CONTROL OF HEREDITY: EUGENICS 391 



more probable that these discrepancies are due 

 to an incomplete analysis of the phenomena in 

 question. It is possible that the divisions of 

 the cell body in these protozoa is not always 

 into exactly equivalent halves in which case 

 variations might take place in the descendants, 

 which might then be heaped up by selection. 



The evidence as to the ability of selection to 

 produce new characters is conflicting and the 

 question is still an open one but for the present, 

 the evidence which is most clear-cut and abund- 

 ant indicates that selection by itself is unable 

 to change inheritance factors or unit charac- 

 ters. Nevertheless selection is of great service 

 in separating good lines or races from poor 

 ones, and this is the chief significance of the 

 artificial selection practiced by breeders. 



The elimination of certain races by natural 

 selection is an important factor in evolu- 

 tion though it may have nothing to do with the 

 formation of new characters or new races but 

 serve merely as a sieve, as deVries has ex- 

 pressed it, to sort the individuals which are 

 supplied to it. Even if selection has no power 

 to make or change characters, it preserves 

 certain lines and eliminates others and thus 



