58 
On the Inheritance of the Finger- Print 
In 1903* Galton set about obtaining a collection of family finger-prints for 
the purpose of studying heredity. The persistence of finger-print type throughout 
life established by Herschel and Galton was to make the inquiry a simple one ; it 
was unnecessary to collect adults only ; classification could take place at any time 
during life. 
As early as 1892 Galton had made a preliminary inquiry into the inheritance 
of finger-print types, but he was not satisfied with either the extent of his material, 
or with the classification of types, which he had at that time evolved. 
We shall give the results he obtained below. They were published in 1892 in 
his work entitled Finger-Prints. 
In this book Galton devotes a chapter to the question of inheritance but at 
that time there was no method of obtaining a correlation coefficient when the 
characters were not quantitative and Galton could only show that type of finger- 
print is inherited but coi;ld not measure the strength of inheritance nor compare 
it with the results obtained from the study of other characters. The method 
Galton used was the following: he found what he describes as (1) Random occur- 
rences, (2) Observed occurrences, (3) Utmost possibilities. He used for the first 
part of the investigation three categories only, namely arch, loop, and whorl. 
The types of 101 (by mistake for 100) couplets of prints of the right forefingers of 
school children were taken from a large collection, the two members A and B 
being picked out at random and formed into a couplet. These were tabled and 
fQund to agree very well with calculated random couplets. 
To study the fraternal relationship the observed occurrences were then taken 
from 105 fraternities and finally the greatest number of correspondences which 
could (without changing the total distribution of types) occur if the kinship were 
as close as possible were found. I now reproduce the results obtained by Galton 
which I take from p. 176 of his book. It will be seen that he only deals with the 
diagonal of his correlation table in this summary of the results. 
A and li both being 
Arches 
Loops 
37 -6 
42 -0 
61-0 
Whorls 
Random 
Observed 
Utmost possible 
1-7 
5-0 
10-0 
6-2 
10-0 
25-0 
We note that in all three cases the observed exceed the random but that this 
excess is not very marked, particularly in loops. Galton says that many other 
cases of this description were calculated all yielding the same general result. He 
* He made an appeal for material in the pages of Dkmetrika, Vol. ii. p. 356, 1903, and the letters 
accompanying the data he received show how active he was in the matter of collecting during 1903 
and 1904. 
