In May 1972, a worker at a nuclear fuel-processing plant in France ran a routine check on a batch of ore and found a number that should not have been possible. Natural uranium anywhere in the crust, on the Moon, even in meteorites, carries the same fraction of the fissile isotope uranium-235: 0.720 percent. The sample in front of him, traced back to the Oklo deposit in Gabon, read 0.717 percent. It was a tiny gap, but in this field a tiny gap is a scream.
The explanation, once French scientists worked it out, was stranger than a measurement error. Roughly two billion years ago, parts of the Oklo ore body had begun running as a natural nuclear reactor, a self-sustaining fission chain reaction with no human hand involved. It ran, by the best reconstruction, for a few hundred thousand years. The missing uranium-235, around 200 kilograms in one zone alone, had been burned up as fuel, two billion years before anyone could build a reactor on purpose.
Why it could happen then and not now
The reason Oklo is a one-time event in Earth’s known history comes down to a slow clock running inside the rock. Uranium-235 is radioactive and decays about six times faster than the more common uranium-238. So the deeper into the past you look, the richer natural uranium was in its fissile isotope.
Today uranium-235 sits at under 1 percent of natural uranium, far too dilute to sustain a chain reaction on its own. But two billion years ago, around the time the Oklo deposit formed, that figure was roughly 3 percent. That happens to be close to the enrichment level used in many commercial power reactors today. The fuel was already, in effect, reactor grade. It simply needed the right conditions to ignite.
A prediction had been waiting in the literature for this. In 1953 two physicists, George Wetherill and Mark Inghram, suggested that some uranium deposits might once have worked as natural reactors. Shortly afterward, in the mid-1950s, the chemist Paul Kuroda spelled out what spontaneous fission in an ore body would actually require. The uranium vein had to be large enough that escaping neutrons would strike another uranium nucleus before leaving the ore. There had to be enough uranium-235. There had to be a moderator, something to slow the neutrons so they could trigger further fission. And there could not be too much of a neutron-absorbing “poison” like boron to smother the reaction. At Oklo, all four conditions lined up.
Water was the throttle
The moderator at Oklo was ordinary groundwater. Water slows fast neutrons to the gentle speeds at which they most readily split uranium-235, and the saturated ore had plenty of it. That detail turns out to explain not just how the reactor started, but how it kept from destroying itself.
When the chain reaction heated the rock past the boiling point, the groundwater flashed to steam and drove itself out of the reaction zone. With no water to slow the neutrons, the reaction stalled. The rock then cooled, fresh groundwater seeped back in, and fission resumed. The pattern repeats the logic of a geyser, which heats, erupts, refills, and waits.
This is not a guess pieced together from the surrounding geology alone. A team led by Alex Meshik at Washington University in St. Louis read the reactor’s operating schedule out of the rock decades later, by studying the isotopes of xenon trapped in grains of aluminum phosphate within a single fragment of Oklo ore. Because different xenon isotopes form on different timetables after fission, the gas preserved a kind of clock. The team’s model pointed to one reactor switching on for about 30 minutes and off for at least two and a half hours, cycle after cycle.
The numbers it left behind
Across the Oklo and adjacent Okelobondo mines, researchers eventually identified 16 separate zones where this had taken place. The reactors were not violent. The total energy released over the whole episode came to about 15,000 megawatt-years, and the average power output was probably under 100 kilowatts, which the researchers note is roughly enough to run a few dozen toasters.
They also left chemical fingerprints. The fission of uranium-235 produces lighter “daughter” elements, and the abundance of those products in the ore is what first proved, soon after the discovery, that a chain reaction rather than some exotic chemistry had drained the uranium. The reactors even bred more than two tons of plutonium-239 from uranium-238, almost all of which has since decayed away.
What did not happen is as striking as what did. Over hundreds of thousands of years of operation, the self-regulating water cycle meant there was, in the reconstructed record, not a single meltdown or explosion. The deposit handled its own radioactive waste in place, which is partly why repository scientists have studied Oklo so closely as a natural model for storing nuclear waste underground.
What this does and does not prove
It is worth being precise about what Oklo establishes, because the bare phrase “natural nuclear reactor” invites bigger claims than the evidence supports. The fission itself is not in doubt. The depleted uranium-235 and the matching spread of fission products are about as direct as geological evidence gets, and they were confirmed within a few years of 1972.
The finer details are reconstructions, and they carry the uncertainty of any model built from two-billion-year-old rock. The neat “30 minutes on, 2.5 hours off” figure comes from one reactor zone and one analytical approach, the xenon study, not from a stopwatch. The total duration is usually given as hundreds of thousands of years rather than a single confident number, and the average power output is an estimate. These are well-supported inferences, but they are inferences, and careful sources keep the hedges attached.
A second caution concerns reach. Oklo is often invoked in debates about whether the fundamental constants of physics have shifted over cosmic time, since the reactor’s behavior depends on nuclear properties that those constants govern. Different teams have drawn opposite conclusions from the same deposit, with some reading it as evidence of no change and at least one pair of physicists reading it as evidence of a small one. That disagreement is a live research question, not a settled result, and Oklo does not resolve it on its own.
A reactor older than complex life
Set against the human timeline, the strangest thing about Oklo may be simply when it happened. Two billion years ago, Earth’s atmosphere was only beginning to fill with oxygen and life was still single-celled. There were no plants, no animals, nothing that could be called a witness. The reactors ran their long, quiet cycles and shut down for good while the planet was still largely microbial.
When the French traced that 0.003 percentage-point shortfall back to its source, they were not discovering a curiosity so much as reading a message left by deep time. The conditions that made Oklo possible have not returned, because the fuel itself has thinned with age. Whether other natural reactors once flickered into existence elsewhere, and left fainter traces, is a question still open to anyone willing to go looking for the right wisps of gas.
