It now becomes evident what offspring may result from the union of parents who have any of the above constitutions. Thus parents both of Group I may have offspring belonging to any group according to which of the five possible constitutions they possess. If the union be represented by AB--AB × AB--AB, then only offspring of Group I can result, since every gamete contains both A and B. The other possibilities may be worked out by the reader if he desire. Similarly, a union of Groups I × II, I × III, or II × III may produce any of the groups, definite limitations being imposed by the detailed constitution of the parents. On the other hand, the remaining group unions that are possible can only produce a more limited variety of offspring. Thus II × II or II × IV can only produce Groups II or IV; III × III or III × IV can only produce Groups III or IV; IV × IV can only produce Group IV. The Mendelian theory of inheritance in general has been subjected to a prolonged and widely ramifying series of tests, and it seems in the present state of knowledge to present a satisfactory and consistent explanation of the facts. For a more extended account of it the reader must be referred to the standard works on the subject.[6] As regards its application to the present case, the test of actual experiment has not yet been carried out on a large scale. A series of observations has, however, been published by J. R. Learmonth, who, taking forty families at random, determined the blood groups of both parents and the children in each family. In this way he tested most of the possible group matings, and, with a single exception, the group inheritance conformed to the theory as set out above. Additional confirmation of the truth of the theory is afforded by the pedigree given on the page opposite. I have recently collected this pedigree, which includes fifty-nine individuals belonging to four generations, and it has not been published before. It will give, perhaps, a more graphic representation of the facts than has been conveyed by the brief summary contained in the foregoing pages. It does not show any variation from the results that were to be anticipated according to the theory. The exceptional result obtained by Learmonth in one of his forty families serves to emphasize the clarity of the theoretical considerations. In this family parents both belonging to Group IV had a child showing the reactions of Group I. There are three possible explanations of this: (1) The observations were at fault. (2) The putative father was not the real father. (3) The Mendelian theory of inheritance is wrong. The Mendelian theory is established on so firm a basis that, in the absence of more numerous exceptions, (3) may be rejected. There is no reason for supposing that the observations were inaccurate, and we are therefore brought to the conclusion that in such a case the child is illegitimate.