The Bucket That Killed

On the morning of March eleventh, nineteen ninety-seven, a barrel of nuclear waste caught fire in the village of Tokaimura, Japan.

The fire started in a building where workers mixed low level liquid waste into molten asphalt, sealing it into steel drums for storage. That day, they were testing a new recipe, with twenty percent less asphalt than usual. Inside one fresh barrel, a slow chemical reaction heated the contents past their ignition point. The temperature controls that were supposed to prevent exactly that were not working. By ten o clock in the morning, the barrel was burning, and the fire was spreading to the barrels beside it.

The workers could not put it out. Smoke and radiation alarms forced everyone from the building. Hours passed. Then, just as people were preparing to go back inside, at around eight o clock that evening, built up flammable gas ignited and exploded. The blast broke windows and doors, and smoke and radiation escaped into the surrounding area. Thirty seven workers were exposed to trace amounts of radiation. A week later, officials detected unusually high levels of caesium forty kilometers to the southwest. The government called it the worst nuclear accident the country had yet seen.

And then, the regulators looked at the other nuclear plant in Tokaimura, a small fuel processing facility run by a company called J C O, and decided it needed no inspection. There had been no reportable events there. Nothing to see. Nothing to fix. The fire was treated as a closed chapter, a problem at one building, with one cause, already understood.

It is worth noting that the nineteen ninety seven fire happened at a different facility, run by a different organization, the Power Reactor and Nuclear Fuel Development Corporation. J C O was a separate company, in a separate building, with a separate safety record. That separation was the whole logic. The fire was P N C's problem. J C O had no reportable events. But the two plants stood in the same village, drew from the same workforce, and operated under the same regulator. The lesson of the fire, that the industry's safety culture had holes in it, applied to both of them. The paperwork said otherwise. The paperwork was wrong.

This podcast was brought to you by flooftags, visit floof tags dot com for one of a kind laser engraved pet tags.

This is the story of what happened two and a half years later, in the plant that needed no inspection.

Tokaimura sits about seventy miles northeast of Tokyo, in Ibaraki Prefecture. By the late nineteen nineties, nearly a third of the village depended on the nuclear industry for work. The J C O plant was small, unremarkable, one of several nuclear facilities in the area. Its full name was the Japan Nuclear Fuel Conversion Company, and its job was to take uranium hexafluoride, a chemical form of uranium, and convert it into uranium dioxide, the form used as fuel in experimental reactors. It was the kind of place that ran quietly, out of the headlines, doing careful work with dangerous material. That was the theory, anyway.

On the morning of September thirtieth, nineteen ninety-nine, three technicians arrived to prepare a small batch of uranium fuel. Their names were Hisashi Ouchi, Masato Shinohara, and Yutaka Yokokawa. Ouchi was thirty five years old. The job in front of them was not one they had done often. The company had not made fuel for this particular reactor in three years, and the three men had not been given specific training for the task.

The uranium they were handling was enriched to eighteen point eight percent. That matters, because the men were used to working with uranium enriched to between three and five percent. Enrichment is the share of the uranium that is the fissile isotope, uranium two thirty five, the kind that splits. At eighteen point eight percent, there was far more splittable material in every kilogram, which meant the margin for error was far smaller, and every safety precaution mattered more. That is precisely when the precautions were set aside.

It is important to be clear about something. The three men were not freelancing. They were not cutting corners on their own initiative. They were following a procedure their company had written, a shortcut designed to move the work faster, and they had been given no training that would have told them why it was dangerous. The procedure was the hazard. The men were its instruments. When we talk about what happened at Tokaimura, the failure belongs to the system that handed three untrained men a bucket and a funnel and called it a process, not to the men who carried them.

To understand what went wrong, you need to understand the two rules that keep this kind of work safe. The first rule is mass. There is a minimum amount of fissile material, called the critical mass, that can sustain a chain reaction on its own. Stay well below it, and nothing happens. The second rule is shape. Even a large amount of uranium is safe if it is spread out in a shape that lets the neutrons escape, a thin sheet, a narrow tube, so that most neutrons fly out into the air instead of striking another uranium atom. A criticality accident happens when both rules are broken at once, enough material, gathered in a shape that traps the neutrons inside.

The approved procedure for this job respected both rules. The uranium solution was supposed to be mixed in a dissolution tank, in batches of no more than two point four kilograms at a time, in a vessel shaped so that neutrons would escape out the sides before they could split another atom. Two point four kilograms, in a safe shape, one batch at a time. That was the procedure. Instead, the workers were following a different procedure, one the company had developed internally to speed up production, and which had never been approved by regulators. Under that procedure, they would mix the uranium by hand, pouring it from stainless steel buckets, through a funnel, into a precipitation tank. A bucket. A funnel. For eighteen point eight percent enriched uranium.

The precipitation tank was never designed for this. It was wider than the approved vessel, wide enough to let the solution pool into a compact mass instead of staying spread thin. And worse, it was wrapped in a jacket of cooling water. Water is an excellent mirror for neutrons. Instead of escaping, the neutrons that reached the edge of the tank bounced off the water jacket and came straight back into the uranium. The tank broke the mass rule, it broke the shape rule, and its water jacket made sure that almost no neutron went to waste. It was, by accident, a nearly perfect machine for sustaining a chain reaction.

The approved procedure would have been unremarkable. Small batches, two point four kilograms at a time, in a vessel shaped to let neutrons escape, with the uranium never gathering in one place in any dangerous amount. A slow, careful, boring morning. The kind of morning that produces no stories at all. That is what safety looks like in this industry, most of the time. Not drama. Not heroics. Just a procedure, followed, that keeps the physics on your side. The men at Tokaimura never got that morning, because the procedure they were given had already traded it away for speed.

With each bucket the men poured in, the amount of uranium in the tank crept closer to the point where it could sustain a chain reaction on its own. Nobody in the room was watching for that line, because nobody in the room believed they were anywhere near it. They poured one bucket. Then another. Then another. Seven buckets in all, about sixteen kilograms of uranium in total, nearly seven times the approved batch limit, all of it sitting in a single tank.

At ten thirty five in the morning, Shinohara poured in the seventh bucket. Ouchi stood beside the tank, holding the funnel. Yokokawa sat at a desk behind a wall, about fourteen feet away. And then, in the tank below them, the uranium went critical.

The men saw a blue flash, the pale blue light given off when charged particles move through water faster than light itself travels through water. It is called Cherenkov radiation, and it meant that inside the tank, uranium atoms were splitting, releasing neutrons, which split more atoms, which released more neutrons, an uncontrolled chain reaction running inside an ordinary steel vessel in an ordinary room. The three men felt ill almost immediately. Ouchi vomited and lost consciousness. All three were rushed to the hospital.

Inside the building, the reaction was still running, and nobody could see it. A criticality is not a fire. There is no smoke, no flame, no noise. There is only invisible radiation, pouring out of a tank, hour after hour, in a room that looked perfectly normal. The plant was evacuated, and the people responsible for stopping it had to work out how to switch off something that had no switch. Every minute the reaction continued, it was exposing anyone nearby, and nobody could walk into the room to do anything about it. That is what made those twenty hours so frightening. Not an explosion. Not a meltdown. Just a tank, sitting in a quiet room, quietly running a chain reaction, while the people outside tried to figure out how to make it stop.

But the reaction did not stop when the men left the room. There was no switch to flip. The uranium was already mixed into the solution, and the chain reaction kept going, pouring neutron and gamma radiation into the building hour after hour. The plant was evacuated. Five hours later, about one hundred sixty residents who lived closest to the facility were evacuated from their homes. They would not be allowed back for two days. Twelve hours after the criticality began, more than three hundred thousand people living within ten kilometers of the plant were ordered to stay indoors. The reaction ran for nearly twenty hours before workers finally stopped it, by draining the cooling water jacket that surrounded the tank, taking away the mirror that had been bouncing neutrons back into the uranium, and adding a boric acid solution to absorb the neutrons that remained.

For the village of Tokaimura, the days after the accident were a slow return to a changed normal. The one hundred sixty evacuated residents came home after two days. The shelter order lifted. But the trust was harder to restore. Nearly a third of the village had depended on the nuclear industry for work, and the industry had just shown them that the plants in their backyard could fail in ways nobody had told them about. The nineteen ninety seven fire had been explained away. The nineteen ninety nine criticality could not be. The village that had lived with nuclear power for decades now had to live with the knowledge that the people running it had been pouring buckets of uranium by hand.

The three men who had been standing beside the tank received the worst of it. Ouchi, who had been closest, absorbed a dose estimated at around seventeen sieverts. To put that in context, a dose of around seven sieverts is considered fatal. Seventeen is not a number the human body is built to survive.

What a dose like that does is quiet and total. The radiation shattered the chromosomes in his cells, which meant his body could no longer produce new cells to replace the ones that died. He was taken to the University of Tokyo Hospital, where doctors fought for eighty three days to keep him alive. He died on December twenty first, nineteen ninety nine. Shinohara, who had received a lower but still lethal dose, died about seven months later. Yokokawa, who had been behind the wall, received a smaller dose and eventually recovered.

In the end, more than six hundred people were exposed to elevated levels of radiation, including plant workers, firefighters, and emergency responders, along with local residents. The accident was rated a level four on the International Nuclear Event Scale. The nineteen ninety seven fire, the one that had been the worst the country had seen, had been a level three.

Afterward, the investigation found what the nineteen ninety seven fire should already have taught everyone. The workers had been given an unauthorized procedure, designed to save time, that stripped away every protection against a criticality. The company had not trained them for the job. The regulator had not inspected the plant. The official findings blamed inadequate regulatory oversight, a lack of safety culture, and inadequate worker training and qualification, at every level, in plain language. J C O lost its operating license, the first time a Japanese plant operator had ever been punished that way by law. The company president resigned. Six officials were charged with professional negligence. By March of two thousand, Japan had begun regular inspections of nuclear facilities, and new training requirements for handling nuclear material.

There is a phrase in this story that deserves a second look. No reportable events. That was the reason J C O was not inspected after the nineteen ninety seven fire. No reportable events. It sounds like a clean bill of health. But an absence of reported events is not the same as an absence of danger. It can just as easily mean that nobody was looking, or that the things worth reporting were happening in procedures nobody had approved, in rooms nobody was inspecting, carried out by men nobody had trained. The regulator looked at J C O and saw nothing wrong, because the system was designed to notice only the things it already knew how to notice. The buckets were invisible to it. The funnel was invisible to it. Until the morning they were not.

And the nineteen ninety seven fire hangs over all of it. A barrel of waste, a broken temperature control, an explosion, thirty seven people exposed, caesium forty kilometers away. A warning, delivered two and a half years early, at the same village, in the same industry, to the same regulators. The response was to look at the little fuel plant down the road and decide it needed no inspection. No reportable events. Nothing to see. Two and a half years later, three men carried buckets into that plant, and poured them through a funnel, one after another, until the uranium in the tank did what uranium does when you give it the chance. The warning had been free. The lesson cost two lives.

This is Nuclear Shadows. I am Morgan Miller. Thank you for listening, and I will see you next time.

A big thank-you to floof tags dot com for sponsoring the show.