MUTATIONS, VARIATIONS, AND REIyATIONSHIPS OF THE OENOTHERAS. 35 



between the deviation in the mean, dM, and that of the coefficient of vari- 

 abihty, dV , not only of the characters studied in 1905, but also those of 

 the preceding year derived in the same way. 



Deviation of means and variabilities, in per cents of the corresponding values for Oenothera 



lamarckiana. 



1904. 



O. rubrinervis 

 Length of leaf . , 



Width of leaf 



I-eaf ratio 



Number of branches. 

 Total branch length . 



O. nanella, 1904. 

 Height 



O. rubrinervis, igo$. 



Length of leaf 



Width of leaf 



Leaf ratio 



Height 



Number of branches. . . 

 Total branch length . . 



Length of ovary 



Length of hypanthium. 

 Length of cone 



0. rubrinervis, igo5. 



Thickness of ovary 



Thickness of hypanthium 

 Thickness of cone 



O. gigas, 1905. 



Length of ovary 



Length of hypanthium. . . 



Length of cone ... 



Thickness of ovary 



Thickness of hypanthium 

 Thickness of cone 



O. lata, 1905. 



Length of ovary . 



Length of hypanthium. . . 



Length of cone 



Thickness of ovary 



Thickness of hypanthium 

 Thickness of cone 



In this table all the deviations which are less than the sum of the probable 

 errors are printed in bold-face figures, and may be left out of the discussion 

 as having no significance. If the signs of the remaining numbers are taken 

 into account it is found that 16 of the 30 characters studied showed a greater 

 variability in the mutant than in the parent, while only 4 were significantly 

 less variable in the mutant. Thus it is found again, as in 1904, that there is a 

 well-marked tendency for the mutant to show a greater variability than its 

 parent, though this relation is far from being absolute. The hint then recog- 

 nized, that the degree of departure from the mean of the parent might be 

 related to the degree of increase in the variability, is also faintly repeated, but 

 is too slight to be now considered of any significance. 



The single study in the correlation of characters in 1904 showed a consid- 

 erably less correlation in the mutant than in the parent species, and we have 

 already seen in the study of leaf characters above that, though that study 

 was unsatisfactory, such evidence as there is seems to accord with the view 

 then expressed that this represents a general fact. In order to gain further 

 evidence on this point the coefficients of correlation of the parts of the buds 

 were calculated, and all the original measurements are given in the correlation 

 tables appended. 



