76 071 the Inheritance of the Finger-Print 
these new coefficients are distinctly higher than we should expect but depend on the 
assumption that inheritance of type is indifferent as to finger. Now if this be true, 
the correlation coefficient between any one finger of a sibling and all the other 
fingers in a second sibling should be the same ; but if inheritance of type depends 
absolutely on the one finger as I have assumed by correlating each finger separately 
there would be no correlation between one finger and any other finger in a sib- 
ship. I decided to test this on the available data in two cases, namely for the 
left thunib and for the left first finger. I did not go further than this for several 
reasons, chiefly because the labour involved is very great, and since I am doubtful 
about the wisdom of combining males and females it seemed waste of time to 
continue till we had a larger sei ies of data for the whole hand, and were not entirely 
dependent on about 360 pairs of brothers and sisters from Galton's earlier data. 
The cross tables were made so that each pair of siblings gave two entries. The 
numbers in each table averaged just over 700. The following coefficients were 
obtained. 
TABLE XIII. 
Left Thumb of first Sibling and Left Thumb of second Sibling 
» First „ „ 
„ „ ,, Second „ ,, 
Third 
„ „ „ Fourth „ ,, 
Left Thumb of first Sibling and Right Thumb „ „ 
First 
„ „ - - „ Second „ „ 
Third 
Fourth 
477 
368 
309 
235 
314 
410 
316 
267 
344 
269 
It will be seen at once that all these coefficients are significant but all are 
lower than the correlation between the same finger in siblings*; that is to say 
inheritance of type is partially indifferent as to finger but not wholly so, and to 
find the true measure of correlation between type of finger-print we should correct 
by using the formula ^" ^^.f' " . Adopting this correction for the right thumb we 
'477 ^ 2'832 
find that R = — ^, _ — = '512 which is a very reasonable value for the coefficient 
1 -I- 5-4684 ^ 
of correlation between siblings. 
I did the same thing for the left first finger, and found the results in Table XIV. 
These results agree fairly well with the results previously found for the resem- 
blance between brothers and I think it is reasonable to assume that when we can 
apply similar corrections to the parental tables we shall reach results corresponding 
equally well. In other words it is extremely probable that finger-prints are in- 
* It will be noted at once that like fingers have the highest, homologous fingers the next highest 
correlation. 
