Beryllium Arrived Nine Years Late

On 11 November 1974, two groups announced the same 3.1 GeV resonance on the same day. The standard telling of the November Revolution is a story about ideas arriving on schedule: charm, GIM, asymptotic freedom, all clicking into place at once. Chris Quigg’s opening keynote to the 4th International Symposium on the History of Particle Physics, posted this week, tells it that way — and then drops in a detail that quietly wrecks the schedule. At the London conference four months earlier, Quigg recalls, “Leon Lederman announced that between 3 GeV and 10 GeV, there were no narrow resonances decaying into lepton pairs.” Lederman had already run the experiment that should have found it. What he lacked was not an idea. It was beryllium.

1968–69, Brookhaven. Christenson, Hicks, Lederman, Limon, Pope and Zavattini put 29 GeV protons into uranium and looked for muon pairs. Muon momentum was measured by range — how far a muon punched through steel. In Lederman’s own Nobel account, there were 18 angular bins and four range bins per muon, “a total of only 5000 mass bins into which to sort the data,” because “multiple scattering in the minimum of 10 feet of steel made finer binning useless.” Mass resolution: 10–15%. The published abstract (PRL 25, 1523) says the spectrum from 1 to 6.7 GeV “exhibits no resonant structure.” The body of the same paper says something else: “in the mass region near 3.5 GeV, the observed spectrum may be reproduced by a composite of a resonance and a steeper continuum.” The J/ψ mass is 3096.900 ± 0.006 MeV. The ratio 3.5/3.0969 is 1.130 — the paper’s guess was high by 13%, which is to say high by exactly one apparatus resolution. They had it, hedged in a subordinate clause, misplaced by the width of their own blur, and contradicted by their own abstract.

1974, MIT/BNL. Aubert et al. repeated essentially the same reaction with a magnetic spectrometer and multiwire proportional chambers. Lederman’s summary is two words long: “The shoulder was refined by the superior resolution into a towering peak.” 1976, Fermilab. Lederman’s group, now at 400 GeV, reported 27 high-mass dielectron events between 5.5 and 10 GeV, with 12 clustered between 5.8 and 6.2 GeV, which “suggests that the data contain a new resonance at 6 GeV.” There is no resonance at 6 GeV. The group nicknamed it Oops-Leon. Early 1977. The rebuild. Hadron absorption was moved out of iron and into “actually 30 feet of beryllium.” Lederman: “The decreased multiple scattering of the surviving muons reduced the mass resolution to 2%, a respectable improvement over the 10–15% of the 1968 BNL experiment.” One month of spring running produced 7,000 pairs above 4 GeV and 800 events in a bump at 9.5 GeV. By September, 30,000 events split it into peaks at 9.4 and 10.0 GeV, “whose widths are consistent with our resolution.”

Here is what the swap actually bought, from the PDG material tables. Iron: radiation length 1.757 cm, nuclear interaction length 16.77 cm — ratio 9.54. Beryllium: 35.28 cm and 42.10 cm — ratio 1.19. Iron costs 8.00 times as many radiation lengths per interaction length absorbed. Run the apparatus numbers through that (my arithmetic on their tables): ten feet of steel is 18.2 interaction lengths and 173.5 radiation lengths; thirty feet of beryllium is 21.7 interaction lengths and 25.9 radiation lengths. The 1977 rig got more hadron killing — up 19% — for 6.7× less scattering material. Multiple-scattering angle goes as √(x/X₀)·[1 + 0.038 ln(x/X₀)], so the muons came out 2.75× straighter; magnetic analysis in toroids instead of range measurement supplied the rest of the factor of six in resolution.

That is the whole story. The figure of merit for a dilepton experiment was never “how much steel can we afford.” It was λ_I/X₀ — a ratio of two numbers sitting in a materials table, minimised by the lightest solid you can machine. Nine years elapsed between a resonance passing through Lederman’s detector unidentified and Lederman discovering a different resonance in four weeks. The intervening work was not conceptual. It was learning to read one column against another. Histories of the November Revolution should be indexed by absorber material. The interesting counterfactual is not “what if Lederman had trusted his 3.5 GeV shoulder” — he had 5,000 mass bins and no business trusting it. It is “what if beryllium had been procurable in bulk in 1968,” in which case the J/ψ is a Brookhaven result from 1969 and charm arrives before the GIM mechanism is five years old, into a theory community that had not yet talked itself into expecting it. The ratio never stopped mattering. The reason the beam pipe threading ATLAS and CMS is beryllium — under a millimetre of a toxic, expensive, hard-to-machine metal, where steel would be free — is the same 8.00. Everything you want to measure has to get out through it.