GENIC INTERACTION 1H7 



Fig. 33. Mean selective values of gene-frequency systems. 



Selection 



h- — I 



Selection 



Random drift 



Random drift 



aBCd 

 (1.125) 



Random drift 



ABcd 

 .250 



abCD 

 (1.000) 



abed 

 (0.875) 



abCd 

 (0.750) 



Mean selective values of gene-frequency arrays along the path of least depression from 

 the lowest peak abCD through an intermediate peak aBCd to the highest peak ABcd, of 

 figures 30 and 32, and along the direct path from the lowest to the highest peak through 

 the four-dimensional col. 



only slight differences in their net selective values. But as can be seen from figure 3 1 , 

 the process can apply indirectly to a pair of alleles (M, M' ) with major differential 

 effects. Random drift may lead to sufficient frequencies of several of the very slightly 

 deleterious modifiers A', B', C, D', E', and F' at some time in some deme to reverse 

 the selection against M' which, with its associated modifiers, will then start to spread 

 through the species. 



Table 43 makes a comparison between the selective process in a homogeneous 

 species 373 and the double process (intrademic and interdemic) in a finely subdivided 

 species. 1422 



The conditions of balance among net selection coefficients, immigration, and 

 random drift may not have held anywhere at any time in the ranges of some species. 

 They almost certainly have held in others (for example in primitive man and many 

 other mammals) . Where they have not held at all, the evolutionary process has been 

 restricted to that of the second theory. The first theory is merely the limiting form 

 of the second under conditions in which all loci are only very weakly heterallelic. 

 The three are really complementary. 



