The Bantu Migrations: Two Thousand Years of Movement

The Bantu expansion is the spread of Bantu-speaking communities across much of sub-Saharan Africa over several thousand years. It is usually drawn as a set of bold arrows sweeping south and east from Cameroon, which is the one thing the evidence does not support. There was no organised march, no single route, and no fixed departure date — and the methods that reconstruct it produce ranges rather than dates.

Understanding how those ranges are produced is the useful part, because this is one of the clearest cases in world history of a major process known entirely without written records.

The evidence is linguistic, archaeological and botanical

There is no chronicle of the Bantu expansion. What exists is a family of several hundred related languages distributed across a vast area, archaeological sequences with radiocarbon dates, plant and animal remains, and — more recently — genetic data. Each method resolves something different, and none of them resolves everything.

Historical linguistics can establish that languages are related and roughly how they branched. Statistical phylogenetics can turn that branching into a dated tree, provided the tree is calibrated against something external. Archaeology supplies that calibration and also supplies the material record of settlement. Our explainer on how historians date events covers why calibration is the load-bearing step in any of this.

Where it started, and how fast it moved

Rebecca Grollemund and colleagues, writing in the Proceedings of the National Academy of Sciences, analysed around 400 Bantu languages and dated the common linguistic ancestor to approximately 4,800 years before present, with a stated uncertainty of about 138 years either side. The starting area in their reconstruction is north-western Cameroon.

Their central finding is about habitat rather than about people. In 96 of 100 reconstructed phylogenetic trees, the main line of movement runs south-east toward the southern edge of the rainforest — that is, along emerging savannah corridors rather than across the forest. Where movement did enter rainforest, it slowed: they calculate that rainforest occupation was delayed by an average of about 300 years compared with equivalent movement within a single habitat type.

That is a finding about terrain shaping human dispersal, and it is the opposite of an arrow drawn straight across a map. It also means the expansion was not one event but a sequence of movements with very different speeds in different places.

What the 2023 classification changed

Scholarship here is active, and the picture has been revised recently. A classification of 424 Bantu languages published in the Journal of African History in 2023 by Grollemund, David Schoenbrun and Jan Vansina rebuilt the tree from entirely new vocabulary data — a 100-word basic lexicon plus sound-correspondence analysis — and calibrated it against four archaeological events, including material culture change at the Shum Laka rock shelter in Cameroon and Urewe pottery in the Great Lakes region.

Two revisions matter. First, they reject the long-standing "deep split" hypothesis that Eastern Bantu separated early from Western Bantu; on their reading, Eastern Bantu emerged from Western Bantu groups considerably later, after populations had crossed the rainforests around 500 BCE. Second, they describe early Bantu speakers as operating eclectic food systems that combined farming with hunting, fishing and foraging.

That second point undercuts the standard shorthand in which the expansion is a farming frontier pushing outward. If the communities involved were mixing subsistence strategies, the expansion cannot be explained by agriculture alone, and the mechanism becomes a genuine open question rather than a settled one.