308 A. FRANKLIN SHULL 



rately and averaged. The comparison of the two groups, with 

 respect to nuclear size, is briefly stated as follows : 



Relative nuclear volume of oocytes in generations pre- 

 ceding male-producers 0.0606 ±0.0018 



Relative nuclear volume of oocytes in generations pre- 

 ceding no male-producers 0.0580 ± 0.0020 



Difference of mean sizes 0.0026 ± 0.0026 



The difference is equaled by its probable error, which is usually 

 interpreted to mean that the difference is insignificant. A large 

 probable error does not, however, prove a result insignificant; it 

 merely does not prove it significant. To avoid any possible 

 error in the conclusion due to a probable error that is large be- 

 cause cells of very different sizes are included, a somewhat similar 

 comparison has been made involving only cells of approximately 

 the same size, namely, those whose drawings were from 2 to 3 

 square inches in area. To increase the number of cells in each 

 group, the generations were not in this case studied separately. 

 Instead, there were distinguished two periods of many male-pro- 

 ducers and three periods of few or no male-producers (table 1). 

 The small percentage of male-producers in generation 9 was 

 ignored. The oocytes (of the size mentioned above) from these 

 periods were combined into two groups which were averaged 

 separately. The following comparison was obtained : 



Relative nuclear volume of oocytes in periods of many 



male-producers 0.0584 ±0.0024 



Relative nuclear volume of oocytes in periods of few 



male-producers 0.0599 ±0.0020 



Difference of mean sizes 0.0015 ± 0.0031 



Here the difference is less than its probable error. It is worthy 

 of note, furthermore, that while in the earlier comparison the 

 mean relative nuclear size was a trifle greater in periods of male- 

 production than in periods of no male-producers, in the second 

 comparison the relative nuclear volume is less in periods of many 

 male-producers. The conclusion to be drawn, therefore, is just 

 what .vould have been drawn from a simple comparison of dif- 

 ferences of mean and probable error in any of the tests applied, 

 namely, that there is no relation between relative nuclear volume 

 of the oocytes and the number of male-producers. 



