The argument most frequently made on behalf of historical fiction is that, if responsibly done, it can stimulate interest in the study of history. But I'd like to offer a stronger argument, based on my experience working in both forms: if properly understood, the writing of historical fiction can be a valuable adjunct to the work of historians in their discipline. A novel can be as accurate as a history in telling what happened, when, and how. It can, and should, be based on careful research and rigorous analysis of evidence. But the distinction and advantage of the fictional form lies in the way it uses evidence and represents conclusions. The novel tests historical hypotheses by a kind of thought-experiment. assume that events are driven by the conditions and forces you believe to be most significant-what sort of history, what kind of human experience, then results? For the thought-experiment to work, the fiction writer must treat a theory which may be true as if it was certainly true, without quibble or qualification; and credibly represent a material world in which that theory appears to work. We should not suppose that thought-experiments, however elegant, make empirical tests unnecessary. But as the history of modern physics shows, without such experiments, the forward movement of knowledge becomes slower and more difficult. Moreover, because the novel imaginatively recovers the indeterminacy of a past time, the form allows writer and reader to explore those alternative possibilities for belief, action, and political change, unrealized by history, which existed in the past. In so doing, the novelist may restore, as imaginable possibilities, the ideas, movements, and values defeated or discarded in the struggles that produced the modern state.