Part of a nuclear weapon is still buried in a North Carolina field.

September 20, 1950.

In 1960 the Strategic Air Command kept part of its bomber fleet in the air at all times.

The reasoning behind it was simple and came from the top of the Air Force: a surprise attack could destroy planes sitting on a runway, but it could not touch a plane that was already flying. The program was called Operation Chrome Dome. B-52 bombers loaded with live hydrogen bombs flew fixed routes over the United States and out over the Atlantic, in shifts, around the clock, so that at any given hour some number of armed aircraft was already off the ground. Seymour Johnson Air Force Base, just outside Goldsboro, North Carolina, was one of the bases that flew the mission.

On the evening of January 23, 1961, a B-52 lifted off from Seymour Johnson carrying two Mark 39 hydrogen bombs, weapons far more powerful than anything used in the Second World War. The flight plan was a long loop off the East Coast with a scheduled air refueling partway through, then a return to base. Eight men were aboard, more than the standard crew, because the mission carried extra personnel that night. None of it was unusual. Crews at Seymour Johnson had flown this pattern many times.

A ground crew had found a fuel leak in the right wing before takeoff, a slow loss measured in a few hundred pounds an hour, and judged the aircraft fit to fly anyway. In the air, the leak got worse. By the time the bomber reached its tanker for the scheduled refueling, the crew was losing fuel much faster than anyone on the ground had accounted for, and the refueling had to be called off. The aircraft commander turned back toward Seymour Johnson early.

The wing kept leaking on the way home. Just after midnight, with the bomber descending toward the base in the last stretch of the flight, the right wing came apart in the air. Without it, the rest of the aircraft could not hold together. The B-52 broke up above the fields around Faro, a stretch of low, wet farmland a few miles from the runway the crew had been trying to reach.

Some of the men ejected. One of them, the copilot, Lieutenant Adam Mattocks, had no ejection seat built into his position. He got out by climbing to an opening in the fuselage as the aircraft came apart around him and jumping free by hand, the only man known to have escaped a B-52 in flight that way and lived to describe it. Three of the eight men aboard did not survive the breakup.

The two bombs separated from the aircraft as it disintegrated, the way they were built to in an emergency. Neither man nor switch chose to release them. They simply came free of a plane that no longer existed as a single object, and after that, each one fell on its own.

One of the two behaved the way an emergency drop was supposed to behave. Its parachute opened. It came down slowly and caught in the branches of a tree in a farmer's field, largely intact, low enough that a recovery team could reach up within hours and bring it down without digging for it.

The other bomb's parachute did not deploy. It fell the way anything falls without one, and it went into a low, saturated field at a speed that drove it deep into the ground before it broke apart. Pieces of it were scattered near the surface across the field, and a high explosive component within the weapon fired on impact, though the nuclear material did not detonate. The section that mattered most, the part of the bomb holding its uranium secondary, kept going down into ground that would not hold its own shape, and every time crews thought they had reached it, it turned out to be deeper still.

The bomb recovered from the tree was the one that told investigators what had almost happened. A weapon like the Mark 39 was built with a sequence of safety devices meant to keep it from detonating outside of a deliberate order to drop it on a target. When technicians went through this bomb's arming system, they found that most of that sequence had already moved into position, the same position it would have moved into if a crew had actually intended to arm and release it over an enemy target. Switch after switch in the chain had gone the wrong way. By the account written up afterward, only one low-voltage switch, of several that were supposed to stand between an accident and a live detonation, was still set to safe when the bomb settled into the tree at Faro. If that one switch had failed the same way the others had, the report concluded, the weapon would have gone off, with a force many times greater than the bomb dropped on Hiroshima sixteen years before.

None of this reached the public at the time. Officials at the base and at the Pentagon told reporters that neither weapon had come close to exploding and that the safety systems had performed exactly as designed. Newspapers ran that account: two nuclear bombs down near Goldsboro, no detonation, no danger to residents, the crash itself under investigation as a mechanical failure.

The digging for the buried section went on for days. The ground at Faro was drained swampland, low and wet no matter the season, and the water table stood close enough to the surface that crews had to keep pumping the hole to keep it from flooding as they went deeper. Engineers eventually decided that continuing to excavate put the men in the hole at risk, since the saturated soil at that depth would not hold a stable shaft. The Air Force called off the recovery. Most of the weapon had been retrieved by then. The uranium bearing section had not, and it was left where it had come to rest, deep under the field.

The story of the switch did not surface until years later, and even then it surfaced by accident. A Sandia National Laboratories engineer named Parker F. Jones, who studied the safety of American nuclear weapons for a living, wrote an internal report on the Goldsboro accident in 1969, eight years after the crash. The report stayed classified for decades. It became public only after a journalist researching the history of nuclear weapons accidents filed a Freedom of Information request and the document was released, and it was reported on in 2013 in The Guardian, more than fifty years after the bomb came down in the tree. Jones' own conclusion, in that report, was that the bomb had come far closer to detonating than the public had ever been told, and that the outcome had rested on a single low-voltage switch behaving correctly.

By then, the government had already settled the matter of the buried debris a different way. It negotiated a legal easement on the land where the weapon's remains lay, a restriction attached permanently to the property that barred digging, drilling, or construction below a certain depth without the Air Force's involvement. The family that farmed the land kept farming the surface. What was left of the bomb stayed underneath it.

No one was charged with anything. The Air Force's own investigation attributed the crash to structural failure in the wing brought on by the fuel leak, a mechanical finding rather than a question of blame, and the type of wing failure involved was one the Air Force later addressed with structural changes across part of the B-52 fleet.

The field outside Faro is still farmland. The family that owns it still works the surface, under the same restriction that has run with the property since the 1960s. The Air Force has periodically tested the site over the years and reported no contamination reaching the surface. No one has ever finished digging for the rest of the bomb.

It has been sixty five years since the wing came off that B-52 over Faro, North Carolina. Part of a hydrogen bomb built to be many times stronger than the one dropped on Hiroshima is still down there, under a working field, exactly where it landed.