"Evidence and Inference in History," Daedalus, v87 n4 (Fall, 1958): 11-39.

Excerpt:

I confess that I feel apprehensive about the subject proposed to me. Even though there may be possible French equivalents to to an accumulation of data available to the scholar, in this case the historian, before he risks making inferences, evidence and inference, the two concepts which guide this colloquium, they are not part of my spontaneous vocabulary. The concept evidence refers, I believe, i.e., operations leading to more or less general propositions which were not included in the facts but nevertheless may be legitimately deduced, inferred, or extracted from them. These two concepts embrace a vast territory. The 1st includes the documents of the historian as well as the experimental results of the physicist or the statistics of the sociologist. The 2nd includes the hypotheses of an Einstein as well as the psychological hypotheses of the Soviet expert who, given the data concerning the great purge, imagines the motivations of Stalin or the impersonal mechanism by which the purge triumphed.

Online:
JSTOR