64 
Oil the Inheritance of the Finger- Print 
so I have decided to give here the result for the first grouping only, which is the 
one that I prefer on the whole as it makes composites a rather more satisfactory 
group in size. I have used mean square contingency throughout and considered 
the correction for the number of cells only when any doubt arose as to whether 0^ 
was significant or not but have not subtracted this correction from 0'^ before 
finding C.^*. 
(4) Before I applied biometric methods to the problem of finger-prints some 
endeavour was made to consider the inheritance of types from the Mendelian 
standpoint. The difiiculties attending any attempt to consider finger-prints from 
this point of view consist not only in the existence of transitional forms, but from 
the fact that there may bo inheritance with change of finger. It would be exceed- 
ingly difficult to collect cases only of both parents having all their finger-prints 
unique in type, for example a father with 10 whorls and a mother with 10 whorls, 
or a father with 10 whorls and a mother with 10 arches. Yet something of this 
kind seems needful, if we admit transference of type from one finger in parent to 
another in child. Galton's later data at any rate will not help us in this matter as 
we have only the prints for two fingers provided. If we assume inheritance is to 
the same finger we can obtain from Galton's data the following tables for inter- 
pretation on Mendelian lines : 
Inheritance of Finger-Prlnt Type on Left Forefinger. 
Father's Type. 
Mother's 
Type 
Arch 
Loop 
Composite 
Whorl 
Totals 
A. 
L. 
C. 
w. 
A. 
L. 
C. 
W. 
A. 
L. 
C. W. 
A. 
L. 
C. 
W. 
m 
Arch 
Loop 
Composite 
Whorl ... 
1 
1 
19 
30 
1 
6 
2 
5 
2 
1 
8 
4 
18 
35 
6 
4 
14 
141 
9 
28 
6 
24 
6 
3 
7 
47 
11 
27 
1 
7 
1 
2 
15 
4 
1 
6 
2 
4 
5 
5 
8 
1 
6 
1 
1 
4 
55 
5 
18 
2 
1 
4 
7 
3 
11 
76 
394 
57 
119 
o 
Totals 
56 
63 
192 
39 
92 
9 
22 
12 
18 
82 
7 
21 
646 
Inheritance of Finger-Print Type on Right Forefinger. 
Father's Type. 
Mother's 
Type 
Arch 
Loop 
Composite 
Whorl 
Totals 
A. 
L. 
C. 
W. 
A. 
L. 
C. 
W. 
A. 
L. 
C. 
W. 
A. 
L. 
C. 
W. 
Child's Type, 
Arch 
Loop 
Composite 
Whorl ... 
8 
5 
1 
6 
27 
5 
2 
3 
14 
3 
5 
13 
31 
5 
1 
25 
147 
19 
31 
4 
23 
3 
5 
10 
32 
14 
28 
2 
5 
1 
4 
15 
5 
3 
3 
4 
10 
2 
12 
2 
5 
46 
7 
27 
2 
2 
9 
3 
13 
82 
371 
72 
127 
Totals 
14 
40 

25 
50 
222 
35 
84 j 8 
27 
17 
16 
85 
4 
25 
652 
* The "oorrectiou for cells" is the mean value of the mean square contingency when the association 
is zero. It forms of course the proper and adequate test of zero association, but Professor Pearson tells 
me that he is now extremely doubtful whether it should be subtracted from the mean square con- 
tingency when the latter is really significant. 
