No. 539] DISTRIBUTION OF PURE LINE MEANS 691 



were grown under conditions more heterogeneous than 

 those to which the 1908 plants were exposed. In the 

 second case, the S.D. might be directly raised, i. e., 

 heterogeneity in the crop may be merely a reflection of 

 heterogeneity in the substratum. 



There is no way of determining whether Eoemer's 

 cultural conditions were more heterogeneous in 1909 

 than in 1908, but it must be noted that in ten of the four- 

 teen cases the coefficient of variation is higher in 1909. 



Two ratios are to be examined, 



Mean Pure Line V aria bility 

 Parental Population Variability' 



Me an Pure Line Variability 

 General Population Variability * 



Consider first the ratio of the mean pure line to the pa- 

 rental population variability. If the offspring of the 

 individual parents are differentiated we should expect 

 to find the mean variability of the pure lines less than 

 that of the parent population, providing, of course, 

 that innate tendencies are not obscured by environmental 

 factors. Table II 10 gives the necessary data. 



Now the remarkable thing about these standard devia- 

 tions is that in the most cases the variability within the 

 individual "pure lines " in 1909 is greater than that of 

 a mixture of all the pure lines in 1908. The excess is 

 very striking in several cases. Of the fourteen com- 

 parisons, thirteen show a higher variability within the 

 pure line than in the population. For the "Viktoria" 



10 For 1908 the population 2 and C.V. are from Eoemer's Table I. The 

 1909 population 2 and C.V. have been calculated by the formulae given 

 above. The mean pure line standard deviations have been taken from 

 Eoemer's Tables II-IIT. None of the constants have been rechecked, since 



of stem in Table II is obviously a printer's slip for 2.13. The mean pure 

 line coefficients of variation are from Eoemer's Table X. These were not 



