No. 549] ON DIFFERENTIAL MORTALITY 519 



in all cases. Only the deviations of the sub-samples from 

 the general population, i. e., the differences obtained by 

 Sheppard's correction was applied to the second moment 

 subtracting constants for the general population from 

 those for the sub-samples producing fertile plants, as 

 given in Table V, require our attention. 



Consider first the differences in mean seed weight. 



They are equally divided between positive and nega 

 tive, the average weight of the seeds which produced fer- 

 tile plants being in 14 cases higher and in 14 cases lower 

 than that of the general population of seeds from which 

 the plantings were made. The average value of the posi 



TABLE V 



tive differences is +.163 units ; of the negative differences 

 — .188 units; of the entire series of 28 comparisons 



