740 PROCEEDINGS OF THE AMERICAN ACADEMY. 



made under identical conditions, but from stock not inbred, had a pro- 

 ductiveness two or three times as great and showed no signs of sterility. 

 The A stock, accordingly, is characterized by low productiveness and an 

 inclination to sterility. If these are effects of inbreeding, they began to 

 show themselves early, the small productiveness as early as the sixth 

 generation, the partial sterility as early at least as the thirteenth genera- 

 tion, and probably earlier. But it is very improbable that the low pro- 

 ductiveness of race A was due to inbreeding, for when the inbreeding 

 experiment was repeated with other stocks of flies no appreciable diminu- 

 tion of fertility occurred. 



It cannot have been due to selection, conscious or unconscious, of 

 parents from the least productive broods, for it was the practice of Messrs. 

 Carpenter and Clark " to start a number of cultures and then breed 

 from the one which succeeded best," which would be in general the 

 most productive rather than the least productive pair. That selec- 

 tion is a very important factor, will presently appear. It certainly has 

 a much greater influence on fertility than does inbreeding, so that selec- 

 tion from the most productive pairs is able to more than offset the effects 

 of inbreeding. But it was not sufficient in the experiments of Carpenter 

 and Clark to bring the A race up to normal productiveness. The low 

 productiveness must have been present in the A stock at the beginning 

 of our experiments and have persisted in spite of selection favoring 

 increase of fertility. 



Selections for High or Low Productiveness. 



The bearing of selection from parental broods more or less productive 

 upon the fertility of the race will be clear from Table II. 



In the A series, between generation 30 and generation 43, pairs were 

 several times taken simultaneously from parental broods of unequal size. 

 The result is shown in Table II. The larger parental brood produces, as 

 a rule, a larger average filial brood, and with a smaller proportion of sterile 

 pairs. Thus, in the seven generations tabulated (which include all avail- 

 able data), if we weight each generation the same, the following aver- 

 age values are obtained from the lowest and highest parental broods 

 respectively : 



Lowest .... 

 Highest 1 



