Early Anthntpoliujical RotearcJu-nt 



r2» 



or say 2 personfi twins in 101 persons. Now in his own data for 94 sets of 

 twins he found that they had 1065 uncles and aunts, and amonpj the.s*- were 

 27 sets of twins or 54 twins in 10<!5 persons, or, say, 1 person a twin in 20. 

 Galton accordingly concludes that twins are far more I'recjuent among the 

 relatives of twins than in the general population or twinning must be an 

 hereditary character. Galton further noted that on the father's side there 

 were 538 luicles and aunts with 28 twins among them, and on the mother's 

 side 527 uncles and aunts with 2(5 twins among them. Hence he concluded 

 that the hereditary tendency to twinnirit,-- was the same in the male and 

 female lines'. 



Two other interesting points are given in this j)aper: (n) the ca.se of a 

 woman in a family remarkable for twins, who.se single children were poly- 

 dactyle, but not the twins, and (6) a pedigree — unfortunately omitting 

 single births — in which the intermarriages of three twinning stocks show 

 eight sets of twins, one triplet, and one (piadruplet'^ 



I have endeavoured in this chapter to trace one strand of Galton's labours, 

 that which shows him jiassing from the human side of geography to anthro- 

 pology, and so to heredity and race-betterment. The reader, who has had 

 the patience to follow my analysis of Galton's papers, will have observed that 

 Galton was becoming more and more conscious that a statistical treatment 

 of both anthropology and heredity was necessary, and he steadily set himself 

 to understand the then existing statistical processes and to develop them 

 where necessary. I have not hesitated to indicate the statistical weakness 

 displayed in some of the papers discussed in this chapter, because it was 

 Galton's growing consciousness in this matter that largely led to his later 

 contributions to statistical theory. 



In 1872 Mrs Galton's mother had died and in 1874 Galton's mother 



' Aswuiuiiig one person in tifty to lj« a twin, Galtou'8 94 .set« of twins lead us to a table of 

 the following character: 



If we assume for a moment that twinning is the result of some practically contiDuous variable 

 exceeding a certain value, the correlation for uncle and nephew would be -1817, rather low for 

 this relationship (-2 to -3), although such a value has occurred for pliysical characters occasionally. 

 It is possible that Ansell's value fi)r the frequency of twin births is somewhat exaggerated. 



' A niimber of peiligrees of twinning stocks were presented some years ago to the Galton 

 Laboratory, and will ultimately be publishiMl, but having been collected because of the twinning 

 frequency, it is not ea.sy to use theni for measuring the intensity of here<lity in twinning. 



p o n 17 



