22 
Variation and Correlation in Brain- Weight 
Standard Unit of 
Giouping 
Brain-weight 
Age 
Stature ... 
Skull length ... 
Skull breadth ... 
Body- weight 
50 grams 
5 years 
3 centimetres 
5 millimetres 
5 millimetres 
5 kilograms 
These standard units were chosen after considerable study and experimenting 
and are believed to be the best values for the data discussed. However, it has 
been shown recently by other workers* that "the system of grouping adopted is 
within wide limits immaterial." An examination of those cases recorded in the 
present paper where it was necessary to use different units of grouping — notably in 
the Matiegka data where the brain-weights were grouped into 100 gram classes 
and the ages into 10 year classes — leads to the same conclusion. 
The standard deviation was evaluated according to the equation a = ^/fi2. In 
obtaining Sheppard's correction was used throughout for the characters brain- 
weight, stature, skull length, skull breadth and body-weight. The uncorrected 
" rough " moment was used in getting the standard deviation in age, as in this case 
there is no approach to high contact at either end of the range. 
The decimals have been retained to three places in the tabulated values in 
order that the significant value of the first place may be seen. It will be under- 
stood, of course, that in the computations the decimals were retained to a larger 
number of places for arithmetical reasons. 
• 4. Brain-weight Types. 
In brain-weight work hitherto it has not been possible to make accurate com- 
parisons of the mean weight of the brain in different races. There were two 
reasons for this: one that the probable errors of the means were not determined, 
and the other that no method was available by which allowance could be made for 
the differences in the mean stature and age of the samples of material to be com- 
pared. Obviously if brain-weight is correlated with stature and age, we should 
expect to get a considerable difference between the mean brain- weight of two 
groups which differed widely with respect to these other characters. Inasmuch as 
a know'le(ige of the degree of correlation between these characters affords a method 
whereby the material can be reduced to a " standard stature-age base," it will be 
worth while to examine the actual differences in brain-weight types in the four 
racial groups here discussed. 
* " Assortative Mating in Man." A Cooperative Study. Biometrika, Yol. ii. -pp. 481 — 498,1903. 
