The question is asked in every language, in every era: "So, dear, when will you give me grandchildren?" Darwin would approve. (Click here to follow Sharon Begley)

At least he would if the "grandma hypothesis" is right. According to this idea, the reason women—uniquely among primates—outlive their child-bearing years is that a female who survives past menopause can contribute to the care of her children's children, improving their chances of reaching adulthood. Natural selection favors behavior that increases an individual's genetic contribution to future generations; surviving long enough to help grandkids is thus an evolutionary adaptation.

Too bad data don't support this intriguing notion. In some studies, a grandmother living nearby was indeed associated with better survival of grandchildren, as the hypothesis predicts. But other studies found no such benefit. Leslie Knapp, a biological anthropologist at the University of Cambridge, and her graduate student Molly Fox wondered if the inconsistency reflected a basic fact of genetics—namely, that because of how the X chromosome is passed down from parents to children, grandmothers are more closely related to some grandkids than to others.

Here's why. A paternal grandmother, like all women, has two X chromosomes. She passes one to her son (who gets his Y chromosome from Dad, which is why he's a he). He then passes grandma's X—the one and only X he has—to his daughter. But Dad passes his Y chromosome to his son, who therefore does not carry his paternal grandma's X. A maternal grandmother, too, passes one of her X's to her daughter; there is a 50–50 chance that that X will be transmitted to the daughter's child, of either sex. A maternal grandmother, therefore, has only a 50–50 chance that her X will be transmitted to a grandchild. A little math shows that maternal grandmothers are related to granddaughters and grandsons equally, for an "X-relatedness" of 25 percent. But paternal grandmothers are twice as close to granddaughters (50 percent) and not at all to grandsons (zero percent), explains Knapp. It may seem arbitrary to focus on X, one of 23 chromosomes, but it has 8 percent (1,529) of all our genes, including some for fertility and intelligence, which affect reproductive success.

Many of those earlier, inconsistent tests of the grandma hypothesis lumped together both kinds of grandmas (maternal and paternal) and both sexes of grandkids. Given the different degrees of X-relatedness, says Knapp, "we decided to look at the data from a genetic perspective. Since it is adaptive to favor those with whom we share the most genes, evolution should favor women who invest in grandchildren in a way that mirrors X-relatedness."

She, Fox, and colleagues analyzed existing data on the survival of 43,000 children in seven traditional societies, from rural farming villages in Japan and Malawi to towns in Germany and Canada, from the 1600s to today. "The most striking effect was of the paternal grandmother," says Fox. In six of the seven societies, having a paternal grandmother nearby improved the survival of granddaughters (50 percent X-relatedness) by up to 4.5-fold, but for some unknown reason decreased the survival of grandsons (zero percent) by 8 to 29 percent. And a boy had a greater chance of survival if he lived with his maternal grandmother (25 percent X-relatedness) than with his paternal grandmother (zero percent). In four of the seven societies, a girl had a better chance of survival if she lived with her paternal grandmother (50 percent) than her maternal grandmother (25 percent).

In other words, the effect of a grandmother perfectly tracked the DNA. "The higher the X-relatedness," the scientists write in Proceedings of the Royal Society B, "the more beneficial effect the grandmother has on that child's" survival. That the correlation held across four continents and four centuries suggests a biological, not cultural, explanation.

But what? There is no evidence grandmothers consciously treat grandsons and granddaughters differently, or a son's children different from a daughter's. The best guess is that grandchildren transmit some signal of genetic relatedness, such as resemblance or a pheromone, which Grandma unconsciously uses to apportion how much she invests in different grandkids. Grandmothers will surely recoil at the very idea, which is why the reader is advised not to leave this column lying around during a multigenerational Thanksgiving.

Sharon Begley is NEWSWEEK's science editor and author of The Plastic Mind: New science reveals our extraordinary potential to transform ourselves and Train Your Mind, Change Your Brain: How a New Science Reveals Our Extraordinary Potential to Transform Ourselves .

© 2009