“I can see no more reason for preferring the theories of fifty years ago than for preferring the observational data of fifty years ago.”