One of two competing theories in human evolution, regarding the origin of fully modern humans, Homo sapiens sapiens. The rival theory is called multiregionalism. Everyone is agreed that hominids, the genus Homo, evolved in Africa several million years ago, and fanned out into the rest of the Old World about a million years ago. The older theory, multiregionalism, says that these outspread pre-humans evolved into modern humans. The new theory, Out of Africa, says that a new burst of evolution happened in Africa only one or two hundred thousand years ago, and this group then moved out into the rest of the world and replaced the existing pre-humans.

The controversy over which is true is very strong today and shows no signs of being definitively resolved. Evidence from molecular biology and linguistics tends to favour Out of Africa, while evidence from palaeontology tends to favour multiregionalism. I get the impression that Out of Africa is now the dominant view and its rival is defended mainly by a small hardcore group who loudly trumpet any new evidence that shows the tide turning back to them; but of course weight of numbers doesn't count as evidence.

The new theory originated in the 1980s when mass comparison of mitochondrial DNA from people of many different ethnic groups all across the planet enabled scientists to draw up a tree of descent for a substantial part of the human race. Mitochondrial DNA or mtDNA is inherited only down the maternal line, so is not mixed up by sexual recombination the way most DNA is. This means the only changes to it come from mutations; and mutations randomly accumulate across time, so degree in similarity in mtDNA is likely to correspond to depth of descent from a common ancestor.

This is uncontroversial in theory; the heart of the dispute is how reliably the comparison can be done. How do we estimate the mutation rate? We could plot it for known migrations in history, but how do we know it went at the same rate everywhere throughout the depths of evolutionary time?

If you draw the tree of descent back far enough you find that the mtDNA in all of us inevitably passes through a single female at some point (so-called Mitochondrial Eve). Multiregionalism claims that this female was a Homo erectus back in Africa before they dispersed; Out of Africa claims she was much more recent, and was a member of the recently evolved Homo sapiens lineage.

The original 1987 study used a less than fully representative sample: 46 European Americans, 34 Asians, 26 native New Guineans, 21 native Australians, 18 African Americans, and 2 (two) native Africans. Even with this, it was strikingly evident that there was much greater genetic diversity within those of African origin. They also did not examine the full range of mtDNA genes. Four years later a new study included 121 native Africans out of 189 people studied, and concentrated on different genes. The "molecular clock" (rate of mutation) was calibrated by assuming humans and chimpanzees to have diverged 4.6 million years ago.

From these studies, the latest common ancestor of all modern humans lived between about 150 000 and 300 000 years ago. This is a little before anatomically modern skulls appear in the fossil record.

Three deep lineages were identified in the Africans. These are called L1, L2, and L3. All other humans in the world were found to belong to one sub-branch of L3. It was estimated that L3 humans crossed into Asia between 100 000 and 50 000 years ago, reached Australia 80 000 to 40 000 years ago, New Guinea 60 000 to 30 000, and Europe (then largely under glaciation so less habitable) 50 000 to 25 000. As these numbers have such uncertainty I've rounded them a bit more to be memorable. Later studies have refined some of these: Asia 75 000 to 55 000, Europe 50 000 to 40 000. The earliest crossing into America over the Bering Land Bridge is problematic, but was more recent than these.

Most of the evidence in favour of Out of Africa also suggests that the replacement of existing proto-human populations outside Africa, such as the Neanderthals and their equivalents in Australia and Asia, was total: there was no interbreeding. (Conversely, any evidence of Neanderthal genes in modern humans is seized on by the multiregionalists.) If this is true, what made them such a superior competitor? In anatomy there is very little difference. This is why palaeontologists generally favour the multiregional theory: in their fossil sequences they see Australian Homo erectus skulls smoothly becoming modern Australian natives, and likewise for the other races.

The most obvious candidate is language, modern language with modern syntax and complexity. No living humans speak primitive languages, under any possible sense of "primitive". Yet there are some grammatical universals and striking similarities across the globe, which suggest an underlying genetic unity to part of our language skills. Tracing the existing language families back becomes harder and harder to do as you go deeper, but a common Proto-World language back at 50 000 or 70 000 or 100 000 years ago feels about the right scale. Modern language must have evolved at some time; we can't tell when by any other means; but this enormous superiority that one subspecies had over another one physically almost identical, does fit neatly if it was a radical change in conceptual ability that differentiated them.

The language families linguists have reconstructed do give some tantalizing matches with the DNA tree, but do not match closely enough to confirm anything. Populations often switch languages entirely, without it showing up in genes or archaeology.

Another finding from DNA studies was that humans underwent one or more genetic bottlenecks between 50 000 and 100 000 years ago: a time when the amount of genetic diversity was sharply reduced, and perhaps only a thousand of our ancestors were alive at that time. (Our ancestors being the ones in Africa, not the soon to be extinct cousins elsewhere.) One possible explanation of this is rapid expansion from a single population ("demic expansion") at that time. But it might also result from existing spread populations being reduced to very small habitats because of some catastrophic loss of viability.

Climate fluctuations happen all the time, but the smoking gun here may be the volcano of Toba, on the island of Sumatra, which around 75 000 years ago exploded cataclysmically, causing a volcanic winter of many years, followed by a thousand-year ice age. Perhaps a major change in brain organization was selected for, allowing one group of humans to survive in the relative warmth of Africa, and spread out from there once the planet became inhabitable again.

Afterthought. However neat the Toba coincidence is, it would be unrealistic to expect an easy resolution. A piece of ochre in the Blombos Cave in South Africa has apparently artistic geometric designs on it, and dates from either "77 000" or "more than 70 000" years ago - but the difference is crucial if Toba is to be involved. How precisely can ochre be dated?

There is of course much debate over what symbolic capability Neanderthals had. The position of the hyoid bone suggests they had a larynx sited for human speech. The Shanidar cave burial suggests they cared for their sick and may have had some form of religion.