512 Universifi) of California Puhlications in Zoology [Vol. 13 



is not true, however. F^Xa descended from normal = 86 N. -|- 10 

 ex. = 11.6 per cent. F^X-c descended from extra crosses = 44 N. -|- 5 

 ex. = 8.1 per cent. 



Summary of Experiments on Selection 



1. Selection for an increased number of extra bristles makes no 

 advance, the tendency being to return to the conditions of four and 

 five bristles. 



2. High-grade parents do not produce high-grade children, in fact 

 there is no defuiite relation between parent and offspring as to the 

 number of extra bristles. 



Discussion 



Certain of the data seem to point to the fact that we are dealing 

 with a ]\lendelian inheritance and that three factors are concerned, 

 giving frequently the 63 to 1 ratio rather than the. simple ^Mendelian 

 ratio of three to 1. The results here agree with the work of Nilsson- 

 Ehle (1909), Shull (1914). and Gates (1915). The intermediate 

 ratios, 5 to 1. 7 to 1, and others, might be explained either by fluctua- 

 tion or alternating dominance. The latter is easily conceivable as 

 shown in the fluctuating percentages of extras in tables III, IV, V. 

 and VI. If the dominance of normal were complete, there would have 

 been more regularity in the percentages. Some of the departures from 

 expectation are of course not mathematically significant, but the 

 various standard deviations have not been worked out. 



The fact that some of the results are so different from those, of Mac- 

 Dowell (1915), as for instance, the failure to develop a line producing 

 practically all extras, may be partially explained by the fact that he 

 was dealing with eleven generations and in this case we have only 

 five. Nevertheless, results were different from the beginning and it 

 seems necessary to find some other explanation than that of the loss 

 of restrictive or inhibiting factors. 



The fact that the family having the greatest number of extras, 

 the Y family, also had the greatest number of other mutations or 

 variations, namely, the club-wing mutant, the beaded wings, the abdom- 

 inal variations and the fly with only one wing, seemed to point to 

 the facts that there were other factors involved and that the germ 

 plasm was in an unstable condition. The increase of sterility among 

 the extra-bristled flies and the fact that thev were not so viable may 



