CHANGE IN THE BAR GENE OF DROSOPHILA 301 



the standard deviations. In the case of factorial units the stan- 

 dard deviation may be used directly as the coefficient of varia- 

 tion. The values for full eye are based upon such a small num- 

 ber of cases wholly because of the tedious character of the count. 

 In the bar stocks counts can be made by placing the flies on their 

 sides on a small block of wood with a surface painted black and 

 so inclined as to bring the eye facets in a horizontal plane under 

 the microscope. The rows of facets can then be followed by the 

 observer and counts can be made with a fair degree of accuracy. 

 In practice there is, however, always a small per cent of error 

 which increases with the facet number. In full eye the large 

 number of facets combined with the curvature of the eye makes 

 this procedure impossible. It is necessary to resort to the mount- 

 ing of the eye on a slide under a cover-glass. The non-chitinous 

 parts are removed with caustic potash and the faceted part 

 mounted in Canada balsam and flattened out under a cover- 

 glass. With a high objective, a mechanical stage, and cross 

 hairs in the ocular, it is possible to count the facets with a fair 

 degree of accuracy, but it is not profitable to count a large num- 

 ber of individual eyes. 



Tables 3 and 4 and figures 2 and 3 give comparisons of the 

 range and variability of the three stocks in graphic form. There 

 can be no question that the stocks are absolutely distinct. Bar 

 and ultra-bar barely overlap in the females and not at all in the 

 males. As shown in the tables, the variability of ultra-bar, here 

 expressed directly by the standard deviation, is significantly less 

 than that of the bar stock from which it was derived, and this 

 difference between bar and ultra-bar remains throughout the 

 selection generations. Figures 2 and 3 bring out the same point. 

 The standard deviation value for full eye of course has no special 

 significance because of the small number of cases. It may per- 

 haps indicate that full eye is less variable than bar. 



