Ethel M. Eldp:rton 
69 
and left finger but is certainly not significantly different for the left hand ; it 
might be just significant if we compare sister-sister and brother-brother results 
for the right first finger; but I think not, considering that the probable error of a 
contingency coefficient is decidedly higher than the probable error of r. Before 
comparing these results with those found for other characters we will consider the 
results from the series in which the whole hand was taken. There are roughly 
380 males and 230 females. I decided to put males and females together if 
possible and work with sibships. I had therefore to see first whether the distri- 
bution of the types of finger-prints differed sensibly, as, if it did, the sexes could 
not be combined. I used the method described by Professor Pearson in Biometrika* 
and found the following values for ■)(^ and P for the different fingers. There 
were five groups and I worked out and P in the double classification, including 
Central Pocket Loops first with composites and secondly with whorls. 
TABLE VL 
^1 
R, 
R; 
U 
Including 
CP. L. 
IncludinR 
C.P.L. 
with Composites 
with Whorls 
F 
x'-' 
P 
9-21 
•06 
9-93 
•04 
10-87 
•03 
12^08 
•02 
4'04 
•40 
4^72 
•32 
1-47 
•83 
2-34 
•68 
5-91 
•21 
4 •.59 
•34 
3-89 
•42 
3^87 
•43 
•97 
•91 
r26 
•87 
4-03 
•40 
5^50 
•24 
3-70 
•45 
4^02 
•41 
6-57 
•16 
5^33 
•26 
It is clear from this table that, whichever grouping is used, there is no 
reason why the sexes should not be put together except for the thumb and first 
finger of the right hand. For the right thumb the odds are roughly 17 to 1 
or 24 to 1 according to the grouping used against the males and females being 
random samples of the same population and for the right first finger the odds 
are roughly 59 to 1 or 33 to 1 against. This sex difference in the first finger of 
the right hand is confirmed by the series in which only the first finger was taken. 
I find from that series that for the first left finger when C. P. L.'s are grouped 
with composites = 2-96 and P = "56 but for the right first finger yj = 9-86 anrl 
P --- ^04 and the odds are about 23 to 1 against males and females being random 
samples of the same population in these two series. That two independent series 
should both support the suggestion that there is a sexual difference between the 
finger-print type of the right forefinger (possibly also of the right thumb) is very 
noteworthy, and the possibility of it should be borne in mind and retested on 
* "On the Probability that Two Independent Distributions of Frequency are really samples from 
the same Population," by Karl Pearson, F.ll.S., Biometrika, Vol. viii. pp. 250 — 254, July 1911. 
