Because these tubers had to be dug up from the soil, small children had difficulty feeding themselves and needed help. Suddenly, grandmothers became useful, investing time and energy in their grandchildren's upbringing and freeing up their daughters to invest in new pregnancies.

The theory appeared to explain why the overall lifespan of older women began incrementally to increase, while their reproductive lifespan remained fixed - a highly unusual phenomenon among mammals. Women naturally stop reproducing at around the age of 45, while chimps do so around 10 years earlier. But whereas women may live on to 70, female chimps don't survive much beyond menopause.

Now Alan Rogers, also of the University of Utah, and his colleague Richard McElreath have compared demographic data on fertility and survival among females in two different populations: the human population of Taiwan in 1906, widely considered a model population based on natural fertility, and the chimpanzees of Gombe National Park in Tanzania so meticulously documented by the Jane Goodall Institute's Center for Primate Studies.

Their analysis makes two assumptions: first, that humans evolved from a chimp-like ancestor with comparable demography to the Gombe chimps; and second, that to reach the current human situation, the human gene pool would have had to resist invasion by mutants that would either increase the fertility or reduce the survival of older women. The researchers then compared the life histories of chimps and humans to look for factors that could have driven us along our anomalous evolutionary pathway, while still maintaining a viable population.

Their main finding supports the grandmother hypothesis, and has to do with a tradeoff between fertility and survival across generations: for each additional child born to a mother who already has a daughter of reproductive age, their model predicts a cost to the daughter of two children. According to Rogers, that means selection would act to limit the mother's fertility. "The maintenance of reduced late fertility is at least plausible under the grandmother hypothesis," he says.

And why isn't the same effect seen in chimps? "Chimpanzee mothers appear to be less necessary to their daughters' fertility than human mothers," explains Rogers. Compared to humans, chimps invest much less in caring for offspring that aren't their own, he says.

However, as the theory stands, it cannot explain the inordinately long postmenopausal survival of women. "A small increase in grandmothering wouldn't help [provide an explanation]," he said. "To produce that effect on mortality, our chimp-like ancestor would have had to invest more in grandparenting than modern humans."

Something is missing from the grandmother hypothesis, he concludes. That missing x-factor could have been something external - a reduction in the risk of predation, say - or it could have been another tradeoff acting within the population that both he and Hawkes have so far overlooked: the tradeoff between a mother's continued fertility and the survival of her own, prereproductive children, perhaps.

Robin Holliday, a retired geneticist from Sydney, Australia who has studied the evolution of human longevity, says that this discrepancy could potentially be cleared up by tweaking some of the assumptions on which the model is based.

For instance, rather than looking at the 1906 Taiwanese population, he thinks Rogers should have used data from certain Amerindian populations in South America, whose life histories far more closely resemble those of the earliest hunter-gatherers. In some of those populations, he says, life expectancy at birth is no more than 20 years, and only 3% of women survive to menopause.

By and large however, Holliday thinks the grandmother hypothesis does bear scientific scrutiny. "When you have young children," he says, "it makes sense to invest in them rather than have more."