14 



Indiana University Studies 



continued thruout the experiment. For some reasons it would 

 be desirable to give the complete data, but it would take up 

 so much space that, instead, a summary generation by genera- 

 tion is given (Table I). This table gives in each generation 

 the total number of offspring, the number of offspring that are 

 normal, i.e. those with four bristles, and the number with 

 five bristles, six bristles, and so on as high as they go. 



The data for the males and females are kept apart with the 

 exception of the normal males and females in the first seven 

 generations. A glance at this table shows clearly that the 

 number of flies having extra bristles has been increased until 

 in the last generations no normal flies are found. It also 

 shows that the number of extra bristles has been increased 

 until the maximum, 15, is reached. Only three females, how- 

 ever, show this extreme number. Fourteen is the highest 

 number reached in the male. In this connection it should be 

 mentioned that the bristle number of the female is always 

 slightly higher than in the males. MacDowell observed this 

 same fact in his experiment and believed it was a question 

 of size, the larger flies having the higher bristle number. In 

 my work this suggestion has no support. Often the smallest 

 flies have the highest bristle number. As will be shown later, 

 there is present a sex-linked factor which when homozygous 

 produces a more marked effect than when heterozygous. Table 

 2 gives the mean bristle number of the parent and offspring 

 in each generation, and also the percentage of flies having five 

 bristles, six bristles and so on. The means of parent and 

 offspring are plotted in Figure 2. The percentage of flies in 

 each class is plotted in Figure 3. Table II and Figure 2 

 show the rather remarkable fact that selection has not only 

 been effective, but that it has been effective for 29 genera- 

 tions. MacDowell reached his maximum results in six genera- 

 tions. In most experiments on artificial selection the effect 

 is immediate and lasts for only a few generations. Castle's 

 experiments on hooded rats is "an exception, where he has pro- 

 duced results up to the seventeenth generation and believes 

 the end has not yet been reached. An analysis of the curves 

 in Figure 2 shows, however, that the increase for 29 genera- 

 tions has not been a gradual one, but that there are three 

 separate and distinct rises. From the first to the eleventh 

 generations there is a distinct rise, the mean reaching 6.567 in 

 the eleventh generation. From the eleventh to the fourteenth 



