Everything we know
about the earliest humans

Written history covers roughly the last five thousand years. Stone tools go back three and a third million. If the whole toolmaking record were a single day, writing would begin about two and a half minutes before midnight.

The chart below keeps real proportions. Every one of its three strips is an ordinary, evenly spaced timeline, so the gaps you see are the gaps that actually elapsed. The catch is that almost everything happens in the last sliver of the top strip — so each strip magnifies the tail of the one above it, and says by how much.

Ordering those sites by age produces a list in which Europe appears late and intermittently. Which continent holds the first of any given innovation is a less stable fact than it sounds — it depends on how narrowly the category is drawn, and it moves whenever something older is dug up or redated. The map is a record of where the surviving evidence happens to be, not a league table.

At Lomekwi 3, west of Lake Turkana in Kenya, stone flakes and cores were struck deliberately 3.3 million years ago. That is around 700,000 years before the earliest fossil assigned to Homo. Whoever knapped them — Kenyanthropus and Australopithecus are both candidates — the behaviour we think of as definitively human was already underway in a body that was not.

The Oldowan industry follows from about 2.6 million years ago at Gona in Ethiopia: simple cores and sharp flakes, effective and unfussy. Then, from roughly 1.76 million years ago at Kokiselei in Kenya, the Acheulean handaxe — worked on both faces to a symmetrical teardrop, made to the same template across Africa, Europe and Asia for over a million years. It is the longest-running design in human history by an enormous margin.

A single human species on Earth is a very recent and very unusual situation. For nearly all of prehistory there were several, overlapping in time and often in place.

The colours answer the obvious question about relatedness. Neanderthals and Denisovans are sister lineages — our common ancestor with both is around 600,000 years ago, and those two parted from each other about 400,000 years ago. Homo erectus and Homo heidelbergensis sit on our own line, or close enough to it that the argument is about detail. Homo naledi, Homo floresiensis and Homo luzonensis are cousins whose branching points are genuinely unresolved; the shared ancestor is probably somewhere near two million years back for each, but nobody should pretend that is settled.

Neanderthals were not a failed draft of us. They buried their dead, made birch-bark tar — which requires heating bark in the absence of oxygen, a genuinely difficult process — collected eagle talons that look like ornaments at Krapina about 130,000 years ago, and in one case went 336 metres into a French cave and built something.

Denisovans are stranger still: a whole population first identified in 2010 from a fingerbone in the Altai, and known mostly from DNA rather than bones. One individual from Denisova Cave, nicknamed Denny, turned out to have a Neanderthal mother and a Denisovan father — a first-generation hybrid, found on the first try, which suggests such pairings were not rare.

For most of the twentieth century, the story of early art was a European story — Chauvet, Lascaux, Altamira. That is no longer where the record sits, and the correction has come almost entirely from Asia, Africa and Australia.

In Africa, engraved ochre from Blombos Cave in South Africa carries deliberate cross-hatching at around 73,000 to 77,000 years ago, and shell beads from Bizmoune Cave in Morocco — pierced, and therefore meant to be strung and worn — date to roughly 142,000 years ago. Ornament, which is a statement about who you are to somebody looking at you, is that old.

And the oldest known picture of a scene is in Indonesia.

The hand stencil is worth pausing on, because it turns up everywhere: Sulawesi, Borneo, Spain, Argentina, Australia. It is the simplest possible self-portrait and requires no drawing skill at all — you put your hand on the wall and spray around it. People separated by oceans and tens of thousands of years arrived at the same idea, which is either a very dull coincidence or a fact about minds.

Music arrives in the record around 40,000 years ago, and in the same German caves as the ivory figures: flutes of bird bone and mammoth ivory from Hohle Fels and Geißenklösterle, drilled with finger holes and playable.

When our ancestors moved out of Africa they met these populations and had children with them. Those genomes are still circulating. Everyone whose ancestry lies outside sub-Saharan Africa carries Neanderthal DNA, and people in Oceania carry a substantial Denisovan component as well.

This is not a curiosity. Some of the inherited variants do real work.

Among the most numerous surviving objects from the Palaeolithic and Neolithic are small carved and modelled figures, a great many of them female. They appear from the Pyrenees to Siberia in ivory, limestone, bone and fired clay, across an enormous span of time.

The assemblages are mixed. This matters, and it is often left out. At Çatalhöyük in Turkey, the great majority of the roughly two thousand figurines recovered are animals or abstract forms rather than the well-known fleshy female type.

What can be said with confidence is that they mattered. People carried the stone for them over long distances, spent many hours on objects a few centimetres high, and kept making them for tens of thousands of years. Nobody does that with something trivial.

The asymmetry among the human figures is real, though. Mixed assemblages cut against a single Mother Goddess; they do not cut against this. Where Upper Palaeolithic people carved a person rather than an animal, the person is usually a woman. Male figures exist and are scarce. Whatever the figurines were for, the oldest images we have of ourselves are mostly of women — and the oldest undisputed one of all, the Venus of Hohle Fels, is a woman carved from mammoth ivory before anyone painted a cave in Europe.

Put the two facts side by side and the subject matter of the earliest art is strikingly narrow. It is animals, and it is women. Those are the categories that got carved, modelled and painted, for tens of thousands of years and across continents — the herds that fed people, and the bodies that made and fed people. Men are present and thin on the ground: human figures are the minority of the assemblage at Çatalhöyük to begin with, and identifiably male ones a minority of those. The painted caves run the same way, full of aurochs, horses and lions, with people rare and men rarer.

Reading the Snake Goddess, symbol by symbol

This figure is the most reproduced object in Minoan archaeology, and almost every element on it has been assigned a meaning at some point. It is worth going through them one at a time and asking, for each, what it actually rests on — because the answers are very uneven, and the object itself is partly a modern reconstruction.

Click any marker on the figure.

For most of the twentieth century the answer was fixed: people entered the Americas about 13,000 years ago, and the Clovis points found across North America marked their arrival. Monte Verde in southern Chile broke it. A waterlogged site preserved wooden stakes, cordage, worked timber and plant remains at around 14,500 years ago — south of the ice, and earlier than the supposed first arrival in the north. After a long fight the site was accepted, and "Clovis first" went with it.

Human footprints preserved in lakeshore sediments at White Sands in New Mexico have since been dated to roughly 21,000 to 23,000 years ago. That work is still being argued over, on the reliability of the material used to date it, and it deserves to be watched rather than banked.

Caral is also notable for what has not been found there. Across roughly a thousand years of occupation, excavation has produced no weapons, no defensive walls and no mutilated remains. It does have elite residences and evident social ranking — this is not a society without status — but it appears to have been held together by ceremony and exchange rather than by force.

How the stone was actually fitted

This is documented rather than mysterious, and the documentation is better than the legends. Andean quarries at Rumiqolqa and Kachiqhata still show the working faces. Blocks were levered free along natural fracture planes, then shaped by pounding — not carving — with river-cobble hammerstones ranging from about 200 grams up to eight kilos, a graded set from rough reduction to fine finishing.

The joints were made by fitting, not by measuring. A block was lowered onto the one below, then lifted; the compressed stone dust left on the lower surface showed exactly where the two touched. Those high spots were pounded down, and the block was lowered again. Repeat until the dust smears evenly and the seam closes. It is slow, entirely reproducible, and experimental archaeology has done it.

Ollantaytambo, where the building stopped mid-sentence

Ollantaytambo is the best single site for this, because the Inca were still working on it when the Spanish arrived and the whole operation was abandoned in place.

The rhyolite came from the Kachiqhata quarries on the slopes of Yana Urqu, four to five kilometres away and on the far side of the Urubamba river valley. The route is still legible on the ground: the transport roads, the earthen ramps, the wooden rollers — and the "tired stones," blocks that were dragged part of the way and then left where they stopped when the work ended.

A half-finished building site is worth more than a perfect one. Nothing has been cleared away, so the ramps, the abandoned blocks and the unfinished joints are all still lying in the sequence they were used in.

Who built them

The Inca ran a labour tax called mit'a: communities owed the state a rotation of workers rather than a share of goods. That is how the roads, terraces and walls were built — not by a permanent servile class, but by people serving a turn and going home. It is the same broad answer as at Giza, arrived at independently on the other side of the world, and it is the answer the evidence keeps returning for large ancient construction.

The textbook sequence used to run: agriculture produces surplus, surplus produces hierarchy, hierarchy produces monuments. The evidence no longer supports that order.

Göbekli Tepe and its neighbours were built by hunter-gatherers, millennia before agriculture reached the region. The excavations answer the obvious practical question directly: the limestone plateau immediately adjacent is the quarry, and it still holds cutting channels and unfinished pillars lying where they cracked, attached to the bedrock. The stone was worked on the spot, with stone.

The organising force appears to have been food and ceremony rather than command. Enclosure backfills are thick with smashed animal bone — gazelle, aurochs, wild ass — the residue of feasting on a scale that implies large gatherings. People came, ate, built, and went.

The same pattern recurs on other continents, independently.

Çatalhöyük, occupied from about 7500 BCE, packed several thousand people into houses entered through the roof, with the dead buried beneath the floors. Twenty-five years of excavation under Ian Hodder found houses near-identical in size, layout and contents, no elite residences, no public buildings set apart from ordinary dwellings, and — from isotope analysis of bone — no differential access to food.

Further north, the Cucuteni-Trypillia culture built settlements of hundreds of hectares in rings of houses around 4000 BCE. Nebelivka, Taljanky and Maidanetske cover more ground than the earliest Mesopotamian cities and are older. They contain no palace, no central temple dominating the plan, and no royal burials. Roughly every sixty to eighty years the inhabitants appear to have burned the settlement deliberately and rebuilt it.

The land that is now the United States holds monumental construction going back five and a half thousand years, and most of it is barely known outside the states it sits in.

The oldest is Watson Brake in Louisiana, an eleven-mound complex raised around 3500 BCE — older than Poverty Point, older than Stonehenge, and built by hunter-gatherers. Poverty Point follows on the same river system. Then, from about 100 BCE, the Hopewell earthworks of the Ohio valley.

Who built them

The Hopewell were not an empire and, as far as the evidence shows, had no centralised leadership and no kings. They were substantially hunter-gatherers, supplementing wild food with a few domesticated plants. What they did have was reach: obsidian from the Yellowstone region, copper from Lake Superior, mica from the southern Appalachians, and shark teeth and marine shell from the Gulf coast, all turning up in Ohio. Archaeologists call it the Hopewell Interaction Sphere, and it is a continental network without a capital.

Further west, the Ancestral Puebloans built the great houses of Chaco Canyon in New Mexico between roughly 850 and 1250 CE — Pueblo Bonito alone rising four storeys and containing over six hundred rooms — together with astronomical alignments and a network of engineered roads running straight across the desert. Their descendants are the living Pueblo peoples of the southwest, which is worth stating plainly: these are not vanished civilisations, and the people who built them have not gone anywhere.

It is sometimes said that the first Egyptian pyramid was the largest and most sophisticated, with no earlier architecture leading up to it — and that the knowledge must therefore have come from somewhere else. The physical record runs the other way, and it runs that way unusually clearly, because the Egyptians did not tidy away their failures.

Before any pyramid there were mastabas: flat-topped rectangular tombs built through the first two dynasties. The first pyramid, Djoser's at Saqqara around 2670 BCE, is essentially six of those stacked one on another. It is a step pyramid, not a smooth one, and at 330,400 cubic metres it is a fraction of what came later.

Between those two came the pyramid at Meidum, begun as a step pyramid and converted mid-project to a smooth-sided one. It collapsed. Sneferu then built the Bent Pyramid, met trouble again, and finished it at a compromise angle with its interior chambers unfinished. Only on his third attempt — the Red Pyramid, laid out at a safe 43°22' from the ground up — did anyone in Egypt complete a structurally sound true pyramid. His son Khufu built the Great Pyramid using what those three had cost to learn.

Who built them

The workers' settlements at Giza have been excavated, and they answer the question directly. There are bakeries and breweries on an industrial scale. The animal bone shows cattle, sheep and goat raised on estates elsewhere in Egypt and shipped in — the builders ate better than most Egyptians ever did.

The builders also left their own names on blocks hidden inside the masonry, and they organised themselves into competing gangs with names like "Friends of Khufu" and "Drunkards of Menkaure," subdivided into units that borrowed nautical terms — starboard, prow.

Prehistory has a vocabulary, and most of it is doing real work rather than showing off. These are the terms used above, with what each one actually means.

Listen & Repeat reads each word and then waits for you to say it back. Speaking drill turns on the microphone and checks whether the words came out clearly. On each card, say it reads the word, and tell me more opens a fuller explanation and a sentence to try.