On 26 September 1991, eight people were sealed inside Biosphere 2, a three-acre enclosure of glass and steel near Oracle, Arizona. They stayed for two years. And the oxygen in their sealed atmosphere did fall, slowly and persistently, until it reached a level low enough to be a genuine medical problem and outside oxygen had to be pumped in.
The word “disappearing” in the headline turns out to be the interesting part. The oxygen did not leak away or vanish. It went somewhere specific, and working out where became one of the central puzzles of the mission.
What Biosphere 2 was trying to be
Biosphere 2 was a privately funded attempt to build a working miniature of Earth’s biosphere, sealed off from the planet, which the project called Biosphere 1. Inside its glass were five wilderness biomes, a rainforest, an ocean with a coral reef, a mangrove wetland, a savannah, and a desert, along with an agricultural area where the crew grew their food and a human habitat where they lived.
The stated goals were a mix of ecology and a longer-range ambition. One was to study how a closed system of this kind behaves as an experimental science rather than an observational one. Another, more speculative, was to learn something useful about the life-support systems that long-duration space habitats would eventually need. The crew of four men and four women, known as biospherians, were to live for two years on what the enclosure could produce and recycle, with as little crossing the glass as possible.
It is worth being plain about the project’s standing. Biosphere 2 was widely criticised at the time, and its scientific value has been contested since. It originated with a group around the figure John Allen rather than within a conventional research institution, and the venture drew accusations of being closer to spectacle than to controlled science. As NPR reported in 2025, the crew’s personal conflicts became the international story, displacing the science. The oxygen episode is the part with the clearest and best-documented physical explanation, which is why it is worth examining on its own.
Where the oxygen went
The atmosphere inside started at the normal 21 percent oxygen. Over the following months it declined steadily. By the time outside oxygen was added, it had fallen to around 14 percent, which is roughly the oxygen level a person would experience at an altitude of some 4,000 metres. The crew, working physically demanding days, were experiencing fatigue and breathlessness.
The puzzle was that the oxygen was not simply turning into the expected amount of carbon dioxide. If it had, the carbon dioxide level would have climbed to match, and it had not. Oxygen was leaving the air faster than the obvious accounting could explain.
The explanation, identified during the mission and confirmed afterward, had two linked parts. The soil in the agricultural areas had been made deliberately rich in organic matter, and it supported a large population of microbes. Those microbes consumed oxygen and produced carbon dioxide at a high rate, faster than the enclosure’s plants could reverse through photosynthesis. That accounted for the oxygen loss. It did not account for the missing carbon dioxide.
The carbon dioxide was being absorbed by the concrete. The structure contained a large amount of exposed, uncured concrete, and concrete takes up carbon dioxide from the air as it cures, locking the carbon into the material as calcium carbonate. Mark Nelson, one of the eight crew members, has described oxygen slowly disappearing for sixteen months with no one knowing where it had gone. So the two halves fitted together: microbes were turning oxygen into carbon dioxide, and the concrete was quietly taking that carbon dioxide out of circulation. The oxygen had not vanished. It had been moved, by way of the air, into the walls.
What the episode actually demonstrated
The oxygen problem is often filed as evidence that Biosphere 2 failed. That is true in one sense and misleading in another, and the distinction is worth keeping.
As a demonstration of a self-sustaining closed habitat, the first mission did fail on its own terms. The enclosure could not hold a breathable atmosphere for two years without intervention, and oxygen was pumped in. A system that needs resupply to keep its occupants functional is not closed. For the space-colony ambition the project sometimes invoked, that is a real and not a cosmetic shortfall.
As an experiment in how closed systems behave, though, the episode produced something genuine. The interaction between rich soil, microbial respiration, and curing concrete was not anticipated at the design stage, and it was diagnosed by tracking the chemistry inside a real sealed structure. That is the kind of result that is difficult to obtain any other way. The lesson was less “humans cannot build closed ecosystems” than “a closed system contains every surface in it, including the ones nobody thought of as part of the ecology.” Concrete is not usually treated as a component of an atmosphere. Inside a sealed three-acre building, it was.
What became of it
The first crew completed their two years and left on 26 September 1993. A second mission began in 1994 but ended early, after about six months, amid management disputes and a change of control. The site passed through different hands and is now operated by the University of Arizona as an earth-science research facility, used for controlled environmental experiments rather than sealed human habitation.
The open question Biosphere 2 left behind is not whether eight people can be kept alive behind glass. It is how completely the behaviour of a closed life-support system can be predicted before it is sealed, and the oxygen decline stands as a fairly precise measure of how large the gap between prediction and outcome can be.
