MEASURE OF RESEMBLANCE OF FIRST CO.USINS 7 



A, B and C are types of male, D, E and F are types of female, G, H, I, A' of male 

 and female cousins. There are thus ten types of simple first cousins. It was con- 

 sidered desirable to keep these ten types distinct in order to ascertain how far 

 resemblance was modified by change of sex in descent from a common ancestor. 

 Special cases of abnormal cousinship in the first degree were not included. All 

 individuals dealt with were adults. The four characteristics : General Health, 

 Ability, Temper and Success in Life, providing 5, 7, 4 and 5 categories, admitted 

 at once of tables of contingency being formed of at least 4x4 groups, and these were 

 at once reduced by the method of mean square contingency. This was done for all 

 the ten types of cousinship. In the case of Temperament there were only the 

 alternatives and the ' Betwixt ' groups. We were thus compelled to use either con- 

 tingency on a 3 x 3-fold grouping or else assume the material to have a Gaussian 

 distribution and apply fourfold table divisions. In the latter case the results will vary 

 somewhat according to the alternative with which the ' Betwixt ' group is associated. 



The Temperament results are, however, in our opinion the least reliable of the 

 series. We believe this to be due to the fact that the ability, success, health and 

 to some extent the temper of an individual are matters of common knowledge or 

 repute ; but that temperament as we have classified it is less generally realised. 

 There is little doubt that some of the recorders had not previously formed a general 

 estimate of temperament, and have taken that of a particular individual as a standard 

 to classify other members of the family by. We should not be inclined accordingly 

 to place much stress on the Temperament results as proving anything beyond the 

 basal principle that temperament is an inherited charactei\ 



In the second series of investigations, it was considered that it would not be 

 without interest to deal with an entirely novel physical character, i.e. novel from the 

 standpoint of inheritance, and accordingly the hand was selected as easily accessible 

 and, at any rate for some characters*, capable of fairly accurate measurement. Other 

 physical characters readily ascertainable were eye and hair colours and general health. 

 Accordingly a cousin schedule was issued with the directions for measurement noted 

 below. See Appendix A, p. 21. Much time and energy had already been spent 

 over an endeavour to reproduce in a cheap manner the eye and hair scales used as 

 standards in the Biometric Laboratory. Ultimately we had to content ourselves with 

 an admittedly imperfect chromolithograph of hair coloursf. For the eye scale we used 

 a hand-painted scale. Miss Mary Beeton kindly painted on a printed blank the 

 irides of 24 eyes, painting one eye at a time in about 100 copies from a standard 

 glass eye. To these scales was added a cheap but quite efficient hand-spanner 

 prepared by the Cambridge Scientific Instrument Company. The scales, spanners, 

 directions and schedules were circulated in the same manner as the similar material 



* R. S. Proc. Vol. 65, pp. 126—151 : "Data for the Problem of Evolution in Man. A First Study 

 of the Human Hand." By M. A. Whiteley and Karl Pearson, 

 t Reproduced Biometrika, Vol. v. p. 474. 



