.70 
On the rnherltance of the Finger-Print 
larger numbers. The biological origin of such a difference in one (or two) out of 
the ten fingers is not obvious, but it is the sort of clue which should not be 
allowed to drop. At the same time I am doubtful if it will be confirmed eventually 
that there is a sex difference. In both series the higher value found for 
is due mainly to a disproportion in the number of arches between males and 
females, but in the shorter and earlier series the males have a larger proportion of 
arches than the females and in the later and longer series the females have a 
larger proportion than the males ' A recount, after an interval of three or four 
yeai's, confirmed the original figures, and I think the discrepancy may be due to 
the fact that in both series we are dealing with family finger-prints and a pre- 
dominance of a certain type of finger-print in a large family of males or females 
may be the source of the divergency. For the right thumb there is no other 
series with which to compare it, but I should certainly lay no stress on the sex 
difference in that case; again arches are mainly responsible for the high value 
found for j^-, and as ai'ches are very rare on the thumb the percentages of arches 
are based on 7 males and 11 females and with such small numbers are liable to a 
large error. This sex diff'erence may or may not exist ; on the available data 
I think it would be unwise to be dogmatic on the point. Whether or not there is 
a sex difference there is a diff'erence in the right thumb and right first finger in 
this data and under the circumstances I was reluctant to put males and females 
together, but the number of pairs of sisters is so small that I decided not to 
keep them separate. I sought justification for this step in the following way. 
I worked out the cori'elation coefficients between each finger and every other 
finger for (a) males alone, (6) males and females together and found that in the 
tables involving the right thumb and the right first finger the differences between 
the coefficients found were not significant and I trust that the results for inherit- 
ance in the case of the right thumb and right first finger will not be upset by 
dealing with sibships and neglecting the diff'erence in sex. Table VII gives the 
differences in the correlation coefficients and the probable error of the difference. 
Again I have used the p.e. of r which is an underestimate of the probable 
error of a contingency table and if the diff'erences are not significant when this 
TABLE VII. 
Eight Thumb 
Eight First 
Finger 
Left Thumb ... 
■001 + -021 
•035 + -036 
Fir-st 
•034+ -036 
•009+ -026 
Second ... 
•069 + ^037 
•031 + ^027 
Third ... 
•016+ •OSS 
•028 + -033 
Fourth . . . 
•053 ± •OSS 
•001 + ^036 
Right Thumb ... 
•049 ± •ose 
First 
•049 + •ose 
Second ... 
•043+ •0.36 
•057 + -028 
Third ... 
•037 + ^034 
•009+ -033 
Fourth . . . 
•056 + •OSS 
•026 ± -033 
