472 



THE AMERICAN NATURALIST [Vol. LI 



The eight young so obtained fall between grades + 1^ 

 and + 2f . From this cross 239 F2 young have been ob- 

 tained ranging from grades — 1 to + 3^. 129 or 53.9 

 per cent, show grades characteristic of the hooded grand- 

 parent. 



The remaining cross was between females of grade + 3, 

 third selection generation and Irish males. Controls ob- 

 tained by tabulating the progeny of animals of similar 

 grade and generation show in 143 young, a range between 

 grades f and -f- 3f . Only twenty-three F., young were 

 obtained, ranging between — 1 and -f 2^. Of these, six- 

 teen, or 69.5 per cent., reproduce the grades characterizing 

 the hooded grandparent. 



One other striking cross is recorded by Castle in the 

 case of the rats. This is a cross between "mutant" rats 

 showing a particularly advanced degree of plus pigmen- 

 tation and pure wild rats. The range of variability of 

 the pure "mutant" race is according to Castle's state- 

 ment (Castle and Wright, 1916, page 173) between grades 

 4-5-^ and +5f, see Table VIII. The 109 Fo "mutant" 

 young show a range of variation between the same grades. 

 According to Castle's statement on page 174, "their range 

 of variation does not fall beyond that of the uncrossed 

 mutant race." It would api)eav as though the evidence 

 of segregation in this case, even to an almost exact de- 

 gree, was clear. This case, together with that of the tyi)e 

 of spotting in mice known as "blaze" (Little, 1914) al- 

 ]-eady discussed, ap]>ear- to show that races differing 

 from each other in only ;i minute degree of a quantitative 

 character may show sourcuntion clearly. 



To sum up the experiments witli rats, it may be said 

 that while it can not be claimed that the evidence is final 

 in regard to the unit nature of the factors involved, they 



