24 MUTATION'S, VARIATIONS, AND RELATIONSHIPS OF THE OENOTHERAS. 



With respect to the number of branches, the figures in this table do not 

 fully represent the tendency in 0. rubrinervis to develop a large proportion of 

 its axillary buds into branches, as many of these buds had started into growth, 

 but for some reason had been arrested before they had attained a length of 

 I cm., under which length none were counted. Comparing these results with 

 those of 1904 it is found that with respect to variability Oenothera lamarck- 

 iana presents essentially the same condition in the two years, though the 

 values of both number and length of branches has increased in the latter year. 

 \r\ Oenothera rubrinervis, on the other hand, although there has been in 1905 

 a great reduction both in number and length of branches and the coefficients 

 of variability are considerably changed, they are still much higher than in the 

 parent species. This will be best appreciated by a comparison of the results 

 for the tw^o years as shown in the following table : 



The greater variability of the mutants was taken in 1904 to indicate that 

 these forms might not yet have acquired, perhaps, that fixity of habit exhib- 

 ited by an older species, and the fact that Oenothera rubrinervis underwent 

 greater change in its mode of branching on being cultivated in a new environ- 

 ment than was exhibited by 0. lamarckiana when subjected to the same 

 change would accord well with this suggestion. 



The increase in the number and length of the branches in O. lamarckiana 

 and the considerable decrease in those of O. rubrinervis must not be inter- 

 preted, however, as in any sense an approach to actual identity in these two 

 species, in regard to branching habit, for in the latter year they were even 

 more distinct in general type of branching than before. Oenothera rubri- 

 nervis had simply lost its long basal branches, while O. lamarckiana had retained 

 them. 



THE SIZE AND SHAPE OF LEAVES OF OENOTHER^V RUBRINERVIS. 



Some of the difficulties attending the statistical study of leaves were pointed 

 out in our earlier paper, and the complete suppression of the basal branches of 

 0. rubrinervis in 1905 has made the continuation of the comparison between 



