46 HEREDITY [OH. 



selection now ceases, the selected race will very 

 slowly revert towards the mean of the general popu- 

 lation. But, as has been seen, this conclusion is based 

 on the assumption that continuous variation is due to 

 the concurrent action of an indefinite number of small 

 independent causes. If, however, Johannsen is correct, 

 we may divide these causes into two classes : the 

 causes which induce ' fluctuation ' as explained above, 

 which agree with Pearson's requirements, and the 

 cause or causes which give rise to the difference 

 between one pure line and another. Now this second 

 group may conceivably consist in a single factor of 

 the nature of a small ' mutation/ and if so, by isolating 

 the pure line this factor is also isolated, and no return 

 towards the mean of the general population need 

 take place. According to Johannsen this isolation 

 can be effected in one generation by selecting the 

 self-fertilised plants which have the highest average 

 yield, instead of selecting the heaviest beans them- 

 selves. 



We thus obtain by experiments such as those of 

 Johannsen a new conception of the possible nature of 

 continuous variation ; it may be due partly to ' fluctua- 

 tion* brought about by the action of environment 

 and not inherited, partly to a series of small step- 

 wise ' mutations/ each of which owing to fluctuation 

 overlaps the next, and can only be isolated when it is 

 possible to breed pure lines. It should be said that 



