Arabia’s Empty Quarter had 55 ancient lakes and no trace of humans, for lack of stone

Deep inside Arabia’s Empty Quarter, the largest sand sea on Earth, the dunes hide the floors of old freshwater lakes, still littered with snail shells and pieces of ostrich egg. What they do not hide is any sign that people camped beside them. A team that crossed the desert found lake beds in abundance and not a single stone tool, although early humans lived on the desert’s edges at the same time.

The explanation, published in Nature, is geological rather than climatic. The interior of the Rub’ al Khali has no rock that breaks into sharp edges, so people who depended on stone tools could not settle there, however much rain fell. If that holds, the maps used to trace how Homo sapiens spread out of Africa have been drawing the habitable world too large.

Those maps follow the water. During wet phases driven by slow shifts in Earth’s orbit, the Indian Ocean monsoon pushed north and turned parts of the Arabian Peninsula into lakes and grassland, a landscape researchers call “Green Arabia”, and interior lakes were treated as magnets for settlement. The new study, led by archaeologist Jeffrey Rose of the University of Algarve in Portugal, with University of Vienna geologist Rudolf Dellmour as first author, is the first to test that assumption on the ground in the interior.

The survey logged 125 localities along its route. Fifty-five of them held lake deposits from a wet phase dated, at nearby sections, to roughly 122,000 to 67,000 years ago. Ground visibility on the wind-scoured flats between the dunes was above 95%. Across the whole route the team found no artefacts, no flaking debris and no rock suitable for making either.

How a desert crossing became an archaeological test

The fieldwork was a single expedition: eight people in three vehicles, carrying about 1,700 litres of water and supplies for six weeks. The track ran roughly 1,500 kilometres through the central Rub’ al Khali, plotted from satellite images to cross a belt of lake beds that earlier researchers had dated by luminescence, a method that measures when buried sand grains last saw sunlight.

At each lake locality the team walked a corridor of about 500 metres around the exposed sediment, looking for artefacts, shells, ostrich eggshell, fossil burrows and any rock that could be worked into tools. Dellmour classified every surface on the spot. Nothing was dug up or removed; the work was coordinated with the Saudi Ministry of Culture and recorded with GPS-tagged photographs and drone images.

The lake evidence was clear: carbonate platforms more than 20 centimetres thick packed with the freshwater snail Melanoides tuberculata, Corbicula clams, and gypsum crystals that grew as the lakes dried out. In a University of Algarve statement, Dellmour put the logic simply: a team that could spot a snail shell the size of a fingernail would also have spotted a stone tool, and there were none.

The stone half of the test was just as clear. About 80 kilometres from the route’s western start, near Mundafan, a known ancient lake site, the last bedrock appeared. Beyond it, the flint-bearing formations dip under younger sand, carbonate and salt deposits, none of which breaks with the smooth, shell-like fracture toolmakers need. Workable stone turned up at only 10 of the 125 localities, all at the two ends of the route, and the nearest usable outcrops lie more than 200 kilometres from the surveyed lake basins.

Water without stone, stone without water

The emptiness is striking because the neighbours were busy. Around the margins of the same desert, people were making stone tools during the same wet phase: at Mundafan about 96,000 to 77,000 years ago, at Jebel Faya in the United Arab Emirates between about 134,000 and 80,000 years ago, and on Oman’s Dhofar Nejd plateau about 109,000 to 95,000 years ago.

To test the pattern beyond one transect, the team built a map that scores land on water and stone together, trained on more than two decades of surveys in Oman and checked against 301 Middle Stone Age find spots there. Across the whole Peninsula, 57% of the land within reach of water sits on formations with no workable stone. The same pattern shows up across 363 documented sites on the Peninsula. Adding stone also improved how well the model separates known sites from random ground: its score on a standard test, where 0.5 means guessing and 1 means perfect, rose from 0.695 to 0.768.

Two other places act as controls. Jubbah, a lake basin in northern Saudi Arabia, sits on sandstone that supplied plenty of quartzite, and people lived along its shores. Duqm in central Oman has good chert but little water, and very few sites. The 10 stone-bearing localities on the transect yielded no artefacts either: stone alone was no more enough than water alone.

Distance explains the rest. In well-studied Middle Palaeolithic sites, 60 to 98% of the stone comes from within about 5 kilometres, and sources beyond 20 to 30 kilometres usually supply only 1 to 2%. Living by the interior lakes would have meant hauling stone more than 200 kilometres between groundwater-fed basins that had no rivers to carry pebbles in.

When lakes returned after the last ice age, the central basin still produced no cultural material from any period. The authors cite similar patterns on Libya’s Messak Plateau, where 91% of Middle Stone Age finds cluster near one kind of sandstone, and in Egypt, Tibet and Clovis-era North America. In the paper’s words, “‘Green Arabia’ was a lithic archipelago whose islands were defined as much by geology as by climate.”

Rose said in the university’s statement that everyone working on early human dispersals in Arabia, himself included, had assumed that finding the water meant finding the people. Twenty-five years of fieldwork in Oman, he said, taught him that people who depend on stone tools never stray far from good stone.

What the empty desert doesn’t settle

Absence is harder to prove than presence. The team walked about 43 square kilometres of corridor in a desert of roughly 650,000 square kilometres, along one line, and a surface survey cannot see under the dunes. The authors accept that future fieldwork may find small sites where people stopped briefly, though they argue such finds would mean short visits, not settlement.

The dating is indirect. Only five surveyed lake localities lie within 6 to 31 kilometres of sections dated in earlier work; the rest are tied to that wet phase by elevation and sediment type. The shells were identified from photographs alone. The 57% figure rests on geological maps at a scale of 1 to 2,000,000 and a yes-or-no classification of each formation, which the authors call approximate but conservative.

The model also does not explain why people stayed away. Small founding populations may never have needed to move into poor land, or foragers may have known the interior lacked stone and avoided it; neither idea can be tested with Arabia’s current climate records. One alternative the team rules out is wooden or bone tools, because those are shaped with stone edges and would still leave flakes behind.

Common questions about the Empty Quarter and early humans

What is the Rub’ al Khali?

The Rub’ al Khali, or Empty Quarter, is the largest continuous sand desert on Earth, covering about 650,000 square kilometres of the southern Arabian Peninsula across Saudi Arabia, Oman, the United Arab Emirates and Yemen.

Was the Arabian desert ever green?

Yes. During wet phases of the Pleistocene, a stronger Indian Ocean monsoon raised water tables and filled lakes across the Peninsula, with grassland around them. The most extensive lakes in the Rub’ al Khali date to roughly 122,000 to 67,000 years ago.

Why did early humans need stone to survive?

Middle Palaeolithic people made cutting and hunting tools by flaking hard, glassy rock such as flint, chert or quartzite, and the methods common in Arabia used a lot of it.

When did Homo sapiens reach Arabia?

A fossil finger bone from Al Wusta in Saudi Arabia shows modern humans were there by about 85,000 years ago. Stone tools at sites such as Jebel Faya in the United Arab Emirates are older, although who made them is still debated.

What comes next

The team has archived its field photographs and analysis code in a public repository so other groups can rerun the stone-and-water model on their own deserts, and it asks dispersal modellers to add raw-material geology to rainfall and rivers. The expedition crossed the Rub’ al Khali between 15 November and 20 December 2023, and the study was published in Nature on 28 October 2026. Its closing line puts the result in one sentence: “Whether early human populations settled a desert landscape depended as much on the stone exposed across its surface as on the water that filled its basins.”

Reference: Dellmour et al., “Stone and water governed early human settlement of the desert,” Nature, 2026. DOI: 10.1038/s41586-026-11180-5

Tags: archaeology, human evolution, stone tools, Arabian Peninsula, Homo sapiens, Rub' al Khali