The Garamantes: Herodotus Wrote Nomads. The Fazzan Survey Found Cities.

In the fifth century BCE, Herodotus described a people living deep in the Sahara whose cattle grazed backward, because their horns curved so far forward that grazing any other way was not possible. The same people, he wrote, hunted „cave-dwelling Ethiopians“ from four-horse chariots, chasing down men who were, in his account, the fastest runners anyone had ever heard of. He called them the Garamantes, and for most of the twenty-four centuries since, that was roughly all anyone had to go on: a strange footnote about a desert people on the far edge of the known world.

Then, starting in the late 1990s, a British-led survey team walked the Fazzan region of southern Libya and mapped what satellite images had already hinted at. Not nomads. Walls. Towers. Streets laid out on a grid. Gardens by the thousand.

A city the size of a Roman town, in the middle of the Sahara

The survey, led by archaeologists David Mattingly and Martin Sterry, recorded 158 previously unmapped settlements, 184 cemeteries, and 30 square kilometres of field systems across the central Fazzan, most of it unknown before the project began. The Garamantian capital, Old Jarma, covered at least 7.5 hectares, with a walled area reaching 9.3. A second town, Qasr ash-Sharraba, was larger still: 15 to 18 hectares, a planned street grid, a 0.7-hectare fort, and a field system of 556 hectares feeding roughly 1,600 individual gardens. The Greco-Roman geographer Ptolemy called Jarma a metropolis in his Geography and tied it directly to Garamantian kings, not wandering herders.

Mattingly and Sterry set those numbers against towns the Roman world already recognised as urban: Byzantine-period Sabratha on the Mediterranean coast at 9 hectares, Gigthis in North Africa at 20. The Garamantian towns sit in the same range, just further from the sea. The two archaeologists also proposed a threshold for what an oasis could manage on its own: up to roughly 4 hectares and 400 to 800 people, a settlement can support itself from its own fields. Beyond that line, it needs outside capital, trade or political power strong enough to hold the extra population together. Most of the surviving Garamantian fortifications date not from a period of decline, as a reader might expect, but from the settlements‘ peak, roughly 300 to 500 CE.

The survey did not just find two large towns. It sorted everything it mapped into a seven-tier hierarchy, from full towns like Jarma and Qasr ash-Sharraba down through fortified villages, villages built around a fortlet, open villages with no wall at all, isolated fortlets, isolated mudbrick buildings and, at the smallest end, isolated oval huts. That range matters for the argument Mattingly and Sterry are making: a single large excavation can be dismissed as an outlier, but a seven-tier settlement hierarchy spread across the 158 mapped settlements and 30 square kilometres of fields above is a description of an entire society, not one impressive dig.

Three readings of the same rock art

Scattered across the Sahara are rock engravings of horses pulling two-wheeled chariots. From the 1950s onward, the French researcher Henri Lhote read them as evidence of a fixed trade route, carrying goods as far as West Africa. That reading is now considered disproved. There is no wreckage west of the Fazzan, no early horse skeletons, and spoked wheels are a poor fit for deep dune sand in the first place.

Mattingly reads the same images more cautiously, as a sign of increased mobility across the first-millennium-BCE Sahara, nothing more, no continuous trade corridor. A third reading goes further, and it is the harder one to sit with. Archaeologist Elizabeth Fentress published a paper in 2011 titled „Slavers on Chariots,“ arguing the images may have more to do with slave raiding than with trade. I have only encountered this argument through the citation in Mattingly’s own 2017 chapter, not in the original paper itself, so what follows is a report that the argument exists and who is reporting it, not a judgment on whether it holds up on its own terms.

Three researchers, the same rock art, three different stories about what it shows.

What the canals actually cost

None of this worked without water, and the water came from foggaras: underground channels that carried groundwater to the surface, sometimes for kilometres. Without them, no gardens, no dates, no grain at the scale these towns needed. Keeping that system running cost an estimated 72,000 person-years of labor, according to Andrew Wilson and David Mattingly, who write plainly that the work was done, in their words, „in large measure… by the use of slaves.“

That is not an isolated guess sitting alone in one paper. The Tariff of Zarai, a Roman customs inscription from 202 CE, names mancipia, enslaved people, near the top of the list, ahead of textiles and livestock. An inscription, not a suspicion. The customs post itself, though, sat not in the Sahara proper but on its northern edge, in the Numidian highlands, and records trade mainly between the fertile Tell region and the pre-Saharan pastoral zone. Whether the people listed there actually came from as far as the Fazzan or from closer frontier populations is something the inscription itself does not say. What it does show clearly is how routine slave trading was along the whole North African periphery under Rome. Classicist Brent Shaw, reviewing the same body of research independently of Mattingly and Sterry, raises a second, harder question in the same review: was the Garamantian rise genuinely independent of Rome, as Mattingly tends to argue, or did it depend far more on proximity to the Roman Empire than the archaeological enthusiasm for the find tends to allow? Nobody has settled that question yet.