304 FOUNDATIONS OF BIOLOGY 



must occur in a pure line for selection to be effective and 

 then, ipso facto, the single pure line becomes two. (Fig. 153.) 



The trend of present work certainly seems to indicate 

 that these conclusions are of general application and that the 

 explanation of the long-accepted feeling that selection is 

 'creative' is to be found in the fact that variations are of 

 three sorts: modifications which are not heritable and com- 

 binations and mutations which are heritable. Most of the 

 variations within pure lines apparently are the result of en- 

 vironmental influences recurrent in each generation, but the 

 germ plasm is homogeneous. The variability within a popu- 

 lation is the composite variability of its component pure lines, 

 but the germ plasm is not the same in all individuals these 

 may be segregated into groups, the pure lines. Thus, very 

 liberally interpreted, the pure line concept is a formal expres- 

 sion of the fact that most of the variations which we recog- 

 nize are either somatic or the result of recombinations of 

 diverse parental genes. Accordingly when the genes of the 

 gametes are identical (as in pure lines) the latter source of 

 variation does not exist, and selection is powerless except 

 when comparatively rarely mutations occur. (Fig. 157.) 



However, some recent work indicates that under certain 

 conditions selection appears to be effective, at least to a 

 limited degree, within a pure line. We have previously seen 

 that certain characters are the expression of multiple genes. 

 In some such cases one gene is, so to speak, the determining 

 gene for the character as a whole, while associated with this 

 gene there is a galaxy of modifying genes which themselves 

 do not find expression without the presence of the determin- 

 ing gene, but merely serve to alter the character expression 

 of the latter. Under such conditions it is possible to modify 

 the character by selection to add or subtract or otherwise 

 change the relationships of the modifying genes to the pri- 



