The random film of the day was Cocoon (1985), Ron Howard’s comedy about a Florida retirement home: old people swim in a pool with an alien cocoon lying on the bottom, and grow younger. I checked the archive: this film had never once been used as a hook. But what caught me wasn’t the rejuvenation—it was the premise on which the whole plot machinery rests: beneath an ordinary-looking pool, alien technology is hidden away, and nobody knows.
That exact thing was built in real life. Inside the Greenland ice sheet, 240 kilometers from the American Thule base, Camp Century operated—officially an Arctic research site, in fact cover for Project Iceworm: a network of under-ice launch silos with nuclear missiles. What got me was a question phrased with engineering-grade dullness: can you hide a large military facility inside a mass of ice? In the end, the answer came not from the enemy, but from the physics of ice itself.
The station was established in 1959, at an elevation of roughly 2,000 meters. No domes, no hangars: tunnels were cut directly into compacted firn using the “cut and cover” method. A snow cutter carved a trench sloping inward—eight feet across at the top, eighteen at the floor—metal forms were set over it, so-called Peter snow was blown onto them, and once the arch had hardened, the forms were removed. The glacier itself was the building material.
In all, they cut twenty-one trenches with a combined length of around 3 kilometers. Inside fit a hospital, store, theater, church, barbershop, library with a laundry—a city for 200 people. Water was not hauled in: they installed a steam generator and melted a shaft up to 100 feet deep into the ice mass, until a dense layer stopped letting water through—then stored reserves in the resulting subglacial pool. Supplies came in on tractor-sled trains from Camp Tuto: in 1959, roughly 1,000 short tons of cargo were hauled there; in 1960, 4,500.
The base was not the secret. The base was shown off, in fact: National Geographic and The New York Times wrote about it; in 1960, The Saturday Evening Post ran a feature; Walter Cronkite flew in to film a CBS segment. Two Boy Scouts—one from Kansas, the other from Denmark—won a competition for a trip to the city beneath the ice. What was secret was the reason all of it had been built.
The city’s power came from the PM-2A—the world’s first “portable” nuclear reactor, the brainchild of the American Locomotive Company. Two megawatts of electrical output plus heating, uranium-235 enriched to 93 percent, first criticality on October 3, 1960. It was brought to Greenland dismantled into components, assembled on site, operated for three years, dismantled, and shipped back to the United States. Back—that is an important detail.
The point of the exercise was economics. The reactor belonged to a U.S. Army program looking for uses for small nuclear installations at remote facilities; its own final verdict sounds like a sentence: “a solution in search of a problem.” Camp Century’s budget put the cost of the camp and its supporting infrastructure at $7.92 million, of which the reactor accounted for 5.7 million—nearly three-quarters of the total. Why they put up with that becomes clear from the arithmetic: a few pounds of enriched uranium replaced thousands of barrels of diesel that otherwise had to be hauled over the ice in tractor trains through the polar night.
Then things aligned in a way that never happens beautifully. On January 3, 1961, the experimental SL-1 reactor in Idaho exploded—part of the same Army project, only not in the Arctic: a control rod was pulled too far, and in four milliseconds the installation went prompt critical; the 12-ton vessel leapt nearly 3 meters into the air. Three operators died. It remains the only fatal accident involving a power reactor in U.S. history. Afterward, Atomic Energy Commission general manager Alvin Luedecke personally barred the PM-2A from operating until an interlock was installed on its central rod. The reason was simple and unpleasant: the reactor in Greenland was one of just two in the country that could be started by a single rod.
While the station churned through everyday life and science, Army planners were calculating the main thing. A 1960 report, “The Strategic Value of the Greenland Icecap,” described a tunnel network 4,000 kilometers long—covering an area the size of three Denmarks, around 130,000 square kilometers. The floors of the launch complexes were to be set 8.5 meters below the surface, silo clusters spaced 6.5 kilometers apart, new tunnels cut every year, so that after five years there would be thousands of positions and missiles could be moved between them. The logic was a shell game with the Soviet General Staff: more targets than warheads, and from there they could not tell which silos were loaded today.
The missiles intended for it were their own, Army missiles: a shortened two-stage Minuteman for which they even invented a separate name—Iceman. Retellings mention 600 missiles and more than 2,000 silos. All of it existed only on paper for one reason: Denmark’s consent. Greenland is Danish territory; nuclear weapons could not be deployed there without Copenhagen’s permission, and that permission was never obtained. It was never even requested.
The camp, meanwhile, was built openly, because openness was exactly what was needed. Thule had been a military base since 1951, and the appearance of something similar nearby would have been hard to explain; a scientific station deep in the ice sheet raised no questions from journalists, Danes, or Boy Scouts with notebooks.
The main mistake was not political. From the outside, a glacier looks like concrete, but it behaves like a very viscous liquid: it spreads from the center toward the edges, and the greater the pressure above a tunnel, the faster its walls move together. The engineers knew this—in a Cold Regions Research and Engineering Laboratory report, there is even an optimal depth, from 60 to 120 feet, where the rate of closure is lowest. Then reality began.
The average rate of vertical tunnel closure at Camp Century was 29.66 inches a year. Measurements showed that the ice’s creep rate doubled as temperatures rose from minus 17 to minus 10 degrees Celsius, and at minus 2 it was 100 times higher than at minus 40—which meant heat accelerated deformation more than any other factor, while the city was heated continuously. By the middle of 1962, the ceiling of the reactor hall had sagged enough that it was raised by five feet. The metal arches above the reactor trenches were each raised twice by roughly six feet over PM-2A’s three years of operation. Some arches had to be left in the snow forever: when the forms were removed too early, the unsupported snow vaults flowed apart literally before their eyes.
There is a particular irony in the rejected alternatives. The engineers considered ordinary above-ground structures on pilings that could periodically be raised with jacks—and rejected them for two reasons: they were more expensive and, above all, “visible for many miles against the white icecap.” Secrecy and construction logic pulled in one direction; the nature of ice pulled in the other. The tunnels were designed for a ten-year service life; the camp operated for five years and was abandoned after eight.
The reactor was removed; everything else was left behind. It was left on the assumption that perpetual snowfall would seal the station forever, and until the 1960s there was nothing strange about that assumption. Estimates of what, precisely, lies there emerged only in 2016, when William Colgan’s group described the burial ground: 200,000 liters of diesel fuel, around 24 million liters of wastewater containing domestic sewage, an undetermined volume of low-level radioactive coolant, and, presumably, polychlorinated biphenyls. The contamination footprint is 55 hectares, roughly 100 football fields. There was enough diesel to drive around the Earth 80 times.
This is where things get most interesting for a reader with a calculator. The same 2016 study gave a threshold: if warming proceeded under the standard scenario, around 2090 the area above the base would shift from net snow accumulation to net melting, and after that the emergence of the waste would become irreversible. Five years later, in a 2021 paper, Colgan himself revised the conclusion using data from an observatory installed over Camp Century: meltwater was not reaching the base and did not penetrate more than a little over a meter deep; there was no risk of the burial ground being washed out before 2100. The same researcher, the same site, two different answers—the difference being that the second time, models had been replaced by sensors actually measuring snow and ice. Unfortunately, we will have to wait about 70 years to find out which forecast is more accurate.
Sources also disagree on details that do not look like details. Publications following the study give 24 million liters of wastewater, while an American Geophysical Union press release gives 240,000—100 times less. A zero got lost somewhere, and both figures are still sitting in databases. The closure timeline is just as fluid: the reactor is dismantled in 1964 in some accounts, in the summer of that same year in others; the camp is “evacuated” in 1965, “closed” in 1966, while the station sign gives its operational span as 1959 to 1967. It appears different agencies counted different events as the end.
The political side remains entirely unresolved. The base is American, the land is Danish, and Greenland has governed itself since 1979. International law, as environmental-law specialist Jessica Green puts it, clearly answers the question “who is liable for future waste” and is silent on the question “who is responsible for waste already buried.”
The secret stopped being a secret not because of spies. In January 1995, the Danish Institute of International Affairs began an investigation into nuclear weapons in Greenland—the trigger was declassified information about the 1968 crash of a B-52 bomber near Thule, which contradicted the government’s official position. The Iceworm materials were published in 1997. Denmark, incidentally, learned of the plans for under-ice silos from those same documents and reacted rather sourly.
Then, in April 2024, the city was found again—physically this time. NASA scientist Chad Greene was flying in a Gulfstream III 150 miles east of Pituffik Space Base, calibrating the UAVSAR side-looking radar for a task mapping the glacier’s internal layers. The radar recorded something other than what it had been asked for: at a depth of around 30 meters, a grid of parallel structures emerged, matching historical plans of Camp Century. Earlier surveys had also seen the base, but that had been a two-dimensional strip beneath the ice; here, an actual map took shape. The agency’s official wording on the discovery is brutally honest: the image’s scientific value has not yet been determined; it is an intriguing byproduct of the calibration.
This is where it is worth returning to how the station began. The first deep ice core used to build climate models was drilled at Camp Century; soil analysis from those samples showed that the site was free of ice roughly 400,000 years ago. The military facility hidden beneath the ice sheet turned out to be a climate archive as well. Since 2017, it has been under scientific monitoring by the Geological Survey of Denmark and Greenland—the very sensors that ultimately rewrote the forecast.
It is possible to hide a large military facility inside a mass of ice—the question was correctly posed, the answer came back affirmative, and for several years no one could reach the city beneath the snow. But hiding it and sealing it away turned out to be different tasks. The ice covered the station, then slowly squeezed it out, then closed over it, and now it is the ice itself that decides when the contents return to the surface.
It is not yet possible to calculate this precisely. Some models point to irreversible melting by 2090; measurements from recent years say nothing will break through before 2100. There is no contradiction in the physics between those answers—there is a difference between a calculation and an instrument, and only time can resolve it. But the one question that does not depend on time—who removes the trash—still has no answer: the law knows how to prohibit future burials and does not know how to assign responsibility for past ones.