The oxygen in Biosphere 2 fell from 20.9 percent at closure to 14.4 on day 475, at an average of about 140 parts per million a day, faster in the dark months. The crew had decided early not to add any until they understood the loss, and held to that for sixteen months, through headaches and altitude sickness, until the doctor could not add a column of figures in January 1993 and liquid oxygen was trucked up from Tucson. That is the symptom, and the main entry tells it. This piece is about the diagnosis, which is the one unambiguous scientific result the building produced.

What should have happened

Life runs on one equation, read in two directions. Photosynthesis takes carbon dioxide and water and makes sugar and oxygen; respiration takes sugar and oxygen and gives back carbon dioxide and water. Mole for mole, every molecule of oxygen that respiration consumes becomes one molecule of carbon dioxide, and the total number of gas molecules does not change. If something inside the building was breathing faster than the plants could compensate, which was everyone’s first guess, then two things had to be true. The carbon dioxide had to rise by as much as the oxygen fell, and the volume of the atmosphere had to stay the same.

Neither was true. The oxygen had fallen by six and a half percentage points, which is 65,000 parts per million. The carbon dioxide, which had started near the outside’s 350, swung between about 1,000 in summer and 5,000 in the second winter, never anywhere near the figure the equation demanded. And the atmosphere was shrinking: the membranes of the two lungs were slowly, steadily falling, which meant there was less gas in the building than there had been. Something was taking gas out of a system that, by the best measurement anyone had made, exchanged less than a tenth of its air with the outside in a year.

Two small sinks were already known. The scrubber, which blew the air through a falling rain of sodium hydroxide and precipitated the carbon as solid calcium carbonate, had run through both winters and captured about 98 kilomoles of carbon dioxide, some 4.3 tonnes. And the plants were growing, locking carbon into wood; the rainforest’s biomass had increased by half before closure and kept climbing. Neither came close to closing the gap.

The counters

Wallace Broecker of Columbia University’s Lamont-Doherty Earth Observatory was by then the most eminent geochemist of the carbon cycle alive, the man who had put the words “global warming” into print in 1975 and spent a career tracing where the planet’s carbon goes. He took the problem on with his graduate student Jeff Severinghaus, who would later build a career reading ancient air out of polar ice. With them were Dempster, who knew the building; Taber MacCallum, who ran the analytical laboratory inside and could take the samples; and Martin Wahlen, who could measure them.

What they did was an audit, and it was possible only because the building was closed. A geochemist on the open Earth cannot account for every atom; a geochemist inside a sealed box can, in principle, because the atoms you began with are all still in there. They measured carbon in every compartment they could reach: the air, the plant tissue, the carbonate the scrubber had made, and the concrete of the structure itself. And they measured not just how much carbon was in each but which carbon. Plants prefer the lighter isotope, carbon-12, and reject a little of the heavier carbon-13, so carbon that has passed through a leaf carries a signature, a slight depletion in carbon-13 that persists through whatever becomes of it. Carbon dioxide breathed out of compost by a microbe carries the plant’s mark. The carbon locked in the original cement, made from limestone laid down in an ancient sea, does not.

The source

The soil was the first finding. Biosphere 2 held 30,000 tonnes of it, and the farm’s beds had been made rich on purpose, with organic matter running to something like 30 percent by Donella Meadows’s account, so that the first crops would not want for anything. Soil that rich is alive, and what lives in it breathes. The microbes were oxidising that organic matter, drawing down oxygen and releasing carbon dioxide, on a scale that nobody had budgeted and that the young plants, in two cloudy winters behind glass that took half the light, could not reverse.

The numbers in a closed system are unforgiving in a way the open Earth’s are not. On the planet the carbon in living plants roughly equals the carbon in the air; in Biosphere 2 the ratio was a hundred to one. On the planet the carbon in soils is about twice the carbon in the air; in Biosphere 2 it was a thousand times more. A concentration of 1,500 parts per million in the building’s air amounted to about 70 kilograms of carbon. Against 30,000 tonnes of breathing soil, the atmosphere was a rounding error.

So the oxygen had gone into carbon dioxide after all. The question was where the carbon dioxide had gone.

The sink

It had gone into the walls. The building’s basement and footings were concrete, some 15,800 square metres of it exposed to the interior air, and the concrete was young. Fresh concrete contains calcium hydroxide, and over years it carbonates: carbon dioxide from the air reacts with it to form calcium carbonate, the same chalk the scrubber had been making on purpose. In the open air that process is slow and harmless. In a sealed building with carbon dioxide at ten times the outdoor concentration, over acres of exposed surface, it was a pump.

The isotopes let them tell the new carbonate from the old: carbon that has been through a plant and a microbe carries a different ratio of carbon-13 from the carbon in the original cement, and the measurements pointed to the concrete as the sink. It was the crew’s missing atmosphere, breathed out by the soil and taken up by the foundations. Severinghaus and Broecker put the quantity at about 600 kilomoles of carbon dioxide, around 26 tonnes, six times what the scrubber had caught. Each of those molecules had carried away an atom of oxygen that respiration had bound to it. A back-of-the-envelope check closes the loop: six and a half percent of 180,000 cubic metres of air is about 500 kilomoles of oxygen gone, and the concrete held about 600 kilomoles of carbon dioxide. The difference is roughly what the growing plants and the scrubber account for. Within the error of a building, the books balanced.

The oxygen had not leaked. It was in the basement, as stone.

What it took

The finding was published in Eos, the newspaper of the American Geophysical Union, in January 1994, under five names. Dempster, who had built the seal that made it possible, wrote the sentence that explains why the method worked at all: “It can hardly be overemphasized how powerful is the tool of material closure. It tells you that the atoms you begin with are in there somewhere.”

Fourteen years later he went back to the oxygen curve and asked what it would have looked like in a leakier building. At the leak rates of the best American growth chambers of the day, a few percent a day, the simulation shows the oxygen holding near 20 percent throughout, the loss washed out by outside air as fast as it occurred. At one percent a day it is still invisible. The crew would have breathed easily for two years and nobody would have known that the soil was eating the air and the walls were eating the soil’s breath. The tight seal did not cause the crisis. It revealed one that a looser building would have hidden until the day it mattered.

Two corrections followed. The concrete was sealed with a coating so that it could no longer take up carbon dioxide. And the farm soil was never changed, because there has never been another closed mission to change it for. The audit stands as the project’s one unambiguous scientific result: a closed-system carbon budget, reconciled to the isotope, in a building the size of three football fields. It is also the clearest demonstration anyone has produced of the rule that every sealed system, from a submarine to a spacecraft to a planet, eventually teaches: the flow you did not measure is the one that is running the show.

Sources

• Jeffrey P. Severinghaus, Wallace S. Broecker, William F. Dempster, Taber MacCallum and Martin Wahlen, “Oxygen loss in Biosphere 2,” Eos, Transactions of the American Geophysical Union 75 (3), 33–37, January 1994. doi:10.1029/94EO00285
• William F. Dempster, “Biosphere 2 engineering design,” Ecological Engineering 13 (1999), 31–42. The oxygen and carbon dioxide charts for the two years, the 600 kilomoles in 15,800 square metres of concrete, the scrubber’s 98 kilomoles, and the sentence about closure. ecotechnics.edu
• William F. Dempster, “Tightly closed ecological systems reveal atmospheric subtleties: experience from Biosphere 2,” Advances in Space Research 42 (2008), 1951–1956. The 140 ppm a day, 14.4 percent on day 475, and the simulated oxygen curves at six other leak rates. ecotechnics.edu
• Mark Nelson and William F. Dempster, “Living in space: results from Biosphere 2’s initial closure,” in Strategies for Mars, 1996. The carbon ratios of plants, soil and air inside against the Earth’s, the 70 kilograms of atmospheric carbon, the rainforest’s growth, and the carbon dioxide record by month.
• Mark Nelson, “Biosphere 2: What Really Happened?”, Dartmouth Alumni Magazine, May–June 2018. The crew’s decision to study the decline before adding oxygen, the symptoms, the doctor’s arithmetic, and the trucks. dartmouthalumnimagazine.com
• Donella Meadows, “Biosphere as theater,” October 28, 1993, for the 30 percent organic soil and the trucks from Tucson. donellameadows.org
• Wallace S. Broecker, “Climatic Change: Are We on the Brink of a Pronounced Global Warming?”, Science 189, 460–463, August 8, 1975.