IN Problems of Analysis,' Max Black formulates and discusses what he calls counter-inductive policy. Essentially the policy amounts to this: Given a sequence of events defined by the property A, given a sample of this sequence having n members, and given that m members of the sample have been found to have the property B, predict that n -n m of the total sequence of A's will be B's. The person who adopts this policy is congenital pessimist, for he believes that the ratio of B's to A's in the total population is exactly the ratio of non-B's to A's in the observed sample. Furthermore, When most A's have been B, he expects the next A not to be a B; and the more A's have been B in the past, the more firmly he anticipates that the next A will not be a Black subsequently remarks, have no wish to suggest that a 'counter-inductionist' would be behaving rationally (though anybody who could explain in convincing detail why this should be so would have mastered the most important problems in the philosophy of induction). In this note I shall attempt to show that the counterinductive policy involves an irregular rule of induction and that rules of this class are unsatisfactory because they lead to formal contradiction. Suppose we have a sequence of A's and three mutually exclusive attributes, B1, B2, and B. In an observed sample of this sequence let us suppose 5/8 of the elements have the attribute B,, I/4 have the attribute B2, and I/8 have the attribute B3. Now the counter-inductionist must prefer B3 to the other two attributes, for he predicts that 7/8 of the A's are B3. But consider the attribute non-A which obviously is not exemplified in the sample. He should conclude that all A's are non-A's, for that is the consequence of the counter-inductive rule. The same may be said for every attribute which fails to be exemplified in the sample, even though many such attributes will be mutually incompatible. Thus, if the population is balls in an urn, and the sample contains red, white, and black balls only, the counter-inductionist would be required to predict that all remaining balls are green, and that all remaining balls are blue, and that all remaining balls are pink, and so on. Surely the counter-inductionist must avoid troubles of the foregoing sort. So let him hold to the counter-inductive policy for only those attributes which are actually exemplified in the sample-i.e., attri-