74 
On the Inheritance of the Finger-Print 
Again if N be the total number of pairs 
i /S {au) = |/S + & > 
where ps,? is the correlation of and and p^^ of and x' g^. Accordingly if 
Pss - To (Pn + /322 + • ■ • + Pio lo). and pss' - yV (/3s.5')> 1-6. are the mean correlations, 
we have 
Thus we obtain finally : 
Here is a mean of organic correlations in the same individual, and and 
joss' are mean direct and cross coefficients of heredity. 
If inheritance be unassociated with likeness of finger, we have p^.^- = p^s and 
accordingly 
= 10^.<,s/(l + 9rs,/). 
If inheritance is only concerned with like fingers p^s' = 0 and 
nnt' = Pss/(1 + 9rss')- 
If there were equal heredity for any characters or function of characters then we 
must have ?•„,,' = Pss, which leads directly to 
the relation suggested by Pearson for cross-heredity, i.e. the correlation between 
two different characters in parent and offspring*. It seems, however, that there is 
extremely likely to be a closer relationship than that of cross- heredity between 
finger-prints on different fingers, and the only means of testing this point is to 
measure directly p^g and pss-. 
Unfortunately Galton's later and longer series will not here help us. We are 
compelled to turn to his earlier series, not gathered for hereditary purposes, but 
containing — so to speak — accidentally some 600 pairs of siblings with all ten finger- 
prints given in most of our individual records. I decided to find f^^ from that data ; 
then assuming that inheritance of type was indifferent as to finger we could obtain 
limits between which the true correlation coefficient of type of finger-print might be 
said to lie. As before I included Central Pocket Loops in Composites. I combined 
males and females and used the correction for grouping. The results obtained are 
given in Table X. There are about 600 cases. The mean (taken from four figures) 
was found to be '6076. We can now proceed to apply this corrective factor to the 
coefficients we have found using the coefficient corrected for grouping. We are 
assuming for the moment that jo,) = pj.^, then the coefficients already obtained must 
be multiplied by 1-54598. The results are given in Tables XI and XII. Now all 
* Fhil. Trans. Vol. 197, p. 290, B. S. Proc. Vol. 62, p. 410, aud Biomrtrika, Vol. ii. p. 385. 
