MEASURE OF RESEMBLANf 'E OK FIRST COUSINS 



17 



but it must be remembered that the series were short, i.e. treated as symmetrical 

 tables we had only 68, and 218 entries, and for male and female cousins 113. The 

 results given in the following- table were reached. 



Table VI. Non-quantitative Characters, Second Scries. 



It will be seen at once that (i) the preponderating intensity of pairs of female 

 cousins no longer exists, (ii) the eye colour values are, however, very high, in one case 

 at least approaching the intensity of the resemblance of siblings, and (iii) the generally 

 higher value obtained in the case of the measurable characters of the same series is 

 maintained. We have already noted that eye and hair scales were used in these 

 observations, 24 eye and 24 hair tints being given. There were 6 categories in the 

 Health graduation, but the " Very Delicate " and the " Very Robust " categories were 

 only very slightly represented, so that for Health merely 3 x 3-fold contingency tables, — 

 " Robust," " Normally Healthy," " Delicate " seemed possible. For eye colour the 24 

 eve tints were first classed as " Pure Blue," " Blue with some orange," "Pure Grey," 

 " Grey with some orange," " Hazel-Green," " Hazel-Brown," " Brown " ; the two greys 

 were then clubbed together, as also the two hazels to form a 5 x 5-fold table for 

 contingency. The 7 x 7-fold table seemed far too fine for the numbers, 68, involved 

 in the male and male cousin tables, and it was desirable to treat all three tables alike. 

 The 24 tints of the hair scale were first grouped into : " Very Dark," " Dark-Brown," 

 " Brown," " Light-Brown," " Fair," " Red." But for the male data only a single 

 "fair" and a single "red" occurred and only three "browns." Accordingly the 2nd 

 and 3rd categories and also the 5th and 6th were grouped together and a 4 x 4-fold 

 table used for the contingency of hair-colour. As samples, the Eye and Hair colour 

 tables for pairs of female cousins are given in Appendix B as Tables LXXX and 

 LXXXI. Now while we frankly admit that this Second Series, whether of 

 measurable or pigmentation characters, has a much too inadequate frequency to be 

 conclusive, still its drift is undoubtedly to confirm the view, that the average 

 resemblance of cousins is higher than that given by the Family Record results. It 

 approaches nearer the value indicated by the more precise of the " Record " characters, 

 and the more accurate of the hand measurements. The numbers in the first series are 

 large as compared with those of the second, and the second series also involves several 



