of the narrative device distracted readers from the only question that matters for detective criticism—one possibly more prosaic than reflections on literary effectiveness, but more in keeping with ethics: Who in fact did kill Roger Ackroyd?

  To take a simple example of the problems the text presents, the presumed murderer, Dr. Sheppard, in order to give himself an alibi, is supposed to have used a dictaphone that can start automatically. Like a radio alarm clock, the apparatus, planted in the victim’s office, can be set to turn itself on and play a recording of the victim’s voice. His voice will be heard by the other occupants of the house after Dr. Sheppard has left the scene, thereby proving his innocence. When he arrives at the house later, summoned by the victim’s butler, Dr. Sheppard is supposed to have discreetly whisked away the apparatus used to give him an alibi.

  Aside from the fact that this audio device is never found by anyone—which is a regrettable quality in a piece of evidence—Poirot’s reasoning comes up against a material impossibility. Preparing such a sophisticated device takes time—especially in 1926, when such a thing would have been technologically very advanced—yet it’s only on the morning of the murder that Sheppard learns that Ackroyd is preparing to accuse him of blackmail. He has just a few hours to get things ready. We have his precise schedule for the day, corroborated by several witnesses, and it reveals no period of time sufficient for him to make such an apparatus. So when is Sheppard supposed to have put it together?*

  An unlikelihood of this kind—and it is far from being the only one in Poirot’s solution—casts doubt on the guilt of the presumed murderer. But I wasn’t content to question our detective’s conclusions. I intended, by reopening the case, to look for the real criminal. Although it is difficult to accuse someone with certainty after so many years, the cluster of clues I gathered together leads ineluctably to one single person, whose name I revealed in the final pages of my book, thus carrying out the additional step beyond the work of those critics who had voiced doubts about the guilt of Oedipus.

  After this first attempt, which was crowned with success but which dealt with a book in which literary criticism took little interest, it made sense to see if such a method could be just as effective with a masterpiece of world literature, studied by many specialists. Shakespeare’s most famous play, Hamlet, perfectly suited this project, not least because it has the structure of a detective story, with the main character leading the investigation to clarify the circumstances round the death of his father, whose ghost has pleaded with him for vengeance.*

  The case presented itself in an entirely different way from The Murder of Roger Ackroyd. Not only
has the play given rise to a considerable number of critical interpretations, but others had already pointed out the improbabilities contained in the argument that has held sway for centuries: that the victim’s brother Claudius, who quickly married his brother’s widow and took his throne, is the murderer.

  To cite only one of the most obvious improbabilities, the famous performance of the play-within-a-play poses an insoluble problem. As we know, Hamlet, the victim’s son, convinced of his uncle Claudius’s guilt, prepares a trap for him by asking a troupe of traveling actors to play the murder scene in front of the presumed murderer, to see how he will react. Claudius reacts as Hamlet expects he will, which tend to support the argument of his guilt. Upon seeing the exact circumstances of the murder reenacted—the murderer supposedly pouring poison into his sleeping victim’s ear—Claudius, annoyed, abruptly leaves the hall.

  It wasn’t until the beginning of the twentieth century and the arrival of a reader—the critic Walter Wilson Greg—who was more attentive than the others, that this version of the facts was faulted. Greg recalls that in Shakespeare’s time plays were often preceded by a “dumbshow,” a pantomime during which the actors silently played out the important moments in the play. And that is indeed the case with our play, put on by the itinerant actors; we are told explicitly that it is preceded by a dumbshow representing for the first time a murder by poisoning.

  Now we can see the problem this presents, and we wonder why it took centuries to see it. If Claudius is indeed his brother’s murderer, how do we explain that he remains seated unperturbed during the first representation of the murder, but gets up, furious, at the second? The hypotheses advanced by Shakespearians—the notion, for instance, that his annoyance grew on him progressively—are hardly convincing. In any case they leave open the other hypothesis: that Claudius does not react to the first representation of the murder because he is innocent, and walks out on the second because he is annoyed by the clamor Hamlet is making in the hall—noise that is clearly indicated in the stage directions.

  Once we can vanquish our internal hesitations and accept the hypothesis of Claudius’s innocence, we must resolve to reopen the file—to ponder whether there exists in Shakespeare’s play another suspect who might have committed the murder. That is precisely what I did in Enquête sur Hamlet, arriving at a different conclusion from the one the play suggests and the majority of Shakespeare scholars confirm. The adoption of this conclusion, when it is finally accepted by Shakespearian criticism, should significantly change performances of Hamlet.

  What can we say about these declarations I’ve just made, that Oedipus, Dr. Sheppard, and Claudius are innocent of the crimes of which they are accused? On the face of it these assertions are wrong, since they’re not in keeping with what the books seem to say. But things are not so simple. That apparent barrier against delirium, textual closure—the notion that a text includes only a limited number of readings*—is a material closure, but not necessarily a subjective closure. How shall we understand this?

  It is important first of all to stress that separating a true statement (“Hamlet is the nephew of Claudius”) from a false statement (“Hamlet is the brother of Ophelia”) is easy to practice but tends to produce unoriginal readings, content to repeat what the text says in more or less similar terms. Without even going as far as interpretative criticism, even the least ambitious psychological analysis quickly goes beyond strict written statements, surmising things that the text might possibly encourage but does not strictly speaking authorize. In short, to keep exclusively to what the text says risks leading to readings that are unarguable but also uninteresting.

  Above all, the world that the literary text produces is an incomplete world, even if some works offer more complete worlds than others. It would be more correct to speak of heterogeneous fragments of worlds, made up of parts of characters and dialogues that are never joined together into a coherent whole. And—an essential point—these weaknesses in the world of the work do not stem from a lack of information, one that studious research, as in the field of history, might hope one day to fill, but from a lack of structure: in other words, this world does not suffer from a lost completeness; it was never complete. What we are dealing with in literature is a gapped universe.

  This incompleteness is especially striking when it comes to descriptions, which enclose some possibilities but leave many others open to the imagination. The remark was made long ago that written descriptions, compared to figurative painting or cinema, leave much more room for the inventiveness of the reader—often regarded as an advantage for literature.

  Every story, furthermore, leaves to the reader’s imagination vast spaces of narrative, in the form of direct or indirect ellipses. In principle the reader does not have to worry about what is going on in these virgin spaces of the story. But just as with descriptions, it is hardly likely that he won’t be tempted to fill them, especially when the text mysteriously alludes to absent events.

  To these descriptive and narrative incompletenesses a third gap should be added, which concerns character. A great number of elements in the characters’ lives, both psychic and factual, are not communicated to us. This uncertainty is closely linked to an essential point that will be discussed later on regarding the special mode of existence of literary characters. These characters, I believe, enjoy a much greater autonomy than we usually think, and are able to take initiatives unknown both to the writer and the reader. When characters have their own will, their own autonomy, it gives the literary universe a greater internal mobility; it also makes the texts through which we view this world all the more open and incomplete.

  This incompleteness of the written world is not absolute, however. It is restricted by the intervention of the reader. The reader in effect comes to fill in, at least partially, the rifts in the text. This work of completion—or, if you prefer, of subjective closure—functions just as well for descriptions as for the ellipses in the thoughts or actions of the characters. It is more or less precise and conscious, depending on the reader, but it always takes place. And once a reader has found his own subjective closure of a work, he will find it impossible, past the level of superficial agreement, to truly communicate with other readers of the same book—precisely because they are talking about the same book.

  Because of this work of completion, it is in fact utopian to think that any objective, or even shared, text exists, onto which different readers could project themselves. Even if this text existed it would unfortunately be impossible to reach it without passing through the prism of subjectivity. It is the reader who comes to complete the work and to close, albeit temporarily, the world that it opens, and the reader does this in a different way every time.

  This subjective incompleteness of the world in the work encourages us to suppose that there exists around each work, produced by the limited nature of statements and the impossibility of increasing the quantity of available information, a whole intermediate world—part of which is conscious and another part unconscious—that the reader develops by inferences so that the work, completed, can attain autonomy: a different world, a space with its own laws, more fluid and more personal than the text itself, but indispensable if the text is to achieve, in the limitless series of its encounters with the reader, a minimal coherence.

  To admit the existence of these many intermediate worlds orbiting literary works has obvious risks: that we may be misled into keeping the expansion going ad infinitum, giving unknown lovers to the Princesse de Clèves or making her die of poisoning. But it is difficult to do otherwise. Beyond the good faith of protagonist or narrator (who can, as in Agatha Christie’s novel or Shakespeare’s play, be caught in the act), the hypothesis of detective criticism is that the writer himself is often misled. His work, in fact, necessarily escapes him, since, incomplete, it closes itself at every reading in ever different ways.

  If we accept this hypothesis, then there exists around the written world opened by the work a multitude of other possible
worlds, which we can complete by means of our images and our words. Denying oneself this work of completion in the name of some hypothetical fidelity to the work is bound to fail: we can indeed reject filling these gaps in a conscious way, but we cannot prevent our unconscious from finishing the work, according to its own priorities and those of the era in which it was written.

  Since this work of completion is inevitable, one might as well do it with as much rigor as possible. For the virtual intermediate worlds in a given work, as numerous as they are, are not strictly equivalent to each other; it is possible to classify them according to their credibility, both on an individual and a collective level. To start with the individual, it is obvious that the operation of completing a literary text will be carried out differently according to the sensibility of each reader, and, in the field of the detective novel, according to the reader’s conception of criminals and crime.

  But the possible worlds vary also in accordance with the times, their conception of criticism, and the evolution of scientific research. As the years go by, our reading of a given work changes; today, we have grown sensitive to certain details of the text that strike our modernity and can lead us, according to the type of completion we bring to bear, to renewed access to the text.

  Thus we could say that the question of the guilt of Oedipus, Sheppard, and Claudius is not intrinsic, but is posed anew for each reader, in the framework of what I have called in my book on Hamlet an inner paradigm—that is, the unique way in which each of us portrays the world and confronts reality, based on the questions posed by our own time.

  Within these personal paradigms, one’s rigorous investigations may unfold with some chance of success. And through them, a fragile form of truth, profoundly anchored in one of those intermediate worlds that extend and complete the text, can hope to come to light for a while.