756 PROCEEDINGS OF THE AMERICAN ACADEMY. 



The character of the successive choices in each line was as follows, L 

 standing for largest, M for medium, and S for smallest : 



The so-called smallest broods of the iV line were larger than the 

 largest broods of the N' line in generations 21 and 25, so that the choices 

 of "smallest" broods in the inline do not mean much. They were not 

 intended as choices of small broods, and in fact those chosen were not small 

 broods, though they happened to be less large than the two other broods 

 of the same line. 



The broods selected as smallest in the N' line were in most cases really 

 small. The choices of smallest broods coming in succession, in genera- 

 tions 20, 21, and 22 respectively, seem to have given the N' line a char- 

 acter of low productiveness, from which, however, it steadily recovered 

 when medium or largest broods were selected. No doubt if the selection 

 of small broods had been consistently followed up in generations 23-27, 

 the N' line would have continued lower in productiveness than the N 

 line, just as the A' line continued lower than the A line (Figure 1), 

 though reared under identical conditions and without conscious selection 

 of broods. 



These experiments make it clear that Drosophila is in fecundity much 

 influenced by external conditions, such as temperature, upon which is 

 dependent the proper fermentation of its food. This is strikingly shown 

 also in Figure 3. But independently of external conditions some families 

 are characterized by high productiveness, others by low. Improved con- 

 ditions increase the productiveness of all, by starting them on a cycle of 

 greater productiveness. But the response is more prompt and vigorous 

 on the part of a race normally high in productiveness. Long continued 

 inbreeding may possibly cause a decline in the fertility of Drosophila 

 (for cross-breeding two inbred stocks certainly causes a rise in fertility 

 of the offspring), but the influence of inbreeding on fertility is less than 

 variations in fecundity due to other causes, so that selection of the more 

 fecund broods would certainly maintain a normal fertility even under 

 the closest inbreeding for fifty or more generations. 



