12 ETHEL M. ELDERTON 



omitting the cousins from the record. On the same ground the marriage of niece or 

 nephew with uncle or aunt seems to be a marriage of exactly the same degree of 

 kinship as a marriage between first cousins. 



The only grandparental data at present reduced for man* are those for eye-colour 

 and the eight cases give a mean value of "32 1. This is somewhat higher than the 

 value ("27) for cousins, and pigmentation data in horses have given an almost equal 

 value. Still other species show rather smaller intensity, and until further data are 

 reduced for the case of man, especially for psychical characters, we are not convinced 

 that the grandparental relationship is definitely more important than the cousinship. 

 At any rate, even with our present values ( - 27 as against "32) it will be seen that it is 

 not reasonable for the purposes of medical or actuarial diagnosis to neglect the cousins, 

 and make a considerable point as to the grandparental constitution. The grandparent, 

 the uncle or aunt and the cousin are practically on the same footing with regard to 

 relationship or intensity of kinship as measured by degree of likeness of character ; 

 and it seems probable that any scientific marriage enactments would equally allow or 

 equally forbid marriage between grandparent and grandchild, uncle and niece, aunt and 

 nephew and between first cousins. This conclusion is reached on the assumption that 

 the undesirability of marriage depends on the closeness of likeness in the gametic 

 constitution, and that on the average the i-esemblance of the somatic characters may 

 be taken as a measure of the average gametic resemblance between any two classes. 



5. We now turn to the details of Table III. 



We first ask whether there is any sensible difference between the intensity of 

 inheritance in males and females. We note that the probable error of any individual 

 result runs from about -02 for cousins of same sex to "03 for cousins of different sexes. 

 Our table shows us that the average for pairs of male cousins is the same as that for 

 pairs of female cousins, i.e. "26. If we could lay any stress on the difference "02, we 

 should assert that cousins of different sexes were more alike than cousins of the same 

 sex. But we certainly cannot, and thus, as far as our data go, we can only conclude 

 that difference of sex makes no difference in degree of likeness. 



In the next place we may consider whether type of cousinship makes any differ- 

 ence in the intensity of resemblance. Our mean values for all the characters range 

 from '22 to - 31 according to the type, and it might be thought that this offered 

 sufficient range to answer the question. As defining the types there are two considera- 

 tions to be noted, (i) a difference of sex in either generation, parental or cousinal, and 

 (ii) a change of sex in descent. Neglecting the first we have : 



No change of sex in descent in : A (SO), E (-27) or K (-29) ; 



One change of sex in descent in : C (-23), F (-28), G (-24) or H ('31) ; 



Two changes of sex in descent in : B ('24), D ('22), or / ('29). 



* The Family .Records of the present series provide unreduced material for seven characters, and this 

 will shortly be dealt with. 



t Phil. Trans. Vol. 195, A, p. 115. 



