Studies on variation and selection. 163 



possibility remains, that amongst Irish rats, as well as amongst Hooded 

 ones, more than one gene differentiate the animals, and that selection 

 favours a change of proportion between animals with and those without 

 such genes. 



It is, for example, theoretically very well possible, that, in the 

 case of the partially albinistic rats, we would have a complete 

 analogon to Nilsson-Ehle's case of the coloured wheats. Supposing, 

 that a number of genetic factors, each by their cooperation tend to 

 make the coloured area on such a rat extend further on the body, 

 it is easy to see, how a light animal, mated to a dark one, may 

 produce dark young, which, in their turn, will give some ligt ones. 

 And if it pleases us, at some arbitrary point in the range of pigmen- 

 tation-grades, to make a division, by calling all the lighter animals 

 by one name, and all the darker ones by another, it may very well 

 be, that we will get about three rats of the darker type to one of 

 the lighter. Especially must this be so, if one of the genetic factors 

 involved happens to have a somewhat greater influence on the 

 development than the others. 



We see, that it is vitally important to know, whether the material 

 used by Castle was pure for all the genetic factors involving the 

 extension of pigmentation, or whether it possibly was not. We know, 

 that generally, animal populations are very far from pure, in respect 

 to all sorts of genes, even for such as give them part of their 

 commercial value, and the lack of which may make them worthless. 

 So for instance, we found that in Holland black sheep, which always 

 lack a genetic factor, present in the whites, though never used for 

 breeding -purposes, are nevertheless produced in practically all the 

 very best folds. The same holds true for red-and-white individuals 

 in the black-and-white Dutch cattle, miscalled Holstein-Frisians. 



Even for such very important genes, a percentage of the animals 

 is always impure, and by the ordinary methods of breeding, this pro- 

 portion is only exceedingly slowly being ameliorated. Weeding out 

 the heterozygotes, whenever they betray themselves, ought to help, 

 and one of the best sheep-breeders in Holland, Mr J. Dyt, on Texel, 

 is now using this method consistently. 



Strict inbreeding, mating of brother and sister, parent and child, 

 together with a weeding out of any couple, consisting of hetero- 

 zygotes, and their possible children, would theoretically be the very 

 best method of making a family of animals pure for any genetic 

 factor its members may contain. For this heightens the chance, of 



