LINKAGE VARIATION IN DROSOPHILA 



differs from the change for purple curved in any female by 3.8 

 units, which is practically the same amount. From this point, 

 that there is a negligible correlative change for the values for 

 black purple and purple curved, we obtain confirmation of the 

 view that this change is a general one which affects different 

 sections of the chromosome independently, but on the average 

 to about the same extent. 



From the percentages of occurrence of the three mutants 

 (table 9) it can be seen that differential viability has been al- 

 most entirely eliminated. In the case of black, 98 flies hatched 

 for every 100 expected — a very close approach to expectation. 

 In the second broods the viability is somewhat poorer, but 

 since the decreases are uniform no changes in linkage are to be 

 expected from that source. 



TABLE 9 

 Viability coefficients (data of table 7) 



The results in table 7 are not balanced here by an equal amount 

 of data for each of the converse experiments. To balance the 

 data of an experiment involving three loci requires four sets 

 of data, instead of two, as in the case of two loci. 



Fortunately, in the case of black and curved there are avail- 

 able data for such a balancing as is given for purple and vestigial 

 in table 3. Only flies in first broods can be used for this balanc- 

 ing, as I shall make clear in the conclusion. 



In a paper by Bridges and Sturtevant (Biol. Bull. '14), there 

 appear records of 7419 such first brood flies in 'coupling' and 



