BREWSTER. — RANGE OF VARIABILITY. 273 



In practice C. V. may be found most readily by separating the given 

 numbers into two groups, as discussed above, and finding the mean of 

 each. The difference between these two means, divided by their sum, is 

 the C. V. of the structure under consideration. 



Application of the Method of Pleasuring Variability to the Problem of the 

 Relation between Individual Variations and Specific Differences. 



Thesis. 

 While it is generally a<rreed that specific characters are more subject to 

 strikinsr variations in individuals than are the characters common to allied 

 species, it is not clear how far this relation extends, I wish to show that 

 what is true of obvious variations and sports is also true of those minute 

 differences between individuals which only careful measurements can 

 detect ; or, in other words, that any measurable quality is, in general, 

 variable in individuals in proportion as it is a distinguishing character 

 of the group to which the individuals belong. 



Evidence. 



Table A is based on the extensive tables of body measurement of 

 twenty races of men, which are given by Weisbach ('78) in the Appendix 

 to his *' Koroerraessunsen," etc. Each of the numbers of the first eight 

 columns is the CV. of a single dimension of a single race, and the ninth 

 column gives the average for eight races of the C. F,'s of each dimension. 

 The values of these C. F.'s only approximate to the true values, for the 

 reason that the number of individuals measured is not sufficiently large 

 to eliminate all accidents of age and sex. 



An examination of Table A shows that these eighteen dimensions are 

 of nearly the same relative variability in each of the eight races. This 

 fact is well brought out by Table B, in which the largest number of each 

 column of Table A is replaced by 1, the next smaller by 2, and so on up 

 to 18, From Table B it appears that certain dimensions — as, for ex- 

 ample, the height of the forehead — are always decidedly variable ; and, 

 on the other hand, it appears that other dimensions — such as the length 

 and breadth of the head — are more constant. 



The last column of Table A, " Mean of 20 Races," gives the coef- 

 ficient of variability, not of any individuals, but of the means of each of 

 the twenty races of "Weisbach's tables. That is, the mean value of a 

 dimension in each race is treated as are the dimensions of individuals in 

 other columns. This column, therefore, shows the distribution of racial 

 differences in the same way in which the remainder of the table shows 



VOL. XXXII. — 18 



