The spread zones, Dr. Renfrew said, are mostly the result of recent dispersals caused by agricultural inventions. The mosaic zones "may be those of the first humans to occupy those areas, at least in Australia and America," he said.
The language spoken by the ancestral human population may never be known, though Dr. Greenberg has tried to reconstruct a few words of it. But some linguists who study the click languages of southern Africa feel they are very ancient. This belief is supported by genetic evidence showing that the Khoisan peoples, the principal speakers of click languages, belong to the most ancient of all the human lineages, based on mitochondrial DNA.
Dr. Anthony Traill, a click language expert at the University of the Witwatersrand in Johannesburg, said that linguistically the languages fell into three separate groups whose relationship, aside from the clicks, was hard to establish. The clicks must be ancient, he said, because "the chances of clicks being invented after being lost is zero." The only use of clicks outside of Africa is in an Australian aboriginal initiation languages in which the clicks are used as meaningless sounds.
"The idea that clicks were lost from all languages other than Khoisan," Dr. Traill said, "is stimulating, but I don't know what to make of it."
Of the three disciplines that bear on human origins — historical linguistics, population genetics and archaeology — only archaeology has a rock-solid method of dating, based on radiocarbon and other kinds of radioactive decay.
But geneticists are now improving their dating methods, even though the dates are still very approximate, to the point that they can begin to correlate their findings with the archaeologists'. The geneticists' first foray into human prehistory was the famous "mitochondrial Eve" article of 1987 by the late Allan Wilson, showing that when people around the world were placed on a family tree constructed from their mitochondrial DNA, the tree was rooted in African populations, in an individual who lived about 200,000 years ago.
Though the methodology of the paper was imperfect, its result was unchanged after the method had been corrected, and geneticists have developed a growing confidence in mitochondrial DNA dates. The mitochondrial DNA trees trace back to a single individual, not because there was only one Eve — the ancestral human population is thought to have contained about 10,000 people — but because the lineages of all the other Eves have gone extinct. The process is easy to visualize by thinking of an island population with 10 surnames. In each generation, some men will have no children or only daughters and their surnames will disappear until only one is left; the Y chromosome and mitochondrial DNA follow the same pattern.
The first major branch points in the mitochondrial Eve tree have been called the daughters of Eve and they fall in a geographic pattern with some daughters of Eve being characteristic of Africa, some of Asia and the Americas and some of Europe and the Near East.
Dr. Richards and his colleagues have analyzed the ancestry of the present European population by looking within the major daughter of Eve branches for subbranches that occur both in Europe and the Near East, from western Iran through Turkey and Arabia to Egypt, because the Near East is the probable source of most of the ancestral populations that entered Europe.
The subbranches from each region were then dated by counting the number of mutations that had occurred in the mitochondrial DNA sequence from the beginning of the subbranch until today. If the subbranch was older in the Near East than Europe, it indicated a migration into Europe. By this method Dr. Richards's team was able to date the migrations into Europe. They also picked up a sizable back-migration from Europe to the Near East.