66 
On the Inheritance of the Finger-Print 
The symmetry of (ii) and (vi) and again of (iii) and (v) are possibly suggestive ; 
while a comparison of (i) and (iv) seems to indicate that the whorl is to some 
degree dominant over the arch, though both whorl and arch must be heterozygous 
in their nature. We have not succeeded in putting any satisfactory interpretation 
on the above equations, although we have made a variety of attempts. It is possible 
that some one better versed than we are in handling Mendelian ratios will be able 
to throw more light on the matter. But we feel that the above equations are in 
themselves forced because : 
((f ) there is no real justification for putting the Composite with the Loop, 
(b) there are no really clear-cut categories, but a great variety of types with 
transitional forms, and 
(c) there exists good evidence to show that the type is not necessarily inherited 
from one finger of the parent to the same finger of the child. 
Failing a Mendelian inteipretation, which appeared to us at all satisfactory, we 
have discussed the inheritance in this paper solely from the biometric standpoint. 
(5) Parental Correlation in Finger-Print Types. 
We will first consider the later and hxrger series of data collected by Francis 
Galton in which prints were taken of tlie first finger only and deal with the 
parental tables. 
TABLE L 
Right first finger 
Left first finger 
Uncorrected 
Corrected 
Uncorrected 
Corrected 
for grouping 
for grouping 
for grouping 
for grouping 
Father aiiH Son 
■291 
■333 
•379 
•434 
Father and Daughter 
■321 
■363 
■236 
■273 
Mother and Son 
■369 
■423 
•366 
•416 
Mother and Daughter 
•366 
•411 
■341 
•386 
Mean Parental \'alue 
■337 
•383 
•334 
■382 
I have not worked out the true probable error of these contingency values. 
If we use the probable error of r it is of the order of '03 ; contingency would 
be higher but as all the values of the coefficients are obviously significant, it 
seemed needless to undertake the labour of finding the true probable error. The 
mean parental value for both fingers is "38 when corrected fur grouping, which is 
a lower value than that generally found for the relationship between parent 
and child. The coefficient between the left fingt-r of fafher and daughter is dis- 
tinctly lower than any other of the coefficients iu the table, and this difference 
exists whatever grouping I use. I was at first very interested to find this smaller 
value for the father and daughter as Francis Galton in his study of finger-prints 
