6o8 fViii. E. Kellicott 



that may underlie the fact that in the same community where con- 

 ditions of Hfe are remarkably uniform, the females should be so 

 much (23 per cent) more variable than the males. May it not be 

 because the females are at the same time more perfectly correlated ? 

 Variations from the type of a given character are not disadvantage- 

 ous because they are backed up by corresponding variations in other 

 characters; the "balance of fitness" is maintained in the more 

 variable individuals. The females remain more variable — are not 

 selected dov^n to the same level of variability as the males — simply 

 because their variability involves not single parts or organs but 

 groups of organs, in fact the entire organization. In other words, 

 the organization of the females, abmodal as well as modal is more 

 nearly a unit, the elements are better organized, less independent 

 of one another, /. e., simply "better correlated." 



Schuster ('03) found a relation between sex and correlation 

 in his crab measurements (Eupagurus) but there it was the 

 males which were the more variable and more highly correlated. 

 There seems therefore frequently to be a relation between sex and 

 correlation as well as between sex and variability. Whether the 

 relation is actually between variability and correlation will be 

 discussed presently. 



b Comparative Degrees of Correlation in External and Internal 



Characters 



In the toad the average degrees of correlation in both sexes of 

 the external and internal characters are .815 and .579, respectively 

 (Table XVIII), and it should be borne in mind that it is the inter- 

 nal characters which are the more variable. In man, to which 

 we are limited for comparative data among vertebrates, the rela- 

 tion is similar, as shown in Table XXII compiled from a number 

 of sources. And here too the internal characters are the more 

 variable. The coefficients of the brain are similar in general to 

 those of the viscera ranging between .17 and .40 while the skeletal 

 characters, excluding the skull, show an average correlation of .70: 

 the skull resembles the brain in this respect. 



In general these coefficients are comparatively high in the toad, 

 both in external and internal characters: I know of no form in 



