A nuclear reactor ran in Gabon with no one to build it

About 1.7 billion years ago, uranium ore at Oklo went critical. Groundwater was the moderator. The fission products are still in the rock.

Natural uranium today is about 0.72 percent uranium-235, too lean to hold a chain reaction in ordinary water. In the Palaeoproterozoic it was closer to 3 percent, which is reactor fuel. At Oklo the ore was rich, the geometry was right, and groundwater slowed the neutrons. Sixteen small zones ran, off and on, for a few hundred thousand years. The power was modest, on the order of a hundred kilowatts, not a bomb and not a plant. When the water boiled off, the reaction stopped. When water seeped back, it started again.

The proof is the ash. The ore is depleted in 235 exactly as spent fuel is, and the stable daughters of the fission products are sitting where the neutrons made them. Paul Kuroda had said in 1956 that such a thing could happen. The French atomic agency found it because their assay of a commercial shipment did not add up. Nothing like it has been found anywhere else. The conditions that allowed it — enough 235 still in the crust, and oxygen enough to concentrate the ore — came and went.