1921-22.] On the Dominance Ratio. 339 
Writing to represent complete dominance, and V=p‘^, R = g^ since 
+ '2pq)i + q^k = 0, 
k i-k pH — qVc _ 
^ _ !€ 
q^ p{p + 2q) 
and since i — k = 2a, we have 
I=i+ ^ 
2pq^ 
1- A 
4:aq^ , 
J= . '2aq{p-q) 
K = k- 
1-A 
. 4apq ; 
or 
If now the survival factors of the three phases are a, h, c, the effect 
of one generation’s selection is given by 
Pi__^ ap + hq 
g'l q^bp+cq qj^ ? 0’ 
since a, b, and c are near to 1 ; 
hence • 
a = p(a -b)-\- q(b — c). 
Now as I — J, J — K, the mean differences in any trait due to a single 
factor, are small compared with the whole variation within the population, 
we must take a — b, b — p proportional to I — J and J — K. In other words, 
a-b = (l-j)y, 
b-c = {J-K)y, 
where y measures the. intensity of selection per unit change in the trait. 
Hence 
a - y(pl — J + qJ — K) 
2a 
7- 
1 - A 
The general case of uniform genotypic selection when the mean values 
of the phases are modified by homogamy, therefore, reduces to the case 
already considered in which homogamy is absent. The total effect of 
homogamy is to increase the effect of selection by the factor 
1-A‘ 
The 
distribution of frequency ratios is unaltered, for although by introducing 
a difference between I and J the selective effect is made more intense when 
