Recent arguments of Watkins, one purporting to show the impossibility of probabilistic induction, and the other to be a solution of the practical problem of induction, are examined and two are shown to generate inconsistencies in his system. The paper ends with some reflections on the Bayesian theory of inductive inference.