Abstract

William Dray has offered a number of criticisms of Hempel's account of historical explanation. While many of these miss the mark because they do not identify the relevant Hempelian claim regarding confirmation, some of the most important ones can be reformulated to provide good evidence that historians can and do confirm their explanations without confirming general hypotheses. ;Michael Scriven has proposed an account of how such general-hypothesis-independent confirmation occurs. He argues that such confirmation is confirmation relative to a set of possible causes of the event to be explained. That is, if a historian has a set of possible causes of an event, he can confirm that one member of the set was the cause without formulating or confirming general hypotheses. ;Scriven's account presupposes that the object of historical explanation is that of discovering which of a number of possible causes is the cause of the phenomenon to be explained, rather than describing how the phenomenon was produced, as occurs often in scientific explanation. I conclude by arguing that we can reformulate Scriven's argument for independent confirmation of historical explanation while avoiding this presupposition. ;With regard to Hempel's second testability condition, I agree that all adequate explanations must be testable. In order for Hempel to use this condition to argue for his covering law account of explanation, however, he must maintain the stronger claim that the confirmation of any explanation must involve the confirmation of general hypotheses. Using the case of historians' practice in confirming the explanations they offer, I argue against this stronger claim. ;The dominant view of the nature of scientific explanation is presented in the covering law theory of explanation. According to this theory, an explanation must employ general hypotheses and a description of initial conditions from which one can deduce a description of the phenomenon to be explained. Carl G. Hempel, a contemporary proponent of this theory of explanation has applied the theory widely, including an application to historical explanation. Hempel has also broadened the covering law account to include statistical or probabilistic explanation. Such explanations make the following claim: that one need only be able to deduce that the phenomenon to be explained is highly likely to occur, given the laws and initial conditions mentioned in the explanans, in order for the explanation to be adequate. ;Hempel has identified two conditions which he argues are necessary in order for an explanation to be fully adequate. First, an explanation must be such that if one had the information contained in the explanans prior to the occurrence of the phenomenon to be explained, it would be rational to expect the phenomenon to occur; one could legitimately have predicted its occurrence. Second, in order for an explanation to be a genuine scientific explanation rather than a pseudo explanation, it must be testable. I argue that the first of these conditions is not a necessary condition of adequacy of explanation. I acknowledge the adequacy of many statistical explanations, but also demonstrate that their adequacy derives not simply from the strength of the likelihood that the occurrence of the phenomenon to be explained, given the explanans, will occur. Rather, I propose that statistical, like other explanations derive their force from the fact that they cite causes of, or indicate causal mechanisms responsible for, the phenomenon to be explained. Moreover, I show that we can have knowledge of causes and causal mechanisms when we could not reasonably have expected the effect to occur