No. 519] THE EFFECTIVENESS OF SELECTION 139 



But how can we apply the pure line idea to organisms 

 whose lines are not pure; organisms that interbreed 

 freely; organisms in which the characters of a given 

 line split off, separate, and become exchanged for those 

 of other lines, in the way characteristic of Mcn.lelian 

 inheritance! 



The pure line idea here becomes a little elusive, a 

 little abstract. But possibly it is still helpful as an 

 instrument of analysis; let us try it. In order not to 

 emphasize purity where impurity is the rule, let us sub- 

 stitute Johannsen's term genotype for "pure line"— 

 defining the genotype as a set of individuals which, so 

 long as they are interbred, produce progeny that are 

 characteristically uniform in their hereditary features, 

 not systematically splitting into diverse groups. 



Now, how can we determine whether the genotype con- 

 cept, with its consequences for the effects of selection, 

 applies to organisms with biparental inheritance! Re- 

 flection shows that if it does, certain general propositions 

 are true: if these propositions are found to hold, the 

 genotypic explanation of the effects of selection is con- 

 firmed. 



1. The first proposition is this: Organisms in which 

 selection has shown itself effective are composed of many 

 genotypes ; of many races that are diverse in their heredi- 

 tary characters. This we know to be true. 



2. Second, from such a mixture of genotypes it is pos- 

 sible to isolate by selection any of the things that are 

 present— perhaps in a great number of different com- 

 binations. 



3. But from such a mixture it is not possible to get by 

 methodical selection anything not present (save when 

 rare mutations have occurred). 



4. Therefore it is not possible to get by methodical 

 selection anything lying outside the extremes of the 

 genotypic characters already existing. 



This is perhaps practically our most im\ 

 sition. For in order that selection shall 



