advantage inspect the opposed arrays, with the intention of seeing how closely any battalion of consequences matches the battalion of facts. In making this inspection, the comparer must evidently give particular attention to the facts from which an invention did not spring; and look closely to see how successfully they are matched by the consequences of the invention. There must be no pressure to force an agreement where none exists; no constraining of the facts or torturing of the consequences to make them look like each other; but simply a fair-minded comparison, followed by a clear unbiased report as to where agreement and disagreement occur.
Preliminary Judgment.—If two rival hypotheses have yielded only identical consequences, all of which agree nicely with the corresponding facts, no decision in favor of either hypothesis can be made, and judgment must be suspended. The comparer must then ask the deducer if he can not find unlike consequences of the rival hypotheses; and if such are found close attention must be given to the degree of success with which they match the corresponding facts. Evidently, then, some consequences have a greater value than others in discriminating among rival hypotheses. If the consequences which are peculiar to one hypothesis match the appropriate facts, while the contrasted consequences of other hypotheses fail to do so, then a higher value may be given to this one of the several rival hypotheses, although before it had no greater value than its now defeated competitors.
Revision.—It will often happen, when confrontation is made and an encouraging amount of agreement is found between the consequences of a certain hypothesis and the corresponding facts of observation, that the agreement is nevertheless in some respects imperfect. It may be that, for some of the facts, no corresponding consequences have been deduced; or that, for certain deduced consequences, there are no corresponding facts. Then the investigator must revise his work. He must return to the stage of deduction, and look closely to see if those consequences which are only partly successful in meeting the facts, were rightly deduced; he must inquire if the absence of a certain consequence, with which some well ascertained fact ought to be matched, is perhaps due to oversight in his deduction. He must examine with particular care all the principles, introduced by memory from previous acquisition, to see if they are safely established and correctly applied; he must be especially careful not to overlook any tacit postulate, which, without being consciously recognized as such, has nevertheless been taken for granted without sufficient proof, and used as an essential basis for some of his deductions. He must go still farther back and modify his inventions in one way or another, in the hope that, after such modification, some one of them may lead to new consequences that will better than before fit the previously unmatched facts: hence it is im-
