Nick Martin and Grant Montgomery, senior research fellows at the Queensland Institute of Medical Research in Brisbane, Australia, specialize in twin studies. They supply the following answer.

There is no doubt that twinning seems to run in some families. Closer investigation reveals that this is mainly confined to nonidentical, or fraternal (dizygotic - DZ) twins, although there are a few rare families with extraordinary numbers of identical (monozygotic - MZ) twins. Furthermore, the increased chance of twinning seems to be entirely a property of the mother, not the father. Of course, there are counter examples to all these rules, and there are even families with large numbers of both identical and fraternal twins. At the moment it is not clear whether these just occur by chance and we get to hear about them due to reporting bias (they always make great front page stories) or whether they represent some new biological mechanism.

The strongest clues we have--that it is confined to fraternal twinning and females--suggest that the primary cause is an increased risk of multiple ovulation, in which a woman releases two or more eggs in a given menstrual cycle rather than the normal one. We tested this by ultrasound scanning of ovaries just before ovulation in mothers who had previously given birth to two sets of fraternal twins (MODZTs). Such women showed a much higher rate of multiple ovulation than control women did. The maturation and release of one or more eggs is controlled by several reproductive hormones, including follicle stimulating hormone (FSH), luteinizing hormone (LH) and inhibins. We checked the level of these hormones in MODZTs and controls at different parts of the menstrual cycle and showed that there were some abnormalities, although these were not very consistent. Perhaps the most intriguing finding is Dutch evidence that MODZTs release FSH in more frequent pulses than normal women.

The best evidence for twinning genes comes from sheep, where mutations in two genes increase twinning by altering the response of the ovary to FSH. Currently, we are collecting DNA from pairs of sisters who have both had fraternal twins, because these will enable us to use genetic linkage analysis to locate the genes responsible for familial fraternal twinning. Once we have found them, and understood the mechanisms by which they work, we may well be able to predict which particular women have a greater chance of having twins.

Meanwhile, the best way of increasing your chances of having fraternal twins if you are female is to have a mother, a sister, or an aunt (on either your mother's or father's side) who has had fraternal twins. If you are male, then marry a woman who fits this description because it seems that the father has no influence on the chance of twins, although he can pass the tendency to his daughters. Another major factor is maternal age; a woman who gives birth at 37 is four times more likely to have fraternal twins than at age 18, although she is also more likely to be unable to conceive, since many women's ovaries are already starting to fail at that age. The third major risk factor is race; West Africans are ten times more likely to have fraternal twins than Chinese or Japanese, with Caucasians intermediate. Some increased chance is also seen in African-Americans.

Interestingly, neither maternal age nor race seems to influence identical twinning, which is almost universally constant at about four per 1000 births. Finally, all the above only applies to spontaneous, or natural, births. The introduction of assisted reproduction techniques (ART), particularly ovarian stimulation, has seen a dramatic rise in the number of twin and higher multiple births. They also slightly increase the chance of having identical twins, which suggests that there may be some connection between the two types of twinning.