Scienti c methods are intimately bound up with rules of inference. Few would characterize science so narrowly as to say that it is only the collection of the perforce nite number of disparate reports we make about what we can directly observe. Such a naive view of science has no role for inferencemaking. In recognizing the need to transcend the meagre amount of data we can collect, we must make inferences, drawing out two broad kinds of conclusion. e rst concerns the potentially in nite number of particular claims about states of a airs, past, future and elsewhere, that we do not, or cannot, directly observe. We wish to know what the rest of the world is like outside the small spacetime region to which our lives are con ned, the only region of which we can have direct observational access through our senses. e second kind of conclusion concerns our inferences to the laws and theories that go beyond what we can observe and that, we hope, give us a deeper cognitive

insight into how the world really works. Even if we are more imaginative and start not from observations or data but from observation-transcendent conjectures, laws, hypotheses or theories, we still need to make inductive inferences to bring our conjectures into a relation with observational data as a “reality check” against our conjecturing.