Possibilities increase under explanation. To decide how plausible any such possibility, causal or practical, is, is not, I have argued, in any strong or simple sense a theoretical matter. None the less, explanations themselves may draw on theory, and to the extent that they do, the plausibility of the possibilities they suggest will bear back on the theory which informs them.

But this is ambiguous. The theory in question may be an account of how, causally or practically, the kind of thing we wanted to explain came about. Or it may be a theory in a more descriptive sense, in which what is first at issue is not how we might explain how something occurred, but how we should see it. If the possibilities that our explanations suggest bear back on a theory of the first kind, revisions to that theory will not necessarily affect our characterisation of what it was that had to be explained. The possibility that the incidence and effects of plague in some parts of early modern Europe could have been alleviated, or that rural fertility in seventeenth or eighteenth-century France could have been lower, has a bearing on pre-existing explanations; but neither requires us to change our characterisation of what it is that we want to explain. Likewise, the possibility that the United States did not move to occupy southern Korea in 1945, or the possibility that once there, it worked to create a government for the whole country, do not require us to change our reports of what it actually did and did not do.