Part of the bomb is still down there. The Air Force bought the field instead of digging.

January 23, 1961.
35.3394° N, 77.9606° W

The airborne alert was arithmetic. A Soviet missile could reach an American bomber base in under half an hour, and a bomber sitting on the ground during that half hour was a bomber that would never fly. The Strategic Air Command's answer was to keep part of the force in the air all the time. Bombers flew long circuits over the Atlantic and the Arctic, took fuel from tankers in flight, and came home after a full day aloft. They carried live thermonuclear weapons the entire way. The routes had code names. Crews rotated through them the way other men rotate through shifts.

The aircraft doing this work was the B-52, and the B-52 was having trouble with its wings. The bomber had been designed for high altitude, and the mission had changed under it. Soviet surface to air missiles pushed the flight profiles lower, and flying low meant hour after hour of rough air. Cracks turned up in wing structures across the fleet. The fuel was carried in the wings. Inspection and modification programs ran through the late 1950s and into the 1960s, and the alert flights continued while that work went on.

Seymour Johnson Air Force Base sits at the edge of Goldsboro, in the flat farm country east of Raleigh. Tobacco, soybeans, hogs, two-lane roads, crossroads communities with names like Faro and Eureka and Nahunta. The base had been there since the war. In 1961, it was home to the 4241st Strategic Wing, and its bombers stood alert and flew it.

On January 23, 1961, a B-52G lifted off from Seymour Johnson for a twenty-four-hour airborne alert mission. There were eight men aboard. The aircraft commander was Major Walter Scott Tulloch. In the bomb bay were two Mark 39 thermonuclear bombs, each in the megaton range, each many times the yield of the weapon used on Hiroshima. They were not cargo in the way a crate is cargo. They were mission load, aboard because the order to use them might come.

The flight went normally into the evening. At the first refueling, the tanker crew looked the bomber over, as tanker crews did, and saw nothing worth mentioning. At the second refueling the boom operator on the KC-135 told the B-52 crew that fuel was coming out of the right wing. The bomber broke away. The mission was cancelled and the aircraft was directed to hold off the coast and burn down to a safe landing weight.

The hold put the bomber in a racetrack pattern over the water while the crew watched the numbers. The right wing was emptying at a rate that matched nothing the gauges were supposed to show. Tens of thousands of pounds went out of that wing in a span of minutes. The crew reported it. The base told them to come home.

They turned inland. On the approach, with the gear coming down, the aircraft became hard to hold. Then the right wing failed. The B-52 rolled over into a spin and broke up in the dark above farmland northeast of the base, several thousand feet over houses and fields and county roads.

Five men got out. Four of them ejected. One of the pilots, Lieutenant Adam Mattocks, went out through the hatch in the roof of the cockpit without an ejection seat, into the airstream of a rolling bomber, and his parachute opened. Three men did not survive. One ejected and died on landing. Two never left the aircraft. The wreckage came down across several fields and burned.

The two Mark 39s came out of the airplane while it was coming apart.

A weapon of that kind is built to be released on purpose, by a crew that intends to release it, and once it is in the air it runs a sequence. Signals from the bomb bay. A lanyard pulled. Fuzes armed. A parachute deployed to give the aircraft time to get clear. Barometric switches closing as the weapon falls. At the front of all of it, a switch in the cockpit that a man has to move with his hand. The sequence does not know why the bomb left the airplane.

The first weapon behaved as if it had been dropped deliberately. Its parachute came out and filled. It descended slowly and came down in trees near Faro, where it hung by its shroud lines with the nose pointed at the ground. When the recovery team reached it the next morning it was intact, upright, and undamaged enough to photograph and haul away. It had run its arming sequence on the way down.

What stopped it was not the parachute and not the trees. A single safe/arm switch inside the weapon had not closed. Years afterward, a supervisor in the nuclear weapons safety department at Sandia National Laboratories went back through the Goldsboro accident and wrote up what he found. His name was Parker F. Jones. He titled the piece Goldsboro Revisited, and his conclusion was that "one simple, dynamo-technology, low-voltage switch stood between the United States and a major catastrophe." The report was written in 1969 and classified.

The second weapon's parachute did not deploy. It fell free and struck a muddy field off a road called Big Daddy's Road, in the low wet ground near Nahunta Swamp, at a speed measured in hundreds of miles an hour. It did not detonate. It came apart on impact and drove its own pieces into saturated soil. There was no crater in the ordinary sense. There was a hole, a scatter of wet earth, and a farm field with something under it.

People around Faro and Eureka knew there had been a crash before they knew anything else. The noise brought them out of their houses. Wreckage burned in the fields, and by morning there were airmen and trucks and rope lines across the farm roads and a guard posted on a stand of trees. Farmers were told which ground to stay off. What was in the trees was not explained to them.

Recovery crews from the base worked the impact site for weeks. They found pieces near the surface, then they dug, and the deeper they went the worse the ground got. The hole filled with water as fast as it could be pumped. They brought in more equipment, shored the walls, and kept going, and the water kept coming. They recovered most of the weapon. They did not recover the secondary stage, the uranium component that makes a thermonuclear weapon thermonuclear. It had gone down into the mud and the water and stayed there.

The Air Force stopped digging. In place of the excavation the government took an easement over the ground, a circle of land around the impact point with restrictions on what could be done beneath it. The surface stayed in agriculture. Nothing was to be dug or drilled below a set depth. The arrangement was recorded and left in place.

The public account was short. The Defense Department said an aircraft had crashed, that two unarmed nuclear weapons had been aboard, and that there had been no possibility of a nuclear explosion. That position was repeated in press statements and in official accident summaries for decades. The surviving crew was told not to discuss the flight. Mattocks's escape through the roof hatch was not part of the account the Air Force gave, and it did not reach the public until he described it himself in interviews many years later.

The airborne alert kept flying. Bombers on alert crashed or burned or lost weapons in Texas, in Spain, in Greenland, and in each case the weapons were described publicly as unarmed. The program ended in 1968, after a B-52 carrying four hydrogen bombs went down on the sea ice near Thule.

No one on the ground was killed at Goldsboro. Three of the eight men aboard died, and they were buried. No charges were filed and no finding of wrongdoing was made. The accident was handled inside the Air Force as the loss of an aircraft, and the wing structural problem was folded into inspection and modification work already in progress.

The Defense Department's position that the weapons could not have gone off was maintained for half a century.

In 2012, the state of North Carolina put up a highway marker beside Route 111 outside Eureka, the same silver and black sign used for battlefields and birthplaces, noting a nuclear mishap at the site. In 2013, the Jones report was released under the Freedom of Information Act to the writer Eric Schlosser and published, and the sentence about the switch went around the world.

The easement is still in force. The field is still farmed. Below it, past the depth where the pumps gave out, the uranium secondary of a Mark 39 hydrogen bomb has been in the ground since a Tuesday morning in January 1961.