In two recent papers Joseph Margolis has sketched a situation, his characterisation of which involves a denial of the ubiquitous contention that knowing that p is logically sufficient for believing that p. There are not many philosophers who would follow him in this denial of what is most usually taken as the only “natural” way of construing knowledge. If Margolis has not succeeded in constructing a counterexample to the official view, and I do not believe that he has, it is (...) of some importance to understand just why he has failed. No epistemological gain is achieved by meeting Margolis with a question begging reply, as I think has been done by his critic, J.A. Barker.The case the Margolis offers is this:… S may not have thought about P at all, may have thought about Q-which is “significantly” related to P, and believes that Q; on that condition, it is conceivably true that S knows that P. (shrink)