1 _DR_ : 1 _RR_. The _DD_ group and the _RR_ group thus produced giving on self-fertilisation pure _D_ offspring and pure _R_ offspring respectively, while the _DR_ groups gave again 1 _DD_ : 2 _DR_ : 1 _RR_. How does Professor Weldon propose to deal with these results, and by what reasoning can he suggest that considerations of ancestry are to be applied to them? If I may venture to suggest what was in Mendel’s mind when he applied this further test to his principles it was perhaps some such considerations as the following. Knowing that the cross-breds on self-fertilisation give 1 _DD_ : 2 _DR_ : 1 _RR_ three explanations are possible: (_a_) These cross-breds may produce pure _D_ germs of both sexes and pure _R_ germs of both sexes on an average in equal numbers. (_b_) _Either_ the female, _or_ the male, gametes may be _alone_ differentiated according to the allelomorphs, into pure _D_’s, pure _R_’s, and crosses _DR_ or _RD_, the gametes of the other sex being homogeneous and neutral in regard to those allelomorphs. (_c_) There may be some neutralisation or cancelling between characters in _fertilisation_ occurring in such a way that the well-known ratios resulted. The absence of and inability to transmit the _D_ character in the _RR_’s, for instance, might have been due not to the original purity of the germs constituting them, but to some condition incidental to or connected with fertilisation. It is clear that Mendel realized (_b_) as a possibility, for he says _DR_ was fertilised with the pure forms to test the composition of its egg-cells, but the reciprocal crosses were made to test the composition of the pollen of the hybrids. Readers familiar with the literature will know that both Gärtner and Wichura had in many instances shown that the offspring of crosses in the form (_a_ × _b_) ♀ × _c_ ♂ were less variable than those of crosses in the form _a_ ♀ × (_b_ × _c_) ♂, &c. This important fact in many cases is observed, and points to differentiation of characters occurring frequently among the male gametes when it does not occur or is much less marked among the maternal gametes. Mendel of course knew this, and proceeded to test for such a possibility, finding by the result that differentiation was the same in the gametes of both sexes[155]. [155] See Wichura (46), pp. 55–6. Of hypotheses (_b_) and (_c_) the results of recrossing with the two pure forms dispose; and we can suggest no hypothesis but (_a_) which gives an acceptable account of the facts. It is the purity of the “extracted” recessives and the “extracted” dominants--primarily the former, as being easier to recognize--that constitutes the real proof of the validity of Mendel’s principle.