The parti-resultantness of requirement: An explanation that failed

Conclusion It appears that some relations may happen to be both defeasible and parti-resultant, relative to certain interesting reference classes. Confirmation is one of them, but, under several of the best available reconstructions of parti-resultantness, requirement is not. Moreover, parti-resultantness is not necessary for being defeasible, nor sufficient to assure it. So the conjecture by Ross and Chisholm that having the former property explains having the latter, especially where duty and requirement are concerned, turns out to be mistaken. It is probably only the persistent and deep-running ambiguity of parti-resultantness itself (uncovered above) that made the prospect of using it to account for defeasibility seem plausible.