Abstract
It is widely recognized that Goodman's grue example demonstrates that the rules for induction, unlike those for deduction, cannot be purely syntactic. Ways in which Goodman's proof generalizes, however, are not widely recognized. Gruesome considerations demonstrate that neither theories of simplicity nor theories of empirical confirmation can be purely syntactic. Moreover, the grue paradox can be seen as an instance of a much more general phenomenon. All empirical investigations require semantic constraints, since purely structural constraints are inadequate. Both Russell's theory of empirical knowledge and Putnam's model-theoretic argument against metaphysical realism illustrate the inadequacy of purely structural constraints