The simulations illustrated show that a plausible model for mate selection can generate data on the similarity of twins and their spouses which are remarkably consistent with a transitive model for the effects of mate selection. That is, biological considerations impose constraints upon the relative values of correlations which are not foreseen, for example, by the some advocates of conventional path models although they might be predicted by common sense. In particular, the correlation between the spouses of twins is expected to be non-zero under a model of phenotypic assortment and turns out to be approximately equal to the product of the twin correlation and the square of the marital correlation. The relative magnitudes of the correlations derived from an empirical study of such relationships should enable models of phenotypic assortment to be tested more rigorously. Including both identical and non-identical twins in the sample studied should permit the inherited and cultural components of the mating system to be identified with more conviction. In the event of one sex playing a more significant role in mate selection for particular traits, such studies should reveal diagnostic patterns of familial correlations as long as male and female twins and their spouses are analysed separately. If the analysis is restricted to phenotypic correlations of the parents, the qualitative findings do not appear to be greatly affected by selection due to assortative mating although a reduction in variance is to be expected if a large proportion of individuals is unable to mate. In such cases twins will also be significantly concordant for mating. The consequences of such varied regimes of assortation for the population structure and the relationship between traits in subsequent generations remain the object of future inquiry. © 1979 The Genetical Society of Great Britain.
