INTRODUCTION

Since 1948, when Hempel and Oppenheim¹ published their pioneering article, various models of explanation have appeared. But each has had its counterexamples, and observers of the philosophical scene may wonder whether models of the kind sought are really possible.

Hempel and other modelists are particularly concerned with explana­tions that answer questions of the form

(1) Why is it the case that p?^{[61] [62]}

The sentence replacing ‘p’ in (1) Hempel calls the explanandum. It de­scribes the phenomenon, or event, or fact, to be explained. The answer to an explanation-seeking why-question of form (1) Hempel calls the ex­planans. It is a sentence, or set of sentences, that provides the explanation. We can speak of the explanans as explaining the explanandum. And we can say that an explanans potentially explains an explanandum when, if the sentences of the explanans were true, the explanans would correctly explain the explanandum.

Thus, if the explanation-seeking why-question is

Q: Why is it the case that this metal expanded,

then the explanandum is

(2) This metal expanded.

If, in reply to Q, an explainer claims that

(3) This metal was heated; and all metals that are heated expand,

then (3) is the explanans for the explanandum (2). And (3) potentially explains (2) if, given the truth of (3), (3) would correctly explain (2).

A model of explanation is a set of necessary and sufficient conditions that determine whether the explanans correctly explains the explanan- dum (where the explanation-seeking question is of form (1)). It can also be described as a set of conditions that determine whether the explanans potentially explains the explanandum. If the conditions are satisfied by a given explanans and explanandum, then the former correctly explains the latter, provided that the former is true.

In what follows, my concern will be with models as sets of sufficient (rather than necessary) conditions for correct explanations; and as providing such conditions for explanations of particular events or facts rather than of general laws. Most of the counterexamples in the literature have been raised against models construed in this way. I shall argue that one important reason for the failure of these models is that their proponents want to impose requirements which, in effect, destroy the efficacy of their models.

2.