Disturbing History - Three Mile Island Meltdown

Episode Date: August 30, 2026

On March 28, 1979, at 4:00 a.m., a relief valve at Three Mile Island opened as designed—and then failed to close. For more than two hours, operators believed it had shut because the control-room indicator showed the electrical command, not the valve’s actual position.By the time the mistake was understood, roughly half the fuel in Unit Two had melted.The men in the control room did not know how bad it was. Neither did the Nuclear Regulatory Commission. The full extent of the damage would not be confirmed until engineers lowered a camera into the reactor vessel three years later and found a five-foot void where the top of the core had once been.In this episode of Disturbing History, Brian goes beyond the familiar story of America’s worst commercial nuclear accident and into the warnings, failures and unanswered questions that came before and after it.Eighteen months before the meltdown, another Babcock & Wilcox reactor at Davis-Besse experienced a nearly identical stuck-open relief valve. Engineers warned that operators could misread the signs of a loss-of-coolant accident and shut down emergency cooling at exactly the wrong moment. One NRC inspector was concerned enough to travel to Washington on his own time just six days before Three Mile Island to warn commissioners that something was badly wrong.The Kemeny Commission later concluded that the correct operating guidance was never passed along to the utilities.The episode reconstructs the accident minute by minute: the closed emergency feedwater valves, the overwhelming alarm system, the printer that fell so far behind that critical records were lost, temperature readings above 2,000 degrees Fahrenheit that operators dismissed as impossible, and a hydrogen burn inside containment that was initially written off as an instrument problem.Then the crisis moved beyond the plant.Conflicting radiation readings, contradictory public statements and confusion inside state and federal agencies turned Three Mile Island into a communications disaster. Governor Dick Thornburgh advised pregnant women and young children within five miles of the plant to leave. Roughly 144,000 people ultimately evacuated the surrounding area.The episode also examines the infamous hydrogen bubble, the federal scramble to obtain potassium iodide because no national stockpile existed, and the radiation readings that appeared, disappeared or were later disputed. And then there was the criminal investigation.Less than two months after the accident, control-room operator Harold Hartman told federal investigators that Metropolitan Edison had been falsifying reactor coolant leak-rate data before the meltdown to avoid shutting the plant down for repairs.Federal prosecutors later sharply criticized the NRC’s handling of the allegations. In 1983, a grand jury indicted Metropolitan Edison on charges involving falsified tests and destroyed records. The company eventually pleaded guilty, becoming the only American nuclear utility convicted of a criminal offense.The final portion of the episode turns to the question that remains the most controversial: what happened to the people living around the plant?Residents reported metallic tastes, skin irritation, nausea, hair loss and dead animals. Official estimates said offsite radiation doses were far too low to cause acute radiation sickness. Decades of epidemiological studies have produced conflicting interpretations, with some finding no convincing cancer link and others identifying patterns that warranted further investigation.Thousands of personal injury claims were eventually dismissed because plaintiffs could not prove exposure above the required legal threshold.Nearly half a century later, the physical story of Three Mile Island still is not finished. Much of the damaged core remains stored in Idaho, some radioactive material is still being removed from Unit Two, and the neighboring Unit One reactor is being prepared for a possible return to service as the Crane Clean Energy Center.After thousands of pages of federal reports, a criminal prosecution, decades of litigation and the dismantling of a melted reactor core, some of the most basic questions surrounding Three Mile Island remain unresolved—including who closed two critical valves that should have been open on the morning the accident began.Email BrianJoin Our FREE NewsletterGet Brian's Books Leave Us A VoicemailVisit Our WebsiteHave a forgotten historical mystery, disturbing event, unsolved crime, or hidden conspiracy you think deserves investigation?Send your suggestions to [e-mail] History is a dark history podcast exploring unsolved mysteries, secret societies, historical conspiracies, lost civilizations, and the shadowy stories buried beneath the surface of the past.Follow the show and enable automatic downloads so you never miss a deep dive into history’s most unsettling secrets.Because sometimes the truth is darker than fiction.

Some stories were never meant to be told. Others were buried on purpose. This podcast digs them all up. Disturbing history peels back the layers of the past to uncover the strange, the sinister, and the stories that were never supposed to survive. From shadowy presidential secrets to government experiments that sound more like fiction than fact,

this is history they hoped you'd forget. I'm Brian, investigator, author, and your guide through the dark corner. of our collective memory. Each week I'll narrate some of the most chilling and little-known tales from history that will make you question everything you thought you knew.

And here's the twist. Sometimes the history is disturbing to us. And sometimes, we have to disturb history itself, just to get to the truth. If you like your facts with the side of fear, if you're not afraid to pull at threads, others leave alone. You're in the right place.

History isn't just written by the victors. victors. Sometimes, it's rewritten by The Disturbed. Before we get into this one, I owe a thank you to a listener named Jason. Jason wrote in recently with a suggestion. He said that between his own love of history and the shows we put out over at paranormal world productions, one topic kept surfacing on him, and the topic was Three Mile Island, and he has a good reason for it. Jason grew up in Pennsylvania, and in the spring of 1979, his family was living in Colonial Park. out in suburban Harrisburg. That put them in Diphon County,

the same county as the island. He remembers his grandmother gathering him up along with his sister, who was about a month old at the time. He remembers his parents saying to head for the mountains. The line from his email that stayed with me is the one about everything that came after. Disaster was averted, he wrote.

But the specter of what could have been and still could be remained in the back of his mind the rest of his life. He also said that I might already have this one queued up to research. As it happens, he was right. I was already several weeks into the Kimmany report when his note came in, which is either very good timing on his part

or proof that this particular story has a way of finding the people it touched. Either way, Jason, thank you for listening and thank you for writing, and this episode belongs to you as much as it does to me. And if there's a story you've been carrying around, I'd like to hear it, something out of your town or your family, or something that just never sat right with you and never got properly explained. The email address is Brian, that's B-R-I-A-N, at paranormal world productions.com. I read everything that comes in,

and more than one episode of this show started life as somebody's email. In 1982, a crew at Three Mile Island lowered a small camera down through an opening in the head of the reactor vessel at Unit 2. The vessel had been sealed since the spring of 1979. Every public statement made in the three years since had described a reactor that suffered some damage to its fuel and a cleanup that was proceeding carefully. What the camera found was a hole,

where the top of the core should have been, where fuel assemblies should have been standing in a lattice 12 feet tall, there was open space. Avoid roughly five feet deep. Below it, a bed of broken rubble that had once been the most carefully engineered object in Pennsylvania. It took another four years of drilling and sampling to establish the rest of it. At least 45% of that core had melted. Something on the order of 62 tons of it had come apart,

and about 19 tons of molten uranium and zirconium and steel, had run down through the bottom of the core and pooled on the lower head of the reactor vessel, which is the last piece of metal between the fuel and the, and the floor of the containment building. An elliptical patch of that steel head, roughly three feet by three feet, reached about 2,000 degrees Fahrenheit, and stayed there for something like half an hour.

That steel held, and nobody has ever been entirely sure exactly why. The Organization for Economic Cooperation and Development ran an international project on that reactor vessel through the 1990s, and one of the plainest lines to come out of it is that it took eight to ten years, years after the accident before anyone in the field understood that the bottom head could have failed at all.

So for most of a decade, the industry and the regulator and the public were arguing about an accident whose worst moment none of them had correctly identified yet. The week in the spring of 1979 is the part everyone remembers, Kronkite, the cooling towers, the governor on television. What I keep coming back to is everything stacked underneath that week. a warning memo written more than a year ahead of time that named the exact mistake the operators would make. An almost identical accident at a sister plan in Ohio that got investigated and then filed away. An engineer who flew to Washington on his own time and his own money six days before to tell two commissioners that something was badly wrong.

And a criminal case, years later, that established as a matter of federal record that the reactor which melted had been kept running on falsified paperwork, for the five months before it happened. So that's what this episode is really about. The meltdown itself runs about 16 hours. The runway leading into it took roughly two years to build, and every foot of it is written down somewhere. Three Mile Island sits in the Susquehanna River

about 10 miles downstream from Harrisburg, near the town of Middletown, Pennsylvania. It's a real island, a couple of miles long, low and flat, with four enormous natural draft cooling, towers that are each about 370 feet tall. There are two reactors on it. Unit 1 went into service in 1974, and in Unit 2 followed a few years later. How much newer than that is worth pausing on? Unit 2 got its operating license in February of 1978. It was declared to be in commercial

operation on the 30th of December that year. The accident happened on the 28th of March 1979. That works out to 88 days. The current owner of the site puts the actual operating time at roughly 81 days, which says something about how much of that 88 it spent shut down. The Smithsonian's account of the plant describes construction delays followed by repeated unscheduled shutdowns from the moment it started running. That reactor never had a smooth stretch. On the 3rd of November, 1978,

it tripped off line from 90% power because an instrument technician opened the wrong breaker, and shut every outlet valve on the condensate polisher's at once. I need to explain the condensate polisher's because they're where this whole thing starts and because they'd already been telling people for a year and a half that they were going to cause a problem. A pressurized water reactor has two separate water loops. The primary loop runs through the core, picks up heat and carries it to a steam generator.

The secondary loop runs through the other side of that steam generator, boils into steam, spins a turbine, condenses back into water, and goes around again. The polisher sit in the secondary loop. They're big tanks of resin beads that strip dissolved minerals out of the feed water so the system doesn't scale up and corrode. At Unit 2 there were eight of them. Seven ran while one was being cleaned out. Cleaning meant blowing compressed air through a bed of resin that had packed itself into a solid mass

to break it up enough to flush it out. The President's Commission staff who looked at that system afterward wrote that resin removal had been a chronic problem since the earliest system tests, failing roughly one time in every 12 transfers. They also wrote that the polisher installation had, in their words, essentially zero operational margin. There was no automatic bypass valve installed. If the polisher's isolated themselves, feed water stopped. It had already happened more than once. In October of 1977, a resin transfer caused a loss of feed water, and it took two operators about an hour to get the bypass valve open by hand.

In May of 1978, the effluent valves closed on their own during a changeout, and when people went looking, they found water in the airlines that operate those valves. And then there is the memo itself. In May of 1978, somebody inside that plant wrote it down. The commission's own staff analysis quotes it. It says it's time to really do something about this problem before a very serious accident occurs and that if the polisher's take themselves offline

at any high level of power, the resulting damage could be very significant. The memo recommended making the bypass valve automatic so that feed water would keep flowing if the polishers dropped out. Ten months later, at 97% power, the polisher's took themselves. offline. The polisher failure needs some care because the popular version of this story is more confident than the record actually is. The leading explanation is that during the air blowing, water got

passed a check valve that had frozen open and worked its way into the instrument airline, and that water then drove the polisher valves shut. That's what the Rogovan report describes, and even Rogovan brackets the passage and uses the word postulated. The president's commission staff went further and said flatly that testing at Unit 2 had not confirmed a reason for the closure. They could reconstruct a plausible path. They could not reproduce it in testing. So the accident that changed nuclear power in the United States begins with a mechanism that was never definitively established. What is established is what the closure did. At 37 seconds past four in the morning, on Wednesday the 28th of March, feed water flow to both steam generators stopped. About a second later, the turbo

Tripped. With no feed water, the secondary side of those steam generators started drying out fast, and a Babcock in Wilcox once-through steam generator holds very little water compared to other designs. The Nuclear Regulatory Commission's own training material says that kind of generator can boil dry in as little as five minutes. Heat kept coming out of the core with nowhere to go. Pressure in the primary system climbed. Three seconds in, a relief valve on top of the pressurizer opened automatically to bleed some of that pressure off, exactly as designed. Eight seconds in, the reactor scrambled on high pressure. Control rods dropped, the fission chain reaction stopped, and pressure started falling.

13 seconds in, pressure dropped through the set point where that relief valve was supposed to shut. The electrical signal holding it open went away. The valve itself never moved, and it stayed open. That valve is called a pilot-operated relief valve, and everybody in this store, story calls it the P-O-R-V. It sits at the top of the pressurizer, which is the tallest point in the primary system. A valve stuck open there is a hole in the reactor coolant system at the highest place in the loop. Water and steam started leaving at a rate Rogoven put at about 110,000 pounds an hour, which works out to something like 220 gallons a minute. For two hours

and 22 minutes, nobody closed it. The reason is a detail that sounds like a small thing and isn't. In the control room, there was an indicator light for that valve. The light was wired to the solenoid. It told the operators whether the electrical signal to close had been sent. It did not tell them whether the valve had actually closed. At 13 seconds, the signal went away. The light went out, and four men who had been trained on that panel read it the way anyone would read it. The Kimmany Commission's finding puts it in one sentence. The indicator showed only that the signal had been sent, rather than the fact that the valve remained open. There was another way to catch it.

The discharge line downstream of that valve has a temperature reading, and a stuck open valve will keep that pipe hot. Shift supervisor William Zhu later told the commission that he'd seen the reading and knew the valve had lifted and expected the downstream temperature to be high and expected it to take a while for the pipe to cool back below the set point, which is a completely reasonable thing to think. Reasonable but wrong.

and there was nothing on that panel to tell him otherwise. Meanwhile, something else had gone bad in a way nobody in the room knew about yet. When feedwaters lost, three emergency feedwater pumps start automatically and push water into the steam generators. They started on schedule and they ran, and they pumped against two closed valves. The valves are called EFV12A and 12B. The operators called them the 12s. They're required by the plant's own rules to be open the insurance. entire time the reactor is running, and on that morning, both of them were shut.

Emergency feed water pumps ran for eight minutes and delivered nothing to either steam generator. Somebody found them at eight minutes and 18 seconds and opened them. Who that somebody was is genuinely in dispute in the federal record, which I find remarkable for an event this heavily investigated. The Kimmany Commission wrote that eight minutes into the accident, someone, discovered that no emergency feed water was reaching the steam generators. Their narrative then credits control room operator Craig Faust, scanning the panel, checking the 12s, finding them closed, opening them. The commission also records that one of those indicator lights was covered by a yellow maintenance tag. The Ruggavan report tells it differently.

In Ruggavent, the man who spots it is Ken Bryant, who had walked over from Unit 1 that morning specifically to watch a reactor trip get handled. Brian comes through the door, looks at the board and yells that the 12s are closed. Zui yells back to open them. Faust confirms that they're both closed. Two federal investigations looked at one event and came back with two different men. The bigger question is why they were shut in the first place, and nobody has ever answered that either. The 12s had been closed two days earlier on the 26th of March as part of a routine test of the emergency feed water pumps. Rogavan calls it 42 hours prior, and then the record simply stops. Ruggavin says in plain language that the special inquiry group has not been able to determine how the block valves got to be closed that morning.

Kimmany says the commission investigation has not identified a specific reason. Kimmany lists the possibilities without picking one. They were never reopened after the test, or they were reopened, and the control room crew shut them by mistake in the first seconds of the accident. Or somebody closed them from a control point outside the control room. Roveman adds a fourth that Kameney doesn't print and then dismisses it for lack of evidence, which is that somebody closed them deliberately to cause trouble. The bottom line is, no matter the reason, both investigations concluded the eight minutes didn't change the outcome.

The Corps was going to be damaged by the open relief valve, regardless of what the steam generators were doing. But it's worth sitting with the fact that in the most exhaustively documented industrial accident in American history, the question of the question of the same. of who shut two valves that were required to be open has never been resolved, and the industry's own analysis group later wrote it up as though it had been.

Everything that happened next follows from one instrument reading and one habit of mind. With the relief valve open, the primary system was losing water. System pressure was falling the whole time, but the pressurizer level indicator, which is the gauge operator's watch to know how much water is in the system, was going up.

It pegged high at about 292 inches. That combination is not supposed to be possible. In every scenario those men had been trained on, pressure and pressure and pressurizer level moved together, and down always meant down. When the level indicator climbed while pressure fell, they were looking at a reading that in their training described a system filling up with water.

There's a condition called going solid, where the pressurizer fills completely and control of pressure is lost entirely. It's dangerous and every operator is drilled to avoid it. Two minutes in, emergency high pressure injection had started automatically, dumping cold water into the core the way it's supposed to during a loss of coolant. Four and a half minutes in, the crew shut down one of those pumps and throttled the other one back hard, then maximized let down to pull even more water out of the system.

The Kimmany finding on this is blunt. The operators were conditioned to maintain the specified water level in the, the pressurizer and were concerned that the plant was going solid. So they cut back high pressure injection. And then the commission wrote the sentence that hangs over the entire accident. If the high pressure injection had not been throttled, core damage would have been prevented. They did exactly what they'd been taught. The instrument told them the system was flooding. The system was actually draining out the top. And this is where it stops being a story about four men in a control room at four in the morning. And starts being a story. And starts being a story.

about who knew and when. 18 months earlier on the 24th of September, 1977, the Davis Best Plant in Ohio had the same accident. It was the same reactor vendor. A spurious signal cut feed water to a steam generator. The pilot operated relief valve lifted nine times and then stuck open on the 10th. Pressurizer level went up while pressure went down.

More than 300 alarms came in. The shift supervisor, a man named Michael Daravan, Later described the crew as being in complete confusion for over 20 minutes. Then Daravan ordered the block valve on that relief line shut, about 23 minutes in. There was no core damage at Davis-Bess. That plant was running at 9% power that day. Three Mile Island was at 97. The Ohio event got investigated by the vendor and by the regulator.

The Kimmany Commission's finding on what came of that is one sentence long, and it's the most damaging sentence in the report. That incident was investigated by both Babcock and Wilcox and the Nuclear Regulatory Commission, but no information calling attention to the correct operator actions was provided to utilities prior to the Three Mile Island accident. It gets worse, because the correct operator actions had already been written down by name. In November of 1977, a Babcock and Wilcox engineer named Joseph Kelly wrote a memo about Davis-Bess. His point was narrow and precise. The operator stopped high-pressure injection when pressurizer level began to recover without regard to primary pressure.

Kelly recommended telling customers about it. Leadership didn't act on the recommendation. A reply came back inside the company that week, which contains an admission. It confirmed that the Davis Best operators had responded in the correct manner considering how they'd been trained and stated that the writer and the company's training department, both assumed that reactor coolant pressure and pressurizer level would trend in the same direction during a loss of coolant accident. That single assumption is the entire accident.

It was written down in an internal memo 16 months early by the people who trained the operators as a plain statement of what they believed. Stay tuned for more disturbing history. We'll be back after these messages. Then in February of 1978, another person, Babcock and Wilcox engineer, Bert Dunn, who ran the group that analyzed emergency core cooling, wrote his own memo. The Kimmany Commission quotes it directly.

Dunn wrote that if the Davis Best Event had happened in a reactor operating at full power, it is quite possible, perhaps probable, that core uncovery and possible fuel damage would have occurred. Dunn's group submitted revised emergency procedures. A manager treated it as routine, delegated the follow-up. And the follow-up never happened. And separately from all of that, a Tennessee Valley Authority engineer named Carlisle Michelson

had run the same problem out on paper in 1977, before Davis Best ever happened. Michelson worked out that a leak in the steam space at the top of the pressurizer would drive the indicated water level up while pressure went down, and he predicted the consequence in riding. Operators would mistakenly terminate high-pressure injection flow based on the response of the pressure riser level. He sent his analysis to Babcock and Wilcox,

to the Nuclear Regulatory Commission, and to the Advisory Committee on Reactor Safeguards. Kemeny's finding on that one is four words longer than the last one, and says the same thing. Again, no notification was given to utilities prior to the accident. There is one more warning in that file. In January of 1978, an official inside the commission itself pointed out the likelihood of erroneous operas.

operation in exactly this kind of event. The Commission did not notify utilities. There's also James Creswell, who I think deserves to be better known than he is. Creswell was a reactor inspector in the Commission's Region 3 office out of Chicago. He read the Davis Best event and understood what it meant, and he pushed it up through his chain of command and got nowhere. So he went around the chain. On the 22nd of March, 1979, he flew to Washington on his day off, at his own expense, and briefed two commissioners and their technical staff on what he thought was going to happen. They found his complaint serious enough to merit further consideration. Six days later, the Corps at Unit 2 started coming apart.

By 5 o'clock in the morning, an hour into it, the four big reactor coolant pumps were shaking hard enough that the crew was worried about the piping. Those pumps were trying to move a mixture of water and steam, and pumps don't like that. At 13 minutes past five, the operators shut down the two pumps on the B-loop. At about 20 minutes to six, they shut down the last two. Ruggavin describes what that meant in one sentence. With no flow, the hot core boils away the water, making more steam and gradually dropping the water level until the 12-foot fuel bundles are more than half uncovered.

The top of the core came out of the water somewhere around, 114 to 120 minutes in. The Idaho National Laboratory's reconstruction has the liquid level down at the mid-plane of the core by about two hours 20. Detectors outside the vessel started reading in a way that only makes sense if fuel is standing in steam. Thermocouples in the hot leg piping showed the steam coming out of the core was superheated and there's no way to superheat the hot leg without uncovering the core. Zirconium is what fuel cladding is made of and a above about 2,000 degrees Fahrenheit zirconium and steam react with each other. The zirconium takes the oxygen out of the water and gives back hydrogen and an enormous amount of heat.

And the reaction feeds itself. Post-accident measurements indicate about half of all the zirconium in that core reacted. That's where the hydrogen came from that everybody would spend the following weekend arguing about. The one instrument that might have made all of this legible was the plant computer, which printed an alarm log. Ruggavon's finding on that printer is the detail I'd put on the wall of every control room designer's office. The alarm printer can only type one line every four seconds. During the accident, several alarms per second were occurring.

Within a few minutes, the computer was far behind real time, and the alarms coming off the printer were for things that had happened several minutes earlier. Then Ruggavan adds one more line about that printer. alarms from one hour 13 minutes to two hours 37 minutes are irretrievably lost. That window is the core uncovering. The plant's own record of the period in which the fuel came apart does not exist, because a printer couldn't keep up. Over a hundred alarms went off in the early stages with no way to suppress the unimportant ones

or identify the important ones. Some of the indicators that mattered most were physically on the back of the control panel. Craig Faust told the President's Commission that he would have liked to have thrown away the alarm panel because it wasn't giving them any useful information. The training simulator these men had used could not reproduce this accident. The Commission made that a formal finding rather than a footnote. The simulator differed from the actual console in significant ways, and prior to the 28th of March, it had never been programmed to produce the conditions the operators faced.

At 6 o'clock, a shift supervisor named Brian. and Mailer came on shift fresh from outside into a control room that had been running on adrenaline for two hours somewhere around 18 minutes past six he looked at the temperatures on that relief valve discharge line and ordered the block valve shut the valve closed at about 22 minutes past six Rogavon's line about him is generous and a little rye for whatever reason and by whatever route Mailer has arrived at exactly the right decision just 20 minutes after coming on the scene fresh from the outside.

Rovevin also records that Edward Frederick, one of the operators, offered a different reading of the same moment. Frederick didn't dispute that Mailer was the one who did it. He said it was done because no one could think of anything else to do. Between the moment the last pumps went off and the moment the block valve shut, the principal core damage occurred. By roughly three hours and 45 minutes in, a mass of molten material broke through the crust holding it and relocated down.

into the lower plenum of the reactor vessel. That happened at about a quarter to eight in the morning. At that hour, the general emergency had been declared for 20 minutes and the first radio report was still half an hour away. Around 10 minutes to 7 that morning radiation alarm started coming in from inside the plant. At about five minutes before seven, George Cunder and William Zoo declared a site emergency, which under the rules means an event that threatens an uncontrolled release of radioactivity to the immediate environment. Zhu called Dauphin County at two minutes past seven.

Two minutes after that, he reached the watch officer at the Pennsylvania Emergency Management Agency and told him there was a high level of radiation inside the reactor room. At around the same time, someone made the first phone call to the Nuclear Regulatory Commission. According to a University of Pittsburgh case study built out of the governor's own papers, that call was routed through an answering call. service. Station manager Gary Miller arrived a few minutes after seven and declared a general emergency at 24 minutes passed. A general emergency means an incident with the potential for serious radiological consequences to the health and safety of the general public. It was the first one ever declared

at a commercial nuclear plant in the United States. At 10 minutes to 8, the head of the state emergency agency called Governor Dick Thornberg, who was at a breakfast meeting. Thornton. had been sworn in on the 16th of January. He'd been governor for about 10 weeks. He later said that when he first heard the name of the place, he had to think for a moment about where it was. The public found out because a traffic reporter noticed something missing. WKBO and Harrisburg couldn't afford a helicopter, so their traffic man, Dave Edwards, drove a yellow Camaro and listened to a scanner. That morning he heard Middletown police and fire units mobilizing, and he noticed there was no steam coming off the cooling towers at Three Mile Island. He called the news director, Mike Pentech,

who was 27 years old and had grown up in the county. Pentech called the plant on the island directly. Somebody in the control room picked up and said, in Pentech's account, that he couldn't talk now, they had a problem, and to call Redding and talk to them. Reading is where Metropolitan Edison had its headquarters, so Pintech picked up and called Redding. Met Ed's communications people told him the reactor had shut down as prescribed when a malfunction related to a feedwater pump occurred at about four in the morning. They described the general emergency as a red tape type of thing. They told him there was no danger off-site and no danger to the general public. By 25 past 8, it was on the radio. At six minutes past nine, the Associated Press moved its first

dispatch, quoting the state police confirming a general emergency and reporting there was no radiation leak. Metropolitan Edison put out its first written statement at 20 minutes to eight that morning. The whole thing ran two sentences. The reactor had been shut down by a malfunction and would be out of service for about a week. The statement did not mention the general emergency and it did not mention radiation. The rest of that Wednesday reads like two companies talking about two different mornings. At 930, MedEd said no off-site radiation had been found and that they didn't expect any. At noon, a METED draft said there had been no recordings of any significant levels of radiation.

At the same hour, the parent company, General Public Utilities, put out a release saying there had been some low-level release of radioactive gas beyond the site boundary. Fifteen minutes later, a statement from the office of the METED president acknowledged increases of several Millerim per hour at the visitor center at the edge of the site. At half-past one in the afternoon, Med-Ed Vice President John Herbane held a press conference, admitted the fuel had failed, and acknowledged readings of seven millerrim per hour at the site boundary.

That same day, Herbin told the Associated Press about the plant's own workers, that he was sure some of them got exposure, but positively none were overexposed. In the same wire story, a commission spokesman said radiation inside the reactor building was registering a thousand times normal, and another commission man, Joseph Fouchard,

said there was a hell of a lot of radiation in the reactor building. Lieutenant Governor William Scranton III held the state's first press conference just before 11 that morning. His prepared statement said that Metropolitan Edison had informed them there had been an incident, that everything was under control, and that there was and had been no danger to public health and safety.

In that same room, minutes later, a state nuclear engineer named William Dornsife made the first public mention of traces of iodine 131 detected off-site. Scranton came back at half-past four in the afternoon and reversed himself in public. This situation, he said, is more complex than the company first led us to believe, and Metropolitan Edison has given you and us conflicting information. The state's radiation protection director reported measurements taken in downtown Harrisburg. Scranton later testified to the President's Commission about dealing with the utility. He wouldn't say they were exactly helpful, he said, but they were not obstructive.

He said he thought they were defensive. The Commission's own Information Task Force found the specific mechanism of the collapse. The company president was telling some reporters about small off-site releases. At the same time, the public information staff was telling other reporters there had been none. That, the task force wrote, is one reason the utility. lost credibility early in the accident. Two days before all of this on the 26th of March, Med-Ed's president had gone on record saying the company's plants were operated in a way

that places top priority on safety. At 10 minutes to two that Wednesday afternoon, while Herban was telling reporters about seven millerrim at the fence line. Something happened inside the containment building that nobody outside understood for two more days. Instruments recorded a sudden spike in pressure and temperature inside. the reactor building. The building sprays came on automatically, which means the pressure had to have reached at least 28 pounds per square inch. Multiple independent pressure instruments saw it.

The building temperature instruments saw it too. The air intake alarm on a reactor coolant pump saw it. Raghavan's account of how that was received is four words that ought to be famous. Initially, the spike was dismissed as some type of instrument malfunction. But the instruments had been working correctly. All that hydrogen from the zirconium reaction had collected in the containment atmosphere and ignited, and it burned for 12 or 15 seconds inside the building holding the reactor. What proved it afterward was chemistry rather than pressure. When they analyzed the air in that building later, the oxygen had been depleted by several percent. Something in that building had burned it. Rogavan traced how long it took for that understanding to travel. Some supervisors concluded

fairly quickly that too many independent instruments had moved for it to be a glitch. But it was not until late Thursday night that control room personnel became generally aware of what the pressure spike meant, and it became common knowledge among management early Friday morning. Some residents and some reporters have described that event as an explosion that shook the control room and blew doors off their hinges, and there are reactor operator affidavits describing a pressure spike. The commission in the industry describe it as a a deflagration, a burn rather than a detonation, and the technical work supports that distinction. What isn't in dispute is that a hydrogen combustion event happened inside the containment building

at 10 minutes to 2 in the afternoon on the first day. The people on site initially wrote it off as a broken gauge, and that the public was told nothing about it while it was being written off. There's a companion detail from that same morning. Between 8 and 9 o'clock, operators took manual readings from thermocouples that sit at the exit of the reactor core. Some of those readings came back above 2,000 degrees Fahrenheit. 2,000 degrees at the core exit means the fuel is not covered and the cladding is reacting. The readings weren't reported to the commission. Victor Stellow, who ran the commission's inspection office, later explained why in six words. They just didn't believe the readings were real. Gary Miller, the station manager,

testified about his own state of mind that morning. He said he didn't believe in his mind that he really believed the Corps had been totally uncovered, or uncovered to a substantial degree at that time. Half the fuel had already melted. Thursday was almost quiet by comparison. Met Ed said a coolant pump was running normally. Herbane and the company president held a press conference at a motor lodge in Hershey, where Herbane said failed fuel was 1% at most and allowed that he wouldn't rule out human error.

On camera that day, Herbane said it's possible to have fuel failure, and what's not possible is the hypothetical accident where the entire core melts and spews molten radioactivity into the area for miles around and kills tens of thousands of people. That, he said, is what we've been telling you is not possible. Asked why radiation was escaping at all. He said that radiation is allowed to be released from the reactor plant in the course of normal operation. Thornberg told people there was no cause for alarm and no reason to disrupt their daily routine. At a joint press conference that night, a commission official from the

regional office said the danger was over. Then came Friday, and Friday was different. At 10 minutes past seven on Friday morning, an operator at Unit 2 named James Floyd made a decision to vent the makeup tank to relieve pressure in it and preserve coolant for later use. The Kimmany technical task force describes the action in exactly those terms. The main report describes the result. It allowed radioactive material to escape into the auxiliary building and then into the air outside. The technical report adds a line without comment. This vent was left open for days. The state was not told in advance. At about four minutes to eight a helicopter measuring above the

plant recorded a thousand miller rim per hour. Five minutes later at a hundred and 30 feet above the Unit 2 vent stack. It recorded 1,200 millerrim per hour. That number needs to be understood correctly because it's been misused in both directions for 47 years. 1,200 millarum per hour is a measurement taken inside the plume, directly above the stack, in the air. It's a measure of what was coming out of the building. It is not a measurement of what anybody standing on the ground received. Nobody was standing up there in it. It was still a real measurement. and that morning it collided with something else. At around the same time in Bethesda, Maryland,

a commission staffer named Lake Barrett was briefing senior officials. He had assumed a release rate and calculated, on the spot, a hypothetical dose rate at ground level. His number came out at 1,200 millerrim per hour, and at that moment, somebody in the room announced that a reading of 1,200 millerrim per hour had just come in from 3 Mile Island. Barrett told the commission afterward that it was the exact same number, and that it arrived within maybe 10 or 15 seconds of his own prediction.

So there were two identical numbers in that room. One of them had been measured in the air 130 feet above a stack. The other had been calculated for people standing on the ground. Nobody in the room caught the difference in time. The technical task force wrote it up afterward in language that is careful and damning at once. The coincidence of the two identical numbers, coupled with an apparent unawareness that the reported measurement was made from a helicopter in the plume above the plant, contributed to an erroneous conclusion regarding the severity of the situation.

And so the commission staff decided to recommend an evacuation. Harold Denton directed a man named Harold Collins to call Pennsylvania. Collins telephoned Oren Henderson at the state emergency agency and recommended evacuating people as far as 10 miles down wind. The commission's own report contains the phrase that shows where the 10 came from. apparently selecting the distance on his own. An hour later, the chairman of the Nuclear Regulatory Commission, Joseph Hendry called Governor Thornburg and told him no evacuation was needed

and suggested he asked people within five miles downwind to stay indoors for the next half hour. Inside of about 60 minutes, the state of Pennsylvania received an evacuation recommendation from the commission's staff and a no evacuation recommendation from the commission's chairman. Thorneberg took the smaller action and issued a shelter advisory. Then, in a second call around 20 minutes to noon, Henry said something to him that changed the day. He told the governor that if his own wife were pregnant and he had small children in the area, he would get them out. Stay tuned for more disturbing history. We'll be back after these messages.

At half past 12, Thornberg issued the advisory, based on the advice of the chairman of the Nuclear Regulatory Commission, and in the interest of taking every precaution. He was advising those who might be particularly susceptible to the effects of radiation. That is, pregnant women and preschool-aged children, to leave the area within a five-mile radius until further notice. He ordered every school inside that radius closed. The advisory covered roughly 5,000 people, about 144,000 left. That gap is the whole communication story in one number.

The state asked 5,000 people. people to go. 29 times that many packed the car. Research done for the commission afterward put the figure at roughly 144,000. Around 40% of everyone living within 15 miles left. Advisory or no advisory. In Goldsboro, across the river in York County, a borough councilman estimated by Saturday evening that 90% of the town was gone. The state had prepared a place for them. Ten miles from the plant, the Hershey Arena was designated as the reception center. The Red Cross put in cots and televisions. The Hershey Medical Center set up a medical station that could handle labor and delivery.

Hershey was told to prepare for 25,000 people, then 20,000. Over the entire five-day emergency, 186 people stayed at the arena. Everybody else went and stayed with relatives. 144,000 Pennsylvanians made their own arrangements and drove. and almost none of them used the shelter the state built. And the honest reason for it is that by Friday, 30th of March, 1979, the people of the lower Susquehanna Valley had stopped believing anything an official told them. At 11 o'clock that Friday morning in the American Legion Hall in Middletown,

John Herbine held the press conference that ended his career. He described the morning's radiation release as being around 300 to 350 millerrim per hour. The reporters in that room already had the 1200 figure. When they put it to him, he said he hadn't heard the number 1,200. Then somebody asked why Met Ed had vented without telling anyone. And Herbond said the sentence that follows him through every book, written about this accident. I don't know why we need to tell you each and everything that we do specifically. The President's Commission wrote the epitaph.

It was that remark that essentially eliminated any credibility both he and Med Ed had left, with the press. By half past five that afternoon, the company was issuing updates saying reports of a meltdown were unfounded. On Saturday morning at his last press conference, Herbane told reporters the hydrogen bubble was not explosive, mentioned the pressure spike,

and disclosed that four workers had been over-exposed. He also said that one of the things the people who live around the plant have to recognize is that we have to get on with our jobs. On Saturday, a presidential assistant named Jack Watson called the president of general public utilities to say that the conflicting statements coming out of Pennsylvania

were making things worse, and to propose that Harold Denton alone brief reporters on the technical situation. The company agreed to the arrangement. From that point forward, the utility was not the voice of the accident anymore. Friday evening, Chairman Joseph Hendry gave Roger Mattson, who ran the commission's division of system safety, a specific job. Find out how fast oxygen is being generated inside that reactor vessel.

And tell me whether the gas bubble sitting in the top of it can blow up. The underlying physics question was a real one. Radiation passing through water split some of it into hydrogen and oxygen. A process called radiolysis. Hydrogen on its own will not burn. Hydrogen mixed with enough oxygen will. There was a large hydrogen bubble in the dome of that vessel, left over from the zirconium reaction and if oxygen were accumulating alongside it the mixture would eventually reach a flammability limit. Mattson ran two teams through the

weekend and called in outside laboratories. Saturday around one in the afternoon he had preliminary answers from four independent sources all with known credentials and they suggested a matter of several days before any combustible mixture could form. At around half-past three he told the commissioners he thought the two to three-day estimate was conservative. Saturday night, consultants came back saying the oxygen percentage was on the threshold of the flammability limit. Sunday, Victor Stella went through the assumptions Mattson's teams had used and concluded they were wrong. The two men had an argument about it at the airport.

Stella turned out to be right. The Chemine Technical Task Force put the correction in a single sentence, and it's a sentence anybody can check against a chemistry textbook. Only one-tenth of a cubic centimeter of hydrogen per kilogram of water will suppress the formation of oxygen. And the hydrogen concentration in the reactor coolant at Unit 2 was about 200 times that level. The water in that reactor was already saturated with dissolved hydrogen. Any oxygen produced by radiolysis recombined with it immediately. Free oxygen could not build up. There was never a combustible mixture, and there was never going to be one.

And the task force stated it flatly that no substance. such explosion was possible. Hendry told a Senate subcommittee on the 10th of April that the possibility of a flammable mixture turned out to have been a misplaced concern. Mattson on the 2nd of May said there was never any danger of a hydrogen explosion and that they had asked the staff the wrong questions and that it was a very bitter pill. To a reporter that same day, he was shorter. We filed up, he said.

The amount of concern was entirely undeserved. There never was any danger. The correction never reached the people who needed it. The Kimmany Commission wrote that the groundless nature of the fear never penetrated the public consciousness afterward, partly because the commission made no effort to inform the public that it had aired. Where I come down on that is fairly simple. The scare was real inside the agency for about 36 hours,

and it was a mistake made by serious people under pressure. And I don't think there's anything scandalous about getting a hard question wrong at three in the morning. The scandal is the silence afterward. They corrected the science and never corrected the record. And for 47 years, the hydrogen bubble has lived in the public memory as the moment Pennsylvania nearly went up, which is not what happened. The commission's own historian Samuel Walker has spent years pushing back on the exaggerated versions of that weekend.

His finding is that reactor experts regarded the bubble as a potential problem that at worst could become a troubling safety issue over a matter of days. and that popular retellings turned it into an imminent detonation. The argument between Mattson and Stella was about whether free oxygen was present in the pressure vessel. It was not about whether they were all about to die. The accident was serious enough without the version we got. What was going on inside the commission that weekend is better documented than most people realize, because the commissioners were recording themselves.

The Washington Post published excerpts from those tapes on the 14th of April, 1979. In them, Mattson says they have an accident that they were never designed to accommodate. The commissioners discuss an idea that ought to be much better known than it is, which was to deliberately cause a different accident, one they understood better than the one they were in. Start up all the reactor pumps, burn them out, blow the seals, and hope they cause a loss of cooling accident. Hendry's assessment of what that would do to the pump seals was that they'd just fall off the head of the dam vessel and give a nice six-inch diameter small break. Commissioner Victor Galinsky, listening to the proposal to manufacture a controlled disaster,

said he found that hard to believe, and that he just didn't think these sorts of things work out. There's a line from those same tapes, quoted in Ernest Sternglass's book in widely repeated sense, in which Hendry describes himself and the governor as a couple of blind men staggering around making decisions. I've seen it in a dozen retellings. I could not confirm the existence.

exact continuous wording against the original transcript, and the book that carries it is written by a committed opponent of nuclear power. So take the phrasing as approximate. The substance of it is not in doubt. Hendry said on that tape that they were operating totally in the blind. On Sunday the 1st of April, the president of the United States walked into the control room of unit one with his wife. Jimmy Carter had been a naval officer in the nuclear program under Hyman Rickover. He had studied nuclear physics at Union College. In early 1953, he was a 28-year-old lieutenant who took a team of men to Chalk River in Ontario to help with the cleanup after a reactor accident there. The legend around that has grown considerably, and it's worth trimming. Carter arrived months

after the accident, not during it. He was not lowered into the reactor. He was not in charge of the cleanup. He did not, in any meaningful sense, save Ottawa. What he asked, actually did was supervise a small crew, disassembling headers and removing equipment above a couple of meters of shielding in two shifts a day of 45 minutes each. That's real work in a real radiation field, and it made him the only president in American history who could read a reactor briefing without an interpreter. Denton briefed him on the hydrogen bubble before he went in. He wore yellow booties taped around the ankles. At three in the afternoon, he spoke at the Middletown Town Hall and set his primary condition.

in coming there was to learn as much as he possibly could and to assure the people of that region that everything possible was being done. He said radiation levels were being very carefully monitored and were quite safe for all concerned and he thanked the governor by name. The mayor of Middletown Robert Reed described the effect afterward People weren't talking to one another, he said, and when the president came, it seemed like everyone came out to see him and it was really a shot in the arm. Whatever anybody thinks of Carter's politics, that visit is the moment the panic stopped. A man who understood the machine walked into the building without a respirator, and the valley took a breath. While that was happening, the federal government was trying to buy a drug that nobody in America had. Potassium iodide saturates the thyroid so it can't absorb radioactive iodine.

The Food and Drug Administration had declared it safe and effective as a preventive measure in late 1978. About three months before the accident, nobody in the country acted on that. At the end of March 1979, there was no stockpile anywhere in the United States, and no pharmaceutical company was marketing medical-grade potassium iodide in the quantities that a threatened population would need. Shortly after 3 o'clock on Saturday morning, the 31st of March, the Mallinckrod Chemical Company in St. Louis agreed to supply roughly a quarter million one-ounce bottles. Malincrot worked with Park Davis in Detroit and a bottle dropper manufacturer in New Jersey, around the clock.

The first shipment reached Harrisburg at about half past one on Sunday morning. The last shipment arrived on Wednesday the 4th of April. The total was 237,0133.13 bottles. None of that supply was ever distributed. Not one dose reached a member of the public. It went into a warehouse where, in the words of one review of the response, it could not be promptly distributed if needed. The advisory had gone out Friday at half-past 12.

The final shipment arrived the day after the crisis effectively ended. There is a claim floating around that the drug was unusable or improperly labeled. I looked for a source and couldn't find one. The verified failure is bad enough. A country with 90 operating power reactors had to phone a chemical company at 3 in the morning and ask them to invent a supply chain. and by the time the trucks arrived, it didn't matter. Pennsylvania's Secretary of Health at the time was a physician named Gordon McLeod.

He had been in the job 12 days. He was not in the notification chain. He found out and had to arrange his own conference call with the state emergency director, the state radiation protection director, and an aide to the lieutenant governor. On that call, he asked whether it might be wise to have pregnant women and children under two, leave the area. The suggestion was rejected Thursday afternoon. He raised it again Friday morning and pressed the governor directly,

and Thornberg's advisory that afternoon covered a wider group than McLeod had proposed. McLeod's two standing complaints afterward were specific and easy to check. Pennsylvania had no stockpile of potassium iodide, and there was no physician on the state body that dealt with nuclear regulation. He also began saying things in public about infant health. He reported that downwind of the plant, babies born with hypothyroidism, went from nine in the nine months before the accident to 20 in the nine months after. He said that in the six months after the accident, 31 infants living within 10 miles of the plant died against 14 in the same six months the year before. The Washington Post covered a related cluster in February of 1980.

13 cases of congenital hypothyroidism among babies born in three Pennsylvania counties during the last nine months of 1979, against an expected number of about three. McLeod said it was impossible to assign a common cause and said he was astonished that the health department had made no announcement and started no investigation. The county breakdown makes the problem obvious. Six of the cases were in Lancaster County, four in Bucks, three in Lee High. Bucks and Lehigh counties are 60 to 100 miles from three mile island. Seven of 13 cases were nowhere near the plume.

McLeod was out of the job before the end of 1979, and he said he was fired for what he'd said about infant health. The exact date and the governor's stated reason are not settled in the sources I could find. What I can say is that the State Health Department's own final report in in 1981 concluded that the accident did not cause local deaths of infants or fetuses and that no subsequent study has supported the infant mortality claim. So there's a health secretary who was frozen out of a radiological emergency in his own state, who was right about the potassium iodide and right about the missing physician, and who then made claims the evidence didn't carry and lost

his job. All of those things are true at once. That's usually how these things actually go. The last piece of that week is the one that, in my opinion, should have gotten the most attention and got almost none. Before the 28th of March, 1979, there was no federal requirement that a nuclear plant have an approved off-site emergency plan in order to operate. There was no such requirement anywhere in the rules. The commission had issued guidance in 1978 proposing a 10-mile planning zone for exposure and a 50-mile zone for contaminated food and milk, and that guidance had not been implemented. The agency's own retrospective description of the plans that existed under the older rules is that emergency plans are vague and sketchy.

What the counties around Three Mile Island actually had, per the Kimini report, were plans to evacuate about 25,000 people living within five miles of the island. Five miles was the outer edge of anybody's imagination. On Friday morning, the state told the counties to prepare 10-mile plans. On Friday night, Harold Denton recommended being ready for 20 miles. And then, in the commission's own words, suddenly six counties were involved in planning for the evacuation of 650,000 people, 13 hospitals, and a prison.

Kevin Malloy ran emergency preparedness for Dauphin County. The commission recorded the specific problem the expansion handed him. Extending the zone meant hospitals, which he had not confronted earlier, because there were no hospitals within five months. The plan stopped at five miles, so nobody in Pennsylvania had ever thought about how to move a hospital. They wrote that plan on Friday night. It hit the newspapers on Sunday. On the 22nd of May, 1979, less than two months after the accident, a Unit 2 control room operator named Harold Hartman, sat down with investigators from the Nuclear Regulatory Commission and told them that Metropolitan Edison

had been falsifying reactor coolant leak rate data for months before the meltdown. and that management knew about it. This is the part of the Three Mile Island story that almost nobody outside Central Pennsylvania has ever heard, and it's the part that turns an accident into something else entirely. A leak rate test is a housekeeping requirement. The plant's technical specifications said

that unidentified leakage out of the reactor coolant system had to stay below one gallon per minute. Operators ran a test on the plant computer, typically over an hour, and if the number came back too high, The rules required the plant to shut down and find the leak. Unit 2 was leaking more than it was supposed to, but in plain terms, shutting down to fix it costs a utility a great deal of money.

A former Unit 2 operator named Martin Cooper described the shift culture in a sworn statement to a commission board in 1986. His shift, he said, typically performed leak rate test early in the shift because they tried to obtain a successful test every shift. If a test came back outside the limits, the practice was to run another one. And if the second came back inside the limits, his shift would throw away the first and keep the second. They rounded off unidentified leakage below one and a half gallons a minute, so the test was deemed to be within specification.

Nobody had much faith in the calculation anyway, he said, because the results were erratic. Cooper denied the two specific manipulations that were alleged, which were adding hydrogen or adding unlogged water to the makeup tank during a test to make the numbers come out right. Hartman's account was that supervisors instructed him to manipulate the reports to get a good leak rate, that the tests were run day and night, and that only the acceptable ones went to the regulator. What the agency did with that comes next.

Hartman came forward on the 22nd of May, 1979. The commission did not open a formal investigation until, March of 1980. That is a delay of 10 months. It opened one then only because Hartman went on television on the 24th of March, 1980, on a program broadcast out of New York, and said it all in public. The office that sat on it was the Office of Inspection and Enforcement, run by Victor Stello. The commission's own office of inspector and auditor later documented that Stello and another official told Commissioner John Ahern in October of 1979 that they had to be. had no indications of criminal violations. That was five months after Hartman's statement.

Stello signed the final enforcement report that left Hartman's allegations out of it. Ten years later, when President George Bush nominated Stello to run the nation's nuclear weapons plants, reporters found out that a federal grand jury had looked at Stello himself over an alleged cover-up of the falsification. Stello himself was never charged with anything. He said the grand jury cleared him. Stay tuned for more disturbing history. We'll be back after these messages. What the grand jury materials produced instead was a set of assessments from the prosecutors that I think stand as the harshest thing any federal lawyer has ever written about that agency in an internal document.

An assistant United States attorney named Rhonda Fields wrote in December of 1985 that a review of the materials put the commission's management practices in serious question as to their competence, judgment, and in some instances their truthfulness and ethics, and that questions remained about why the commission did not act at certain times. David Queen, the lead Justice Department prosecutor, was less diplomatic. He said he could not stand silent and allow the charade carried on by the commission to be treated as anything but that. He said they were the only institution that had made the slightest damn effort to see the thing through.

He said the commission had not conducted any meaningful investigation. Meanwhile, in May of 1983, at a public meeting of the five commissioners, the agency's own director of engineering for the region containing Three Mile Island, got up and said it out loud. I can tell you for a fact that the records were falsified. The president of general public utilities objected at that meeting to the suggestion of falsification and said the Justice Department had found nothing after three years of investigation. The indictment came six months later.

On the 7th of November 1983, a federal grand jury in Harrisburg indicted the Metropolitan Edison Company on 11 counts. The charges covered falsifying tests that showed whether excessive water was leaking from the cooling system, systematically destroying the records of those tests, violating the plants operating license, violating federal regulations, and making false statements to the Nuclear Regulatory Commission. The period charged was October of 1978 through the 28th of March, 1979. That date range runs right up to the morning of the accident and stops there, because that's the morning the reactor stopped. The Philadelphia Inquirer called it the first prosecution of a utility under the Atomic Energy Act. On the 29th of February, 1984, Metropolitan Edison pleaded

guilty to one count and no contest to six more. The company paid, a $45,000 fine and put a million dollars into a fund for emergency preparedness in the surrounding communities. The bulletin of the atomic scientist recorded what that made them, the only nuclear utility in the nation ever convicted of a criminal offense. And the conviction had nothing to do with melting the core. It was for the paperwork that kept the core running during the five months before it melted. Not one individual was ever prosecuted for the leak rate falsification. The only defendant on those 11 counts was the corporation. One man at Three Mile Island did go to federal court though, for something else. James Floyd, the Unit 2 supervisor of operations, the same man who

ordered the venting on Friday morning, was indicted in June of 1984 and convicted that November of cheating on his commission licensing examination by having another employee complete portions of it. The company had already been fined $140,000 for submitting false statements about that certification. Then there was a second cheating matter and people mix these two up constantly, so they're worth separating. In July of 1981, while hearings were underway on whether to restart the undamaged unit one, it came out that operators had cheated on re-qualification examinations. The hearings reopened under a special master named Gary Milholland, whose report found cheating and wrongdoing, found the company had submitted material false statements and urged

prosecution. When the retake examinations were given, more than a third of the operators failed them. The licensing board reversed much of the special master's findings and recommended restart anyway. In January of 1985, Governor Thornburg called for the removal of the judge running those hearings, saying he had gone out of his way to prove himself incapable of fair and impartial decision-making. The trigger was a letter that judge had written to the federal judge sentencing James Floyd, asking for leniency. The commission voted three to two against further hearings. Commissioner James Asselstein wrote in dissent that by its decision,

the commission had violated the trust of the people of central Pennsylvania. On the 29th of May, 1985, the commission voted four to one to let Unit 1 restart. Asselstein cast the single vote against. The governor of Pennsylvania opposed it. Both United States senators from Pennsylvania opposed it. More than 200 people demonstrated, and over 80 were arrested. A statewide referendum three years earlier had gone two to one against restarting.

Unit 1 came back online in October of 1985 and ran, by most measures, very well, for another 34 years. There was still the matter of what was inside Unit 2. The building had been sealed since the accident with about 43,000 curious, of Krypton 85 floating in its atmosphere. Krypton 85 is a noble gas. It doesn't stick to filters. It doesn't react with charcoal. And there is no practical way to capture it.

The only options were to let it sit forever, or let it out. The commission decided to let it out. A federal appeals court cleared the way, and on Sunday the 29th of June, 1980, they started venting. The first attempt lasted four minutes before a monitor mistook Krypton for particular. and shut it down. Harold Denton's comment on that was that the monitor had seen Krypton and thought it was

particulates, and that it just seemed like everything done there was a special operation. The reaction in the valley was about what anybody would expect from people who'd been through the previous spring. Three Mile Island Alert, the local watchdog group, reported that 43 of 91 callers to its hotline said they'd leave during the venting. A pregnant resident named Andrea Hand told a reporter it was fear, that she was pregnant and didn't trust the authorities. A pro-nuclear group calling itself friends and family of Three Mile Island handed out refreshments to the reporters covering it. The venting itself worked as

planned. On the 23rd of July 1980, two metropolitan Edison workers named William Burrell and Michael Benson became the first human beings to enter that reactor building in 16 months. They wore breathing apparatus, a diver's dry suit under a rain suit. Several, pairs of rubber gloves, firemen's boots, and throw away hard hat lights. They stayed inside for about 20 minutes. They took 29 photographs and six swipe samples and carried out a five-gallon bucket of whatever was on the floor. Each of them picked up about 190 miller rim. What they found was a building that looked fine. There was one damaged stairwell door. Rust and dirt on the floors, water marks from the building sprays, cesium on the surfaces.

Nothing they saw told them the core was gone. The basement under about six feet of contaminated water with a ring on the concrete walls where it had stood wasn't explored for years. The camera went into the vessel in 1982. The head came off in July of 1984. The plenum came out in May of 1985. Defueling started that October and ran into 1990, and they took out something like 100 tons of damage fuel in 340-odd stainless. steel canisters. In July of 1986, they started drilling core boars straight down through the debris to get cross sections. In October, they brought in a flat-faced drill to break up a solid

mass in the middle. By May of 1987, the cavity was cleared except for a ring of solidified material around the edge that the crews called the donut. The numbers that came out of that work are the ones I read at the top. At least 45% of the core melted. Idaho National Laboratories analysis puts peak temperatures at a minimum of 2,800 degrees Celsius, which is over 5,000 degrees Fahrenheit. Roughly 20% of the core material escaped the core boundary as liquid and solidified in what the report describes as lava-like formations. About 19,000 kilograms of it ended up on the lower head.

The international investigation of that vessel found the hot spot. An elliptical region roughly 80 centimeters by 100, which held around 1,100 degrees Celsius for as long as half an hour before it cooled. Away from that patch, there's no evidence the vessel steel ever got above 727 degrees. Four things saved it, according to the Idaho analysis. The hot region stayed relatively small. Crusts formed fast against the cold vessel surfaces and protected the instrument nozzles that penetrate the bottom head. The molten material froze inside those nozzle gaps before it could reach the outside.

and the debris bed turned out to be coolable. The Rovevin counterfactual on this is short. If that block valve had stayed open and the operators had done nothing, then within 30 to 60 minutes a substantial portion of the fuel would have melted, perhaps as much as half of all of it. Rogavan also judged, and this cuts against the popular version, that even in a full melt, the most likely outcome was that the containment building would have held. The federal investigations, the industry,

and the anti-nuclear movement. Everybody spent the early 1980s arguing about what almost happened, and it took eight to ten years, and a camera and a drill before the field understood that the real near-miss was three feet of steel in the bottom of a vessel that nobody was looking at. Now that brings us to the argument that's never ended. The official number comes from an interagency group

that pulled together the Department of Health, Education, and Welfare, the Environmental Protection Agency, the Nuclear Regulatory Commission, and the Department of Energy in May of 1979. Their finding was that about 2.5 million cures of radioactive noble gases, and about 15 cures of radioiodines were released, and that this produced an average dose of 1.4 millarrem to approximately 2 million people in the area. The maximum dose to a person standing at the site boundary was put at under 100 milarem. For scale, natural background radiation in that part of Pennsylvania runs somewhere around 100 millerrim a year. 1.4 millerrim is roughly what a person adds by moving from a wood frame house to a brick

one. That number sits very badly next to what people in Middletown and Goldsboro and Eddard said happened to them. A woman named Jean Tremmer in Lisburn, about 10 miles west, said that on the evening of the 30th of March, she was hit by a blast of heat and rain. and afterward her skin went red and itchy as though she'd been badly sunburned. Three weeks later her hair turned white and fell out. A man named Bill Peters in Edders said he had a severe sunburn and a burning throat

the day after it started and that he could taste metal in the air. He described diarrhea, nausea, blisters on his lips and inside his nose. He said that when he came back from evacuating, he found his German Shepherd dead with burnt white eyes. and four of his five cats dead the same way, and the fifth one blind. Those accounts come from Sue Sturgis at the Institute for Southern Studies and through a 1982 book by Harvey Wasserman and Norman Solomon. Both advocacy sources. They're also named people making specific claims about their own bodies and their own animals,

and there were a lot of them. The farm accounts go on for pages, a woman less than three miles from the island whose nanny goat aborted twins, eight days after the accident, and who lost two more pregnant nannies, 26 newborn rabbits and 19 guinea pigs. A man in Newberry Township directly across the river who said 27 chickens and 11 hogs were dead by mid-May, that his corn came in mushy and half-formed, and that the bark peeled off a 23-year-old walnut tree. A dairy farmer with two cows aborting and 10 calves dead after birth.

A farmer in Edders who said he used to have 25 robins on his place, and counted one in 1980. A veterinarian named Robert Weber, who had practiced in Mechanicsburg since the 1940s, testified to the Pennsylvania Public Utility Commission in March of 1980 that he was seeing still births in pigs at two a week against a normal rate of two a year and performing two cesarean sections a week on goats and sheep.

In February of 1981, with the plant shut down, he said that there were no down cows or animals with hypertension anymore, and that he hardly got a call to go over there. The Pennsylvania Department of Agriculture surveyed 100 farmers in May of 1980 and reported that five of them had problems. A Baltimore newspaper investigated that survey and called it worthless, reporting that at least 40% of farmers had complained, that 35 said their views were misrepresented,

that three said they reported problems and were logged as having none, and that some of the people surveyed weren't farmers. I couldn't independently verify that investigation. And it comes from the Wasserman and Solomon book, so I'd weigh it accordingly. The federal study released in November of 1980 concluded the radiation did not cause the animal deaths or deformities and attributed them to nutritional deficiencies and infectious disease.

Its strongest argument is a good one. If radiation had done this, there would have been many anemic animals, more frequently, in the path of the plume, And there weren't. This is the hard center of the whole dispute. The symptoms people reported, the metallic taste, the reddened skin, the nausea and vomiting and hair loss are the recognized signs of acute radiation syndrome.

Acute radiation syndrome requires doses on the order of 100,000 millerim. The official maximum off-site dose is under 100. That's a factor of 1,000. So either the official dose calculations are wrong by 30,000. three orders of magnitude, or those symptoms had a different cause. Both sides have known that for 45 years. Nobody has ever managed to close it. At a commission meeting marking the 25th anniversary, the agency's own historian said on the record that one stack monitor went off scale during the release, because it was calibrated at a very low level, and that there were 20 monitors placed

around the site before the accident in accordance with the regulations. That is 20 fixed monitors for a region holding 2 million people. The Environmental Protection Agency didn't arrive until the 1st of April, four days in. They set up 11 stations, then 31 within two days. The peak releases had already happened, and there's one documented instance of a measurement simply disappearing. At 10 minutes past 2 in the afternoon on Thursday the 29th of March, a helicopter at 3,000 feet over the plant stacks recorded 3,000 millerrim per hour. That's more than twice. the famous Friday reading. A radiation specialist named Michael Slobodian confirmed that the data was telephoned to commission headquarters in Bethesda, and in his words, was never heard again.

It did not become public until the Washington Post reported it in August of 1979. Victor Stello blamed communications, said they were terrible that day, and said the exposure received by the public was really not much. Again, a reading taken 3,000 feet up in a plume is not a ground-level dose, and anybody who claims otherwise doesn't understand the measurement. But a peak reading that gets phoned into the agency and then vanishes inside it for four months is a fact about the record-keeping, and the record-keeping is what every dose calculation ever made about this accident rests on.

The epidemiology is genuinely contested, and it's contested in a more interesting way than either side usually admits. The settlement money paid for the first study. In 1981, the operators settled the class action for $25 million. $20 million for economic harm within a 25-mile radius and $5 million to create the 3-mile Island Public Health Fund. That fund paid for a Columbia University team led by Maureen Hatch to look at cancer among the roughly 160,000 people living within 10 miles. Their 1990 paper found that new cancer diagnoses in that area went, from 1,722 in the five years before the accident to 2,831 in the five years after.

That works out to a 64% increase, and they concluded that the pattern of results does not provide convincing evidence that radiation releases from the plant influenced cancer risk during the limited period of follow-up. They thought changes in cancer screening were the likelyer explanation, with stress as a possible contributor. They didn't say the accident. caused no cancer. They said the pattern didn't convincingly show that it did. In 1997, an epidemiologist at the University of North Carolina named Steve Wing reanalyzed the same data and published his

results in environmental health perspectives. He found positive associations with estimated dose for all cancers, for lung cancer, and for leukemia. When Hatch and her colleagues replied, they confirmed that Wing's replication of their numbers matched theirs within rounding error. The two teams had the same arithmetic and disagreed about what it meant. Hatches group argued Wing had misrepresented their conclusion and had leaned on anecdotal symptom reports. Wing argued that the original team had started from an assumption about small releases and let it drive the interpretation.

Wing also said something about the dosymmetry that I think is the most useful sentence in the entire debate. The dose estimators, he said, could not say whether it was one or ten or a hundred. because of the limited information about the releases. So they built a relative scale, ranking census tracks from lowest to highest, instead of absolute doses. That is what the science actually had to work with.

The University of Pittsburgh ran the other major cohort, using a registry the state built by interviewing more than 32,000 people, within five miles in the weeks after the accident. Evelyn Talbot's mortality studies through 1998 found no significant overall increase in cancer deaths, with a slight elevation in lymphatic and blood cancers. Her own summary is that the radioactivity released does not appear to have caused an overall increase in cancer deaths. Then in 2011, the Pittsburgh Group published a cancer incident study on that cohort. All cancers combined showed no elevation. But they did note that leukemia in

men was shown to have increased. Their conclusion says the increased risks were small and mostly not statistically significant and then says that additional follow-up is warranted, especially to explore the increased risk of leukemia found in men. That sentence is in the literature, but it rarely makes it into the summaries. At Penn State, an ear-nose and throat surgeon named David Goldenberg has published a series of papers on thyroid cancer around the plant. Stay tuned for more disturbing history. We'll be back after these messages.

A 2012 study of more than 26,000. thousand thyroid cancer cases found no statistically significant difference in incidents in the three mile island area then in 2017 his group looked at the molecular profile of 44 papillary thyroid cancers and found that patients in the at-risk group had a much lower rate of the b r a f mutation than controls along with other markers associated with radiation induced thyroid cancer That pattern is what radiation-associated thyroid cancer characteristically looks like. Goldenberg's own statement of what that means is careful, and I'm going to quote it rather than paraphrase it, because paraphrasing it is how it gets abused.

While no single marker can determine whether an individual tumor is radiation-induced, he wrote. These data support the possibility that radiation released from Three Mile Island altered the molecular profile of thyroid cancers in the population surrounding the plant. That word possibility is doing real work there, and the sample was 44 patients. He also named the reason this whole field is stuck, which is that the population around the plant is small. The required population for detecting low-level radiation effects is large, and people move. The committee commission said the same thing in 1979 in blunter terms. Among the two million people within 50 miles, they wrote, there is no conceivable statistical method by which fewer than 10 additional deaths would ever be detected.

That's the commission conceding that its study design cannot find the thing it is routinely cited as having ruled out. Absence of detection isn't detection of absence, and they said so themselves. And in 2023, a scoping review co-authored by Talbot and Goldenberg together, which is to say by both ends of this argument, listed the major limitations of 40 years of research. They listed limited radiation exposure assessment and exclusion of vulnerable populations. That assessment comes from the epidemiologists themselves in a peer-reviewed journal, and not from anybody's advocacy group. The people who believed they'd been hurt got their day in court, and then they didn't.

About 2,000 individual personal injury claims were consolidated before United States District Judge Sylvia Rambo. 10 were chosen as trial plaintiffs. Their diagnoses are in the court record. Acute lymphocytic leukemia in one of them and chronic myelogenous leukemia in two more. Thyroid cancer in a fourth and a herthal cell carcinoma in a fifth. An osteogenic sarcoma of the right leg. A breast cancer brought by an estate. An adenocarsinoma of the ovaries and a bladder cancer. In 1995, the Third Circuit had established the framework. Federal regulations set the standard of care, and a breach happens when releases exceed the federal limit at the plant boundary. And in that

proceeding, the defendants conceded that radiation levels at the three-mile island boundary had exceeded permissible federal standards after the accident. That concession almost never gets mentioned. The operators admitted in federal court that they went over the off-site limit. It didn't matter, because the case turned on dose. In June of 1996, Rambo granted summary judgment. Her reasoning was that the scientific literature shows no established dose response relationship below 10-rim, which is 10,000 miller-rim, so a plaintiff had to show at least that much exposure to make out a case on causation. The evidence, taken in the light most favorable to the plaintiffs, showed four of them under 10 miller-rim,

five between 15 and 25 and 1 at a maximum of 75. The highest estimate among the 10 was three quarters of 1% of the threshold. She wrote that the plaintiffs had failed to present evidence sufficient to create a material factual dispute on the issue of dose, and that the paucity of proof alleged in support of their case was manifest. She noted they'd had nearly two decades to muster evidence. She sanctioned their lawyers for missing deadlines. And then she wrote a sentence that I don't think a judge writes unless it's bothering her. Thousands of individuals who believe that they have suffered adverse medical effects, she wrote,

will not have an opportunity to have their claims heard by a jury. The Third Circuit affirmed as to the ten trial plaintiffs in 1999, revived the roughly 2,000 others on the grounds that their individual evidence differed, and then the whole thing ended in the early 2000s when the appeals court upheld summary judgment against the rest. 17 businesses along Route 30 were the last claims standing. The Amuts went a different route. Marjorie Amit, a psychologist and her husband Norman, an engineer, went door to door through three hilltop neighborhoods they believed sat in the plume path.

Among roughly 450 people, they counted 19 cancer deaths between 1980 and 1984, which is many times the expected rate. In June of 1984, they petitioned the commission to investigate, whether the releases had been larger than reported. The Commission rejected it in December, relying on an analysis by the Centers for Disease Control. Its objections were that the interview methods were unsystematic and prone to bias, that there were no control groups,

that the cancer deaths were unverified, and that the latency between exposure and diagnosis was too short by accepted medical standards. Those are fair objections to a door-to-door survey run by two residents. The Third Circuit refused to disturb it. The court did not find that the commission's refusal to reopen the record was arbitrary or capricious. So the survey was rejected because it wasn't science, and the science was constrained by dose estimates that its own authors said could be off by a factor of 10 or 100, and the lawsuits died because nobody could prove a dose. Every road in this story runs back to the same missing thing, which is a reliable measurement of what came out of that,

that building between Wednesday and Friday. There were 20 fixed monitors, and one of them went off scale, and a peak reading got phoned in and lost. I don't know what dose those people got. Neither does anybody else, and that's the answer nobody wants. The cleanup took until December of 1993 and cost about $973 million, and more than a thousand skilled workers went through that building. The federal government paid a share, the industry paid a share, and the company, that owned it very nearly went under. They processed 2.23 million gallons of accident water

and evaporated it between 1991 and 1993. Over objections, finishing six months late. The last of it put about 658 cures of tritium into the air over central Pennsylvania. The damaged fuel went to Idaho. 82 metric tons of heavy metal, the remains of 177 fuel assemblies, repackaged into 29 carbon steel canisters inside 29 concrete storage modules at a site in Butte County. Under a settlement agreement with the state of Idaho, the Department of Energy is required to have all of it out of Idaho by the 1st of January, 2035. As of the last public status I could find, the facility and the canister design needed to do that. Don't exist yet. About 1% of the Corps never left Pennsylvania. Less than 1,125 kilograms of fuel-bearing material is still inside Unit 2, most of it in the basement,

in what has been described as the highest radiation environment of any retired nuclear plant in the country. Crews trained on mock-ups before they started going after it with remote equipment. The plan filed in October of 2025 puts that work finishing in 2029, the buildings coming down in the early 2030s, and the site filed. closed out around 2052. Meanwhile, a few hundred yards away, Unit 1 is coming back. It shut down in September of 2019 for economic reasons, after the legislature declined to subsidize it. There was nothing wrong with the machine. In September of 2024, Constellation announced a 20-year power

agreement with Microsoft and a plan to restart the reactor under a new name, the Crane Clean Energy Center, at a cost of around 1.7. billion dollars. The Nuclear Regulatory Commission approved storing fresh fuel on the island in July of this year. The company's most recent guidance is a restart in the second half of 2027. So on one island at the same time, one reactor is being taken apart because of what happened in 1979. And the reactor next to it is being rebuilt to run until at least 2054. I keep going back to those two valves. The emergency feed water block valves 12A and 12B closed on the morning of the 28th of March 1979 when the plant's own rules required them to be open. Three separate bodies went looking

for an answer. The utility investigated it, and so did the Nuclear Regulatory Commission, and so did the President's Commission. Roggevin's finding is one sentence, and it's the last word anybody has written on it. The Special Inquiry Group has not been able to determine how the block valves got to be closed that morning. Both federal investigations concluded, those eight minutes didn't change the outcome, and I have no reason to doubt them. The Corps was coming apart because of a relief valve at the other end of the plant. But that's the piece of the file I keep opening. This is the most examined industrial accident in the history of the United States. Something like 6,000 pages of federal reports came out of it, along with a grand jury and two

decades of litigation, in a core that was photographed and drilled and taken apart in pieces and shipped across the country in stainless steel. And nobody in 47 years has ever been able to say who shut those two valves.