FACTORS OF INHERITANCE. 53 



animals. No thoroughly established case has yet been 

 brought to light in which the factors have been proved 

 not to segregate. Many of the supposed instances of 

 blended inheritance, with the formation of series of inter- 

 mediates, have been shown to be due not to failure of 

 segregation, but to incomplete dominance, multiplicity 

 of factors, direct effect of environment, and other com- 

 plications which blur the result. Yet there are still some 

 cases known, such as the colours of human races, which 

 have so far defied analysis into segregating factors ; and 

 it seems possible, in the light of recent work, that the 

 rule is not universal, and that segregation may sometimes 

 be incomplete or even not occur at all. 



Now most of these experiments have been made on 

 domestic varieties, differing from one another by only 

 a few well-marked unit characters ; differences due prob- 

 ably to the dropping out of certain hereditary factors, to 

 "retrogressive mutations." We may now inquire what 

 happens if two natural " species " are crossed. This is a 

 most interesting question, and one directly affecting the 

 general problem of evolution, since the experiment might 

 possibly reveal some hidden but genuine distinction 

 between a " species " and a " variety." The variety 

 might mendelise but not the species ; as, indeed, was 

 suggested by De Vries. Such, however, does not appear 

 to be really the case ; but the problem is much compli- 

 cated owing to the large number of factors of varying 

 importance which may distinguish two even closely 

 allied natural " species." In so far as some of the unit 

 characters are conspicuous and dominant, it may be 

 possible or even easy to trace out the segregation in later 

 generations ; but the task may become too difficult if 

 the distinguishing characters are small and very numer- 

 ous. In such a case thousands of " intermediate " 

 heterozygotes may be produced, each with its particular 

 inheritance, and a bewildering number of intermediate 

 forms result. This is after all perhaps not a matter of 



