What is not in dispute
The Nuclear Materials and Equipment Corporation ran a plant at Apollo, Pennsylvania, processing highly enriched uranium under contract. Its president was Zalman Shapiro, an American chemist of considerable standing.
In November 1965 the Atomic Energy Commission audited the plant's books and found that 178 kilograms of uranium-235, as highly enriched uranium, could not be accounted for since operations began. Of that, 93.8 kilograms could not be attributed to any known loss mechanism — not to waste streams, not to residue, not to anything the plant could point at.
Department of Energy records compiled later make Apollo the worst case in the country. NUMEC's inventory loss was the largest of any commercial site in the United States: 269 kilograms before 1968, and a further 76 kilograms after.
Those numbers are the whole reason this case exists. A loss of that size is not a rounding error in a plant handling weapons-grade material, and nobody has ever argued it was.
The visit
On 10 September 1968, Rafi Eitan arrived at Apollo with three other Israeli nationals.
The Atomic Energy Commission's security office had cleared the visit on NUMEC's assurance that Eitan was a chemist and his companions were electronics specialists. He was not a chemist. From 1963 to 1972 he was deputy chief of operations for Mossad, and in 1968 he was running a covert operations unit on special assignment to LAKAM, Israel's scientific intelligence bureau — the organisation whose job was acquiring exactly this kind of material. He is better known now as the man who ran Jonathan Pollard.
That the visit happened, that it was approved on a false description, and who the visitor actually was are all documented. What the visit achieved is not.
What the CIA told the regulators
In 1976, about fifteen people sat in a conference room at the Nuclear Regulatory Commission's headquarters in Washington to hear from Carl Duckett, the CIA's deputy director for science and technology.
Duckett told them the Agency had concluded, in the late 1960s, that Israel had the uranium from Apollo and was using it to build weapons.
This is the hinge of the affair and it is worth being exact about what it is and is not. It is a senior intelligence official stating his agency's assessment to regulators, which is a long way from rumour. It is not a finding of fact, a court judgment, or a disclosed chain of evidence. The CIA believed it. The CIA has believed things.
Supporting it, and arriving later in the public record, is environmental sampling near Dimona reported as showing highly enriched uranium with a signature associated with the Portsmouth, Ohio enrichment stream that fed NUMEC.
The other explanation, which also has physical evidence
The case is usually told as suspicion against denial. It is better understood as two explanations, each with something solid underneath it.
Shapiro's account, which he gave to Congressman Morris Udall in 1978 and never departed from, was that the uranium was lost the way material is lost in a plant that handles it badly: into air ducts, into cement, into wastewater. He was never charged with anything, and he died in 2016 denying diversion.
What makes that more than an assertion is what is in the ground at Apollo and Parks Township. Radioactive and chemical waste from the plant was buried there in ten trenches between about 1960 and the early 1970s — some 33,000 cubic yards of it. The Army Corps of Engineers has been excavating it at a cost that has run to roughly $400 million and upward, around ten times the original 2002 estimate.
That is the argument stated in dollars. An enormous quantity of contaminated material demonstrably went into the building, the plumbing and the soil of that site, and the nation is still paying to dig it out. Some fraction of the missing inventory is in those trenches, and the only question is what fraction.
What was never done
No one was prosecuted. The FBI investigated; the Atomic Energy Commission investigated; the matter was reopened under the Carter administration; and no charge was ever brought against Shapiro or anyone else.
That absence is read in both directions, and both readings are available on the record. It is offered as evidence of a cover-up — a case killed because pursuing it would have meant publicly accusing an ally of stealing weapons material. It is equally consistent with prosecutors looking at an accounting discrepancy at a filthy plant and concluding they could not prove where the material went, because they could not.
How much uranium that is
The figures only mean something against the amount needed to build something with.
The International Atomic Energy Agency works to a "significant quantity" — the approximate amount of material for which the possibility of manufacturing a nuclear explosive device cannot be excluded. For highly enriched uranium that figure is 25 kilograms of contained uranium-235.
The 93.8 kilograms the 1965 audit could not explain is therefore not a quantity with ambiguous implications. It is several significant quantities. The wider inventory loss across the plant's life, 269 kilograms before 1968 and 76 after, is more again.
This is why an accounting discrepancy at a company in western Pennsylvania reached the deputy director of the CIA, and why it was still being reopened a decade later. Nothing about the case turns on whether the loss was large. It was large. The entire argument is about where it went.
The reopenings
The matter did not rest with the 1965 audit, and the people who returned to it were not conspiracy writers.
The FBI investigated. The Atomic Energy Commission investigated, and its successor agencies inherited the file. The Carter administration reopened the question in the late 1970s, by which point the CIA's assessment was known inside government and the political cost of pursuing it was obvious.
The most substantial modern treatment is by two people with standing to give one. Victor Gilinsky was a commissioner of the Nuclear Regulatory Commission; Roger Mattson was a senior NRC official who worked on the case. Their reassessment, published in the Bulletin of the Atomic Scientists in 2010, went back through the declassified record and concluded that the evidence for diversion is stronger than the official handling of it ever acknowledged.
That is not the same as establishing it, and they do not claim otherwise. But it is the reason this case is not filed with the inventions: the people arguing it hardest are former regulators working from documents, and what they are complaining about is how the question was dropped.
What the record supports
A large quantity of weapons-grade uranium went missing from an American plant, more than from any comparable site, and a significant part of it has never been explained. A senior Mossad officer visited that plant under a false description of who he was. The CIA concluded that the material reached Israel and told regulators so.
Against that: the plant was a documented mess, its president's explanation is consistent with what was later dug out of the ground, and the cleanup bill is the physical record of how much material did not leave the site.
No investigation established a diversion, and none disproved one. The file is open in the specific sense that matters — not that nobody looked, but that everybody looked and the evidence has never closed the gap between a missing inventory and a destination.