Dhar Tichitt escarpment, Tagant, Mauritania
Dhar Tichitt
About 2000 BC to the 4th century BC (dry-stone villages, and a chronology that was redrawn)
Along the sandstone cliff of the Hodh, more than 400 villages were built in dry stone, some of them compounds inside compounds, while the basin below still held lakes. People kept cattle and sheep and grew pearl millet. One chronology walks them down the cliff as the water fell. Another says the cliff, the slope, and the lakeshore were used together, and that the big compounds are not a simple ladder of rank.
Why this is extraordinary
South of the Tagant, a sandstone cliff looks out over a basin that once held lakes. On the lip of that cliff, and on the slopes and the isolated rocks below, people built in dry stone. More than 400 villages are strung along the dhars of Tichitt, Oualata, and Néma. The famous ones are mazes of compounds, up to about 175 enclosures inside a larger wall. Most are smaller, fewer than 20 compounds, and some are only rings of stone with almost no enclosing wall. From about 2500 to 2000 BC until the 4th century BC, the communities on this refuge kept cattle and sheep or goats and grew pearl millet. The millet is known because stems and grains were pressed into wet clay and left their shapes in the pot. Grinding stones are everywhere. Little clay animals, cattle and the smaller herd, turn up as well. On the plateau, walled plots have been read as fields, the soil carried up from the low ground, and small round garden beds compared with the stone-ringed gardens of the Dogon much later. That comparison is a resemblance. It is not a genealogy.
The fight is the order. Excavations in the late 1960s drew a sequence in which the main mass of the large settlements belongs to the second half of the second millennium BC, and the villages step down the cliff as the lakes sink. A falling lake is not in dispute as a climate fact. What later workers refused was the mechanism: one people, one path, each height a new century. They pointed at sites on the plateau, the slope, and the old shore that can belong to the same centuries, and at a study area of a few dozen sites being asked to speak for a region of hundreds. The radiocarbon dates of that first generation were conventional, with the wide plus-or-minus that the method then carried. A 2026 restudy of pottery and of charcoal saved from those 1968 trenches says the old numbers were too blunt to choose between the stories, and it adds AMS dates, the tighter form of radiocarbon, on that charcoal. The restudy is plain that gaps remain. Clearer centuries do not by themselves crown a chief.
Size is the other argument. A plan of Dakhlet el Atrouss I is published as a centre whose compounds differ in ways that look social and economic. A big enclosure can be a ranked household. It can also be a family that mended its wall for longer. Dry stone invites addition. No azimuth of a compound toward a solstice is in the sources used here. The sky in this room is only the climate, the lakes and the drying, and that is weather, not a sighted star.
The ruin in the first photograph is village v.72, dark sandstone walls on the plateau. The second photograph is the living town of Tichitt under the same cliff. The houses with flat roofs are not the Neolithic compounds. What this changes: if cliff and basin were in use together, Tichitt is a long stone-village tradition of herders and millet farmers, not a single retreat. If the downhill ladder returns inside the new dates, the first sequence was right and only imprecise. The bull’s-eye at Richat is a geological dome, not one of these villages.
Why it matters
These are among the earliest large stone villages in West Africa. If the order is a march downhill away from a dying lake, the architecture is a climate story. If the sizes of compounds are a social ladder, the same walls are a ranked society. The dates were too coarse, for a long time, to decide.
The dhars are the cliffs between the Tagant plateau and the Hodh depression. From about 2500 to 2000 BC, agropastoral communities settled them and stayed nearly two millennia, until the 4th century BC. More than 400 villages with thick dry-stone walls stand on the lip of the cliff. The iconic ones enclose up to about 175 domestic compounds, and some much discussed villages are larger still. Small settlements, fewer than 20 compounds, are more numerous. Open sites sit on the baten, the low ground, where lakes lasted into the second millennium BC. Pottery temper carries impressions of domesticated pearl millet. Grinding stones are abundant. Terracotta figurines of cattle and of sheep or goats occur. Walled plots on the plateau have been read as fields whose soil was carried up from below, and small stone-ringed gardens compared with later Dogon practice. A first detailed sequence, from excavations in the late 1960s, put the main occupation of the large settlements in the second half of the second millennium BC and walked the villages down as the lakes dropped. Later workers called that march too mechanical, and argued that the cliff was not used the same way by everyone or in every century. A 2026 restudy of charcoal kept from the 1968 digs says the old conventional radiocarbon dates were too blunt to separate those stories, and it reports new AMS dates without closing the gaps in the evidence. No solstice azimuth of a compound wall is published.
Photo tour
2 photographs, unedited. On a phone, swipe the row. Choose one to look closely.
- Village v.72 is one ruin on the plateau. It is not Dakhlet el Atrouss, and it is not a plan of rank.
What is still unexplained
These sit beside the consensus account, with the same visual weight. None of them is marked solved.
Location and dating
- Location
- The Dhar Tichitt escarpment, Tagant, southeastern Mauritania. The pin is the town of Tichitt, 18.444 north, 9.493 west, at the foot of the cliff. The ruin in the photograph is village v.72 on the plateau, not the living town.
- Period label
- Current consensus — open to revision From about 2500–2000 BC until the 4th century BC for the Tichitt–Oualata–Néma settlement, with the main large villages often placed in the second half of the second millennium BC. The new AMS dates refine that without erasing it.
- How this dating was done
- Radiocarbon on charcoal, first conventional dates with wide errors, then AMS on samples kept from the 1968 excavations. Millet is identified from impressions in pottery temper, which dates a crop only as well as the sherd’s layer.
- What could change it
- The 2026 AMS series is the change already in print. What would change it again is dates from villages outside the original 40-kilometre study strip, especially the very large ones that the first maps left aside.
What is actually there
- More than 400 dry-stone villages line the cliffs.
- Large examples enclose up to about 175 compounds. Smaller hamlets are more common.
- Pearl millet impressions occur in the pottery. Cattle and sheep or goats were kept.
- One chronology follows a falling lake. Another treats cliff and basin as contemporary.
- A 2026 restudy says the old radiocarbon errors were too wide for those stories.
- No solstice azimuth is published. The photograph of the modern town shows the cliff, not a measured wall.
Open questions
| Question | Status |
|---|---|
| Did the villages march downhill as the lakes shrank? | Open. The first sequence says yes. Later work says the cliff was not used in one order. |
| Do big compounds mean ranked households? | Open. Plans show difference in size. Size is not yet a title. |
The current narrative
Where publication stands today. It is the best-supported reading, and it is open to revision. It is not a verdict.
The current narrative — open to revision
Cliff and basin used together
Agropastoral villages stood on the plateau, the slopes, and the old lakeshores across the same broad centuries, from about 2000 BC to the 4th century BC, and the first downhill march was an artefact of blunt dates.
What supports this
- Later surveys found large and small stone sites at more than one height, not in a single descending order.
- The 2026 restudy says the conventional dates could not separate the competing sequences.
- Millet impressions, livestock figurines, and grinding gear fit a mixed farm and herd economy rather than a single refugee moment.
What pushes back
- Lakes in the basin did shrink. A climate pressure is not imaginary.
- The late-1960s sequence was built from excavations, not from a glance at the cliff.
- New AMS dates refine the old charcoal. They do not, in the restudy’s own conclusion, close the social questions.
What would settle it
It is the centre because more than one generation of fieldwork has refused a single march downhill, and the date critique explains why that march looked orderly. It moves if the new AMS series actually restores the old order.
Emerging narratives
Other published readings. Each may be right. None is adopted here unless the evidence carries it.
An emerging narrative
A ranked society following the water
The large compounds are the upper end of a hierarchy, and the settlement shifted down the cliff as the lakes failed.
What supports this
- The first excavated sequence tied site type to height and to a falling lake in the second millennium BC.
- Plans of centres such as Dakhlet el Atrouss I are published as socially differentiated.
- Walled fields and storage, in jars and granaries, fit a community that held surplus.
What pushes back
- Contemporary use of plateau and basin is the published alternative, and the dates that enforced the march were too coarse.
- A larger compound can be a longer-lived household. Rank is an inference from area.
- The very largest villages were not all inside the first study strip, so the ladder was drawn on a sample.
What would settle it
It stays because hierarchy and climate are both still discussed, and the plans are real. It is not the centre while the chronological machine that made one ladder has been called too blunt by the restudy of its own charcoal.
Should the current view be reevaluated?
Open to revision
The villages, the millet, and the broad two-millennium span are secure enough to teach. The order of the sites and the meaning of a large compound are the published argument. Open belongs to that argument.
What drives this
- A downhill sequence and a mixed-height sequence, both published.
- Conventional radiocarbon errors too wide to referee, with new AMS dates that clarify and do not finish the job.
- Compound size read as rank, and also as household history.
This is the museum’s own weighing of the evidence. It is not a visitor vote, and it is separate from the slider.
Weigh it yourself
Anonymous. Nothing here is stored except on this browser. Your lean does not change the assessment beside it.
Visitor lean so far — not evidence. On this browser only; a shared tally is not collected.
If the contested reading is right
What would follow if a reading that is not the consensus turned out to be true, and what could settle it. This explains the stakes. It does not adopt the reading.
Contemporary villages of herders and millet farmers, not one march downhill
West Africa’s early stone villages would be a long regional system along the cliff, not a single convoy retreating from a lake, and rank would have to be shown by more than the area of a wall.
Why it is still worth a test
The new AMS series set beside the plans of the largest villages that the first strip left out.
Sources
- Dhar Tichitt, Walata and Néma Les nouvelles de l’archéologie, the span and the village count
- Revisiting Tichitt Azania, 2026, new AMS dates on the 1968 charcoal
- Dakhlet el Atrouss I African Archaeological Review, 2022, difference inside one centre
Image credits
- Compounds of village v.72 on the Dhar Tichitt–Oualata escarpment. Dry stone, no mortar. The modern town is not this ruin. . Sylvie Amblard-Pison, CC BY 4.0. Shown unedited.
- The living town of Tichitt and the escarpment behind it. The prehistoric compounds are on the cliff, not these houses. . Wilson McMakin, CC BY-SA 4.0. Shown unedited.