18 
Variation and Correlation in Brain-Weight 
was that usually followed : the pia and arachnoid were not removed before 
weighing. The special value of this material of Matiegka's lies in the fact that it 
is the most extensive series of brain-weights in which skull characters are so 
recorded as to make it possible to determine their correlation with the weight of 
the brain. In addition, data are given for the determination of the correlation 
between brain-weight and stature, and brain-weight and age. Unfortunately 
Matiegka's material is open to criticism in certain important respects. In the 
first place detailed tables showing the exact records for each individual are not 
given, but instead the material is thrown into the form of correlation tables. In 
making up these tables the units of grouping were not altogether wisely chosen. 
This results in introducing a certain error into the absolute values of the constants 
deduced from these tables. The exact cause, nature and amount of this error 
will be pointed out later in the paper. Again, the skull-length and skull-breadth 
were measured in a very unreliable manner. Regarding the method of making 
these measurements Matiegka says {loc. cit. p. 46): " Bei dem von mir verar- 
beiteten Materiale und zwar in beiden Instituten wurde gewohnlich auch die 
Ldnge und Breite des SchddeldacJtes d. i. an dem behufs der Hirnentnahme 
vorgenommenen Horizontalschnitte gemessen. Das so erlangte Langenmass ist 
daher bedeutend kilrzer als die grosste Schddelldnge, die Breite wohl haufig etwas 
hleiner als die grosste Schiidelbreite*." The effect of this procedure on the 
biometric constants is discussed later in this paper. My chief reason for including 
this somewhat questionable material was for the sake of comparison, and because 
reasonably long series of brain weighings are not so plentiful as to allow one who 
would study the subject a great range of choice in material. 
The data on the brain-weight of the English were very kindly placed at my 
disposal by Prof FT. H. Donaldson. He was able to obtain, some years ago, a 
copy of the original detailed manuscript tables which the English anatomist 
Marshall had compiled from Boyd's original data collected at the St Marylebone 
Infirmary and the Somerset County Lunatic Asylum. These detailed tables were 
never published, though it was on them that Marshall's well-known paper, " On the 
Relations between the Weight of the Brain and its Parts and Stature and Mass 
of the Body in Man f," was based. For the privilege of examining and using this 
unique and valuable material I am very grateful to Prof Donaldson. 
From the biometrical standpoint the raw material available for a statistical 
study of the weight of the brain is peculiarly complex. The great bulk of the 
statistics consists of material gathered at the autopsies performed in large general 
hospitals or other public institutions of similar character. The result of this is 
that the only large collections of brain-weight data available are not representative 
" random " samples of the general population. Instead they represent a group 
of the population which has been subjected to a rather stringent selective process 
of a peculiar kind. It would be very difficult, if not impossible, to formulate all 
* Italics Matiegka's. 
t Jour. Anat. and Physiol. Vol. xxvi. pp. 445 — 500, 1892. 
