1. Although neither Rogel nor Morgan explore how natural selection may have acted on the psychology of rapists, their interest in indices of fertility is congruent with some evolutionary approaches to rape. Because rape exposes the rapist to an immediate risk of injury and, later, possible retribution, an evolutionary perspective suggests that, if rape constituted a viable reproductive strategy in ancestral populations (a contested possibility, Archer and Vaughan 2001), natural selection may have favoured the evolution of mechanisms that would have led rapists to maximize the likelihood that rape would result in conception, as it is only thus that the rapist would have obtained fitness benefits potentially outweighing the fitness costs of rape to the rapist (Shields and Shields 1983; Thornhill and Thornhill 1983).
2. By introducing a steady dose of exogenous hormones, oral contraceptives can be expected to eliminate cyclic changes in the presumed proximate determinants of phase-dependent behaviour. Following precisely this reasoning, Chavanne and Gallup (1998), Petralia and Gallup (2002), and Bröder and Hohmann (2003) treat oral contraceptive users as a control group, finding support for their respective arguments in the observation that the behaviour of such women does not vary on the factors of interest across the menstrual cycle. It is therefore significant that the only women in Rogel's study to exhibit a reduction in rape frequency on day 14 (the probable day of ovulation in nonusers of oral contraceptives) are individuals who were taking oral contraceptives at the time of the assault (1976: 76), as this is precisely the opposite pattern from that which we would expect given Chavanne and Gallup's account of Rogel's findings.
3. The adaptive significance of even the limited reduction in frequency during days 10–11 among those not using oral contraceptives is questionable given the fact that this pattern is driven primarily by women age 17–20, yet fecundity peaks considerably later (Wood 1989).
4. Thornhill and Palmer (2000: 100) pick up without citation an argument laid out by Moore (1996), namely that, because paternity tests were not performed in the Holmes et al. study, it is possible that the probability of conception due to rape was inflated by the inclusion of pregnancies resulting from consensual coitus with another male. Moore cites Hammond et al.'s (1995) finding that 60% of pregnancies ascribed to rape were actually conceived with a consensual partner, and notes that 64% of the victims in the Holmes et al. study were married or cohabiting at the time of rape. From these figures Thornhill and Palmer conclude that the probability of conception following rape is 2%. However, Moore's argument does not result in such a drastic reduction, since 47% of the rapes that resulted in pregnancy were ascribed by the victim to a boyfriend or husband (Holmes et al. 1996: 322). Crudely, this means that only approximately 17% of the remaining victims were likely to be married or cohabiting at the time of rape. If 60% of the pregnancies contributed by these individuals were mis-ascribed, then the gross number of rape-related pregnancies should be reduced by approximately 10%. Holmes et al. report 20 cases of pregnancy out of 404 rapes of women age 12–45; adjusting the former figure produces a total of 18 cases of pregnancy, or a 4.5% probability of pregnancy per act of rape.
5. This figure was arrived at after controlling for the distribution of copulations across the cycle, hence it is unaffected by the fact that proceptive female sexual behaviour increases as a function of conception risk (reviewed in Regan and Berscheid 1999:46-50; see also Clayton et al. 1999; Gangestad et al. 2002), a pattern that would otherwise bias the comparison against the prediction that the probability of conception following rape will be lower than the probability of conception following consensual coitus.