differences. The r-strategy is associated with relative ecological instability and tends to promote such qualities as small body size, rapid sexual maturation, short life span, and profligate reproduction. The K-strategy is associated with environmental stability and gives advantage to large body size, delayed maturation with parental care, longevity, and economical reproduction. By extrapolating the r/K between-species distinction to a continuum of within-species variability, human aggression is viewed as part of the r end and law abidingness as part of the K end of the dimension.

A second research thrust has been the study of familial homicide by Daly and Wilson (1988a,b). Based largely on the concept of inclusive fitness, they predict that homicide (and presumably other aggression) should vary inversely with the degree of genetic relatedness of assailant and victim and with the reproductive capacity of the victim.

This present round of sociobiological/evolutionary research is more empirical than many previous attempts to extrapolate from other species to human behavior. The research is also relatively new, so a large body of well-controlled studies has yet to be developed to permit assessment of the predictions. For example, Daly and Wilson present Canadian homicide data suggesting that filicide victimization is considerably greater for infants who have yet to reach their first birthday than for older children. To what extent is this due to inhibition because of the reproductive and temporal investment already made in an older child, to what extent is it a consequence of maternal postpartum psychoses, and to what extent is it due to the physical vulnerability of the young infant to a punitive blow that might injure but not kill an older child?

For theories about the r/K-strategies, the relevant multivariate data that would permit one to assess the proportion of variance in violent criminal participation attributable to the latent r/K-variable have not been reported. Presumably, individual differences in these strategies have some heritable component. The requisite twin or adoption data must be gathered and subjected to multivariate analysis.

Racial differences in arrests, homicides, etc., have sometimes been interpreted in terms of mean genetic differences among human groups (Rushton, 1988a,b). From a genetic perspective, race differences fall into the general category of group differences. There

are two distinct questions that may be asked about genetics and violence in different groups. First, is the heritability of violence large within each group? Within-group heritabilities may be derived through adoption or twin studies of each group. Second, to what extent are differences in violence between groups due to the genetic differences between them? This question addresses between-group heritability.

Empirical data suggest the possibility of important within-group heritability in European populations and some American populations of European ancestry. There are no comparable twin or adoption data to document within-group heritability among American minorities.

Even in the presence of such data, it would be incorrect to infer that genetic differences contribute to differences between groups. Within-group heritabilities are insufficient to predict the extent to which mean group differences are genetic (DeFries, 1972; Loehlin et al., 1975). For example, the demonstration of within-group heritability among Danes and among Iowa whites does not imply that differences in crime rates between Denmark and Iowa are due to gene pool variance. Differences in prevalence and incidence of violence between Northern and Southern Ireland might be explained more easily by social and political milieus than by DNA variation.

Wilson and Herrnstein (1985) consider it inconceivable that the rapid historical changes in homicide could be accounted for by changes in allelic frequencies. The two- to threefold increase in homicide rates that occurred between the 1960s and 1970s (Zahn, 1989) is almost certainly due to environmental factors, not all of which have been identified. Hence, it is plausible that large mean group differences could be perpetuated by environmental factors. Despite the fact that there are analytical models in genetics that attempt to account for such types of environmental diffusion (e.g., Boyd and Richerson, 1984; Cavalli-Sforza and Feldman, 1981; Lumsdem and Wilson, 1981), little is known about their empirical validity, let alone their direct application to human violence. It is not clear that it would even be possible to research the genetics of group differences in violence without first identifying these mechanisms for environmental diffusion and their prevalence and impact in different groups.

On the genetic side, it is also improbable that various human