Raymond Pearl 
57 
These are comparatively speaking slight changes, amounting respectively to 1'3°/^ 
and '8 7o of the mean brain-weight for the whole adult period. In the same 
length of time (10 years) males lose an average of •34" in stature or approximately 
•5 of the mean stature *. The diminution in brain-weight seems thus to be 
relatively somewhat more rapid than the diminution in stature. 
The fact that the regression is linear implies that this diminution of brain- 
weight with advancing age so far as can be judged from the material here 
discussed is sensibly equally distributed over the whole adult period. To be sure 
the slope of the lines is in all but one case (the Swedes) less when the individuals 
between 20 and .50 are treated separately, but the difference is a small one when 
the probable error is considered (cf Table IX or XIV). Splitting the material up 
in this way of course greatly reduces the already statistically small series, so that 
the probable errors of the coefficients of correlation for the " young " series become 
large. It is significant, however, that with a single exception (the otherwise 
abnormal Bavarian female series) all the coefficients of correlation between brain- 
weight and age for the " young " series are negative. The evidence presented by 
each of the four series here discussed, which are representative and have been 
shown to be statistically trustworthy, and which if not as large as might be desired 
by the statistician are among the largest ever collected by the neurologist, is 
mutually accordant, and leads to but one conclusion : namely, that after age 
15 — 20 there is a steady though very gradual diminution in the weiglit of the brain 
with advancing age. This conclusion I have emphasized because it is not in agree- 
ment with that reached by some eminent neurological authorities. I would further 
emphasize the fact that the conclusion is by no means final in the mind of the 
biometrician. It simply marks the point to which the material at present 
available leads. Very much larger series of brain- weighings are needed before 
the precise form of the regression can be finally determined. But the fact that four 
distinct series of fair length taken by different observers on different races lead 
to such closely accordant results as those shown in Table XIV may be taken as 
very strong presumptive evidence of the essential correctness of the conclusion 
reached. 
It is noteworthy that there is no tendency towards more rapid diminution on 
brain-weight after age 50 to 60. The rate of diminution shows no such marked 
changes at this time as has been maintained to occur by many of the workers on 
the subject. 
With the exception of the Swedish series the females in all cases show a 
higher correlation between brain-weight and age than the males. This difference 
seems in the case of the Hessians and Bavarians (total series) to be probably 
significant. According to Powys' (loc. cit.) figures the slope of the regression line 
between stature and age is slightly steeper for the females. It is difficult to 
assign any reason for this greater correlation between brain-weight and age in 
* Powys, A. 0.: " Data for the Problem of Evolution in Man. Anthropometric Data from Australia," 
Biometrika, Vol. i. pp. 30 — 49. 
Biometrika iv 8 
