A small discrepancy in uranium measurements led investigators back to Oklo, near Franceville in south-eastern Gabon. The discovery in 1972 opened a much older history: natural nuclear fission in rocks roughly two billion years old. This case follows the laboratory anomaly, the geological explanation and later attempts to reconstruct what happened underground. It is a documented scientific investigation whose original puzzle was resolved through physical evidence.
The measurement that prompted an investigation
In a near-contemporary account, IAEA director general Sigvard Eklund described routine analyses at Pierrelatte in France in June 1972. A measured uranium-235 concentration of 0.7171% differed slightly from the expected 0.7202%. Repeated checks established that the discrepancy was real. Investigators traced the material to Oklo; by early August they were considering ancient chain reactions, and preliminary findings were announced in September. Eklund’s account locates the discovery within a working mine operated by COMUF, with Gabonese and French participation.
Those figures describe the initial investigation, not a uniform composition throughout the deposit. Eklund reported substantial differences between samples. His 1975 article also used age estimates of around 1,800 million years. Later studies commonly describe the phenomenon as about two billion years old. Preserving the date of each estimate prevents an early calculation from appearing to contradict a later geological chronology.
Checking the anomaly by another method
Robert M. Moxham’s 1976 USGS study compared measurements from mass spectrometry with gamma-ray spectroscopy. A particular sample contained about 0.51 atom percent uranium-235 by the former method and gave approximately 0.54 atom percent by the latter. Both indicated depletion. The values were close, rather than identical; the comparison offered another way to examine an unusual ore sample.
This is a useful distinction between evidence and interpretation. An unexpected number first required checking. The eventual explanation also had to account for the material’s isotope pattern. A measurement from one specimen should therefore be read with its analytical method and sample identity, rather than treated as a single number describing the entire mine.
Why this happened in ancient Gabon
Uranium’s isotopic proportions change over geological time because its isotopes decay at different rates. The geological question was how suitable deposits formed while the relevant isotope was naturally more abundant. François Gauthier-Lafaye and Francis Weber’s 2003 study connected uranium concentration with changing atmospheric oxygen, circulating waters and organic material in the Franceville basin. Their model places the ore-forming processes within the evolution of the ancient environment.
The publisher’s abstract and introduction make clear that age alone was insufficient: the geological history also mattered. Their explanation linked mineral deposits to conditions in the basin, rather than treating every ancient uranium occurrence as a former reactor. This is why the sandstone, ore and basement relationships in the accompanying schematic are informative, although that illustration cannot supply the detailed history of any particular reaction zone.
Reconstructing pulses from trapped gas
A different question concerned the way activity varied. In 2004, Alex Meshik, Charles Hohenberg and Olga Pravdivtseva reported analyses of xenon and krypton in minerals from reaction zone 13. Their original paper’s abstract describes unusually concentrated fission-produced noble gases in a uranium-free phosphate associated with the ore. The isotopic structure of xenon supported a reconstruction of approximately half-hour active intervals separated by dormant periods of about two and a half hours.
These are inferred ancient timings, obtained from surviving material and a physical interpretation. They are not observations made while the reactions occurred. The authors associated the cycle with changes involving water and heat. Their result gives a particular mineral sample considerable historical value, but should not be enlarged into a claim that every zone followed an identical timetable. The author-hosted first page and indexed abstract were consulted for this account; access to the complete four-page paper was not established.
A model has to explain different reaction zones
Salah-Eddine Bentridi and colleagues’ 2011 investigation addressed a difficulty with some smaller zones. Earlier calculations had explained larger examples more readily. Their work used geological observations and computer simulations to investigate how the smaller systems could begin and evolve. The authors explicitly noted that their reconstruction was not an exhaustive, unique description.
An essential limitation was that researchers observed the rocks in their final, altered state. Their starting geometry had to be reconstructed. That separates a surviving observation from an inferred earlier arrangement. The paper also distinguished its static calculations from future work on explicit changes through time. Those qualifications leave meaningful research questions without making the existence of ancient natural fission doubtful. Different scales of explanation can be secure to different degrees: identifying a process is distinct from reconstructing every stage of its history.
What remained in the minerals
In 2018, Evan Groopman and colleagues used isotope imaging to locate fission-related caesium and barium signatures within tiny ruthenium-bearing aggregates from zone 13. Their PNAS paper estimated that the aggregates closed to the relevant isotope system about 4.98 ± 0.56 years after the end of criticality. This result concerns mineral formation and element capture after shutdown; it is separate from the shorter operating cycles investigated in 2004.
The detail of the paper matters. Most fission-related caesium had been lost from the reactor material, while particular aggregates retained evidence of capture. Some isotope relationships were affected by later caesium loss, which the authors associated with a much later intrusion. Their dating calculation used a relationship that avoided dependence on the compromised caesium measure. The study therefore documents both preservation and disturbance, rather than an untouched underground container.
A 2021 follow-up by Groopman and colleagues compared caesium and technetium behaviour using coordinated isotope imaging. Its university-hosted abstract reports that nearly all technetium was retained within the reactor, whereas nearly all fission-related caesium was lost from the fuel. The metallic aggregates preserved a limited but important caesium record. The authors inferred incorporation during activity and long subsequent sequestration, with uneven distributions between aggregates.
That comparison prevents a misleading generalisation. Different elements behaved differently, and a mineral phase that retained one signature did not demonstrate universal retention. Oklo offers evidence for studying long geological histories; applying it to a particular modern storage system requires its own assessment. This article uses the 2021 abstract for that comparison and does not claim a full-paper review.
An established discovery with continuing questions
An original JAEA workshop notice for March 2025 described a research collaboration begun in 2024 involving Nagoya University, JAEA, Strasbourg, Khemis Miliana and Washington University. Its stated aims included unresolved questions about reaction mechanisms. The notice documents a research programme and planned presentations; it does not itself establish the results of those presentations.
Oklo’s value is the connection between a carefully checked anomaly and an accumulating material record. Discovery, geological reconstruction, gas-isotope timing and mineral retention answer related questions using different evidence. Keeping their dates, methods and limits visible makes the case more remarkable, while allowing later research to revise particular reconstructions.
Sources and related cases
Source trail: Eklund, IAEA Bulletin (1975), three-page discovery account; Moxham (1976), USGS publication abstract; Gauthier-Lafaye and Weber (2003), publisher abstract and introduction; Meshik, Hohenberg and Pravdivtseva (2004), PRL abstract; Bentridi and colleagues (2011), original study; Groopman and colleagues (2018), full PNAS paper; Groopman and colleagues (2021), author-institution abstract; JAEA workshop notice (2025).
For other African cases involving investigated natural phenomena, see Lake Nyos and the Namib fairy circles. Return to Other Mysteries or all case files.
Map location: Approximate area: Franceville city reference, rounded from the GeoNames city coordinates. Oklo is near Franceville; the pin supplies broad regional context and does not identify the mine entrance, a reactor zone or an overseas laboratory. Coordinate reference.