()()<) 



THE AMERICAN NATURALIST [Vol. XLIX 



the latter period, and were not uniformly selected in the 

 former. By the introduction of the progeny test as an 

 essential part of the selection the whole process of the 

 creation of a highly fecund race of hens was transferred 

 from the realm of blind chance to that of precise and defi- 

 nite control. And it becomes increasingly clear that 

 mass selection, on the basis of performance (phamotypie 

 appearance) alone, is in its essential nature a blind and 

 haphazard process, no matter with what precision and 

 stringency it is carried out, just so long as the correla- 

 tion between the gametic and somatic conditions of the 

 character selected is not perfect. And it is an outstand- 

 ing result of the Mendelian investigations of the last 15 

 years that the gamete-soma correlation is very rarely, if 

 ever, perfect, 9 



It appears on this view that selection for high egg pro- 

 duction in the fowl is effective when it is real. That is, 

 if one selects genetically high producers by means of the 

 trap-nest phis the progeny test, he succeeds very rapidly 

 in fixing a high producing strain. If on the other hand 

 he merely selects high layers by the trap-nest record 

 alone, he is not really selecting genetically high pro- 

 ducers except in a portion of the cases. Under these cir- 

 cumstances he makes no progress in building up a highly 

 fecund strain. To be effective in changing the average 

 productiveness of a flock of poultry selection must pick 

 out those birds as breeders which carry the factors for 

 high fecundity genetically, i. e., as an integral part of 

 their hereditary make-up, and not any other birds. 



With the above interpretation of the results of seven- 

 teen years' continuous selection of the character fecundity, 

 all the facts known to the writer are in complete accord. 

 No other interpretation of the results of this experiment 

 has yet been suggested which will meet all the facts. 



» Complete citations on this point would make a tolerably full bibliog- 

 raphy „t M.-nd.-lism. The motlio.lolo^i.-:il ,.r the strictly quantitative aspects 



F. K. Weldon, Biometrika, Vol. I, pp. 228-254, 1902, and R. Pearl, Biol. 



