Accepting this evidence as on the whole satisfactory, there is still something more to be said. As is well-known there are many grades of color-blindness. We do not know whether these grades are due to fluctuating (individual) variations—assuming it to be due to one gene: or whether there are several genes that differ in the degree to which they produce the defect. In fact we know now of a good many cases in other animals where there are several mutations of the same gene. For instance, in Drosophila there is a series of ten such multiple allelomorphs for eye colors that range from pure white to deep wine-red. There is still another possible interpretation of the different kinds of color-blindness—one which _a priori_ would seem to be the most probable—namely, that the differences are due to other modifying genes that affect the extent to which the character develops. While in the great majority of cases, the scheme of color-blindness is that shown by the diagram, we know that occasionally the machinery may be changed to give a somewhat different result. It is possible, for example, that a color-blind man married to a perfectly normal woman may rarely produce a color-blind son. A few years ago such a result would have appeared to upset the entire scheme of sex-linked inheritance, today we understand how such cases may arise through a process that is called non-disjunction, which is best illustrated by numerous cases well worked out in Drosophila. My second illustration has a more obvious chemical basis. Hemophilia is also sex-linked in inheritance. It is known to be much more common in men than in women, the explanation for this is the same as in the other case. In affected individuals the blood fails to coagulate quickly and the difference in chemical composition of the blood is, in contrast to normal, the inherited character. [Illustration: FIG. 8. Representing the kinds of individuals expected when an individual of the blood group type AaBb marries individual of the same blood type, namely AaBb. Sixteen kinds of individuals are possible in the ratio of 9:3:3:1. These belong to four blood types, namely, class IV that contains at least one A and one B; class II that contains at least one A but no B; class III that contains at least one B but no A; and class I that contains neither A nor B.]