I draw on phenotypic reconstruction in paleobiology to argue that the success of historical scientists is their capacity to achieve plausible, sophisticated hypotheses about the past despite fragmentary evidence is in part explained by their capacity to generate new streams of evidence. Historical scientists are 'methodological omnivores?': in addition to exploiting a diverse range of techniques and theories, they construct purpose-built epistemic tools to further their reach into the past. Further, historical investigation frequently proceeds in a 'scaffolded' manner, that is, before some evidence can gain relevance, a set of hypotheses must be on the table. I argue that this has an upshot for functional ascriptions in paleobiology. Where in other contexts it is necessary for scientists to keep different senses of function separate, their conflation sometimes aids paleobiological investigation.