On the evening of 21 August 1986, a release of gas from Lake Nyos in northwestern Cameroon killed at least 1,700 people. The lake’s surface subsequently looked calm, while communities in the valleys below faced devastating losses. The investigation had to explain an unusual combination: rapid unconsciousness, deaths extending beyond the shore, reported sounds and odours, disturbed lake water, and little evidence of an ordinary hot volcanic eruption. The answer emerged through chemistry, lake measurements and medical investigation, while the precise initiating disturbance remained uncertain.
The night and its aftermath
The United States scientific team’s 1987 final report reconstructs events around 9.30 pm. Survivors described rumbling sounds, warmth and sudden collapse; some reported an odour resembling rotten eggs or gunpowder. An observer above the lake described a rising pale cloud and a water surge. These are accounts collected afterward, rather than instrument readings made during the release. The report says survivors in the valley did not see a visible cloud.
People returning to consciousness found relatives and livestock dead. The pattern varied sharply with position: losses were concentrated in low ground, while animals at higher elevations could remain alive. Investigators found a reddish-brown lake, floating vegetation and evidence of water surges, but household objects had not generally been thrown into disorder. Their task was to account for that uneven distribution and for both the lake disturbance and the absence of widespread blast damage. The report’s reconstruction should not be expanded into a claim that every witness experienced the same symptoms.
A source beneath the lake, but no demonstrated hot eruption
George Kling and colleagues’ April 1987 paper in Science distinguished the origin of the gas from the mechanism that released it. Carbon and helium evidence pointed to a deep magmatic source. That did not require magma to have erupted into the lake that night. Gas could reach groundwater and accumulate in the lower water layers over time. In the team’s September samples, carbon dioxide made up roughly 98–99 per cent of the dissolved gas.
The physical evidence challenged a substantial hot injection. Measured temperatures were relatively low; the chemistry lacked the enrichment in reactive volcanic constituents that the authors expected from such an event. Depth soundings did not reveal the disrupted lake floor expected from a large new vent. Their medical findings supported carbon dioxide asphyxiation. They interpreted many survivors’ lesions as consequences of prolonged immobility, with some injuries associated with falls or nearby heat sources. These findings helped explain why apparently dramatic “burn” accounts did not establish a hot volcanic cloud. The paper nevertheless described the trigger for the gas release as unknown.
What the Japanese investigation measured
A separate 1987 study by Minoru Kusakabe and Takashi Ohsumi examined the lake about fifty days after the disaster. Its published abstract reports that the water remained stratified in temperature, chemistry and density, apart from a shallow surface layer. Carbon dioxide and bicarbonate dominated the dissolved constituents, with substantial dissolved iron in the deeper water. Carbon and helium isotopes again indicated a mantle contribution.
The authors regarded low chloride and sulphate concentrations as evidence against a sudden injection of volcanic gas. They proposed that gas-rich deep water had moved upward and released bubbles as pressure fell. Their suggested initiating process involved increased input of warmer, carbon-dioxide-rich groundwater during the rainy season. That proposed trigger should retain its conditional status. The accessible abstract sets out a hypothesis supported by their observations; it does not provide a recording of the first movement on 21 August. This account uses that abstract, rather than claiming an independent reanalysis of the complete dataset.
The alternative proposed by the Italian mission
The early scientific response was not unanimous. The Italian technical mission’s 1989 paper reports that its investigators arrived seven days after the disaster. Its abstract attributes the outburst to a phreatic explosion beneath the lake bed: an interpretation involving explosive processes below the water, rather than simply release from a previously accumulated dissolved reservoir. The authors stated that this fitted their field observations and the geological setting.
That is a concrete disagreement about mechanism. A volcanic landscape and gas with a deep geological origin did not, by themselves, settle whether a fresh explosive event had occurred. The distinction depended on evidence such as temperature, water chemistry and the condition of the lake floor. Only the Italian paper’s published abstract has been inspected here, so its complete measurements and reasoning are not presented as independently audited. Its inclusion records an actual contemporary scientific alternative, rather than inventing a debate between science and an unspecified supernatural explanation.
Following changes instead of relying on one visit
The next stage was repeated measurement. In their 1989 study of thermal structure and water chemistry, Kling, Michele Tuttle and William Evans compared observations from January, March and May 1987 with September 1986. Their abstract describes changes concentrated in the bottom waters, including increased dissolved constituents and a small temperature rise at depth. They considered geothermal heat and the addition of spring water as explanations. A sequence of measurements could test continuing input and rebuilding stratification in a way that a single post-disaster sample could not. This was evidence about the lake’s subsequent behaviour, not direct observation of its condition immediately before the release.
Dangerous water in the region’s oral traditions
Anthropologist Eugenia Shanklin approached a different archive. Her 1989 study of Grassfields legends and myth asked whether destructive-lake traditions might preserve memories of earlier events. The abstract compares a Kom migration story involving a lake that destroys the Bamessi with an Oku story in which a lake leaves its bed and destroys Kijem. It also considers local classifications of water as beneficial or dangerous and narratives associated with spiritual agency.
Shanklin described the comparison as suggestive rather than conclusive. That qualification is essential. An oral tradition can be evidence of how communities understand water, landscape and past loss without supplying the date, gas composition or mechanism of a particular prehistoric disaster. Equally, the existence of a physical explanation for Nyos does not make the traditions irrelevant to the region’s history. Here the paper’s abstract supplies the comparison; the complete ethnographic interviews and full article have not been independently examined.
2001–05: testing controlled degassing
Understanding the stored gas made intervention possible. A 2005 paper by Kling, Evans, Greg Tanyileke and colleagues examined twelve years of lake measurements and the effects of controlled degassing, begun at Nyos in 2001. The system brought gas-rich deep water up a pipe. Once started, expanding bubbles helped drive the flow, allowing the gas to separate at the surface. The engineering question was whether this could remove gas while preserving the lake’s stability.
The researchers measured temperature, conductivity, dissolved gases and gas pressure, and compared water-column structure before and after piping. They found reduced gas content without the feared general destabilisation. Their modelling also showed why one pipe’s removal rate would decline as the gas pressure driving it fell; they argued for additional removal capacity. The distinction between measured change and modelled future performance matters: the paper’s projections were forecasts made in 2005, not observations of later years. Its value here is the documented transition from investigating a disaster to testing a specific mitigation method. It is not a statement of present-day operating conditions.
What Nyos explains, and what it does not
Lake Nyos is a case in which an initially baffling disaster acquired a strong physical explanation. Gas origin, long-term accumulation, release mechanism, the precise trigger and later risk management were separate questions. Evidence could resolve much of that sequence while leaving uncertainty about its first disturbance. The sources linked here document that process without turning an unconfirmed trigger into fact or treating early scientific disagreement as a permanent absence of explanation. Sources checked on 5 November 2026. For another African landscape investigated through competing physical hypotheses and repeated fieldwork, see the Namib fairy-circle research history.
Map location: Approximate area: Lake Nyos reference, rounded from the Smithsonian Global Volcanism Program’s separately listed Nyos lake coordinates (6°26′ N, 10°18′ E). This locates the lake area, not the full gas-affected valleys, a victim’s home or a current safe-access route. Coordinate reference.