LINKAGE VARIATION IN DROSOPHILA 6 



The mutant races used in these experiments will be fully 

 tlescribed in a series of papers by Morgan and Bridges. In 

 this paper I am considering only the small fraction of data in 

 which we have records of more than one brood from a single 

 female. 



THE SECOND CHROMOSOME DATA 



The first case in which this relation for the second chromosome 

 was clearly shown was that of purple and vestigial given in 

 table 1. Of the eight females whose tests are given in the 

 table, seven showed a decrease in the percentage of crossing- 

 over and one (F) showed an increase, which, however, was 

 smaller in amount than any of the decreases. Likewise, the 

 totals for the second broods when compared with the totals 

 for the first broods showed a decrease of nearly three units in 

 the amount of crossing-over. That differential viability has 

 little or nothing to do with this difference is evident from the 

 regular totals. This -is shown even more clearly by the converse 

 case ('repulsion'), where the difference is entirely negUgible. 

 In all of these experiments contrary classes are affected similarly 

 and to a like degree, which would not be the case if viability 

 were causing the change. 



Of the six females tested (table 2) all showed a decided drop, 

 and the totals show a rather greater drop (5 units) than in the 

 'coupUng' case. 



In obtaining data upon any linkage case it is best to have half 

 the data in the form of 'couphng' and half as 'repulsion' experi- 

 ments. It has been showai by comparative breeding tests that 

 differential viability can be to a great extent ehminated by 

 careful attention to the conditions of breeding — particularly 

 by breeding in pairs in large culture bottles with the right amount 

 of well prepared food. We may offset even this remaining 

 disturbance by balancing the viabilty of a certain class against 

 itself. For example, let us say that in the case of purple vestigial, 

 the class vestigial is poorly viable. If, then, vestigial occurs in 

 an experiment as a crossover class, that class will be too small 

 and a false linkage value will be obtained. The remedy is to 



