362 H. G. MAY. 



numbers in the upward selected lines and successively lower 

 numbers in the downward selected lines, and in each case the 

 three lines were very close together. The average facet number 

 was raised from 98.0 to 139.5 and lowered from 98.0 to 83.7. 

 The maximum was raised from 182 to 213 and the minimum 

 was lowered from 45 to 36. This increase in the range was not 

 due to a larger number of counts, as 500 flies were counted from 

 the stock and only 450 from the selected lines. 



THE EXPERIMENTS OF MACDOWELL. 



In the same year E. Carlton MacDowell published the 

 results of selections for higher bristle numbers in a race of 

 Drosophila with extra bristles on the thorax. For six generations 

 the number of extra bristles increased, but failed to rise any 

 higher in forty additional generations. The response to selection 

 was not so definite as in the experiments of Zeleny and Mattoon. 

 The variation in the averages from different lines was very large. 

 While the average of a large number of lines increased with each 

 generation, that of any particular line often decreased for a 

 generation. This wide range of fluctuations in the averages 

 from different lines was partly explained by the existence of a 

 correlation between the number of extra bristles and the size 

 of the fly. 



There is, however, another possible cause that was apparently 

 overlooked. MacDowell selected for extra bristles in two 

 arbitrarily limited rows on the dorsal side of the thorax, but 

 states that even when no extra bristles were present in those 

 rows "these flies frequently showed extra bristles on other 

 parts of the thorax." The factors for extra bristles evidently 

 controlled noL only the number of extra bristles in those rows, 

 but also the number of extra bristles on other parts of the thorax 

 and possibly over the entire body. Selection, then, was made 

 for only a small part of a variable character. A high number 

 of bristles in the area under observation would in general indi- 

 cate a high number of bristles on other parts of the body, but 

 might actually be accompanied by a low number of extras else- 

 where. Low offspring from high parents could be accounted 

 for in that way. The efficiency of the selection would also 



