282 CALVIN B. BRIDGES 



a crossover by an equal amount of data in which this same 

 class is a non-crossover. It is often not convenient or possi- 

 ble to have complementary crosses of equal weight; but what- 

 ever is done in that direction, however little, is of advantage, 

 and even a partially balanced result is to be preferred to one 

 from only one type of cross. With improvements in culture 

 methods, inviability effects have been very much reduced 

 everywhere. Also with the great increase in the number of 

 mutations, there is now provided an abundance of forms which 

 show only negligible inviability. Our regular work utilizes only 

 these viable forms, and except for very special purposes those 

 mutants which show more than a slight inviability are avoided. 



The first back crosses of purple vestigial had shown a marked 

 inviability for vestigial and a slight inviability for purple. The 

 new back crossefe showed practically no inviability for purple 

 and a very, moderate amount for vestigial, but still enough to 

 repay the added labor required by the balancing cross. As in 

 the first back cross test of the female, the linkage shown was 

 fairly strong. Since the linkage shown by second broods proved 

 to be different from that of firsts, only first broods will be con- 

 sidered for the moment. The 'coupling' experiment (table 7) 

 gave a total of 2839 first-brood flies, of which 305, or 10.7 per 

 cent, were crossovers. The 'rep^ulsion' first broods (table 8) 

 gave a total of 2335 flies, of which 303, or 13 per cent, were 

 crossovers. When the first-brood data from both these experi- 

 ments are combined so that the inviability is balanced, the 

 crossover value is 11.8 (table 9). 



These two component crossover values differed slightly from 

 each other and from the value (9.1) obtained in the original 

 experiment. It may be questioned whether the difference in the 

 crossover values was entirely due to inviability. Slight differ- 

 ences of this order, but many of them undoubtedly significant, 

 are continually appearing in our work. Other known causes of 

 linkage variation besides inviability are: differences in the age 

 of parents (Bridges, '15), or of the temperatures at which the 

 experiments are conducted (Plough, '17), or mutant 'crossover' 

 genes (Sturtevant, JNIuUer, and Bridges), and probably to several 



