DIRECTION AND FREQUENCY OF MUTATION 215 



that the dh-ection of selection has an influence upon the direction 

 of mutation. In these experiments facet counts were made of all 

 individuals so that the data have a high degree of accuracy. 

 The results are shown in figure 2, in which the vertical scale is a 

 logarithmic one so far as facet counts are concerned. A unit of 

 the scale represents a factorial change which causes a change of 10 

 per cent in facet value or approximately the same as that produced 

 by a difference of one degree Centigrade. The middle horizontal 

 line is the mean value of the unselected population which serves as 

 the zero of the scale. The upper horizontal line is the position of 

 full eye on this scale and the lower horizontal line the position of 

 ultra-bar. The upper zigzag line represents the fluctuation of the 

 high selection line for forty-two generations and the lower zigzag 

 line the fluctuation of the low selection line. The vertical 

 dotted lines give the origins and directions of the observed 

 mutations. It is to be noted that there is one mutation to full 

 in the parental generation, P, before the beginning of selection 

 and that there are four mutations to full and one to ultra-bar 

 from the high selection line and three to full and one to ultra-bar 

 from the low selection line. Since there are 14,327 individuals 

 represented in the diagram with eight mutations to full, it follows 

 that the mutation rate to full eye is 0.00056, which is of the 

 same order as that of the whole group of bars, 0.00061. The four 

 fulls in the high line and three in the low line are as close an ap- 

 proach to equality as can be obtained in this number. 



The two mutations to ultra-bar are too few in number to 

 furnish any accurate rate of mutation. The fact that two of the 

 three observed mutations of this character come in the 14,000 

 individuals in which facet counts were made as opposed to one in 

 the 71,000 individuals in which facet counts were not made makes 

 it seem still more probable that, as indicated above, some mutants 

 of this kind were overlooked in the later groups. It is interesting 

 to note that one of the two mutations to ultra-bar in the selection 

 lines came from the highs and the other from the lows. 



The data at hand, in so far as they go, give no support to the 

 view that direction of selection has an influence upon the direction 

 of mutations. 



