Personal Identification and Description 



167 



After a few remarks on scars, which consist chiefly in noting how few he 

 had found which destroyed the patterns to any considerable extent, and how 

 even in these cases with "rolling" generally enough is left for sound identi- 

 fication (see our Plate VI, p. 154), Galton turns to another matter, which 

 needs possibly more criticism or at least an ampler treatment. He considers 

 that there are certain main types of finger-prints, "arches," "loops," "whorls," 

 etc. There are also, he admits, transitional forms which create difficulty in 

 classification, but he says the result of statistical observation shows these 

 intermediates to be relatively few. He considers therefore the finger-print 

 types to be analogous to ordinary genera, and in order to illustrate this he 

 takes the case of the loop, and (a) counts the number of ridges in AH (see 

 our Fig. 21 (iii) and (iv), p. 163), (b) measures the index VY/OI, and (c) the 

 index AO/AH. Using both hands, and populations numbering 140 to 176 

 individuals only, he forms six frequency distributions, reducing them to 

 percentages. For example: 



Percentage Number of Rid yes in 166 Right Thumb Loop Prints. 



Galton does not apply an individual test for normality of distribution to 

 these rather abnormal-looking distributions*, but reducing them to their 

 medians and quartiles (see our Vol. n, pp. 385-6, 401) compounds them 

 together to form a single average "ogive" curve (see our Plate II, p. 31). 

 His final comparison is as follows: 



Considering that we have 965 observations to start from, this does not appear 

 on the face of it a very good agreement, and even Galton (p. 22) contents 



* He merely places his observed values alongside the normal curve results, and says that 

 considering the paucity of observations "there is nothing in the results that contradicts the 

 possibility of much closer conformity when many more observations are dealt with." (p. 19.) 



