Passing now to the third and perhaps most important methodological point, we may say a few words as to what method should be adopted for the drawing of conclusions from evidence considered as reliable. This is neither a logical proceeding, nor is it a kind of induction. Properly speaking, a witness's statement may be either accepted or rejected. But in this book importance has been laid on presenting the evidence in a quite definite way. Evidence is not used in order to exemplify or to prove a given assertion on a special point. Such a proceeding appears to be rather dogmatic, for usually in such cases the author gives preference to an _a priori_ opinion, and looks afterwards for its confirmation in the ethnographic first-hand literature. Owing to the contradictory character of the latter, practically anything can be proved from it. In the present book the author merely sets forth the problem; for instance, such quite general questions are asked, as: How are wives obtained in Australia? What is the treatment of the wife by her husband? What are the sexual relations in general? and so forth. On each of those general topics evidence is afterwards collected, without prejudice or preference given to any type of opinion. There is, therefore, much less risk of bias or one-sidedness; the whole care is to make the best of the evidence thus collected; and a series of statements upon a given subject is presented. Each of them gives information on several points at once; at any rate each of them may usually be analyzed into a series of simpler statements. And this analytical operation will be our first task. There is always one or more assertion sufficiently general, or simple, which will be contained in all or in the majority of our statements and will be contradicted by none. These may be considered as established by our evidence. On other points there will be contradictions. Often these contradictions will be only apparent, due to a confusion in terminology, or to the defective way in which the writers have expressed themselves. Here recourse must be had to our first form of criticism, to the ascertainment of the exact meaning of each statement (verbal criticism). If that fails, the contradictions must be recognized as real ones. In case they cannot be attributed to any local differences, we must try to eliminate them. And on this point recourse must be had to the criticism of the statements from the point of view laid down above (p. 25). Some of the statements may be discarded as untrustworthy; the correct interpretation of others may be determined; and thus the contradictions will vanish. Sometimes this is impossible; the contradictions remain irreducible. Then they must be simply pointed out, and there is nothing further to be done. Undoubtedly much greater service is rendered to science by pointing out really irresolvable contradictions and obscurities than by establishing fallacious certitude.