The textbook story sends iron from Anatolia to Africa last

For most of the twentieth century the history of iron was taught as one invention with many copies. Smelting began in Anatolia around the middle of the second millennium BCE, moved through the Levant and the Middle East between 1300 and 1000 BCE, went east to India and west towards Europe, and crossed from southern Arabia to the Ethiopian highlands before the middle of the first millennium BCE. In North Africa the Phoenicians brought iron to their settlement at Carthage in the late ninth century BCE. Meroe had it by the fifth century BCE, arriving either up the Nile through Egypt or across the Red Sea, and became a large producer. Africa south of the Sahara entered the story last, receiving the technique from these northern points of contact. That version is still commonly taught. Ehret objects on dates. The chronology of the savanna sites does not fit a technology working its way down from the Mediterranean fringe.

Late in the Archaic Age the traffic grew heavier between the islands and the Isthmo-Colombian mainland of Central and South America. Goods moved, and so did people and ideas. Keegan and Hofman describe these routes as lifelines running back to the communities the islanders had come from, which is a precise way to put it. A small population scattered across islands needs contact to stay viable, socially and culturally and in the plainest biological sense, and the mainland connection supplied it. The archaeology of the period looks different from one island to the next, and the authors read that variation as people adjusting to the particular place they had landed in. What this era built was the framework the later ceramic age villages would use. The settled farming communities that eventually filled Hispaniola did not appear out of nowhere. They inherited routes, relationships and habits of movement that had been in service for centuries.

The spread of the early iron smelting sites carries two implications. The first concerns place. The pattern of the distribution makes it likely that the people who worked out the technology lived somewhere in the fifteen hundred kilometres of woodland savanna between central Nigeria and the far northwestern fringe of East Africa. The second concerns time. Sites were already scattered across that enormous stretch three thousand years ago, and a technology takes generations to travel that far, so the founding workshops should date centuries before 1000 BCE. Both expectations have since been tested. In the last two decades archaeologists have found earlier African ironworking than anything previously recorded, and the finds sit inside the predicted zone. Ôboui, in the western Central African Republic, is one of them. This is the pattern that keeps repeating in Ehret's field. A prediction is made from vocabulary and distribution, someone digs in the indicated place, and the dates come back older than the standard account allows.

The evidence is indirect and it points to small states in Aïr, in West Africa, by 1000 BCE or earlier. The Nok region of Nigeria appears to have held an early strong state during the first millennium BCE, and its wealth most likely came from trading the metals found in its territory. Nok is famous for something else. Its terracotta heads and figures are among the earliest known figurative sculpture in West Africa and they sit in museums around the world, captioned as art and nothing more. The political economy behind the workshops goes unmentioned. Sculpture of that quality in that quantity implies patrons, and the patrons were living off control of metal and the trade in it. Together with Aïr, the finding puts political complexity in the Sudan belt at 1000 BCE or earlier. That is contemporary with the first Iron Age states around the African Great Lakes, and centuries before the gold trade that made Ghana, Mali and Songhay rich.