Out of Africa, into India

Foundations

One line of evidence, before the story

Here is a fact that decides a great deal.

Genetic diversity is far higher within Africa than everywhere else in the world combined. Two people from different populations in Africa can differ more from each other than a Norwegian differs from a Han Chinese. And the diversity found outside Africa is essentially a subset of the diversity found inside it.

That pattern has one straightforward explanation. A small group left, carrying only a sample of what the parent population held, and everyone outside Africa descends from that sample. It is the signature of a founder effect, and it is why the African origin of our species is not seriously disputed.

If you have ancestry outside Africa, your ancestors passed through a bottleneck narrow enough to still be visible in your genome.

Not once, and not simply

The clean version says one exodus, around sixty thousand years ago. The reality has at least two complications.

There were earlier attempts. Fossils of anatomically modern humans in Israel are around 120,000 years old, and there are older claims elsewhere. Those people were real. But they left little or no detectable ancestry in living non-Africans. Whether they died out, retreated, or were absorbed, the genetic evidence says the population that fills the world today comes from a later movement.

And it was not a march. The mental image of a purposeful migration is wrong at every scale. These were foragers, and a band moving its range twenty kilometres in a generation covers the distance from East Africa to Southeast Asia in a few thousand years without anyone ever intending to go anywhere. Nobody set out for Australia.

The route, and the puzzle at the end of it

The evidence favours a southern route: out of the Horn of Africa, across the mouth of the Red Sea when sea levels were low, then eastward along the coasts of Arabia, Iran, South Asia and Southeast Asia.

Coastlines make sense for foragers. Shellfish are predictable, do not run away, and can be collected by children. The route requires no new technology, only continuing to do what already worked.

The problem is the far end. Australia was occupied by about 65,000 years ago, on the evidence from Madjedbebe in the Northern Territory. Even at the lowest sea levels, reaching Australia required crossing open water. That means boats, or something like them, sixty-five thousand years ago.

And it creates a tension. If the main expansion left Africa around sixty thousand years ago, arriving in Australia by sixty-five thousand is difficult. Either the Australian date is too old, or the departure was earlier than the genetics suggests, or the coastal route was travelled much faster than anyone expects. This is unresolved, and it is one of the sharper open questions in the field.

The volcano that did not do it

For twenty years, one story dominated the middle of this period, and it has come apart.

About 74,000 years ago, Toba in Sumatra produced the largest volcanic eruption of the last two million years. Ash from it blankets South Asia in a layer you can put your hand on. The proposal was that it caused a volcanic winter severe enough to reduce the human population to a few thousand individuals, a bottleneck through which everyone alive passed.

It was a compelling story, and the evidence has moved against it.

The decisive work was done in India. At Jwalapuram in Andhra Pradesh, excavators found stone tool assemblages directly beneath the Toba ash and directly above it. The tools belong to the same technological tradition. Whoever was making them before the eruption was still there afterwards, doing the same thing.

Climate reconstructions have also scaled down the estimated cooling, and the genetic evidence for a bottleneck at that date is weaker than was once claimed. The largest eruption in two million years appears not to have come close to ending us, which is worth knowing both as a correction and as a caution about how satisfying catastrophe stories are.

India was already occupied, and had been for a very long time

Before any of this, hominins were in the subcontinent.

Attirampakkam, in Tamil Nadu, has Acheulean handaxes dated to about 1.5 million years ago, as lesson 6.3 described. That is not modern humans; it is a much earlier hominin, probably Homo erectus or a relative. The same site preserves a long sequence afterwards, and its later layers have been used to argue for an early transition to Middle Palaeolithic technology in India, which remains debated.

So when modern humans arrived, they entered a landscape that had held people, on and off, for over a million years. South Asia is not a place the human story passed through. It is one of the places the human story happened.

What the genetics of India actually says

This is where popular accounts go wrong most often, in several directions at once, so it is worth being careful.

The 2009 result. A large study of variation across Indian groups found that most of the variation could be described using two inferred ancestral components, labelled Ancestral North Indian and Ancestral South Indian. Almost every Indian group is a mixture of both, in proportions that vary regionally, generally with more of the northern component in the northwest and more of the southern component in the south and among certain tribal groups.

Two cautions matter. These are statistical constructs, ways of summarising patterns of variation, not two ancient peoples who had names and territories. And the mixture is old and thorough: the great majority of Indians are mixed, and the proportions form a gradient rather than a set of boxes.

The one important outgroup is the Andamanese, whose ancestry sits distantly from other South Asian populations and who serve as a reference point for reconstructing the southern component.

The 2019 result. A much larger study using ancient DNA from Central and South Asia added time depth. It found evidence that steppe pastoralist ancestry, originating in Central Asia, spread into South Asia during the second millennium BCE, a movement also visible in Europe.

And it published a genome from an individual buried at Rakhigarhi, associated with the Indus Valley civilisation. That individual carried no steppe ancestry and no Anatolian-farmer-related ancestry of the kind that accompanied the spread of farming into Europe. The implication is that farming in South Asia was not brought by a large migration of Anatolian farmers, and that the Indus population predates the steppe-related arrival.

One genome is one genome, and ancient DNA preserves poorly in a warm climate, which is why South Asia has a handful of ancient genomes where Europe has thousands. This is a young field in the region and the picture will change.

The recent part, which surprises people

The genetic evidence also carries a finding about social structure that is often assumed to be far older than it is.

The pattern in Indian genomes indicates a long period of extensive mixing, during which populations across the subcontinent exchanged ancestry substantially, followed by a shift to strong endogamy, marriage within groups, which the genetics places within roughly the last two thousand years.

In other words, the deep past of the subcontinent is a story of mixture, and the strict boundaries between groups are a comparatively recent overlay. That is a scientific result about population history, drawn from patterns of shared ancestry and the length of shared genetic segments, and it is worth stating precisely because it is so easily bent in either direction politically.

Everywhere else

The rest of the spread is quicker to state. Europe was entered by around 45,000 years ago, later than Australia, which is worth remembering when the story is told from Europe. Northeast Asia and Siberia followed. The Americas were reached by at least 15,000 years ago, and probably earlier, across or along the coast of the land bridge exposed by low sea levels.

By around ten thousand years ago, humans occupied every continent except Antarctica, along with island groups reachable by the boats of the time. The remotest Pacific islands waited another several thousand years and required navigation of a completely different order.

One species, out of one continent, in a geological instant.

Deeper

How founder effects are detected

The signature is not only reduced heterozygosity but its geographic structure: genetic diversity declines with distance from East Africa along plausible migration routes, and linkage disequilibrium increases correspondingly, both of which are expected from serial founder events. Non-African populations also carry a subset of African haplotypes rather than novel ones at appreciable frequency. Together these support a single principal dispersal rather than multiple independent origins.

Why the Australian date is difficult

The Madjedbebe age relies on optically stimulated luminescence dating of the sediments containing artefacts, which dates when quartz grains were last exposed to light. Objections have focused on whether artefacts could have moved downward through the sandy deposit into older layers, which single-grain dating and analysis of artefact orientation were used to address. Independent confirmation from other sites of comparable age would settle it; at present the date is accepted by many and contested by some.

What the Toba climate models now say

Early estimates of six to ten degrees of global cooling for several years have been revised downward substantially by later modelling that better accounts for sulphate aerosol microphysics and the eruption’s sulphur yield. Records from Lake Malawi spanning the eruption show ash but no corresponding signal of severe or prolonged cooling in the local proxy record. The genetic bottleneck once attributed to Toba is now generally attributed to the out-of-Africa founder effect itself, which is a different event at a different date.

Reading admixture from segment lengths

Because recombination breaks chromosomes into shorter segments each generation, the length distribution of ancestry-informative segments dates admixture: long segments indicate recent mixture, short ones older. This is the method behind both the estimate of steppe-related admixture in the second millennium BCE and the estimate that widespread endogamy began within the last two thousand years. The dates carry real uncertainty, typically expressed as ranges of several centuries, and depend on the assumed generation time and recombination map.

The limits in this region

Ancient DNA recovery depends on cool, stable conditions. Almost all South Asian ancient genomes come from a small number of sites and individuals, and the Rakhigarhi genome required screening many samples to obtain one usable result. Palaeoproteomics and improved extraction from petrous bone have helped, and the gap relative to Europe remains large. Statements about South Asian prehistory from genetics should be read with that sample size in mind.

Frontier

When the successful expansion actually left

The tension between the Australian archaeological date and genetic estimates for the out-of-Africa expansion is unresolved. Possible resolutions include an earlier dispersal that contributed a small but real fraction of ancestry, faster coastal movement than modelled, or a revision of the Australian dates. Additional well-dated sites along the southern route, particularly in Arabia and South Asia, would be the most direct test.

Filling the South Asian ancient-DNA gap

The single Indus genome is the clearest illustration of the constraint. Systematic screening of skeletal collections from South Asian sites, with methods optimised for degraded tropical samples, is the main route to resolving questions about the peopling of the subcontinent that are currently answered from modern DNA alone.

Transcript and shot list

Target runtime about 8 minutes 30 seconds. Built as a journey with corrections: follow the route, and stop at each point where the popular story is wrong, ending in South Asia. Shot types: ANIM is rendered from code with Manim, AI is a generated shot from Higgsfield with the shared style reference, REAL is agency imagery with credit, CAM is the narrator.

[00:00] S01 · ANIM · A diversity plot: African populations spread widely, all non-African populations clustered inside a small subset of that range.

Start with one fact, because it settles a lot. Genetic diversity is far higher within Africa than in the entire rest of the world combined. Two people from different African populations can differ from each other more than a Norwegian differs from a Han Chinese. And the diversity found outside Africa is essentially a subset of the diversity found inside it.

[00:30] S02 · ANIM · A large population; a small group leaving with only part of the variation; the world filling from that sample.

There is one straightforward explanation. A small group left, carrying only a sample of what the parent population held, and everyone outside Africa descends from that sample. If you have ancestry outside Africa, your ancestors went through a bottleneck narrow enough that it is still visible in your genome.

[01:00] S03 · ANIM · Levantine fossils at 120,000 years marked, then greyed out as leaving no detectable ancestry; the main expansion at 60,000 highlighted.

Now two corrections. It did not happen once. Modern human fossils in Israel are around a hundred and twenty thousand years old, and there are older claims elsewhere. Those people were real. They left little or no detectable ancestry in anyone alive. Whether they died out, retreated or were absorbed, the population that fills the world today comes from a later movement.

[01:30] S04 · ANIM · A purposeful migration arrow crossed out; replaced by a band shifting its range 20 km per generation, the aggregate covering continents in a few thousand years.

And it was not a march. The image of a purposeful migration is wrong at every scale. These were foragers. A band moving its range twenty kilometres in a generation gets from East Africa to Southeast Asia in a few thousand years without anybody ever intending to go anywhere. Nobody set out for Australia.

[02:00] S05 · ANIM · The southern route: across the mouth of the Red Sea at low sea level, then along the coasts of Arabia, Iran, South Asia and Southeast Asia; shellfish collection shown as a predictable food.

The evidence favours a southern route. Out of the Horn of Africa, across the mouth of the Red Sea when sea levels were low, then east along the coasts. Coastlines make sense for foragers: shellfish are predictable, they do not run away, and children can collect them. It needs no new technology. Just continuing to do what already worked.

[02:35] S06 · REAL · The Madjedbebe rock shelter, northern Australia (research image; confirm credit line). S07 · ANIM · Sea level at its lowest; open water still separating Australia; a date of 65,000 years; then a timeline conflict flagged against the 60,000-year genetic estimate.

Except at the far end there is a problem. Australia was occupied by about sixty-five thousand years ago. And even at the lowest sea levels, getting to Australia meant crossing open water. Which means boats, or something like them, sixty-five thousand years ago. And it creates a real tension: if the main expansion left Africa around sixty thousand years ago, arriving in Australia at sixty-five is difficult. Either the Australian date is too old, or the departure was earlier than the genetics says, or that coastal route was travelled much faster than anyone expects. It is not resolved.

[03:20] S08 · REAL · The Toba ash layer exposed in a South Indian section (research image; confirm credit line). S09 · ANIM · The eruption at 74,000 years; the old story of a volcanic winter and a bottleneck to a few thousand people; then tools below and above the ash at Jwalapuram, same tradition.

Then, in the middle of all this, a story that dominated for twenty years and has come apart. About seventy-four thousand years ago, Toba in Sumatra produced the largest eruption of the last two million years. Its ash blankets South Asia in a layer you can put your hand on. The proposal was a volcanic winter severe enough to cut the human population to a few thousand people. And the work that undid it was done in India. At Jwalapuram in Andhra Pradesh, excavators found stone tools directly below the ash, and directly above it. Same technological tradition. Whoever was making them before the eruption was still there afterwards, doing the same thing.

[04:05] S10 · ANIM · Revised climate models showing much smaller cooling; the genetic bottleneck reattributed to the out-of-Africa founder effect at a different date.

Climate models have also scaled the cooling down substantially, and the genetic bottleneck that was attributed to Toba is now generally attributed to the out-of-Africa founder effect itself, which is a different event at a different date. The largest eruption in two million years appears not to have come close to ending us. Which is worth knowing as a correction, and as a caution about how satisfying catastrophe stories are.

[04:40] S11 · ANIM · Attirampakkam at 1.5 Ma with Acheulean handaxes; a long occupation bar across the subcontinent; modern humans arriving into an already-inhabited landscape.

And India was not empty when any of this happened. Attirampakkam, in Tamil Nadu, has handaxes from about one and a half million years ago. That is not us; it is a much earlier hominin. So when modern humans arrived, they walked into a landscape that had held people, on and off, for over a million years. South Asia is not somewhere the human story passed through. It is one of the places the human story happened.

[05:15] S12 · ANIM · The 2009 analysis: two inferred components, ANI and ASI, shown as a gradient across Indian groups rather than as boxes; the Andamanese placed as a distant outgroup; a caption: statistical constructs, not ancient peoples.

Now the genetics of the subcontinent, where popular accounts go wrong in several directions at once. A large study in 2009 found that most Indian genetic variation could be described with two inferred ancestral components, labelled Ancestral North Indian and Ancestral South Indian. Almost every Indian group is a mixture of both, in proportions that vary regionally. Two cautions. These are statistical constructs, ways of summarising patterns, not two ancient peoples who had names and territories. And the mixing is old and thorough: the great majority of Indians are mixed, and the proportions form a gradient, not a set of boxes.

[06:00] S13 · ANIM · The 2019 ancient-DNA study: steppe pastoralist ancestry spreading into South Asia in the second millennium BCE; then the Rakhigarhi genome with no steppe ancestry and no Anatolian-farmer ancestry; a caveat: one genome, warm climate, few samples.

Then, in 2019, a much larger study using ancient DNA added time depth. It found steppe pastoralist ancestry from Central Asia spreading into South Asia during the second millennium BCE, a movement also visible in Europe. And it published a genome from an individual buried at Rakhigarhi, an Indus Valley site. That individual carried no steppe ancestry, and no Anatolian-farmer ancestry of the kind that came with farming into Europe. Which implies farming in South Asia was not brought by a large migration of Anatolian farmers. One genome is one genome, ancient DNA preserves badly in a warm climate, and South Asia has a handful of ancient genomes where Europe has thousands. This picture will change.

[06:50] S14 · ANIM · Segment-length analysis: long shared segments indicating recent mixture, short ones older; a timeline showing extensive mixing followed by a shift to endogamy within roughly the last 2,000 years.

And the genetics carries one finding about social structure that surprises people. The pattern in Indian genomes shows a long period of extensive mixing across the subcontinent, followed by a shift to strong endogamy, marriage within groups, which the genetics places within roughly the last two thousand years. The deep past of the subcontinent is a story of mixture. The strict boundaries are a comparatively recent overlay. That is a result about population history, read from the lengths of shared genetic segments, and it is worth stating precisely because it is so easily bent in either direction.

[07:35] S15 · ANIM · The rest of the spread: Europe at 45,000, later than Australia; Siberia; the Americas by at least 15,000; every continent except Antarctica by 10,000 years ago; the remote Pacific much later.

The rest is quicker. Europe was entered around forty-five thousand years ago, which is later than Australia, and worth remembering when this story gets told from Europe. Northeast Asia and Siberia followed. The Americas by at least fifteen thousand years ago and probably earlier. By around ten thousand years ago, humans were on every continent except Antarctica. The remotest Pacific islands waited thousands of years more and needed navigation of a completely different order.

[08:05] S16 · AI into ANIM · A coastline at dusk with a small group walking along it; pull back to the world map, then the calendar.

One species, out of one continent, across the whole world, in a geological instant. And nobody involved knew they were doing it. Next lesson: what life was actually like for the people doing the walking.

Shot list

One AI shot, thirteen code-rendered shots, two research images. Estimated Higgsfield use: 2 to 4 generations including retries.