January 1961: the night one low-voltage switch held for North Carolina.

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

Wayne County grew tobacco. It had done that for generations. The land around Goldsboro is flat and low, cut with drainage ditches and small creeks, and the soil holds water. Farmers there plant in the spring, prime leaves through the summer, and cure the crop in barns that still stand along the county roads. In the winter, the fields are bare and brown and the ground is soft from the rain.

The Air Force came to Goldsboro during the Second World War. The base closed after the war and reopened in the 1950s, and by the end of that decade it was a Strategic Air Command installation. That meant B-52 bombers. It meant runways long enough for an eight-engine jet carrying a full load of fuel, and it meant a steady payroll in a county that did not have many of them.

Strategic Air Command had a problem it had been working on for years. A bomber on the ground could be destroyed on the ground. If the Soviet Union launched first, the planes parked at bases like Seymour Johnson might never get off the runway. The solution SAC tested and then expanded was airborne alert. A share of the bomber fleet would be in the air at all times, fully fueled, fully armed, flying long loops toward the edges of Soviet reach and back again. If the order came, those crews would already be on their way.

The weapons they carried were real. On many of these flights, the bomb bay held two Mark 39 thermonuclear bombs. Each one weighed several tons. Each had a yield in the range of megatons, which is to say each one was hundreds of times more powerful than the bomb dropped on Hiroshima. The Mark 39 had been designed with a series of safety devices meant to keep it from going off by accident. The bomb would not detonate unless a sequence of arming steps occurred in order. Some were supposed to happen only when the weapon was released on purpose. One was a switch the crew controlled from the cockpit, marked for ready and safe.

The flights were routine by 1961. Crews trained for them, flew them, came home, and flew them again.

John F. Kennedy had been sworn in as president on January 20. Three days later, on the evening of January 23, a B-52G from Seymour Johnson took off on an airborne alert mission with eight men aboard and two Mark 39s in the bomb bay. The flight plan took it out over the Atlantic. The mission was expected to last about a day, with midair refueling along the way.

One of the eight was Lieutenant Adam Mattocks, the aircraft's third pilot. The B-52G had ejection seats for most of its crew. The third pilot's position did not have one. If the crew ever had to abandon the aircraft, the man in that seat would have to find another way out.

Around midnight, the bomber met a tanker off the coast for refueling. The tanker crew saw something the bomber crew could not see from inside the cockpit. Fuel was streaming out of the B-52's right wing. They told the pilot.

The bomber broke off the refueling. The crew was directed to fly a holding pattern out over the ocean and burn off fuel before coming back to land. A leaking wing tank on a jet that size is a serious problem but not necessarily a fatal one, and the procedure was to get the weight down and bring the aircraft home. The leak did not hold steady. It got worse. The fuel was going out faster than the engines were burning it, and the imbalance between the wings began to pull on the aircraft.

The pilot turned back toward Seymour Johnson. By the time the bomber was descending toward the base, it had lost so much fuel on one side that the crew could not keep it level. The aircraft began to roll. At low altitude, a few miles from the runway, the pilot ordered the crew out.

Five men got clear of the aircraft. Adam Mattocks, with no ejection seat, went out through the top hatch after the others had gone and came down under his parachute. One of the men who left the aircraft died after landing. Two others never got out. Three of the eight were killed.

The B-52 came apart in the air just after midnight on January 24. The breakup tore the bomb bay open. Both Mark 39s fell out over the farmland near the community of Faro, north of Goldsboro.

The two bombs did not fall the same way.

The first one behaved the way a bomb released on purpose would behave. As it separated from the aircraft, the sequence built into it began to run. Its parachute deployed. It drifted down and came to rest in a field with the parachute caught in the trees and the nose of the bomb pushed into the soft ground. It was intact.

The second bomb's parachute did not open. It fell free and struck the wet ground at high speed. The casing broke apart on impact, and much of the weapon drove itself deep into the field, into waterlogged clay, and kept going.

In the morning, the Air Force was in Faro. Explosive ordnance disposal teams went to both sites. At the first bomb, they went through the arming mechanisms one at a time to see what had happened during the fall. The findings were worked out that day and in the review that followed. Of the four steps required for the bomb to detonate, three had been completed. The parachute had opened, the timers had run, and other parts of the sequence had armed themselves as the weapon separated from a breaking aircraft. The remaining step was the ready and safe switch. It was still set to safe.

The second site was harder. Crews began digging where the bomb had gone in. The ground was saturated, and groundwater filled the excavation almost as fast as it could be pumped out. They dug anyway, with pumps running and the walls of the pit sliding. They recovered pieces of the weapon from inside the hole as they went down. Among the parts the disposal team pulled out of that mud was the second bomb's ready and safe switch. It was in the armed position.

The digging went down dozens of feet. Much of the weapon came up, including the components that mattered most for an immediate explosion. The thermonuclear secondary, the stage containing uranium, did not. It had gone deeper than the crews could follow in that ground. After a long effort, the Air Force stopped digging and filled the hole back in.

The public was told that a bomber had crashed and that it had been carrying unarmed nuclear weapons. The Air Force position, stated then and for many years afterward, was that there had never been any real danger of a nuclear explosion.

Inside the weapons establishment, the question stayed open. In 1969, an engineer at Sandia National Laboratories named Parker F. Jones reviewed what had happened at Goldsboro. Sandia was responsible for the safety of American nuclear weapons, and Jones supervised work on that subject. His memo was titled "Goldsboro Revisited or: How I Learned to Mistrust the H-Bomb." In it, he concluded that the Mark 39 did not have adequate safety for the airborne alert role, and that on the night of the crash a single low-voltage switch was what had stood between the United States and a catastrophe. He also noted that the same kind of switch could be activated by an electrical short in a damaged aircraft.

The memo was classified. It stayed classified for more than four decades.

The families near Faro went on farming. The field where the second bomb went in was put back into crops. The government bought an easement on the land that limits how deep anyone may dig there. The easement exists because of what is still underneath.

No one was disciplined in connection with the design of the weapon or the decision to fly it on airborne alert. The Air Force report on the crash addressed the aircraft's structural failure and the fuel leak in the right wing.

Airborne alert continued for years after Goldsboro. It ended in 1968, after another B-52 carrying hydrogen bombs crashed near Thule Air Base in Greenland and spread radioactive material across the sea ice. By then, there had also been Palomares, Spain, in 1966, where a B-52 collided with its tanker and dropped four hydrogen bombs, two of which broke open on impact.

Parker Jones's memo reached the public in 2013. The journalist Eric Schlosser obtained it under the Freedom of Information Act while researching his book on nuclear weapons accidents, and it was published that year. It was the first time the government's own safety engineer's assessment of Goldsboro could be read outside the weapons laboratories.

In 2012, the state of North Carolina put up a highway historical marker near the crash site. It is titled Nuclear Mishap. It says that a B-52 carrying two hydrogen bombs crashed nearby in 1961 and that three crewmen died.

The easement on the field is still in force. It prohibits digging below a set depth. It does not prohibit planting, and the field is planted every spring.