448 EXPEBIMENTtS WITH PLANTS 



bore (in addition to tiie normal leaves) one four- leaf 

 and one five -leaf, he sowed its seed and found that 

 about half the resulting plants bore 

 (in addition to normal leaves) four- 

 and five-leaves. The best four plants 

 were saved and their seed sown: 

 this time 80 per cent of the offspring 

 had the four- and five-leaves 

 and a few six- and seven - 

 leaves made their appear- 

 ance. This process of selec- 

 tion was continued until 



2")!. Four-, five- and seven- . n n ii nc 



leaved Clover. practically all the onsprmg 

 were of the new type (i. e., three- to seven- 

 leaved, Fig. 251), or, in other words, the seed 

 came true. One might suppose that it would 

 now be possible to go on and make plants 

 with eight-and nine-leaves. The attempt, 

 however, proved fruitless. The limit of selec- 

 tion was reached with seven, and it was im- 

 possible to go beyond it. On the ground of 

 many similar experiences, he comes to the 

 conclusion that all plants have a limit which 

 is quickly reached by selection, and here its 

 power ends.^ The improved race is not a species: 

 it has no constancy, and when left to itself quickly 

 returns to the original type. 



1 He believes that it is even possible to tell beforehand how much a 

 plant can be improved by selection and where the limit will be 



