First. While the chance is heavily against any one defect establishing itself, there is always the possibility that some one defect may establish itself. It must be remembered that while many defective strains may be lost, one would notice only those that had taken root. It is the presence of these that may give us an exaggerated idea of the generality of such occurrences. Second. If the human germ-plasm is continually mutating to produce one or another kind of specific defect, this will increase the chance for any recurrent defect to finally establish itself. That particular mutations do recur in other animals is now abundantly established by evidence that comes from several sources. Third. There is a growing impression that a good deal of feeblemindedness and insanity are environmental rather than hereditary traits; poverty, malnutrition, and especially syphilis are said to play a considerable rôle in their production. It is unsafe therefore to conclude that the human germ-plasm is as badly contaminated as some pessimists seem to think. If we turn now more directly to special kinds of human inheritance we shall find a great deal of evidence showing that the same laws of inheritance that hold for animals and for plants apply to man. It would be surprising if this were not the case. On the other hand, when we scrutinize the pedigrees that have been published to illustrate heredity in man, we shall find many of them very unsatisfactory in two main respects. (1) The number of offspring in a family is usually too small to serve as a sample of the germ-plasm of the parents. (2) Therefore, since recourse must be had to many families for sufficient data, it is essential that the diagnosis of the defects of the parents and of the children is correct. A single mistake may throw the result into confusion. In cases where the defect is structural, a correct classification may be possible, but in other cases, especially where psychological defects are involved, the diagnosis is difficult and the results, in consequence, less certain. Often the best that we can do in the case of man is to try to find the simplest Mendelian formula to which the evidence will fit. If one factor-difference will not suffice, then two must be tried; if two will not do, then three must be tried, etc. Now I need hardly point out that we can explain almost anything if we are allowed enough factors. It is, at best, a dangerous practice, one to be used only with great caution and the conclusion stated as provisional and checked in every possible way.