338 ^ Proceedings of the Koyal Society of Edinburgh. [Sess. 
rational explanation, or that the value of the dominance ratio had been 
raised by the prevalence of epistacy (non-linear interaction of factors), a 
suggestion for which no direct evidence could be adduced. 
In the light of the above discussion in which we have deduced the 
distribution of allelomorphic ratios from the conditions of equilibrium 
with selective influences, from which condition it is probable that natural 
species do not widely depart, we find that the value J for the dominance 
ratio is produced by the asymmetry of the distribution, and in such a 
manner as to be independent of the activity of the selective agencies, 
provided that this exceeds a certain very low level. When differential 
survival to the extent of only about 1 per cent, in a generation affects 
the different Mendelian factors, in a population of only a million, and 
far more for more powerful selection, or a larger population, the dominance 
ratio will be very close to its characteristic value of J. 
The importance of the fact that this ratio is independent of the 
intensity of selection, lies not only in the fact that the intensity of 
selection is usually incapable of numerical estimation, but in the fact that 
factors having effects of different magnitudes on the soma, which are 
therefore exposed to selection of varying intensity, and contribute very 
different quota to the variance, are all affected in the same manner ; 
those factors which by their insignificance might be exposed to selective 
influences which are not large compared to the effects of random survival 
will be precisely those which have little weight in computing the 
dominance ratio. 
9. Assortative Mating. 
With assortative mating it has been shown (3, p. 414) that the 
deviations from the mean of the three phases of any factor have, owing 
to association with similar factors, mean genotypic values given by the 
formula 
T _i_ ^ 
1 ? -I- 
iF-m 
^1-A 
p 
1 
2pq 
K = A: - — 
iP-kli 
1- A 
<1 
when i, j, k are the deviations in the absence of association, A measures 
the degree of association produced by assortative mating : p, q are the 
gene frequencies, and P, R the corresponding phase frequencies for the 
homozygous phases. 
