Final report of the E821 muon anomalous magnetic moment measurement at BNL
The complete account of Brookhaven’s g−2 experiment: how the muon’s magnetism was measured to half a part per million, and why it didn’t quite match theory.
G. W. Bennett et al., Muon g−2 Collaboration, including S. Giron. Physical Review D 73, 072003 (2006).
This is the experiment’s full record, almost a hundred pages: the storage ring, the magnetic field, the detectors, the analysis and the result.
The muon’s magnetism is set by its g-factor. Dirac’s theory says g = 2. Quantum effects, every particle that can briefly pop in and out of the vacuum, push it slightly higher. Measuring that small excess, a = (g−2)/2, tests the whole Standard Model at once.
E821 measured it to 0.54 parts per million, about fourteen times better than CERN had. The Standard Model prediction came out 2.2 to 2.7 standard deviations lower: not proof of new physics, but a tension strong enough to launch a new experiment at Fermilab. That experiment, and this one, are the work recognized by the 2026 Breakthrough Prize.
My part Co-author. I helped build the electromagnetic calorimeters described in Section II.H.1.
Open any part
I. Introduction
Why the muon’s anomaly matters, and the history from Dirac’s g = 2 to the CERN experiments.
II. Experimental method
Polarized muons circulate in a 14-meter superconducting storage ring with an extremely uniform magnetic field. Their spins turn slightly faster than their momentum; that difference frequency, together with a precise field map, gives the anomaly.
II.H.1. Electromagnetic calorimeters (my part)
Twenty-four lead/scintillating-fiber calorimeters around the inside of the ring caught the decay positrons and measured their energy and arrival time. Counting high-energy positrons over time reveals the spin precession.
III. Beam dynamics
How the stored beam moves and oscillates, and how those motions are kept from biasing the result.
IV. Data analysis
The field measurement and the precession-frequency fits, done by independent teams with hidden offsets until the end, and every systematic error accounted for.
V. The Standard Model value
The theory side: QED, electroweak and hadronic contributions. The largest theoretical uncertainty comes from hadronic vacuum polarization.
VI. Discussion and conclusions
Experiment and theory differ by 2.2 to 2.7 standard deviations, depending on the hadronic input.
Each part opens to a short summary. Where a free copy exists, the link takes you to that section of the paper.