is universal or any fact necessary. We know that things are so and so, but not that they must be, or will always be. Our knowledge is made up of actual perceptions and of inferences and hypotheses suggested by them. This latter constituent, for which Kant substitutes a system of mind-created, universal and necessary principles, is in fact mere guess, assumption, or postulate. Alike in science and in ordinary knowledge (for the two differ only in the matter of systematization) you may trace the constant rise and fall of such ordering postulates. What we now call general principles or ultimate laws or notions are the exceptional cases of happy survival. They live for us, not because we can see into their necessity, but solely because in the growth of experience no perceptions have conflicted with them, while they in the meantime have served as ordering or systematizing centres of ever widening fields of fact. Darwin could not endure a scientist who, while observing, was not constantly forming hypotheses. He only demanded of the individual what the race has done since language began to be used. The universe which spreads infinitely about and within us is an abysmal mystery. Knowledge consists of the observations and the verified guesses man has made of his more immediate environment. Only the dogmatist, ignorant of what Bishop Butler calls the doubtfulness in which things are involved, can to-day believe that any of our knowledge must be universal or necessary. In an age of omniscient rationalism the case was different, and to such an age Kant belonged. Though he deemed natural theology impossible, he declared the foundations of physical science immovable. For us, the one like the other is made up of facts perceived and of hypotheses to account for them. Some branches of knowledge have more of the perceptional element, like mathematics, others more of the speculative element, like theology; but no absolute line of separation can be drawn between them, and while neither can be pronounced invalid (for there is no higher court) each must be accredited according to the amount of evidence it adduces. If the Newtonian method is good in one domain of knowledge it is good in another. It is vain to claim
