To the Editor: The article by Kenneth S. Kendler, M.D. (1), in which he reported that heritability accounted for 43% to 75% of the variance in social support, illustrates several potential pitfalls in attempting to link biological (i.e., genetic) with social variables. First, Dr. Kendler cannot definitively say much about the effect of heredity on social relations because he has no method for measuring environmental variability or its impact on social support measures. A good analogy would be height, which within certain environments would look largely genetically determined but, measured across diverse environments, its genetic load would markedly decline. Second, Dr. Kendler cannot determine if there are environmental variables that correlate with monozygosity but not dizygosity. Environments may not be equivalent for identical and fraternal twins. There is a rich literature suggesting that identical twins are often treated differently from fraternal twins (2). Third, Dr. Kendler assumes that genetic influences combine in an additive way so that the more similar the genotype, the more similar the behavior. However, it is probable that genetic influences interact to contribute to a specific behavior. Therefore, while identical twins exhibit similar behavior because they share the same combination of genetic influences, fraternal twins may show disproportionately less similarity in behavior—i.e., they have 50% common genetic influences but not 50% of the same combination of influences. Fourth, Dr. Kendler wonders whether he is measuring genetic factors that influence subjective perceptions of the social environment or genetic factors that influence the ways twins objectively interact. However, he ignores the linguistic nuances of social network terminology such as "friends," "confidants," "demands," and the like. For example, he assumes that variables such as number of friends or confidants are more objective than variables examining the emotional nature of relationships. This is not necessarily true. My research (3), for example, has found that within certain environmental niches, attribution of friendship differs considerably from more traditional conceptions, not because of individual subjective (perceptual) differences but because of norms and rules within the cultural subgroup. Fifth, the atheoretical nature of deriving social support factors—i.e., by using factor analysis—raises questions as to why one would expect to find an underlying biological component to each of these statistically generated factors. It would be interesting to know if any of these social factors had similar loadings in other environmental contexts. Finally, although varimax rotation minimized correlations among factors because at least some of the variables within the six factors are likely to be highly intercorrelated (e.g., it is probable that number of friends correlates with number of confidants), many of the models that he tested may be measuring commonalities among variables within factors. This may explain, in part, why all of the genetic models had uniformly high explained variances.