12 
A NOTE ON ANTHROPOMETRY 
each of many individuals are of more value than numerous 
measurements taken from fewer individuals. The average of 
a thousand cranial indices is a more accurate figure than 
the average of twenty. And there is great difficulty in co- 
ordinating more than a few different kinds of data. Accuracy 
of mathematical expression of the difference between two 
classes of things diminishes as the number of features regarded 
as differential increases. 
It is also most important to classify individuals as accurately 
as possible. In Africa it is very difficult. Not everyone 
is a Swahili who calls himself one. Anthropometrically a 
man is only a Swahili if his ancestors so far as he knows lived 
on the coast and spoke Swahili. This definition excludes 
many who are Swahili in their own opinion. A Swahili, 
for instance, whose paternal grandparents were Digo and 
maternal ancestors Makua, should have an entry to himself. 
A similar difficulty exists in the case of other tribes. Many 
Kikuyu are of Dorobo ancestry. A simple way of discovering 
the fact is to ask if they eat game. Classification, also, should 
be as detailed as possible. The Wagunia, for instance, should 
have a page to themselves and not be classed with Mombasa 
people. 
A note on the interpretation of data may be of service. 
A pure race will vary little from its own average in respect 
of any one physical feature. All will, that is, have nearly 
the same stature and nearly the same shape of head. Curves 
drawn to represent such data will be sharp and symmetrical. 
In the case of races containing two or more types these curves 
will be differently shaped. It might be thought that the 
offspring of a broad-headed man and a long-headed woman 
would have an intermediate-shaped head. But the law 
governing the transmission of parental differences is that 
the children more often take after one parent or other in 
respect of each character. Only frequently repeated blends 
are liable to result commonly in intermediates. Thus the 
curve of a mixed race shows evidence of diversity of origin 
in irregularity of outline. It can easily be understood that 
a broad flat-topped curve would indicate a fusion of different 
types, while a curve with a shoulder in it would show the 
