Abstract

I consider two alternative solutions to the problem of computing the values of theoretical quantities, and, thus, of testing theoretical hypotheses, viz., Sneed's structuralist eliminationism and Glymour's bootstrapping. The former attempts to solve the problem by eliminating theoretical quantities by means of the so-called Ramsey-Sneed sentence that represents the global empirical claim of the given theory. The latter proposes to solve the problem by deducing the values of the theoretical quantities from, among others, the very hypothesis to be tested. I argue that in those cases where the theoretical quantities are not strongly Ramsey-eliminable-which seems to be the case for most of the actual physical theories-eliminationism does not succeed in computing the values of theoretical quantities and is compelled to use bootstrapping in this task. On the other hand, we see that a general notion of bootstrapping-which, though implicitly, is present as a subreasoning in structuralism-provides a formally correct procedure for computing theoretical quantities, and thus contributes to the solution to the problem of testing theoretical hypotheses involving these quantities.