Roughly 350 million Africans speak one of about 500 Bantu languages, from Swahili on the east coast to Zulu in the south. At least as many people — including many African Americans and Afro-Latinos — carry Bantu ancestry. The migration that produced both numbers is one of the largest and longest in human history, and it left almost no written record.

What it left instead was pottery, language trees, iron furnaces and DNA. A feature in Knowable Magazine, republished by Ars Technica, traces how three disciplines converged on the same route.

The clue it started with

Vasco da Gama rounded the Cape of Good Hope in 1497. A crew member who had spent time in west-central Africa could reportedly make himself understood in southeastern Africa — the languages were that similar, 4,000 km apart. Nineteenth-century linguists gave the family its name, from muntu, "human being." Twentieth-century archaeologists found matching pottery styles spreading north to south, with northern sites older. By the 1990s the migration reading was under attack: couldn't technologies and languages spread between people who stayed put?

Genetics settled it. A 2009 analysis of 2,432 Africans found Bantu speakers across west, central and southern Africa far more closely related than geography would predict. A 2024 study of 1,763 genomes found the greatest diversity in west Africa, declining with distance from it — the signature of a population expanding from one origin. Carina Schlebusch of Uppsala University: "This is a typical signature of a population expanding from one place to another."

Through the forest, then a fork

The route has been revised. Older accounts had the migration splitting before entering the Congo rainforest. Computational linguistics on roughly 400 Bantu languages in 2015 found deeper splits and far greater diversity inside the rainforest — evidence of a long stay there. Languages to the east and south look like they descend from a less diverse subset. A 2023 genomic comparison found Angolan Bantu speakers closer to Mozambican ones than to Cameroonian ones, pointing to a common ancestor south of the forest. The current picture: the groups crossed the rainforest together and split only at its southern edge, 2,500–3,500 years ago.

The archaeology is still thin where it matters most. At Shum Laka, the best-studied rock shelter in the Grassfields, pottery from 7,000–6,000 years ago is suggestive but not conclusive.

What the mixing looked like

The Bantu expansion was not replacement. Some central African Bantu-speaking groups carry about an eighth rainforest hunter-gatherer ancestry; some southern groups carry a similar share of Khoisan ancestry. The pattern is lopsided, and it is legible in which chromosomes you look at: mitochondrial DNA, inherited maternally, shows more Khoisan contribution than expected, while Y chromosomes show far more Bantu. One reading is Bantu men marrying Khoisan women — at a lower bride price, or not by choice. "Whenever there is a power difference between two groups, it tends to melt down to this," says geneticist Ananyo Choudhury.

The language kept the receipt: the clicks of Khoisan languages survive inside Zulu and Xhosa. Linguist Koen Bostoen's reconstruction is that the children of these pairings were bilingual at first, then lost their mothers' languages while leaving a substrate behind.

Iron, on a natural draft

At Great Zimbabwe — capital of a kingdom from roughly 1000 to 1700 AD — the furnaces needed no bellows. They were built to draw air in and exhaust it through a chimney, some nearly an adult's arm span wide and holding over 2,100°F. Archaeometallurgist Ezekiel Mtetwa has also found bone fragments added to the smelt: plausibly ritual, incidentally useful, since calcium accelerates the separation of iron from ore. A 12th-century Muslim cartographer rated iron from this region above India's.

What is still unsettled

Whether Bantu speakers invented iron smelting or learned it from neighbours. How much the language trees can bear: a 2026 study that allowed for contact between Bantu languages could not reproduce the earlier diversification patterns, and Igor Yanovich of the University of Vienna argues linguists should document the languages thoroughly before using them as a historical instrument. And money: much of this work runs on European grants, and Shadreck Chirikure of Oxford is blunt that without African government funding, "you are not generating fresh ideas."

The full account is Knowable Magazine, DOI 10.1146/knowable-110326-1.