CROSSING. 91 



atively enormous. As can be seen from Fig. 13 which illustrates 

 the weight in pro-centages of a group of Fg animals, there is a 

 very strong indication for the existence of one or two such 

 genes. The variation curve, far from being smooth is decidedly 

 two-topped. As can be seen from Fig. 14 the shape of the var- 

 iation curve is not the same at different ages. 



From this it is apparent that at different ages, different genes 

 play a weight-influencing role. 



When hybrids of this first generation are mated back to 

 either parent, we can expect much clearer evidence for factor- 

 segregations. For, whenever we make this cross, we can limit 

 our attention to those genes which the other parent contribu- 

 ted to the list for which the hybrids are heterozygous. And 

 when this mating-back to either parent species is continued for 

 several generations, eventually purity will be reached, the re- 

 sult of the series will be an individual pure in respect to the gen- 

 ot37pe of one parent species. When in every succeeding gener- 

 ation we pick one individual at random, we know that that in- 

 dividual has one chance in two to be homozygous for every 

 gene in respect to which its hybrid parent was still heterozy- 

 gous. In other words, heterozygosis is reduced one half in every 

 succeeding generation. Eventually in every series a point will 

 be reached, where the hybrid individual mated back to one of 

 pure strain differs from it in just one gene. Whenever this gene 

 happens to have an appreciable influence upon the develop- 

 rnent, and therefore upon weight, this heterozygosis may show 

 itself in a Mendelian segregation into two weight-classes. 

 Whether we ever succeed in demonstrating such cases is a 

 matter of pure chance. The only thing we could do was to have 

 as many separate series of back-mating experiments running 

 as we could afford. 



Two sets of series only were continued for any length of time, 

 namely those, where we mated back hybrids between Japanese 

 mice and large whites to the large Robertson strain, and where 

 we did the same with descendants of a Chinese mouse. I have 

 selected as an illustration the offspring of two sisters in the 



