142 



THE AMERICAN NATURALIST [Vol. XLIV 



idea is appraised at essentially its true value, but much 

 in its action is left obscure. The reader is surprised at 

 the accounts of experiments in which selection does ac- 

 complish marked results, though according to the general 

 theory, it should not; one is left puzzled in judgment as 

 to what we may expect from it. With the sharply formu- 

 lated pure line concept as a guide, most of this obscurity 

 disappears. 



And then, to keep us from resting on our oars ; to give 

 us humility and spur us to further work— we come to the 

 one case in which the pure line idea fails to bring clear- 

 ness. This is de Vries's experiment with buttercups. 

 Here, after selection the extreme was moved far beyond 

 that before selection. Before selection the extreme num- 

 ber of petals was eleven; after selection it was thirty- 

 one. Before selection no single individual had an aver- 

 age number of petals above six ; after selection the aver- 

 age of all was above nine, and some had an average of 

 thirteen! It is true that there are " mitigating circum- 

 stances" here; the work was not done with pure lines, 

 and the variations dealt with are not of the ordinary 

 fluctuating sort (as de Vries points out) ; change in cul- 

 tural conditions doubtless played also a large part. Pos- 

 sibly repetition with thorough analytical experimentation 

 will show that something besides 'selection has brought 

 about the great changes. But at present the case stands 

 sharply against the generalizations from the pure line 

 work. It is the only such case that I have found. 



To sum up, one finds not only that his own results and 

 those of many other modern workers give the pure line 

 interpretation, but also that all other cases that had 

 seemed to point the other way yield readily to the geno- 

 typic analysis— save one. If we ride rough shod over 



draw a tentative conclusion as follows: 



The pure line or genotype idea is the one to see clearly 

 and grasp firmly in experimental investigations on selec- 

 tion. Many even of the modern experiments remain 



