"Now obviously the propositions of the system have reference to matters of empirical fact; if they did not, they could have no claim to be called scientific."