710 



THE AMERICAN NATURALIST 



[Vol. XLVI 



continue to insist that the prolonged application of that 

 system is bound to result in improvement. 



It seems to me that it must be recognized frankly that 

 whether or not continued selection of somatic variations 

 can be expected to produce an effect on the race depends 

 entirely on the mode of inheritance of the character 

 selected. In other words, any systematic plan for the 

 improvement of a race by breeding must be based and 

 operated on a knowledge of the gametic condition and be- 

 havior of the character in which improvement is sought, 

 rather than the somatic. Continued mass selection of 

 somatic variations as a system of breeding, in contrast 

 to an intelligent plan based on a knowledge of the gametic 

 basis of a character and how it is inherited, seems to me 

 to be very much in the same case as a man who, finding 

 himself imprisoned in a dungeon with a securely locked 

 and very heavy and strong door with the key on the 

 inside, proceeded to attempt to get out by beating and 

 kicking against the door in blind fury, rather than to 

 take the trouble to find the location of the key and unlock 

 the door. There is just a possibility that he could finally 

 get out in a very few instances by the first method, but 

 even in those cases he would be regarded by sensible men 

 as rather a fool for his pains. 



Of course what has been said is not meant to imply that 

 selection, on the basis of somatic conditions may not have 

 a part in a well considered system of breeding for a par- 

 ticular end. In many cases it certainly will have. Thus 

 in the case of fecundity in the fowls, selection of mothers 

 on the basis of fecundity records is essential in getting 

 male birds homozygous with respect to L x and L 2 . But 

 the point which seems particularly clear in the light of 

 the present results is that blind mass selection, on the 

 basis of somatic characters only, is essentially a hap- 

 hazard system of breeding which may or may not be 

 successful in changing the type in a particular case. 

 There is nothing in the method per se which insures such 

 success, though that there is inherent potency in the 



