It is a test of true theories not only to account for but to predict phenomena. We ought to accept a theory if it predicts things which observation subsequently confirms, and if no rival theory is equally successful.