I started from the scholarly tradition that examined epic poetry and mythology as if it might have some underlying basis in fact. Heinrich Schliemann assumed the Iliad was true, and found what he claimed was Troy and Mycenae; Arthur Evans believed there was something to the myth of the Minotaur, and uncovered the Palace of Knossos on Crete; M. I. Finley and others had traced the route of Ulysses in the Odyssey; Lionel Casson had written about the real journeys that might underlie the myth of Jason and the Argonauts. Thus it seemed reasonable, within this tradition, to imagine that Beowulf, too, had originally been based on an actual event.
That event had been embellished over centuries of oral retelling, producing the fantastic narrative we read today. But I thought it might be possible to reverse the process, peeling away the poetic invention, and returning to a kernel of genuine human experience - something that had actually happened.
This idea of uncovering the factual core of the narrative was appealing but impractical. Modern scholarship offered no objective procedure to separate poetic invention from underlying fact. Even to try would mean making innumerable subjective decisions, large and small, on every page - in the end, so many decisions that the result must inevitably be still another invention: a modern pseudo-historical fantasy about what the original events might have been.
The insoluble problem prevented me from proceeding. Of course, in writing a novel, I intended to create a fantasy of my own. But fantasies demand strict logic, and I was troubled by the logic behind what I wanted to write. Since a real scholar could not do what I intended to do, I found I could not pretend, in writing, that I had done so. This was not a failure of imagination or nerve. It was a purely practical problem. Like the scholar, I had no basis for deciding which elements of the Beowulf narrative to keep, and which to discard.
Although the idea of working backward seemed untenable, I remained intrigued. I asked a different question: suppose, for a moment, that the practical problems that troubled me did not exist, and the process could indeed be carried out. What would the resulting narrative look like? I imagined it would probably be a rather mundane recounting of some battles that occurred more than a thousand years ago. In fact, I suspected it would probably resemble most eyewitness accounts of famous events, as written by people who are unaware of the significance of the events they are seeing.
This line of thinking eventually led to the solution to my problem. Clearly, I wanted an eyewitness account. I could not extract it from the existing Beowulf narrative, and I did not want to invent it. That was my impasse. But at some point, I realized I did not have to invent it - I could discover it instead.
Suppose, I thought, a contemporary observer had been present at these battles, and had written an account of the events that were later transformed into a poem. Suppose, too, that this account already existed, but had never been recognized for what it was. if this were so, then no invention on my part would be necessary. I could merely reproduce the eyewitness narrative, and annotate it for the reader.
The concept of a preexisting manuscript bypassed the logical problems which had earlier impeded me, because a found manuscript would not be my creation - even though I would create it. Of course such thinking is absurd, but it happens all the time. Often actors cannot act without a prop, or a false moustache, or some other artifice to separate themselves from the character they are portraying. I was engaged in a similar process.
What sort of narrative would be most desirable? I concluded the most useful account would be written by an outsider - someone not part of the culture, who could report objectively on the events as they occurred. But who would this outside observer have been? Where would he have come from?
On reflection, I realized I already knew of such a person. In the tenth century, an Arab named Ibn Fadlan had traveled north from Baghdad into what is now Russia, where he came in contact with the Vikings. His manuscript, well-known to scholars, provides one of the earliest eyewitness accounts of Viking life and culture. As a college undergraduate, I had read portions of the manuscript. Ibn Fadlan had a distinct voice and style. He was imitable. He was believable. He was unexpected. And after a thousand years, I felt that Ibn Fadlan would not mind being revived in a new role, as a witness to the events that led to the epic poem of Beowulf.
Although the full manuscript of Ibn Fadlan has been translated into Russian, German, French and many other languages, only portions had been translated into English. I obtained the existing manuscript fragments and combined them, with only slight modifications, into the first three chapters of Eaters of the Dead. I then wrote the rest of the novel in the style of the manuscript to carry Ibn Fadlan on the rest of his now-fictional journey. I also added commentary and some extremely pedantic footnotes.
I was aware that Ibn Fadlan's actual journey in A.D. 921 had probably occurred too late in history to serve as the basis for Beowulf, which many authorities believe was composed a hundred and fifty years earlier. But the dating of the poem is uncertain, and at some point a novelist will insist on his right to take liberties with the facts. And Eaters contains many overt anachronisms, particularly when Ibn Fadlan meets up with a group of remnant Neanderthals. (One of the oddities of this book is that the intervening decades has seen a scholarly reevaluation of Neanderthal man; and the notion that there might have been a few still around a thousand years ago in a remote location does not seem quite so preposterous now as it did then.)
But certainly, the game that the book plays with its factual bases becomes increasingly complex as it goes along, until the text finally seems quite difficult to evaluate. I have a long-standing interest in verisimilitude, and in the cues which make us take something as real or understand it as fiction. But I finally concluded that in Eaters of the Dead, I had played the game too hard. While I was writing, I felt that I was drawing the line between fact and fiction clearly; for example, one cited translator, Per Fraus-Dolus, means in literal Latin "by trickery-deceit." But within a few years, I could no longer be certain which passages were real, and which were made up; at one point I found myself in a research library trying to locate certain references in my bibliography, and finally concluding, after hours of frustrating effort, that however convincing they appeared, they must be fictitious. I was furious to have wasted my time, but I had only myself to blame.