On the Justification of Induction