Models and lawlikeness

Do analogical models ever play an essential role in scientific explanation and confirmation, or is their role (at most) heuristic? For many years scientists and philosophers have debated this question. I argue that such models may sometimes play an essential role. My argument is based on a proposal to augment Goodman's theory of projection in order to make it easier for novel predicates (extensions) to acquire entrenchment. The heart of this proposal is the claim that analogical models may, under certain conditions, be the medium whereby entrenchment is passed from well established predicates to new and unfamiliar ones.