January 1976: From the “Oops-Leon” to the Upsilon Particle

The history of particle physics is littered with spurious findings, blips in the detector that disappeared on second inspection. Most of these false discoveries have been buried under the deluge of real discoveries. But the “Oops-Leon,” which would have been a new particle with a mass of 6 giga-electronvolts (GeV), remains unusually vivid.

During the late 1960s, the cohort of elementary particles was small but growing. Physicists had already discovered three types of leptons — electrons, muons, and neutrinos — and three of quarks — up, down, and strange. But the quarks were still just mathematical entities, not yet spotted in experiments, and the three remaining quark flavors — charm, top, and bottom — had not yet been proposed.

In 1967, Leon Lederman and his colleagues began a new experiment at Brookhaven National Laboratory, aimed at finding new particles. They slammed 30-GeV protons into neutron-rich uranium, which would decay first into virtual photons and then into pairs of electrons or muons — which could, in turn, be scrutinized by sensitive instruments for signs of new particles. To filter out the unwanted collision debris, the researchers used 10 feet of steel from World War II ships, which only muons could pass through.

The team expected the data to show a smoothly falling distribution of muon pairs and knew any bump could signal a new particle. But when Lederman’s group saw a faint bump at 3 GeV, they convinced themselves it was nothing. “The signal [was] incontrovertibly there,” says Dan Kaplan, an emeritus professor at the Illinois Institute of Technology who worked with Lederman later. “But they would have had great difficulty publishing this distribution as a discovery.”

Seven years later, the 3 GeV particle, called the J/ψ meson, made from a charm and anticharm quark, was definitively discovered. Lederman knew he had missed out on a Nobel-worthy discovery and was determined not to miss out again.

At the newly operational Fermilab, Lederman and his team planned a repeat of the 1967 experiment, but at much higher energies of 400 GeV. For the experiment, called E288, the researchers would measure the results of particle collisions — and, if they found new particles, “publish these and become famous.”

Notably, there was no explicit goal to find a heavier quark. Although physicists had found the up, down, strange, and charm quarks, models that predicted six quarks had not taken hold.

By August 1975, the E288 team began to see a distinct bump around 6 GeV, finding 12 events where the predicted background was four. The team calculated there was a 2 percent chance a signal that large would result from a statistical fluctuation alone. The team, it seemed, had stumbled on a brand-new particle, and it needed a name.

Chuck Brown, a member of the E288 team now retired from Fermilab, recalls a late-night shift with Jeff Weiss, poring over available Greek letters. “Iota was rejected since it resembles a question-mark — in hindsight, it would have been a better choice,” John Yoh, the experiment coordinator, wrote in 1997. Then Walter Innes suggested that, if a particle named “Upsilon” turned out to be a mirage, they could simply call it an “Oops-Leon.”

Incredibly, another Fermilab group also found evidence for a resonance at about 6 GeV using a different method and submitted their findings two days beforehand. Curiously, their result has been largely forgotten.

But data quickly cast doubt on both results. A few months later, Lederman’s team repeated the experiment with muons and found no evidence of excess at 6 GeV, and an examination in 1977 finally ruled it out. The Upsilon was an Oops-Leon after all.

What went wrong? A decade earlier, Arthur Rosenfeld pointed out that “trials factors” — essentially, running an experiment multiple times — would lead to statistically significant “discoveries” that were flukes. To combat this, Rosenfeld proposed waiting until results met a strict statistical standard known as “five-sigma.” But the standard was not adopted in earnest until the discovery of the top quark in 1995, argues experimentalist Tommaso Dorigo.

Although the E288 researchers included the effect of trials factors, a five-sigma standard might have prevented them from claiming a discovery. But the team also made a mistake: They assumed the background scaled as the inverse cube of mass, whereas it actually fell exponentially. It wasn’t 4, but more like 8. Over the increased background, 12 events are much less significant.

Undeterred, the E288 team continued taking data. In November 1976, Yoh noticed a small excess around 9.5 GeV, already more statistically significant than the Oops-Leon. In an act of faith, he wrote “9.5 GeV” on a bottle of champagne and kept it in the lab freezer.

But as the team was collecting more data to support the 9.5 GeV bump, it suffered what could have been a devastating setback. Just before midnight on May 20, a fire ravaged their electronics. “My birthday is May 21,” Kaplan remembers. “I was saved a midnight shift by that.” Lederman, who had dealt with a similar problem before, phoned a Dutch fire expert in the middle of the night. Through State Department contacts, Lederman got a visa approved in record time. The expert arrived on May 21, and the electronics were salvaged.

With the experiment back up and running, they gathered enough data to convince even the most cautious members of the collaboration that a particle at 9.5 GeV did, in fact, exist. The discovery was published in Physical Review Letters without review, under a new policy designed to deal with the priority disputes plaguing high energy experimental physics.

“People were searching around in the unknown,” says Brown. “It's, in some sense, just luck that we stumbled on it. I mean, that was not the goal of the experiment.”

The Upsilon was three times heavier than any previously discovered particle, but what it implied was even more astonishing: the existence of a fifth quark, the bottom quark. While some theorists had proposed six-quark models, the idea was nascent until the Upsilon propelled it forward into the modern Standard Model.

Though the Oops-Leon was a fluke, its legacy remains intertwined with the real Upsilon.

“There's an important theme here,” Kaplan says. “You have to have the good fortune to have the opportunity to fail gracefully and be given a second chance.”