Years ago I read in the daily paper(!) an account of a new theorem in partition theory. The key was that the guy had had the idea of proving theorems that were not for all cases but for all but a finite number of cases. The theorem had something about the number of ways to partion (prime?) numbers, and all I remember is that when the number of partitioning bins was either 17 or 19 (I forget), the number of ways was 237 "for all numbers except in a finite number of cases". You can see how confused I am. The fault is not the original story, which I remember as being pretty clear, but the long memory gap. Does this ring a bell for anybody?