Pearl, Correlation in Brain- WeiHit. 



469 



accounted for solely by differences in other characters is shown 

 in the following table. 



TABLE I. 



'This assumes that the mean stature of the Bavarians would be the same in 

 the "young" as it is in the "total" group. The error introduced by this pro- 

 cedure is practically negligible. 



^Here again, on account of lack of data, the mean stature is assumed to be 

 the same in "total" and "young" groups. 



^The mean age for the Bohemian "young" series was calculated by taking the 

 first four age classes of the "total" series. The age as given is probably a little 

 too high, but the error cannot be great. See complete paper on this point. 



The values in the first column of this table were obtained 

 by the use of the indicated characteristic equations given farther 

 on in this paper (p. 478). Each of these represents the probable 

 mean brain- weight (in grams) of an array of the specified type in 

 respect to the other characters, sex, age, and stature. It shows, 

 in other words, what would be the effect on brain-weight of se- 

 lecting a group of Hessians of either sex to the age and stature 

 type shown by the adult Bohemian or Bavarian population 

 with which we are dealing. The table shows that even after 

 Hessians, Bavarians, and Bohemians are put on the same basis 

 with reference to sex, age and stature they still exhibit consid- 

 erable differences in the weight of the brain. 



3. Variation in Brain-zvcight. In degree of variability the 

 characters brain-weight and skull capacity are sensibly equal. 

 The coefficients of variation for brain-weight (and skull capaci- 

 ty) are intermediate in value between those which have been 



