334 
Proceedings of the Eoyal Society of Edinburgh. [Sess. 
Since the logarithm does not increase very rapidly, we may say approxi- 
mately that A is proportional to - ^ 
It will be seen that to maintain the same amount of variability, as in 
the case of equilibrium in the absence of selection (section 4), the rate 
of mutation must be increased by a factor of the order o-a Jn. Even in the 
low estimate we have made of the intensity of selection on the majority 
of factors, this quantity will usually be considerable. The existence of 
even the slightest selection is in large populations of more influence in 
keeping variability in check than random survival. 
A further effect of selection is to remove preferentially those factors 
for which a is high, and to leave a predominating number in which a is 
low. In any factor a may be low for one of two reasons : (1) the effect of 
the factor on development may be very slight, or (2) the factor ma}^ effect 
changes of little adaptive importance. It is therefore to be expected that 
the large and easily recognised factors in natural organisms will be of little 
adaptive importance, and that the factors affecting important adaptations 
will be individually of very slight effect. We should thus expect that 
variation in organs of adaptive importance should be due to numerous 
factors, which individually are difficult to detect. 
Owing to this preferential removal of important factors the above 
solution only truly represents an equilibrium of the variability of the 
species under absolutely uniform conditions of selection when the new 
mutations which arise have the same frequency distribution of relative 
importance as those removed by selection. It must be remembered, 
however, that the change of variability even by selection is a very slow 
process, and that gradual changes in the physical and biological environ- 
ment of a species will alter the values of a for each factor, so tending to 
neutralise the tendency of selection to lower the value of era. Nevertheless, 
a will be on the whole numerically smaller for factors in the current stock 
tlian it is for fresh mutations. 
7. Uniform Genotypic Selection. 
If the heterozygote is selected to the same extent as the dominant, 
or h = a, it is easy to see by writing down the first generation, that a 
genetic ratio p : q, becomes in one generation by selection ^ 
writing I + /3 for ^ , 
E Lt A ■ 
q 1 ' 
