NATURAL SELECTION—DOBZHANSKY AND ALLEN 367 
over, any genotype which possesses a higher net reproductive efficiency 
has a higher Darwinian fitness, and is, by definition, favored by 
natural selection. Higher Darwinian fitness usually goes together 
with superior adaptedness to the environment; however, the correla- 
tion is not perfect. 
A single example will suffice to illustrate the occasional miscarriages 
of natural selection. Dunn (1953) found a recessive gene in the 
house mouse which is lethal when homozygous. A population of mice 
in which this gene occurs in a certain proportion of individuals pro- 
duces, then, some inviable embryos. The gene is clearly deleterious. 
But this gene possesses the curious property that a male which is 
heterozygous for it and for its normal allele yields more spermatozoa 
carrying the abnormal than the normal gene. This automatically 
confers upon the abnormal gene an advantage in the population, and 
causes it to spread until its lethal effect in homozygotes checks its 
propensity to increase in frequency. Dunn has found that the lethal 
is actually common in many “normal” mouse populations, outside of 
genetic laboratories. Up to a point, therefore, natural selection 
favors the spread of a lethal gene in mouse populations because this 
gene happens to be able to subvert the male reproductive processes in 
its own favor. The reproductive success of a genotype is, in this case, 
opposed to adaptive success of the population. 
This discrepancy between reproductive and adaptive success arises 
because the former has but one dimension: the rate of perpetuation 
of a gene from generation to generation relative to that of an alter- 
native gene. Adaptation is multidimensional, and herein lie some 
unresolved problems about natural selection, particularly as it occurs 
in the human species. The pioneers of Darwinism were already aware 
that, in a social animal, the qualities which promote success in an 
individual are not necessarily those which are most useful to the 
society in which the individual lives. <A gene for altruism (if such 
existed) might be discriminated against by natural selection on the 
individual level, but favored on the population level. The outcome 
of selection would, therefore, be difficult to predict. One might specu- 
late that it would depend on the population structure of the species. 
A gene for altruism might be lost in large undivided populations, but 
might become frequent in a species subdivided into numerous compet- 
ing colonies or tribes. Moreover, adaptedness to a certain environ- 
ment, however perfect, need not go together with adaptability to 
changeable environments (flexibility, according to Thoday, 1953). 
For example, it is to be expected that of all the genotypes which are 
successful in times of abundant food supply, relatively few will be 
adaptable to periodic starvation; genotypes which can resist a large 
variety of infections are not necessarily the most successful ones in 
disease-free environments. 
