portant to regard each invented hypothesis as an elastic conception, whose form may be changed as necessity demands. The investigator must even return to his field of observation and reexamine the facts, particularly such as do not match with the well defined consequences of a partly successful hypothesis; and he must search his field with the sharpest scrutiny to see if any facts, previously unnoticed, really do occur in the manner indicated by unmatched consequences. In every way, the utmost care must be taken not to allow oneself to be satisfied with imperfect or incomplete agreements.
If these various recommendations are carefully carried out, the danger, often feared, that an investigator may, Procrustes-like, force the facts to fit the needs of a favorite hypothesis, is practically ruled out; for if the investigator has several unlike hypotheses in mind, and has deduced several unlike series of consequences from them, it will evidently be impossible for him to force his facts to agree with all of them, however much the facts may be trimmed or stretched.
Irregular Order of Procedure.—In practise the several processes that have been necessarily considered in systematic succession, are carried on in a much more irregular fashion. It has already been pointed out that invention may advisedly go hand in hand with observation. It is evident that, after a hypothesis has been invented, any time spared from observation may be devoted to deduction; and it often happens that the consequences of the hypothesis may grow to a greater number than that of the classes of observed facts then accumulated. Confrontation and comparison may be made repeatedly as observation advances, and revision is always in order the moment there seems to be occasion for it. The active-minded investigator, thus continually reviewing the different aspects of his problem, may gradually come to feel that one hypothesis, modified as far as needs be from its original form, appears to deserve greater acceptance than any of its rivals; then arises the great question: Is this hypothesis really true? Is it surely a correct counterpart of the invisible facts of the past? Clearly it is essential that an investigator, on reaching this stage in his work, should fully understand the nature of scientific proof.
Final Judgment.—It is at this advanced stage of an investigation that the exercise of a sound judgment is needed, in order to estimate the measure of confidence that may be given to an apparently successful hypothesis. The most important point to emphasize now is that, in such problems as we are here dealing with, the only available method of testing the truth of any hypothesis is to measure the agreement of its deduced consequences with the appropriate facts of observation. In this respect scientific proof is altogether unlike geometrical proof, in which the correctness of a theorem is never tested by its agreement with observable facts, but only by the continuity with which successively de-
