A dilemma may be raised of this kind. The universality of the law of causation--in other words, the uniform course of nature--is the fundamental principle on which all induction proceeds, the great premise on which all our science is founded. But if this law itself be the result only of experience, itself only a great instance of induction, so long as nature presents cases requiring investigation, where the causes are unknown to us, so long the law itself is imperfectly established. How, then, can this law be a guide and a premiss in the investigations of science, when those investigations are necessary to complete the proof of the law itself? How can this principle accompany and authorise every step we take in science, which itself needs confirmation so long as a process of induction remains to be performed? Or how can this law be established by a series of inductions, in making which it has been taken for granted? Objections which wear the air of a quibble have often this advantage--they put our knowledge to the test. The obligation to find a complete answer clears up our own conceptions. The observations which Mr Mill makes on this point, we shall quote at length. They are taken from his chapter on the _Evidence of the Law of Universal Causation_; the views in which are as much distinguished for boldness as for precision. After having said, that in all the several methods of induction the universality of the law of causation is assumed, he continues:-- "But is this assumption warranted? Doubtless (it may be said) _most_ phenomena are connected as effects with some antecedent or cause--that is, are never produced unless some assignable fact has preceded them; but the very circumstance, that complicated processes of induction are sometimes necessary, shows that cases exist in which this regular order of succession is not apparent to our first and simplest apprehension. If, then, the processes which bring these cases within the same category with the rest, require that we should assume the universality of the very law which they do not at first sight appear to exemplify, is not this a real _petitio principii_? Can we prove a proposition by an argument which takes it for granted? And, if not so proved, on what evidence does it rest?