On the displacement of single-locus polymorphisms

And where, along the evolutionary track, some mutant gene did arise which was advantageous in its heterozygous degree of expression but deleterious homozygously, that gene, though temporarily multiplied as a make-shift arrangement, would usually have become replaced, after a while, by mutant genes of less deviant expression, occupying the same or other loci, which gave an equivalent advantage when they were present homozygously. For such genes would make possible a more uniform, and therefore (when all individuals were averaged) a closer approach to the adaptational optimum.