CONCLUSIONS
Our discussion suggests the following conclusions: (i) If the explanans is not to contain singular sentences that entail the explanandum, then it will be an empirical not an a priori question whether the explanans if true correctly explains the explanandum.
(ii) This empirical question will involve determining the truth-values of certain singular sentences (either explanation-sentences or something akin to them) that do entail the ex- planandum; otherwise factors will be citable in the explanans which are not explanatorily operative with respect to the explanandum-event.Can there be a model of explanation? Specifically, can there be a set of sufficient conditions which are such that if they are satisfied by the explanans and explanandum the former correctly explains the latter? Our discussion suggests that there can be no such model if, like D-N theorists, we insist that it satisfy both the a priori and the NES requirements. Moreover, it also suggests that we will not be successful in discovering a model in which (a) the NES requirement is satisfied, and in which (b) it is not an explicit condition of the model that some singular sentence be true that entails the explanandum.
It does not follow from this that an explanans which appeals to causal factors, laws, dispositions, desires and beliefs, statistically relevant factors, or essential properties cannot correctly explain an explanandum. However, models of explanation of the sort I am considering do not simply list kinds of factors that can be explanatory. Their proponents want to supply sufficient conditions for correct explanations. If it is demanded that these conditions satisfy the NES and a priori requirements, or NES plus ( b) then I am suggesting that such models will not be forthcoming.