In most cases we have to deal with mixed evidence, and frequently it has become habitual to change the problem to be solved according to our convenience, or at least to set aside some one thing. Suppose that the issue deals with a discovered, well-retained footprint of a man. We then suspect somebody and compare the sole of his shoe with the impression. They fit in length and width, in the number of nails and in all the other possible indices, and we therefore assert: It is the footprint of the suspect, for “whose footprint?” is the problem we are troubling ourselves to solve. In truth we have only shown that the particular relations, in the matter of length, breadth, number of nails, etc., agree, and hence we regard the positive part of the evidence as sufficient and neglect the whole troublesome negative part, which might establish the fact that at the time and in the region in question, nobody was or could be whose foot could accurately fit that particular footprint. Therefore we have not proved but have only calculated the probability that at the time there might possibly not have been another person with a shoe of similar length, breadth and number of nails. The probability becomes naturally less as fewer details come to hand. The difficulty lies in finding where such probability, which stands for at least an assumption, must no longer be considered. Suppose, now, that neither shoe-nails nor patches, nor other clear clews can be proved and only length and width agree. If the agreement of the clews were really a substantiation of the proof by evidence, it would have to suffice as positive evidence; but as has been explained, the thing proved is not the point at issue, but another point. The negative portion of the evidence will naturally be developed with less accuracy. The proof is limited to the assertion that such shoes as were indicated in the evidence were very rarely or never worn in that region, also that no native could have been present, that the form of the nails allowed inference of somebody from foreign regions, one of which might be the home of the suspect, etc. Such an examination shows that what we call evidence is only probability or possibility.