62 Variation and Correlation in Brain- Weight 
the outlying points were left out entirely in the graphical representations. The 
non-biometrical reader should understand that each plotted point carries weight in 
proportion to the fraction which the number of individuals in the array on which 
it is based is of the total number of individuals. With this caution in mind the 
substantial linearity of the regression through the bulk of the observations, which 
alone are of importance, becomes evident. 
The regression equations showing numerically the relation between brain- 
weight and stature, may now be given. In these equations W denotes the 
probable brain-weight of an array of individuals of mean stature S or s in centi- 
metres. As before, the equations involving S are based on the " total " series, 
and those involving s on the " young " series. S is the standard deviation of the 
array having the probable mean brain-weight W. 
(9) 
Swedish 
6 
w= 
:915-054-|-2-859,S', 
2 = 
104-533 
(10) 
6 
w= 
919-374 + 2-914S, 
2 = 
107-405 
(11) 
? 
]V= 
= 421 •994 + 5-234;$', 
2 = 
94-421 
(12) 
? 
w= 
:451-643 + 5-121s, 
2 = 
99-294 
(13) 
Hessian 
6 
w= 
:913-592-h2'857*S', 
2 = 
110-787 
(14) 
6 
w= 
:950-214 + 2-723s, 
2 = 
109-628 
(15) 
? 
w= 
: 834-624 + 2-714,5', 
2 = 
100-643 
(16) 
? 
w= 
= 894-031 +2-460 s, 
2 = 
100-161 
(17) 
Bavarian (J 
1F= 
= 836-667 + 3-127 >S', 
2 = 
118-452 
(18) 
? 
TF= 
627-157 + 3-824S, 
2 = 
103-562 
We may now proceed to the general conclusions to be drawn from the data 
given above regarding the relation of brain-weight to stature. In the first place 
it is to be noted that the correlation coefficients between these two variables 
are for each series of the same general order of magnitude as those between brain- 
weight and age, though of course positive instead of negative, as they are in that 
case. Or it may be concluded that during adult life, at least, brain-weight is 
only a very little more closely related to general size of body, in so far as we may 
take stature as a measure of this, than to age. When we turn to the regression 
coefScients, however, it is seen that a unit of stature connotes a somewhat larger 
change of brain-weight than a unit of age. Thus, in the Swedish males, an 
increase of 10 cm. in stature connotes an increase in brain-weight of 28'59 gr., 
while an increase of 10 years in age connotes a decrease of 19"39 gr. in the 
brain-weight. 
As in the correlation of brain-weight with age, the females have uniformly a ■ 
higher degree of correlation between stature and brain-weight than the males. 
I am inclined to think that the same general sort of explanation may be given in 
this case as in that of the brain-weight and age correlations. The correlation 
between brain-weight and stature and the regression of brain-weight on stature 
are sensibly the same for the early adult period of life as for the total adult 
period. Without any question, I think, this correlation is to be regarded in its 
origin as a growth correlation, the relation between the two variables being 
