44 ORGANIC ETOLUTION 



might soon become definitely established in two groups of 

 the species, making a constant distinction which might 

 become a specific character. Now, as no part of an organ- 

 ism varies independently of the rest, there would doubt- 

 less, in establishing the two groups which differ in time 

 of breeding, also be established as constant certain other 

 characters associated with the difference in breeding time. 

 Amonor some of the higher animals sexual selection, 



O O 



that is, the exercise of choice in mating, prevents promis- 

 cuous intercrossing and so may serve to preserve from 

 swamping certain divergent characters which may be asso- 

 ciated with such choice. To this point we will refer again. 



Anything which divides a species into groups will be 

 likely to prevent free intercrossing, and so tend to pre- 

 serve characters associated with the different groups. We 

 will come back to this point soon. 



The recently rediscovered work of Mendel has a bear- 

 ing upon the question of the persistence of variations. 

 Mendel showed a half-century ago, and recent workers 

 have more fully established, certain facts of heredity in 

 the case of hybrids between distinct species, and crosses 

 between widely divergent varieties of the same species. 

 Castle's work in breeding mice, which closely agrees with 

 Mendel's observations, shows the point clearly. Castle 

 bred white mice and common gray mice together and got 

 the following results. The offspring developed from the 

 first cross were all apparently normal gray mice. When, 

 however, a male and female from this first lot of young 

 were bred together very interesting results were obtained. 

 Three-fourths of the young of this second lot appeared to 

 be normal gray mice, but one-fourth were found to be 



