MENDELISM AND SELECTION 



flowers in the buttercup is a heritable varia- 

 tion and supposed it to be a unit-character, 

 and so to conform with Mendel's law. 



Now, if the tendency to produce double 

 flowers were a simple Mendelian character it 

 could exist in only three conditions, — that of 

 a recessive, that of a homozygous dominant, 

 or that of a heterozygous dominant. But re- 

 cessives and homozygous dominants are pure, 

 that is, they form only one type of gamete, 

 and selection therefore from among their pro- 

 geny could produce no new type. As regards 

 the heterozygous dominant type, this would 

 itself be unfixable, and selection could accom- 

 plish nothing permanent except by isolating a 

 homozygous type. But such types should all 

 be in evidence within two generations; there- 

 fore, if a completely and permanently double 

 type had not been discovered within the five 

 generations covered by the experiment, such a 

 type was not to be expected at all from the 

 material in hand, unless either a wholly new 

 unit-character were introduced or an existing 

 one were profoundly modified. De Vries con- 

 siders changes of both these sorts possible, 



109 



