Personal Identification and Description 



189 



Chapter XI (pp. 170-191) discusses the suhject of Heredity in finger-prints. 

 This is a most difficult problem ; it is not only that certain fingers favour certain 

 classes of patterns, but that certain patterns classed in different broad groups 

 are closely associated with each other. If we classify merely in arches, loops and 

 whorls, we may find two kinsmen who have really kindred patterns, e.g. one 

 having a plain arch and the other a nascent loop, classified as being as widely 

 apart, as if the one had shown a tented arch and the other a twined loop. 

 Again, supposing an extremely rare pattern occurs on the ring finger of one 

 kinsman, and on the forefinger of the second, are we to dismiss this from our 

 consideration of hereditary resemblance? It is almost inconceivable that a 

 mere Arch-Loop-Whorl classification, especially if confined to a few fingers, 

 can provide a true measure of inheritance although it may demonstrate that 

 heredity is a factor of finger-print determination. Galton, in his first series 

 of observations, confines himself to the fraternal relationship (boys and girls) 

 of 105 pairs, dealing with right hand forefinger only and using the simple Arch- 

 Loop-Whorl system. As we have remarked, this may show the existence of 

 heredity, but it cannot really measure its intensity. He obtains the following 

 table : 



Observed Fraternal Coiqrtets. 

 First Child 



Galton then pays attention only to the numbers occurring in the diagonal 

 column, i.e. identical prints in the fraternal couplet with the Arch-Loop- 

 Whorl classification. The numbers in round brackets are what are to be 

 randomly expected, the numbers in square brackets, the highest values 

 attainable for resemblance, on the hypothesis of independence of the marginal 

 totals. In every case the observed values lie between the random and the 

 highest values and Galton takes this as evidence of heredity. 



It will be seen that Galton is here aiming if rather ineffectually at some 

 process like the modern method of contingency. If we apply now his method 

 of centesimal grades we find for the degree of resemblance : 



Arches: 39'8 ; Loops: 18-8°; Whorls: 20-2°. 



Of these the last two are probably equal within the error of random 

 sampling. The first shows about double the relationship of the other two. 

 I do not believe this is due to a greater intensity of the force of heredity in 

 arches than loops, but solely to the fact that arches form a relatively rare and 

 homogeneous group, while loops and whorls are conglomerates and the use of 



