While browsing around at Scribblingwoman, I found Ken McGoogan's letter to the Arctic Review. McGoogan, author of Ancient Mariner: The Arctic Adventures of Samuel Hearne, writes that

For several years now, while teaching courses in “creative nonfiction” and moderating seminars with titles like"Where Fact Meets Fiction," I have been wrestling with such questions [about the"line between likely imaginings and wholesale invention”]. And Potter and Owen are right in perceiving that I have a beef with the way history is conventionally written. I believe we can draw nearer the elusive truth – and also make history more vivid and accessible -- by combining scholarship with imagination, and by augmenting traditional “analytical narrative” with the techniques of fiction (scene, dialogue, point of view).

Perhaps McGoogan is just using"the way history is conventionally written" in an excessively loose fashion, but I fail to see how his chosen solution is any less conventional. Herodotus, anyone? The invented speech? Like it or not, however, such tactics stopped signifying as"history" centuries ago--as the anti-Jacobin polemicist, novelist, and pedagogical theorist Elizabeth Hamilton discovered when she tried to resuscitate them for The Memoirs of the Life of Agrippina, Wife of Germanicus (1804). It's not that the relationship between history and fiction should necessarily be like that of a sphere, projecting against a plane; after all, aren't many historians and biographers good at scene-setting, narrative tension, and so forth? But he's right that most would not recognize the following as"history":

For example, I cite a passage in which Williams reveals that two bitter enemies confronted each other aboard a discovery ship. Because the records and log books have not survived, he moves on without elaborating, as mandated by the pseudo-scientific conventions of analytical narrative:

“Here, where even the dullest novelist would leap to reconstruct the confrontation scene, Williams refuses to speculate, to go beyond the dubious evidence of primary documents. He declines to dramatize his best guess, well-informed though it certainly is. One can’t help wondering what would happen if superlative historians like Glyn Williams stopped pretending that they practice science and accepted that they write literature. Maybe history would begin to regain its audience?”

Subversive, you see? Reviewing Ancient Mariner, Owen asks: Where do we draw the lines? The writer of historical fiction, I believe, has taken out a license to change dates, names and venues, and to invent, combine or kill off characters, whatever; the writer of historical nonfiction, on the other hand, must work within the known facts, changing and ignoring none of them. I take the position that, having assimilated the relevant journals, letters, biographies and histories, the non-fiction writer can then use imagination and craft to bridge gaps in the record.

This seems to me to be a distinction without a difference. Indeed, Gore Vidal normally describes his own novelistic practices very much as McGoogan is here describing nonfiction. There are all sorts of other things wrong in these paragraphs--for starters, popular history (especially military history & Civil War) has a huge audience, and, to my knowledge, most historians aren't under the impression that they're the reincarnation of Ranke (surely being beholden to one's evidence isn't the same thing as doing"pseudo-science"?!)--but I'll let the historians handle it.

UPDATE: The morning after,"distinction without a difference" seems unnecessarily abrupt. To elaborate, McGoogan exaggerates the"license" allowed to historical novelists, who can change only so much before their work becomes either risible or alternate history. (Anyone who has ever snickered through a bad historical film knows of what I speak.) Hence the"ever since Scott" tendency to push major historical figures off to the margins. At the same time, as I said, McGoogan's definition of historical nonfiction actually sounds remarkably close to how many historical novelists would define their own practice. I've mentioned Gore Vidal; my father has also on occasion pointed out that Mary Renault's Funeral Games (the power struggles in the wake of Alexander the Great's death) is very much tied to the historical evidence, even to the extent of speculating why some of the evidence has vanished.

Ralph weighs in on invented speeches in the comments. He sounds against 'em.