Women who give birth to girls are meek and mild; tough, "dominant" women produce boys. This peculiar idea comes from Valerie Grant, a New Zealand psychologist who teaches medical students at the University of Auckland. For more than 30 years, she has tried to prove that a mother's personality determines the sex of her child.

Since 1966, she has given personality tests to a total of 178 pregnant women. At best, her results showed a slight statistical association between what she calls "maternal dominance" and a propensity to have male offspring. But time and again, her data were "never as strong as I had expected".

She tried altered the timing, as well as the content, of her "measuring instruments". Belatedly, she realised that the popularity of assertiveness training in mid-80s Auckland "was undermining my instrument". Before then, she says, the question "If you were served the wrong food in a restaurant would you say a) say nothing b) don't know, c) send it back?" would reliably sort the dominant from the non-dominant women. But post-assertiveness, everyone she asked said they would send the food back.

So why didn't all these assertive women only have boys? Grant has a convenient answer. Environmentally induced behaviour cannot alter a woman's inborn "biological dominance" - said by Grant to be the degree of dominance with which she is "comfortable".

Yet in the next breath, Grant allows that changes in a woman's personal style can alter an offspring's sex. How else could she account for mothers who produce children of both sexes? Their dominance must have changed between pregnancies, Grant contends. Hormones are cast as the mediator: the amount of testosterone in a woman's blood provides the key to a baby's sex, she claims.

This book comes from a mainstream academic publisher, yet its treatment of psychometrics, ethology and endocrinology seem patchy and outdated. It has seen the light of day, I suspect, thanks to a current vogue in sociobiology for research into sex ratios.

This field took off in the mid-1980s, when zoologists studying red deer reported that dominant hinds were more likely to have male offspring. The scientists proposed an evolutionary advantage: in "harem-style" deer societies, all females, but only the top males, get to mate. As fawns largely inherit the dominance status of their mother, natural selection favours the production of male offspring by high-status hinds. In human societies, however, men do not have to be company directors or bank managers to father children.

Grant skates over such complexities because she is unaccountably convinced that dominant women are best suited to rearing boys, while their passive peers make ideal mothers for girls. This is a central tenet of her argument, yet she makes no attempt to justify it.

Why should the meek inherit the girls? Grant simply suggests it's all for the best. Forget sex selection and designer babies. Women who have given birth to all-girl or all-boy families should "view themselves as specialists in one sex or the other and be proud of that".