For much of my life, I have worked on frontiers. Night, fog, armed guards, tension. Walk just a few paces down the snow-covered Friedrichstrasse in Berlin, through a musty East German checkpoint, and you move from a world called West to a world called East. Nothing changes, and everything changes. Or a sandbagged border post between Milosevic’s Serbia and liberated Kosovo: fresh-faced Canadian soldiers pass you tenderly from one darkness to another. But also—and sometimes almost as tense—the frontiers between politics and culture, between continental Europe and the Anglosphere, between academia and journalism, left and right, history and reportage.

I like crossing frontiers. So much is revealed at them. And thinking about “Witness Literature”* I want to explore the frontier between the literature of fact and the literature of fiction. I deliberately use the less familiar “literature of fact,” rather than that lumpen term from a publisher’s catalog, “nonfiction.” “Literature of fact”: the phrase is beautiful, and contains the key word—fact. But first, what of its other half, that large word “literature”?

It seems to me self-evident that these adjacent territories of fact and fiction both belong to literature, as France and Germany both belong to Europe. “Literature” is often taken to mean invented worlds. The twentieth century sustained the nineteenth’s romantic privileging of the creative imagination. For a hundred years, the Nobel Prize for Literature was mainly, though not exclusively, awarded to novelists and poets. But who could possibly argue that the works of Thucydides, Macaulay, and Nietzsche, that Orwell’s Homage to Catalonia and Naipaul’s Among the Believers, are not literature? Wherever the boundary of literature lies, it is not here.

The frontier between the literature of fact and the literature of fiction is open, unmarked. Some very fine writers stray across it quite casually, as one does when traveling in the Masai Mara—no border posts, same shrubland, same dust, same lions, but suddenly you are in Tanzania, not Kenya. These frontier crossings come in many forms. In the reportage of that master traveler Ryzsard Kapuscinski, we find haunting claims that would certainly not survive the attentions of a fact-checker at The New Yorker. (I open his Shah of Shahs at random, and read: “The Iranian Shiites have been living underground, in the catacombs, for eight hundred years.”) With Kapuscinski, we keep crossing from the Kenya of fact to the Tanzania of fiction, and back again, but the transition is nowhere explicitly signaled.

Paul Theroux’s travel book The Great Railway Bazaar, which is full of amusing incidents and wonderfully entertaining dialogue, concludes with an elaborate plea for its own strict, reportorial accuracy. He describes in detail the four thick notebooks in which he wrote things down as they happened, “remembering to put it all in the past tense.” On this railway trip through Asia, he writes, he had learned “that the difference between travel writing and fiction is the difference between recording what the eye sees and discovering what the imagination knows. Fiction is pure joy—how sad that I could not reinvent the trip as fiction.” At which I found myself thinking, “Well, you did, you did.” Perhaps I am wrong, but even the production of four weather-stained notebooks containing words identical to those on the printed page would not dissuade me, for the invention can come at the moment of recording.

The historian Simon Schama begins his stimulating and avowedly experimental Dead Certainties (Unwarranted Speculations) with a compelling eyewitness account of the Battle of Quebec by a soldier who fought in it. At the end of the book, Schama reveals that this account was fiction, “constructed from a number of contemporary documents.” So you were in Tanzania after all. Schama suggests that history as storytelling, as literature, must reclaim the ground it has lost to history as science, or pseudoscience. I entirely agree; but from this particular literary device it is not a long step to the postmodernist conclusion that any historian’s “story” is as good as any other’s.

Sometimes the frontier transgression comes not in the text itself, but in the context established by the writer. According to the biography of Jerzy Kosinski by James Park Sloan, Elie Wiesel was initially lukewarm to Kosinski’s novel The Painted Bird, which tells of a Jewish child in hiding alone during the war in a Polish village, thrown into a slurry pit by anti-Semitic Polish peasants, and struck dumb by the experience. Then Wiesel gathered from Kosinski himself that the book was closely based on Kosinski’s own childhood experience, and so he hailed it as a “chronicle” and “a poignant first-person account.” The novel was celebrated as a “testament,” a work of witness. Later, it turned out that Kosinski was never in hiding alone, thrown into a slurry pit, or struck dumb. The work was discredited on the very grounds that had established it. Kosinski’s self-justification was interesting for our purpose. “I aim at truth, not facts,” said the novelist, “and I’m old enough to know the difference.”


Now I want to mount a defense of this frontier, so open, ill-marked, often transgressed; a difficult defense, against the spirit of the times (“Who cares? It’s all entertainment anyway”), yet one that seems to me of the first importance precisely when it comes to the moral and artistic quality of witness.

Of course Kosinski had a point. Just as literature extends both sides of this frontier, so does truth. Truth is the other continent to which the states of fact and fiction equally belong. “Ah,” you may say, “but these are two different kinds of truth.” Yet that is exactly what needs to be examined, for in saying that both belong to literature we are suggesting—and I think rightly—that in many ways it is actually the same kind of truth.

Nor shall we naively suppose that “witness” can be found only on one side of the line. “You who harmed an ordinary man…,” writes Czesl/aw Milosz, in one of his most famous poems, “do not feel safe. The poet remembers./You may kill him—another will be born./Deeds and words shall be recorded.” The poet remembers: Poeta pamiåüeta! Poems and novels are an essential part of the literature of witness. But I do suggest that any meaningful notion of witness depends on having a clear delineation of this frontier, and knowing which side you are supposed to be on at any one time. Ian McEwan’s novel Atonement imagines a novelist who tries at the end of her life to atone for a terrible thing she did as a child by telling the truth about it. But since she does so in a novel, no one can know what is invented and what real. She cannot atone, because she is God in this invented world.

Words like “witness,” “testimony,” “evidence,” and, of course, “fact” have their sober offices in a court of law. And witnesses in literature, as in law, often testify to a particular kind of fact: the fact as something someone has done, often to someone else. In sixteenth- and seventeenth-century English the most common meaning of the word “fact” was “an evil deed or crime.” “He is…hanged,” wrote a sixteenth-century authority, “neere the place where the fact was committed.” (The usage survives in the English phrase “accessory after the fact.” In German, Tatsache means fact and Tatort, scene of the crime.) When we say “witness literature” we think first of witnesses to those “facts” that are committed by human beings on other human beings, whether in war, apartheid, holocaust, or gulag.

Now in defending this line, we must start by conceding much to those who would blithely stroll across it. For a start, all historians, journalists, and lawyers know that witnesses are wildly unreliable. They forget, lie, exaggerate, get confused. That’s why, so biblical scholars tell us, the Bible reflects the Jewish law of multiple witness. And Jesus chose twelve witnesses to record his acts. But (as this example suggests) even multiple corroboration achieves only a very rough approximation to the original reality. I have spent some time in recent years talking to Serb and Albanian witnesses to atrocities in Kosovo. Turning from Serb to Albanian, and back again, I have often wanted to say to them what Chaim Weizmann’s father, a famous village peacemaker, reportedly used to exclaim after hearing one side of an argument: “From what you tell me, I can see that you are entirely in the wrong. Now I shall hear the other side; perhaps you are in the right after all.”

Moreover, the evidential basis on which history is written is often extraordinarily thin. Sometimes, we have only one witness. During the Velvet Revolution in Prague, in 1989, crucial decisions were taken by a group around Václav Havel, meeting in a curious glass-walled room in the subterannean Magic Lantern theater. Most of the time, I was the only outsider present, and certainly the only one with a notebook open, trying to record what was being said. I remember thinking: if I don’t write this down, nobody will. It will be lost forever, as most of the past is, like bathwater down the drain. But what a fragile foundation on which to write history.

Of course, others who were there will add their recollections. But what use are recollections? The problem with memory is at the heart of the problem with witness. When I set out some years ago to explore my Stasi file, I thought: “This is the perfect way to test the credibility of secret police files. After all, if I know anything at all, I know what I myself did and said.” But as I read the file, talked to the people who had informed on me and the secret police officers who had spied on me, I found that I didn’t really know even that. Or rather, what I thought I knew kept changing with every new revelation. We don’t simply forget; we re-remember. Memory is a rewritable CD that is constantly being rewritten. And rewritten in a particular way: one that both makes sense of the story to us and makes it more comfortable for us. Isn’t it curious how, if two people separately describe to you an argument between them, both seem to have won?


Philosophers have long been on to this in their different ways. Thomas Hobbes wrote that “Imagination and Memory are but one thing.” One of Nietzsche’s deepest apothegms reads: “‘I did that,’ says my memory. ‘I can’t have done that,’ says my pride, and remains adamant. Finally, memory gives way.” Schopenhauer ascribed it to vanity rather than pride. More recently, the neuroscientist Michael Gazzaniga has suggested, after studying patients whose left and right brain hemispheres are disconnected, that human beings have what he labels “the interpreter” located specifically in the left brain, whose job it is to string together our experiences into narratives that seem to make sense. In short, we all have a novelist in our heads. A novelist called Memory ceaselessly redrafting the short story we call “My Life.”

Yet that is only half the ground we defenders of the line must concede, the better to advance. Suppose for a moment that there was no involuntary exercise of the creative imagination through memory. Suppose we had a perfect, impartial, scientific record of what really happened. (With the new technology of videocameras we can get closer to it than ever before, though with only two of the four senses deployed by a human witness. Cameras cannot smell or touch.) Even then, we would still have almost nothing—and much too much. To study five years of the French Revolution in just one corner of Paris you would have to sit for five years in front of a screen.

To create the literature of fact, we have to work like novelists in many ways. We select. We cast light on this object, shadow on that. We imagine. We imagine what it is like to be that old Albanian woman weeping over the body of her murdered son, or what it was like to be a fourteenth-century French serf. No good history or reportage was ever written without a large imaginative sympathy with the people you are writing about. Our characters are real people; but we shape them like characters, using our own interpretation of their personalities. Then we talk of “Michelet’s Napoleon,” “Taine’s Napoleon,” and “Carlyle’s Napoleon,” for each Napoleon is in some important sense the author’s creation.

The property of deliberate imagining is certainly not confined to the Tanzania of fiction. Imagination is the sun that illuminates both countries. But this leads us into temptation. A voice in your ear whispers, “You know that Kenyan in the slouch hat really did say that awfully funny thing you think he almost said. Just write it down. No one will ever know. And look, just across the frontier there is that gorgeous flower—the one missing novelistic detail that will bring the whole story alive. Pop across and pick it. No one will notice.” I know this voice. I have heard it. But if we claim to write the literature of fact, it must be resisted.

Why? For moral reasons, above all. Words written about the real world have consequences in the real world. If, in my book The File, I had identified as a Stasi informer someone who was not, in fact, a Stasi informer—and I nearly did—that man could have found his life ruined. Friends would have shunned him, he would probably have lost his job—and worse. (At least one person exposed as a Stasi informer committed suicide.) On a larger scale, the Balkan wars of the last decade have been fueled by bad history, written by all sides. As the historian Eric Hobsbawm observes, “The sentences typed on apparently innocuous keyboards may be sentences of death.” There is also a moral obligation to the victims, whether living or dead. How would we feel—how would the survivors feel—if we learned that events described in Primo Levi’s If This Is a Man had been deliberately invented, or ornamented?

These moral reasons are sufficient; but there are artistic ones too. Writers often cross this frontier because they think their work will be enhanced as a result. Reportage or history will become Literature. Paragraph for paragraph, that may be true. But as a whole, the work is diminished.

We also need to ask “How?” (often an even more difficult question than “Why?”). How does one determine when this frontier has been crossed, given everything I have said about the unreliability of witnesses, the involuntary creativity of memory, and the necessity of deliberate imagining? A simplistic, nineteenth-century positivist an- swer about scientific truth won’t do. For the truth achieved by the literature of fact is in many ways the same as that achieved by the literature of fiction. If we are convinced that human beings might have acted, thought, or felt in this way, it is in large measure as a result of the writer’s art and imagination.

I would suggest that, as well as satisfying all the truth-tests that apply to fiction, the literature of fact must pass two further, special truth-tests: those of “facticity” and of veracity. First, facticity. Are those things in the text that claim to be facts actually facts, or are they merely, to use Norman Mailer’s vivid coinage, factoid? Dates, places, events, quotations. Did the informer identified in my Stasi file with the incongruous code-name “Smith” actually sign a formal undertaking to work for the secret police as an “unofficial collaborator” or did he not? Everything else—causes, motives, consequences—is, strictly speaking, speculation; this is fact. (As a matter of fact, I know “Smith” did sign, because I have studied the original document.) Many alleged facts can be externally verified. The discipline of history and the craft of reporting have developed rules, procedures, specialist skills for testing evidence. Some even merit the label “scientific.” (An analysis of the ink used in “Smith’s” pledge, for example.) To pass this basic test of facticity does not make a text true, but to fail does make it untrue.

Yet much of the time, especially with “witness literature,” the witness is alone at the scene he or she describes. Alone with his or her eyes, conscience, and imagination. If we find witnesses accurate on things we know, we are more likely to believe them on things we don’t; but sometimes, there is little that we can know or check. What test works here? The best I can come up with is the quite unscientific litmus of veracity. Do we feel, as we read the text, that the writer is making what George Orwell, in praising Henry Miller, called “a definite attempt to get at real facts”?

For me, the model of such veracity is Orwell’s own Homage to Catalonia. Actually, Orwell got some of his externally verifiable facts wrong—not least because most of his notes were stolen during a secret police search of his hotel room in Barcelona. But we never for a moment doubt that he is trying to tell it exactly as it was. And when we reach his plea of veracity at the end of the book, it is the very opposite of Paul Theroux’s. Orwell writes, in that wonderfully plain, conversational style that he worked so hard to achieve, “In case I have not said this somewhere earlier in the book I will say it now: beware of my partisanship, my mistakes of fact, and the distortion inevitably caused by my having seen only one corner of events.” In effect, he says, “Don’t believe me!”—and so we believe him.

Veracity is revealed in tone, style, voice. It takes us back to the artistic reasons for defending this line. You can often tell just from internal, stylistic evidence when a writer has strayed. Take a now notorious example: the book published in 1995 as Bruchstücke (in English, Fragments) by Binjamin Wilkomirski, which purported to be the memories of a man who survived the Nazi death camps as a Polish Jewish child. It is now established beyond reasonable doubt that the author was a Swiss musician of troubled past and disturbed mind, originally called Bruno Grosjean, who had never been near a Nazi death camp—but had imagined himself into that past, that other self. Reading Fragments now, one is amazed that it could ever have been hailed as it was. The wooden irony (“Majdanek is no playground”), the hackneyed images (silences broken by the sound of cracking skulls), the crude, hectoring melodrama (his father squashed against the wall by a transporter, dead women with rats crawling on their stomachs). Material which, once you know it is fraudulent, is truly obscene. But even before one knew that, all the aesthetic alarms should have sounded. For every page has the authentic ring of falsehood.

Compare this with the great books of true witness. Of course there are large variations in tone and style between these works. Many nonetheless have a certain voice in common: one of pained, sober, yet often ironical or even sarcastic veracity, which speaks from the very first line. Take, for example, and contrast with Wilkomirski, the first line of Primo Levi’s If This Is a Man: “It was my good fortune to be deported to Auschwitz only in 1944, that is, after the German Government had decided, owing to the growing scarcity of labour, to lengthen the average life-span of the prisoners destined for elimination; it conceded noticeable improvements in the camp routine and temporarily suspended killings at the whim of individuals.” How could we not believe this?

The facts need not always be ugly. “I will bring you,” writes the English poet Craig Raine, “the beauty of facts.” It is, no doubt, a rather Anglo-Saxon sentiment. Yet facts, like artifacts, can be beautiful. On a white shelf at my home in Oxford, I have two objects. One is a rounded natural stone, some three inches high, of a delicate gray color tinged with a very pale pink, moulded into contours by the cold sea washing across a pebble beach on the northeasternmost tip of mainland Britain, at the Duncansby Stacks, where I picked it up during a contented afternoon spent with my family. The other is a jagged piece of the Berlin Wall, made of a gritty composite barely deserving the name of stone, with a patch of garish graffiti on one side. They sit there, the rough and the smooth, the unnatural and the natural, facing each other, rather brightly lit on the white shelf—a stone poem for the literature of fact.

It may seem a grave limitation for any writer to leave the facts as facts, but self-limitation is a key to art. On this frontier we should stand.

This Issue

November 7, 2002