164 ELMER ROBERTS 



lected series A occurred during the summer of 1914 and 1915 

 when there was extremely warm weather. The connection be- 

 tween high temperature and size of wing was unknown at first 

 and consequently no temperature records were kept during the 

 first part of the experiment. Thereafter records were kept and 

 a series of flies was subjected to high temperature in an incubator, 

 the outcome of which is discussed later in this paper. It is 

 significant to note that during the production of the F29 genera- 

 tion in selected series A, higher temperature occurred than in 

 either the F28 or Fro generation. The males are more affected 

 than the females, as can be seen from the means, standard devia- 

 tions, and coefficients of variability in the generations noted 

 above (tables 1 and 2). 



2. 'Crossed-in-and-selected' series C 



This series was obtained as described in section II, by the 

 crossing of ten vestigial-winged males from the Fie generation 

 of selected series A with long-winged females. Selection was 

 started among the vestigial-winged segregates appearmg in the 

 second subsequent generation. Table 3 exhibits the means, 

 standard deviations, and coefficients of variability of males and 

 females separately. The means of the males and females com- 

 bined are given in table 6. 



a. Effect of selection. Text figure 2 is constructed from the 

 means of the generations given in table 6. By an inspection of 

 this graph, one cannot say that selection has been effective. If 

 tlj.e experiment had been brought to an end in the twenty-first 

 generation, probably one would have concluded that selection 

 had been effective, but such fluctuations as are found in the F22 

 ' and F29 generations prevent such a conclusion. However, it 

 may have been that the somatic appearance was not always an 

 index of the germinal constitution, in which case there is a pos- 

 sibility that selection may have been effective, but that environ- 

 mental factors suppressed the full expression of the germinal 

 potency or constitution. Under the conditions of this experi- 

 ment, selection had no visible effect upon the expression of the 

 character in question. 



