2IO 



C. B. BRIDGES AND A. H. STURTEVANT. 



chance deviations; and one might therefore be led to suppose 

 that it is due to an' actual variation in the strength of linkage. 

 While this conclusion may be correct, it is not necessary, since 

 there is another important factor which must be considered 

 namely, the effects of differential viability. In order to get 

 definite data regarding the manner of action of this disturbing 

 element we have made some crosses in which it may be studied 

 without the complication of linkage. If curved flies heterozygous 

 for black (B c v b c v ) be mated to blacks heterozygous for curved 

 (b C v b ), the same four classes of flies as in the above tables 

 should be produced, but now in equal numbers. If equality is 

 not shown the deviation cannot be due to linkage, but must 

 probably be attributed either to the error of random sampling, 

 or to differential viability. The results actually obtained in 

 these experiments are shown in the following table: 



Black 1 (het. for c t ,) X curved (het. for b). 

 I 



489 



49S 



384 



407 



That there really is no linkage present is evident from the 

 totals, since the complementary classes give totals of 879 and 

 896, respectively. The deviation of these numbers from equality 

 is less than half the standard error a very close approximation 

 to expectation. These totals show some effects of differential 

 viability, especially in that the curved flies were less numerous 

 than those with normal wings. A study of the individual bottles 

 brings out some other interesting points. It is obvious that 



1 The first five cultures below were from curve 9 X black d", the others from 

 the reciprocal cross (black 9 X curved cf ) 



