DIHECTION AND FREQUENCY OF MUTATION 209 



more than 46,000 flies were examined. Throughout the whole 

 period there was not a single case of mutation to bar or ultra- 

 bar. Attention is called to the fact that various different kinds 

 of stocks were used and that in the cases listed in table 1 the 

 majority of the stocks are fulls derived by mutation from bar 

 or ultra-bar. Their recent origin from the bar stocks might 

 have led one to believe that they would possess a high degree of 

 instability. A single 'oval full' mutation different from any of 

 the bar mutations was the only change observed. The oval con- 

 dition, however, is due to an accessory factor and is not in the 

 bar allelomorphic series. 



Full therefore shows no tendency to form bar or ultra-bar since 

 the time of the single appearance of the former, as noted by 

 Tice in 1913. 



2. Bar eye 



Tables 2, 3, 5, and 6 and figures 2, 3, and 5 give the data for 

 bar eye. Attention is called to the great variety of bar stocks 

 used, including unselected bars, high selected bars, low selected 

 bars, reverse selections from high bar, bar mutations from ultra- 

 bar, emarginate and a high or ' broad ' bar recently received from 

 Prof. T. H. Morgan. The differences between these various 

 stocks are of two kinds: 1) differences in origin, some being 

 derived by mutation from ultra-bar, while the majority came 

 directly from full, and, 2) differences due to accessory factors. 



The total number of observed individuals is 85,008. Among 

 them there were noted 52 separate mutations to full, or one in 

 1636, which gives a mutation coefficient of 0.00061. There were 

 also three mutations to ultra-bar, or one in 28,363, giving a 

 mutation coefficient of 0.00003. It is probable that very few 

 mutations to full were missed because of the ease with which full 

 or heterozygous individuals may be distinguished from bar. 

 Mutants from bar to ultra-bar may, however, be missed, as 

 noted above, unless facet counts and breeding tests are made. 

 The true number of these latter mutations may therefore be 

 appreciably larger than is indicated in the table which includes 

 only the tested cases. Among the mutations to full also a con- 



