In another instance of this co-evolution, the horse made chivalry a possible culture (hence the name, of course: compare the French cheval') and enabled the Mongols to achieve one of the largest, best-controlled empires in human history. Under the Khans it was said that a virgin could walk unmolested from Seville to Hang Chou. Only in the twentieth century was that again achievable, with luck and possibly a harder search for the virgin. The Spanish took horses to America, where humans had killed off several equine species some 13,000 years before,38 and changed the lives of all the North American Indian tribes -and the cowboys, of course. And, a little later, Hollywood.

The horse did wonders for the genetics of humans, too. Just as they say that the invention of the bicycle saved East Anglia from an incest implosion, so the people that had come out of Africa were only a tiny part of early Homo sapiens's genetic diversity. All recent studies of the DNA

genetics of human populations agree that the genetic diversity outside Africa is only a tiny fraction of the diversity that is still found on that continent. Those who left, to go as far as Australia or China, to Western Europe or via the high Arctic to America, are less diverse in total than many small indigenous African peoples. With the arrival of the horse, it became possible for traders to carry goods - and gene alleles - for very long distances, very effectively. So the out-of- Africa humans have inherited a relatively small part of the African gene-pool: they are genetically impoverished, but well stirred.

At the end of the twentieth century there was, for some years, a belief that Homo sapiens was a polyphyletic species. This word means that different groups of Homo sapiens evolved from different groups of Homo erectus in different places. This, it was thought, might account for the racial differences, especially differences in skin pigmentation, that seemed to fit geography pretty well. From DNA studies, we now know this theory can't be true. On the contrary, there was a bottleneck in our population as we came out of Africa -humanity was reduced to rather small numbers -and all of us living today, all of the out-of-Africa 'races', were extracted from that small population. All the Homo erectus died out. The evidence so far looks as if there was only one exodus, of a minimum of some 100,000 people. We were all there in potentia in that tiny population, Japanese and Eskimos and Norsemen and Sioux and Beaker people and Mandarin Chinese; Indians and Jews and Irishmen. In the same way, all the current kinds of dog were 'present' in the original domesticated wolf (assuming it was indeed a wolf) -that is, they were in the wolfs space of the adjacent possible -and we've pulled out Saint Bernards and chihuahuas and labradors and King Charles spaniels and poodles from that local region of organism-space.

There was, about thirty years ago, a brief fashion for the concept of 'mitochondrial Eve', and many media reports seem to have picked up the idea that there was just one woman, a veritable Eve, in that ancestral bottleneck. This is nonsense, but the media reports were written up to encourage the belief. The real story, as always, was a little more complicated, and it goes like this. There are mitochondria in the cells of people, indeed of most animals and plants. These are the billions-of-generations, descendants of symbiotic bacteria, and they still have some of their ancient DNA heredity, called mitochondrial DNA. Mitochondria from the mother go into the embryo's cells, but those from father do not: they die, or go only into the placenta. In any event, mitochondrial inheritance is very nearly all maternal. The mitochondrial DNA accumulates mutations over time, with important genes changing less (presumably because the resulting babies, if any, were defective) and some DNA sequences changing quite quickly. That enables us to judge how far back it is to the common ancestor of any pair of women, from the accumulated differences in several DNA sequences. Surprisingly, nearly all such pairs from very different women converge on to a single consensus sequence, about 70,000 years ago.

A single woman, the ancestor of us all.

Eve?

Well, that was the story that the media latched on to, and you can see why. However, it doesn't hang together. The occurrence of just one mitochondrial DNA sequence doesn't mean that there was just one woman with that sequence, or that she was the ancestress of all the other women whose DNA was sequenced. Evidence based on the current diversity of various genes shows that there were at least 50,000 women in the human population 70,000 years ago, and many of them will have had that particular DNA sequence, or one that cannot be distinguished from it with the evidence remaining today. The lineages of the women who did not have that sequence continued for some time, but eventually died out: their 'branch' of the human family tree doesn't reach all the way to the present day. We can't be certain why those lineages died out, but in mathematical models such effects are commonplace. Perhaps the women carrying sequences like today's sole survivor were more 'fit', or they simply came to outnumber the others by chance. It is even possible that the choice of the contemporary women to test was in some way biased, and that more than one mitochondrial DNA sequence is actually present in today's women.

How do we know that there were at least 100,000 humans 70,000 years ago, and not, as in the stories, just two 6,000 years ago? Many (about 30 per cent) of the genes in the cell nucleus have several versions in today's human population. Like most 'wild' populations (not bred in the laboratory or for dog shows), each individual human has two versions of about 10 per cent of his or her genes, different versions received from father and mother in sperm and egg. Humans have roughly 30,000 genes, of which about 3,000 will be represented by two versions in the average person. For some genes, notably those of the immune system that give each of us a very specific lock-and-key individuality, making us susceptible to some ailments but resistant to others, there are hundreds of versions of each gene (of four important ones, anyway). The (common)

chimpanzee has a set of these immune variants that is very like the human: in one list of 65 variants of one immune gene, only two were not exactly the same. We don't know about the DNA of enough bonobos yet to see if the story is the same for them, but the smart money says that it will be, possibly even more so. The gorilla set seems to be a little different again (but only about thirty gorillas have been tested).

At any rate, all of these immune gene variants had to come out of Africa in that 'bottleneck'

population that produced all the ex-African human populations. It is unreasonable to suppose that each individual inherited different versions of each variable gene from their parents: some will have carried only one version, the same from both parents, and no one can have carried more than two. The humans that came out of Africa have about 500 immune variants, at least, in common with chimpanzees, out of about 750 possibilities. The humans who stayed in Africa have more: they weren't subject to the bottleneck. There are many other genes where several ancient versions (ancient because they're common to us, chimpanzees, perhaps gorillas, maybe other species) have come through; 100,000 people is a reasonable minimum to carry all those. If you want to be critical and get that number down a bit, you could argue that a few variants from African populations may have been mixed in later, for example via slavery to the US, or to Mediterranean peoples and then via Phoenician sailors to the rest of us. Still, the evidence does not point to an Adam and an Eve, unless they came with a lot of servants, slaves or concubines.