Studies on variation and selection. 



157 



How can we ever know the answer to this question? It is 

 evident that to study the influence of selection on a genetic factor, 

 we can never get this genetic factor away from organisms carrying it. 

 But we can at least take pains, to make the disturbing influences 

 as slight as possible. We must, if possible, work only with one 

 genotype, with a group of organisms all having the same set of 

 genes. For we know, that mary genes influence each others action 

 on the development. 



This requirement would be fulfilled by starting the selection- 

 experiment with one single plant, homozygous for all its genes. The 

 difficulty lies in the question whether such plants exist. We can not 

 easily be sure that a plant is homozygous for all its genes. 



If it does give a homogeneous descendance, this only proves, 

 that it was pure for those genes which were factors in its development. 

 We can not be sure of those genes as were not. The only plants, 

 which are certainly so pure, are those, which have developed partheno- 

 genetically from one single gamete. The authors have experiments 

 well under way, which tend to prove that such a thing is possible. 

 Continued self-fertilization tends to make a strain of plants pure for 

 those genes, for which some members are still heterozygous. And 

 continued strict inbreeding in animals does the same thing. If from 

 the daughterplants from self-fertilized seed of one plant, a single one 

 is chosen, and so on for some generations, the family will automati- 

 cally become pure. A plant, heterozygous for a certain gene, will 

 produce 50 % offspring pure for it, either with or without, so that 

 the chance of choosing a homozygote is 1:1. Once a homozygote 

 chosen, the family is pure for this factor, and so it is for all those 

 factors, for which in the family there may still be heterozygotes. 



It is for this reason, that families, begun from one single plant, 

 of a habitually self-fertilized type, are as nearly "pure lines" as we 

 can get in practice. If, therefore, selection should be able to modify 

 any character in such a family, it would be an indication that it 

 could change the nature of a gene. Now Johannsen has shown, 

 that selection on such characters as the length-width index of beans, 

 repeated through a number of generations, could not alter these 

 characters. His experiments are very conclusive. Still, objections 

 have been made to them Belling, in the American Breeders Magazine, 

 made the objection, that in the experiments of Johannsen different 

 seeds were selected, instead of plants, differing in the average form 

 of their grains. And an other objection, made by Castle and others, 



