12 



EXPERIMENTS WITH DROSOPHILA AMPELOPHILA. 



ancestry as the abnormal strain, and about one- third of 1 per cent of 

 wild Drosophila ampelophila were found to be abnormal. For four 

 generations after branching ^from the abnormal strain (five, counting 

 mating 405) not a single abnormal individual was found, but in the next 

 generation 1 fly out of 216 (0.5 per cent) had a very slight abnormality. 

 In succeeding generations the percentage increased for a time, in spite 

 of artificial selection to the contrary, and then diminished to zero under 

 the same treatment. The abnormalities were all small, never greater 

 than " medium." Table 6 summarizes the history of this strain for 40 

 generations. 



TABLE 6. Fluctuation in percentage of abnormal individuals in a normal strain. 



It is to be noted that the percentage of abnormal individuals is greater 

 in this strain artificially selected for normal venation than it is in'nature. 

 At first thought one would say that this is an effect of the environment 

 to which they were subjected. If this be true, environment may have 

 played a part in the production of the abnormal strains. I think, how- 

 ever, that it is not true. A sufficient explanation seems to lie in the 

 fact that a form of selection exists in nature which is keener than the 

 artificial sort, even when the latter is carried out under a lens. 



THE EEFECT OP SELECTION. 



Of late years there has arisen considerable skepticism concerning the 

 cumulative effect of selection except as a means of isolating "pure 

 lines." Jennings (1908) says : "Certainly, therefore, until some one 

 can show that selection is effective within pure lines, it is only a state- 

 ment of fact to say that all experimental evidence is against this." 

 Whether or not the present material has a bearing upon the question 

 thus clearly put depends upon the definition of a pure line. If a pure 

 line be defined as one from which nothing else can be gotten by selec- 

 tion, further discussion is not necessary. On the other hand, if inbreed- 

 ing (for the most part, brother X sister) for 10 or 15 generations and 

 rigid selection (in this case, with respect to wing-venation) may be 



