W. E. Agar 325 



positive parental correlation, since the majority of mothers bom in the 

 first 15 weeks are correlated with offspring bom in the same period, and 

 correspondingly for the last 12 weeks. It is evident therefore that in 

 order to reduce the effect of environmental modification, and allow true 

 genetic differences (if such exist) to manifest themselves, a correction 

 must be applied to the original measurements which will counteract the 

 modifications produced by environment. The correlation table was there- 

 fore prepared in a new way. The difference between each individual 

 ratio and the mean ratio of all individuals born within five days before 

 and after it, was found, and the correlation table was reconstructed with 

 these values. In this way the effects of environment are more or less 

 smoothed over, since each individual, instead of being compared with the 

 mean for the total population is only contrasted with those individuals 

 born at about the same time, and therefore subjected to more similar 

 conditions. The method does not provide an entirely satisfactory correc- 

 tion factor, since 10 days is in some cases not long enough to include a 

 satisfactorily large number of individuals, and yet is long enough to 

 allow of a certain amount of change in the environment. 



The parental correlation as found in this way is '115 + '040, somewhat 

 less than three times its probable error. 



Moreover, if we turn our attention to the few individuals with the 

 highest ratios (and since the mean for the hybrid clone is less than that 

 for either of the parental clones, these are the only cases which need be 

 considered as evidence of segregation), we find that the highest hybrid 

 ratios were one of 2*75 and two of 2*45 (cf. Fig. 3). Unfortunately no 

 offspring were obtained from these three individuals. There were four 

 mothers of ratio 2*35, and these had 14 daughters with mean ratio 2*01. 

 There were also four mothers with ratio 2"25, and these gave 11 offspring 

 also with a mean of 2*01. These means are slightly above the means for 

 the hybrid clone, and hence are responsible for part of the positive corre- 

 lation coefiicient, but they are below the mean for the clone H, and much 

 lower than the lowest ratio found in the RR clone. If we examine, not 

 only the daughters, but the total descendants of these eight mothers, we 

 find that the four with ratio 2 '35 gave a total of 27 female descendants 

 with a mean ratio of 1*96, and the four parents with ratio 2*25 produced 

 42 female descendants with a mean ratio of 1*99. As the mean of the 

 whole hybrid clone is 1*97, it is plain that the high ratios of these 

 females, though it chanced to re-appear to a slight degree in their 

 daughters, was not maintained in their remoter female descendants. 

 These eight mothers also produced together a total of 48 male offspring, 



