Abstract

This paper begins by adopting a broad criterion of scientific methods, in which a method is some practice in science designed to achieve the goals of science; this allows us to deal with a wider range of questions than has been common in much past philosophy of science, and to discuss some of the more detailed methodological questions of concern to scientists. On this definition, apriori approaches prove insufficient, since many methodological questions require some empirical input to be decided. Laudan has suggested that many such questions can be decided by testing methods empirically, but this encounters the problem that the empirical evidence is often indecisive. A more productive strategy is to utilise scientific theories: there are a wide range of these that are relevant to methodological questions and they are easier to validate. But this does not mean that empirical testing has no place. Instead, our experience with methods can be used to detect specific problems with the methods and to modify them accordingly.