Somebody put four foam cushions on a heating vent.

That is where it starts. Not with a missile, not with a Soviet fighter, not with a catastrophic structural failure. Three cloth-covered cushions stowed on a hot-air outlet under a seat before take-off, and a fourth pushed under the seat once airborne, on a B-52 carrying four hydrogen bombs over the Arctic.

Six hours later that aircraft hit the sea ice of North Star Bay at a shallow angle, seven and a half miles from Thule Air Base in Greenland, and the conventional explosives in all four weapons went off.

Why there was a bomber over Greenland at all

Operation Chrome Dome started in 1960: nuclear-armed B-52s permanently in the air near Soviet borders, at its height twelve aircraft aloft at any moment. From 1961 a further, quieter task was added — the Hard Head mission, in which a bomber flew a figure-eight at 35,000 feet over Thule keeping the Ballistic Missile Early Warning System in visual contact.

The logic was grim and sound. If the communications link between NORAD and Thule went dead, somebody needed to be able to say whether the radar had been destroyed by a Soviet first strike or whether a cable had simply failed. A pair of human eyes at 35,000 feet was the answer.

Robert McNamara tried to cut the programme in 1966 — BMEWS was working, missiles had made bombers redundant, and it would have saved $123 million a year. SAC and the Joint Chiefs resisted, and a compromise left four bombers on alert daily. SAC quietly kept one of them dedicated to watching Thule, without telling the civilian authorities, whom it had decided did not have the need to know.

Six hours, then five minutes

HOBO 28 launched from Plattsburgh with seven men aboard — five regular crew plus a substitute navigator and a mandatory third pilot, Major Alfred D’Amario. It was D’Amario who put the cushions on the vent.

The flight was routine until an autopilot fault forced a manual air-to-air refuelling from a KC-135. An hour later the aircraft commander, Captain John Haug, sent co-pilot Leonard Svitenko down for his rest period and D’Amario took his seat — and, crucially, his ejection seat.

The crew were cold. D’Amario opened an engine bleed valve for extra hot air. A heater malfunction meant the air arrived in the ducts barely cooled. Over about thirty minutes the cabin became uncomfortably hot, and the cushions caught fire.

Somebody smelled burning rubber. The navigator searched the lower compartment twice before finding the fire behind a metal box. Two extinguishers made no difference.

At 15:22, ninety miles south of Thule, Haug declared an emergency. Within five minutes the extinguishers were empty, the electrical power was gone and the cockpit was so full of smoke the pilots could not read the instruments. D’Amario looked down, saw the lights of the base directly below, and the crew went out.

Six ejected. Svitenko, with no seat, tried to bail out through a lower hatch and was killed by head injuries in the attempt.

The empty aircraft flew on north, turned through 180 degrees, and struck the ice at 15:39.

Contemporary U.S. Air Force film of the Crested Ice clean-up operation on the ice at Thule.

What was left

An aerial survey found six engines, a tyre and small debris on blackened ice. Wreckage was spread over an area a mile by three miles, with parts of the bomb bay two miles north of the impact point, indicating the aircraft had broken up before it hit. The ice was punched open, briefly exposing about fifty metres of open seawater. South of the impact, a patch 120 by 670 metres was black where the JP-4 had burned.

The four B28FI weapons did not produce a nuclear explosion — they cannot, without the precise implosion sequence — but their conventional high explosive detonated, scattering plutonium, uranium, americium and tritium across the ice. Plutonium was later measured at up to 380 milligrams per square metre.

The search for the survivors, meanwhile, was largely organised by a civilian: Jens Zinglersen, the Thule representative of the Royal Greenland Trade Department, who raised and led dog-sled teams across the ice. He was given the Air Force Exceptional Civilian Service Medal a month later.

Dr. Freezelove

The clean-up was called Project Crested Ice. Everybody on it called it Dr. Freezelove.

It was commanded by USAF general Richard Overton Hunziker, and the camp built on the ice beside the wreckage was named after him. Arctic night lasted until 14 February; operators carried battery packs inside their coats to stop them freezing. A “zero line” marking the contaminated area was established within four days, and two ice roads were cut back to the base, the second because the first had fatigued the ice.

The Danish government made two demands that shaped everything: the blackened area had to be physically removed, and no nuclear material was to remain in Greenland. So graders scraped up contaminated snow and ice, it was boxed, hauled to a “Tank Farm” near the base, and sealed into steel tanks for shipping — weapons debris to the Pantex plant in Texas, the tanks to Savannah River in South Carolina.

Hunziker reported that 93 percent of the contaminated material had been removed. More than 550,000 US gallons of contaminated liquid and thirty tanks of miscellaneous material went south. The last tank was loaded onto a ship on 13 September 1968.

The bomb that may or may not be down there

This is the part that has never quite closed. Recovery on the ice produced four bomb reservoirs, one nearly intact secondary stage, and parts amounting to two more secondaries. Parts equalling one secondary were never accounted for.

In August 1968 a Star III mini-submarine was sent to look for what remained on the seabed, specifically fissile material. It suffered technical problems and the whole debris field was never searched. A contemporary document instructed that the true purpose be handled carefully: for discussion with the Danes, it said, the operation should be referred to as a survey of the bottom under the impact point.

In 2008 the BBC published documents obtained under the U.S. Freedom of Information Act suggesting investigators had realised within weeks that only three weapons could be fully accounted for. In 2009 the Danish Institute for International Studies compared those documents with an earlier Department of Energy release and concluded the opposite: there was no missing bomb, and the underwater search had been for the uranium-235 core of a secondary. That remains the official Danish position.

The Thule accident and its long political afterlife, told in full.

Thulegate

Denmark had been a declared nuclear-free country since 1957. A nuclear-armed American bomber crashing onto Danish territory was therefore not only an accident but a constitutional problem, and at the time both governments said the aircraft had diverted to Thule in a one-off emergency.

Documents declassified in the 1990s showed that was untrue: nuclear-armed flights over Greenland were routine. The 1995 scandal was named Thulegate. A Danish Institute of International Affairs report published in January 1997 concluded that the United States had acted in good faith — and laid the blame on Prime Minister H. C. Hansen, who in a 1957 letter noted that the question of “supplies of munition of a special kind” had not been raised and that he had nothing further to add. The report read that as a deliberate green light.

The same report confirmed the United States had stored nuclear weapons in Greenland until 1965, and revealed Project Iceworm, a plan to put as many as 600 nuclear missiles under the ice cap.

Then there are the Danish clean-up workers. A 1995 survey found 410 cancer deaths in a sample of 1,500. A Danish study found cancer incidence roughly 40 percent higher among Crested Ice workers than among comparable Thule workers, and about 50 percent higher than the general population, without being able to attribute it to radiation. Nearly 200 former workers sued the United States in 1987; they lost, but forced the release of hundreds of classified documents, which revealed that American personnel — probably more exposed than the Danes — had never been medically monitored at all. In 1995 the Danish government paid 1,700 workers 50,000 kroner each.

What changed

Chrome Dome ended immediately. Coming two years after Palomares, where another B-52 had scattered hydrogen bombs over Spain, airborne alert had become politically and operationally impossible to justify.

Weapons design changed too. Investigators concluded the high explosive in the bombs was not chemically stable enough to survive an aircraft accident, and that the safety circuits became unreliable in a fire. Lawrence Livermore developed the “Susan Test” to simulate crash forces on explosives; by 1979 Los Alamos had produced insensitive high explosive. One physicist later suggested the Palomares and Thule weapons probably would not have detonated at all had IHE existed.

And there is one more thing worth sitting with. The political scientist Scott Sagan pointed out that HOBO 28 came down in Baffin Bay. Had it come down on the BMEWS array instead, NORAD would have seen the radar go silent and the monitoring aircraft go silent at the same instant, with no nuclear detonation detected — which is precisely what a surprise conventional attack on Thule would have looked like.

A satellite link that would have resolved the ambiguity was not installed until 1974.

A short account of the Broken Arrow at Thule and what was recovered from the ice.

Sources: Wikipedia (1968 Thule Air Base B-52 crash), Danish Institute of International Affairs (DUPI) 1997 report, Danish Institute for International Studies 2009 report, BBC, Lawrence Livermore National Laboratory, U.S. Air Force / Wikimedia Commons.