RELATIVES ON THE SUPPOSITION OF MENDELIAN INHERITANCE. 
425 
Thus if a group be chosen so that x — t, 
y of group is dJj/, 
z of group is c x c 2 t 9 
- r* • l 1 “P A , 
z or sibs is CjCg — 
also 
y of sibs is |^[1 + c 2 (l + 2 A)]£, 
y of sibs mates is fc T [] + c 2 (l + 2A)]c iy u./, 
z of sibs mates is fcj[l +c 2 (l + 2A)]A£. 
z of nephews is ^: 1 [2c,(l + A) + {1 +r 2 (l 4- 2A)}A]f, 
+ -c,). 
Again for cousins, if a group be chosen so that x = t, we have 
y of uncles is |^iC 2 ^?— + JqAfl - c 2 )Jt, 
Hence 
giving the correlation 
and 
hence 
giving the correlation 
2 of uncles is ^c 2 ^ ~ ^ , 
z of uncles mates is + ^c x A(l _ 
2 of cousins is ^ + x V^iA 2 (l -c 2 )Jf, 
iA»(i 
The formulae show that these two correlations should differ little from those for 
grandparent and great-grandparent, using the values already found, and putting 
Ci = 1 we have 
Stature. 
Span. 
Cubit. 
Grandparent . 
. -3095 
■2612 
•2378 
Great-grandparent . 
. -1891 
T503 
T353 
Uncle 
. -3011 
•2553 
■2311 
Cousin . 
. -1809 
•1445 
T288 
23. On the third supposition, that the marital correlation is due primarily to an 
association in the essential genotype z, we obtain results in some respects more 
intelligible and in accordance with our existing knowledge. 
From the fundamental equations 
/x = CjCgA, 
P = i(<5 1 C 2 + A> 
we may deduce 
V2 = 2^-/4, 
A = /*/(2p-/t), 
whence the following table is calculated : — 
Stature. 
Span. 
Cubit. 
Standard Error. 
(*■ 
. -2804 
T989 
T977 
0304 
P 
•5066 
•4541 
•4180 
•0115 
f 
. -5433 
•5351 
•4619 
•0160 
C 1 C 2 
•7328 
•7093 
•6383 
•038 
A 
. -3826 
•2804 
•3097 
•028 
i(l + A) 
. -6913 
•6402 
•6549 
•014 
