In large modern cities such as Berne the picture is quite different. The phone book contains thousands of names, none of which is particularly common. Again, the rules of chance and time are at work. Cities contain so many people that it is unlikely that any name, or any gene, will go extinct just because its few carriers fail to pass it on. Such places attract immigrants, so that new names (with their associated genes) come in all the time and the population becomes more diverse. A simple but effective way to measure how distinct a community might be is to count the number of surnames in relation to the number of people. If more or less everyone has a different label then the community is open to migration from many places and is, in effect, so big that accident is unimportant. A glance at the New York telephone directory compared to that of, say, Oslo shows at once that the two have had different histories. The USA has a higher proportion of all global names than anywhere else. That reflects its chronicle of immigration from all over the world.

Shared names mean shared ancestors which in turn means shared DNA. A population in which many people carry the same gene (or the same surname) because they have inherited it from a common ancestor is said to be inbred. To some extent we are all inbred as we are all to some degree related. Everyone has two parents, four grandparents and so on. If all were unrelated, the number of ancestors would double each generation to give an absurd number of ancestors within a few centuries. In fact, related people married, and the lines of descent have merged and blended. As a result we all have many ancestors in common.

Pcrh.ips tlu- most inbred individual ever recorded was,m aristocrat, I Jropjtra-Bcrenikc HI, aunt of the Cleopatra enamoured of Anthony. She may have had identical copies of lull her genes because they descended from a single ancestor. As the ancient Egyptians saw the pharaohs as their gods' posterity they were anxious to keep the deities' blood-line as pure as possible with mating among relatives (sometimes, even, between brother and sister). The story is confused by difficulties in reading the hieroglyphs showing degrees of pharaonic relatedness.

Levels of inbreeding vary greatly from place to place. The incidence of marriages between people with the same name is quite a good indicator. This was noticed by George Darwin, son of the more famous Charles (who married his own cousin). He estimated from surnames that the proportion of cousin marriages (the closest legal form of inbreeding) among British aristocrats, by definition a small and exclusive group, was about four and a half per cent — more than twice that in the general population of his time. The pattern of family names shows that the British population as a whole is, on the average, more outbred than much of the rest of Europe. Even in remote and rural East Anglia, just one in fifty of those present at the end of the eighteenth century had been there in the seventeenth, evidence how much movement there had been in comparison to Switzerland or Italy.

A small village offers little choice when it comes to picking a spouse. As a result, relatives marry and the population becomes inbred. Sometimes the married couple have each received a copy of a harmful recessive gene from their common ancestor. As a result their children are at increased risk of having two copies. George Darwin found that Oxford and Cambridge oarsmen, a healthy group, were less likely to have issued from a cousin marriage than were their more indolent peers.

There are constraints on how close a relative one may marry. Brother with sister is forbidden everywhere but even first cousin marriages may be illegal (as in most US states in the nineteenth century and in Cyprus today). This social imperative may have arisen, in part at least, from a fear that the children might be less healthy. As childhood mortality was in any case so high when the taboos were formulated (so that a small increase because of genetic disease would not be noticed) perhaps they have no biological basis at all.

The death rate does increase and development slows in the children of close relatives. Cousins share a grandparent in common. If he or she carried a harmful recessive (as almost everyone does) their children and grandchildren are more likely than average to inherit two copies. In some Japanese villages before the Second World War, up to a third of all marriages were between cousins. The huge survey of the population of Hiroshima after the atom bombs showed that the children of cousins walked and talked later than others and did worse in school. Part of this was due to the relative poverty of their parents but part reflects their heritage. The same is true in India and in Pakistan, where up to half of all marriages are still between cousins or between uncle and niece. The picture is confused here because such marriages tend to retain wealth within the family and to increase the number of children the parents can afford. Nevertheless, these too survive less well than the children of unrelated parents. First-generation Pakistani immigrants to Britain are also rather inbred. Just one birth in fifty is to such parents, but about five per cent of all inborn disease among British children is to those with Pakistani parents.

It is important not to overstate the dangers of inbreeding. Parents who are cousins have rather more than a ninety per cent chance of a perfectly normal baby, compared to more than ninety-five per cent for unrelated parents. Inbreeding has an effect, but it is dwarfed by the improvements in child health which have taken place in the past few decades. The map of human genetic diversity, based as it is on thousands of points across the genome also gives an insight into inbreeding: a child of' a marriage between a couple with a common ancestor is likely to have double copies of long sections of identical sequence. In populations known from the records to share ancestors in common, this is often the case; but, quite often, children not otherwise known to be inbred also show the same decrease in variation in parts of their genome. The effects of forgotten inbreeding long ago can, it seems, persist for many generations.

In part because of this effect, isolated populations often show high frequencies of inherited abnormalities which are rare elsewhere. Most of the gypsies of South Wales belong to one extended kindred and half their marriages are between relatives (which makes them one of the most inbred peoples on earth). One Welsh gypsy in four carries a copy of the gene for phenylketonuria, which is four hundred times more frequent in this group than in Wales as a whole. A long history of social and sexual isolation has had an effect on their genetic health. Other close-knit family groups, such as the Bedouin of Israel, Jordan and neighbouring countries, may show high levels of inbreeding with, in some places, more than half of all marriages among cousins (and a concomitant local increase in diseases such as inborn deafness). Attempts by geneticists to promote outbreeding on health grounds have had limited success.

The effects of marriages of relatives may be subtle. A few women suffer from recurrent abortion. They often become pregnant, but the foetus is lost. The problem is found among the Hutterites, a religious group who originated in the Tyrol in the sixteenth century. In the 1770s, they moved to Russia, where they flourished and multiplied ten-fold from their original community of a hundred or so. A century later, bigotry was renewed, and the Hutterites migrated to America. All thirty thousand alive today, many of whom live in South Dakota, trace their descent from fewer than ninety founders and nearly all marry within the group. Over the years they have all become quire close relatives, and the more inbred a Hutterite woman might be the longer the interval between her children. Hutterite women who find it difficult to have children share, it transpires, a high fraction of their genes with their husbands. This may reflect the malign effects of inbreeding on the embryo.