Further Observations on the Selective Elimination of Ovaries in Slaphylca. 183 



means from the theoretical lines bear the impress of the errors of 

 random sampling. 



Now it is quite evident that, if there be a proportionally greater 

 failure of ovaries at the tip of the inflorescence, the differentiation 

 of the ovaries just demonstrated will account for some of the difference 

 observed between the eliminated and the matured ovaries. It seems 

 doubtful, however, whether it can explain all of it. With a regressive 

 coefficient of only p = . loo, there would have to be an immense 

 preponderance of the eliminated pods from the more distal — and 

 extremely distal — regions of the inflorescence to account for the 

 difference in mean of . 500 observed in the study on Selective 

 Elimination. 



(b) Position on the Inflorescence and Radial Asymmetry 



and Locular Composition. 

 From the viewpoint of the student of selective elimination, it is 

 highly important to know whether radial asymmetry and locular 

 composition are in any degree dependent upon the position of the 

 ovary upon the inflorescence axis. The coefficients for r^^ and r^^ 

 should at once yield the information we need. We have: 

 For Position and Radial Asymmetry: 



1906, Table V, r = — . 0869 + . 0149. 



1908, Table VI, r = — . 0359 zb • 0106. 



1909, Table VII, r =^ — . 0357 + . 0148. 

 For Position and Locular Composition: 



1906, Table VIII, r = — . 0882 + . 0148. 



1908, Table IX, ;---(-. 0260 +. 0106. 



1909, Table X, r ^ — . 0134 + . 0148. 



The more distal ovaries seem to be a little more symmetrical 

 than the more proximal. Possibly there are also fewer locule with odd 

 numbers of ovules, but two of the constants are not even statistically 

 significant and such a conclusion carries very little weight. 



We know from the earlier study that number of ovules per locule 

 and radial asymmetry are positively correlated. It may be in part 

 due to this that we find low negative values for ;-pa. 



(c) Number per Inflorescence and Radial Asymmetry 



and Locular Composition. 

 To make quite sure that tliere are no correlations with characters 

 other than position on the inflorescence which may affect these 



