STURTEVANT 



in major modifiers, though it is still 

 possible to interpret some of the minor 

 differences observed as being due to 

 uneliminated diversity in modifiers. 



The temperature control used was 

 not very exact, but maximum-mini- 

 mum daily records show that 25 °C 

 was maintained to within about ± 1 °C, 

 and even these deviations were prob- 

 ably of short duration. For the main 

 bodv of the experiments it has not 

 proved possible to detect any sys- 

 tematic effect of the fluctuations in 

 temperature that did occur. The few 

 experiments in which such an effect 

 is perhaps present will be specified 

 when described. 



The facets of the smaller eyes,— up 

 to and including homozygous bar,— 

 were counted directly under the bin- 

 ocular microscope, usually on ether- 

 ized flies, but in some cases on alco- 

 holic specimens. The eyes larger than 

 this were not found to be workable by 

 this method. Such specimens were 

 killed and cleared in KOH. The sur- 

 face of each eye was then removed and 

 mounted on a slide. By the aid of a 

 camera lucida a drawing was made, 

 representing each facet by a dot, and 

 these dots were then counted, each dot 

 being marked by a check as counted. 

 In all cases the right eye alone was 

 used. 



The main series of data is shown in 

 table .23. 



The table shows that homozygous 

 infrabar is about like bar over round 

 in facet number, but the two types can 

 be separated by a peculiarity common 

 to all the larger infrabar and double- 

 infrabar types, namely, a roughened 

 appearance of the eye, due to irreg- 

 ularities in the rows of facets. This 

 peculiarity is not present in bar eyes, 

 and is almost completely recessive in 

 bar over infrabar. In infrabar over 

 round (which is not far from round in 

 facet number) the roughness is vari- 



139 



able in extent, and may be not at all 

 evident,— in which case the type can 

 not be distinguished with certainty 

 from homozygous round. In other 

 stocks, where the modifiers are differ- 

 ent, it often happens that infrabar over 

 round is regularly conspicuously 

 roughened and is easily distinguishable 

 from round. This roughness of the 

 eyes may be taken as evidence that the 

 infrabar gene is qualitatively different 

 from bar, rather than being merely a 

 fraction of bar. 



The table shows that in general when 

 bar and infrabar are both present in an 

 individual, the infrabar produces al- 

 most as great an effect in reducing 

 facet number as would another bar, 

 though in the absence of bar the infra- 

 bar is far less effective. For example, 



B B' B 



g = 68, g7 = 348; but J, = 74. And 



in general, BB' is practically as eflFec- 

 tive as BB throughout the table. In 

 two cases the observed differences, 

 though surely not significant, indicate 

 that BJ5* is more eflfective than BB 

 (that is, the combinations with BB 

 and with B*). Similar relations are 

 shown in other parts of the table. 



The most striking relation shown 

 by table 23 is that the relative position 

 of identical genes affects their action 

 on facet number. There are three simi- 

 lar comparisons to be made: 



-^ = 68.1 versus — = 45.4 

 o + 



-^ = 73.5 versus—— 



D + 



50.5 



B* B^B^ 



^ = 292.6 ' versus — — = 200.2 

 B^ + 



B' 

 7 This value for -^ is different from the 



one of table 23. It is based on a series reared 



u • u B'S' . , .... 



at the same time as the — — with which it 



is here compared. The difference between 

 this value and that of the table is probably 

 due to temperature. 



