of androgen on aggressive behavior). Prepubertal castration does not preclude a male from occupying a high-ranking position in a social hierarchy, and androgen therapy does not inevitably enhance aggressiveness or dominance in captive rhesus monkeys. Perhaps androgens influence patterns of "aggressive play" (Symons, 1973; Owens, 1975), which enable the monkeys to practice patterns that will be useful in adult life?

Roberts (1990) has recently reviewed the general parallels between violence in young primates and those seen in our own species. Some of the parallels are extremely striking. Predictors of adolescent, teenage, and adult violence in humans are complex (Farrington, 1989). The best predictors appear to be measures of economic deprivation, family criminality, poor child rearing, school failure, hyperactivity-impulsivity-attention deficient, and antisocial child behavior. In looking for biological predictors, however, Hamburg (1971) failed to find a clear relationship between testosterone levels and aggressive behavior in postpubertal boys. In spite of this, Benson and Migeon (1975) reviewed the physiological and psychological changes occurring around adolescence and puberty in human males. They noted marked changes in serum levels of LH and follicle stimulating hormone (FSH), as well as sex steroids from the testes and adrenal cortices, and tentatively implicated these in the development of a "rebellious attitude" around this time. Hays (1978) provided a review of strategies for studying psychoendocrine aspects of puberty. She speculated that changes in hypothalamic LHRF, which occur around puberty, may (by altering mood) induce sexuality and hostility in our species. She also felt that the involvement of androgens in behavioral changes could be studied by comparing pubertal status of highly aggressive and nonaggressive boys of the same ages. Hays noted that developmental changes are evident with respect to thyroid releasing factor, TSH, prolactin, and somatotropic hormone and that these factors may alter mood and hence aggression. She pointed out four important conclusions from the (then) available data:

1. Mood changes induced by hormones may be consequences of the instigation of "drives" that have no socially acceptable outlet in young people.

2. Development may involve changes in behavioral sensitivity to hormones as well as changes in the hormones per se.

3. All studies should consider the effects of circadian rhythms on hormonal secretion.

4. Interactions between hormones may prove more important than titers of single hormone.

These points are equally valid today.

Archer (1990) has reviewed recent developments in this area. He notes that the only studies that measure hormonal levels and aggressiveness at or soon after puberty in human males are those of Olweus et al. (1980), Mattson et al. (1980), Olweus (1986), and Susman et al. (1987). The basic data from these nonlongitudinal studies are extracted in Table 5. Olweus (1986) noted particularly that the items involving responses to provocation from his physical and verbal scales correlated best with plasma testosterone. In contrast, Eccles et al. (1988) studied hormones and affect in early adolescence, and recorded that increased levels of androgens were not correlated with increased levels of anger/impatience or aggression.

There has been great interest in the claim that pseudohermaphroditism with changed gender identity and role at puberty may be a consequence of a deficit in 5a-reductase activity (an enzyme that converts testosterone to dihydrotestosterone). There are conflicting claims about the relevance of these data to the debate about the biological versus environmental determination of gender in our species (Aron et al., 1990; Gotz et al., 1990; Money, 1990).

What of other hormones? Magnusson (1987), in a Swedish study of 82 boys at 13 years of age, found negative correlations between aggressiveness/restlessness and urinary secretion of adrenomedullary epinephrine under both active and passive conditions. This supported the more general finding of a positive correlation between good social and personal adjustments and elevated epinephrine excretion. Even more intriguingly, there was a strong inverse relationship between epinephrine excretion by these 13 year olds and their adult delinquency at age 18 to 26. More detailed statistical analysis showed, however, that the changes in epinephrine levels were more closely related to motor restlessness than to aggression. Highly aggressive individuals who are not highly restless have about the same epinephrine levels as subjects who are neither restless nor aggressive. Magnusson (1988) reviewed these data and warns that the traditional mechanistic model tending to view epinephrine excretion as the cause of both conduct in a current perspective and adult delinquency in a longitudinal perspective is probably inadequate. He suggests considering