98 



Life and Letters of Francis Galton 



coats, and to remember that granular pigmentation varies enormously within 

 the range of coat-colours described as chestnut by hackney breeders. Galton 

 assumes that full red pigmentation counts for 1 '0 and takes chestnut to be 

 0*8 ; bay, 07; brown, - 4; and black, O'l. Then by using the results of the 

 several lines in A, he concludes that each chestnut parent contributes 

 40 units to the offspring, each bay 337 units, each brown 25 - 3, and each 

 black 10 "4. He is now able to deal with the crosses in B, C, D, E, F and G. 

 He finds that there should be in the offspring of the matings : 



But is this really conclusive ? It is possible that there are almost the same 

 amounts of diffused pigment in the different coats and that the visible 

 colours arise from the relative amounts of granular pigmentation. Had 

 Galton put the amount of red pigment = - 8 for all coat-colours, he would 

 have got his theoretical and observational numbers in perfect agreement. 

 But I do not think this would justify the assumption that the amount of 

 red pigment is the same for all coats. I think then that we cannot assume 

 his far rougher agreement is any proof of the numbers he has selected, or 

 indeed of his theory of average parental contributions*. After the experience 

 we now have had in the coat-colour of mammals, I feel fairly convinced that 

 it is necessary to supplement the macroscopic classification by microscopic 

 examination, for the categories formed in the former manner contain a great 

 range of both diffuse and granular pigmentation. 



Prepotency in Trotting Horses. Still another paper of the same year 

 occurs in Nature, July 14," 1898 (Vol. lviii, pp. 246-7). It is entitled " The 

 Distribution of Prepotency," and deals with data for American Trotting 

 Horses. Galton was much occupied at this time with the idea of the pre- 

 potency of individuals. He believed that some favoured individuals had a 

 power of impressing their exceptional characters on their offspring, and that 

 this prepotency was of the nature of a " sport." The American Trotting 

 Horse data provided, he considered, a method of testing this belief. Wallace's 

 Year Books give lists (i) of the sires of offspring any one of which has 

 succeeded in trotting one mile in 2 minutes and 30 seconds or less, or who 

 has "paced" (ambled) the same distance in 2 minutes and 25 seconds or 

 less ; (ii) of the dams of two such offspring, or else of one such offspring and 

 one such grandchild. Galton selected from these lists of sires and dams those 

 foaled before 1870 and therefore who would be at least 25 years of age in 

 the Year Book for 1896, which he was using. He considered that this would 

 give at least 20 years of breeding age to the parents and 5 years of attempted 



* I took the relative proportions of red to be r u r 2 , r 3 , r t , and determined their values to fit 

 li, G, D, E, F, G by least squares instead of guessing their values; the ratio of the r's was 

 1-00 : 1 04 : 1-14 : 1-06, almost a ratio of equality ! 



