According to these, all we know and infer, in so far as it is unmathematical, results from experience, and our conviction of it and our reasoning about it, means by which we pass the bounds of sense-perception, depend on sensation, memory and inference from causation. Our knowledge of the relation of cause and effect results also from experience, and the doctrine, applied to the work of the criminalist, may be formulated as follows: “Whatever we take as true is not an intellectual deduction, but an empirical proposition.” In other words, our presuppositions and inferential knowledge depend only upon those innumerable repetitions of events from which we postulate that the event recurred in the place in question. This sets us the problem of determining whether the similar cases with which we compare the present one really are similar and if they are known in sufficiently large numbers to exclude everything else. Consider a simple example. Suppose somebody who had traveled all through Europe, but had never seen or heard of a negro, thought about the pigmentation of human beings: neither all his thinking nor the assistance of all possible scientific means can lead him to the conclusion that there are also black people--that fact he can only discover, not think out. If he depends only on experience, he must conclude from the millions of examples he has observed that all human beings are white. His mistake consists in the fact that the immense number of people he has seen belong to the inhabitants of a single zone, and that he has _failed to observe_ the inhabitants of other regions. In our own cases we need no examples, for I know of no inference which was not made in the following fashion: “The situation was so in a hundred cases, it must be so in this case.” We rarely ask whether we know enough examples, whether they were the correct ones, and whether they were exhaustive. It will be no mistake to assert that we lawyers do this more or less consciously on the supposition that we have an immense collection of suggestive a priori inferences which the human understanding has brought together for thousands of years, and hence believe them to be indubitably certain. If we recognize that all these presuppositions are compounds of experience, and that every experience may finally show itself to be deceptive and false; if we recognize how the actual progress of human knowledge consists in the addition of one hundred new experiences to a thousand old ones, and if we recognize that many of the new ones contradict the old ones: if we recognize the consequence that there is no reason for the mathematical deduction from the first to the hundred-and-first case, we shall make fewer mistakes and do less harm. In this regard, Hume[135] is very illuminative.