Inside a squat, silver-domed hall in Batavia, Illinois, a superconducting ring 50 feet across has spent the past decade whipping muons around at nearly the speed of light, watching each one wobble like a tiny top before it dies. The ring itself is older than the experiment. It was barged and trucked from Brookhaven National Laboratory on Long Island to Fermilab in Illinois in 2013, a land-and-sea journey that ended a Brookhaven measurement begun in the 1990s and started a Fermilab one that finally published its last word in 2025.
The muon is the electron’s fat cousin. About 207 times heavier, unstable, gone in 2.2 microseconds. It carries spin, and it carries an internal magnet, and when you drop it into a magnetic field it precesses — wobbles — at a rate set by a number physicists call g.
In a universe with no quantum weirdness, g would be exactly 2. It isn’t.
A ring older than the puzzle it measures
The g−2 experiment began at CERN in the 1960s, moved to Brookhaven in the 1990s, and moved again to Fermilab in the 2010s. Three generations of physicists, one shared question: how much does the muon’s magnetism drift from that classical value of 2, and does the drift match what the Standard Model predicts?
The 2026 Breakthrough Prize in Fundamental Physics went to all three generations at once. The $3 million award was split across the hundreds of collaborators who built, rebuilt, and ran the storage ring across five decades and two continents.
Fermilab’s stretch of the story is the reason the ring made that unusual trip through the Atlantic and up the Mississippi. Brookhaven’s Alternating Gradient Synchrotron had measured the muon’s anomalous magnetic moment with high precision, and the answer sat about three standard deviations away from theory. Not enough to claim a discovery. Enough to obsess over. Fermilab’s more intense muon beam offered a way to sharpen the number by roughly a factor of four — if the same 50-foot magnet could be moved intact.
It was moved intact. Cracking the ring would have destroyed the uniformity of the field, which had been coaxed into shape over years.
What the muon actually does
Inject a muon into a uniform magnetic field and it spirals. The spin axis, the little internal arrow of its magnetism, doesn’t stay pointed the same way. It precesses around the field lines, the way a spinning top precesses around gravity when it starts to tilt.
The rate of that precession is what g−2 measures. Not g itself, but the tiny sliver by which g exceeds 2. That sliver is called the anomalous magnetic moment, written aμ = (g − 2)/2, and its value hovers near 0.00116592.
The last few digits are where the wars are fought.
Every one of those digits comes from the quantum vacuum. Virtual particles — electrons, photons, quarks, W bosons, anything with the right quantum numbers — flicker into brief existence around the muon and tug on its magnetism before winking out. Because the muon is heavy, it feels heavier virtual particles too. That is exactly what makes it such a sensitive detector for particles no one has yet named. As Ars Technica put it in April 2026, the muon is light enough to be plentiful, heavy enough to probe the Standard Model’s fine print.
Three runs, one number
Fermilab’s collaboration announced its first result in April 2021, using data from a single run. The Batavia value agreed with the older Brookhaven value and widened the gap with theory to about 4.2 standard deviations. That is close to the 5-sigma threshold physicists demand for a discovery.
A second run, announced in 2023, roughly doubled the precision. The gap held.
The third and final measurement, published in 2025 with the full six years of data, sits as the most precise measurement of the muon magnetic anomaly ever taken — precise to about 127 parts per billion. To picture that: measure the distance from Chicago to New York and be wrong by less than a tenth of a millimetre.
The experimental number, by then, was settled. What refused to settle was the theoretical prediction it was being compared against.
The strong force is the hard part
Working out what the Standard Model actually predicts for aμ means adding up every quantum tug on the muon. The electromagnetic part is easy. Physicists have calculated it to five loops of Feynman diagrams. The weak force part is tractable.
Then there is the strong force. Quantum chromodynamics — the theory that binds quarks into protons and neutrons — becomes harder to compute the further apart the quarks get. Pull them apart, and the field can spawn new particles from the vacuum. Perturbation theory, the workhorse of quantum field theory, stops working.
For decades, physicists sidestepped this by using experimental data as a shortcut. They took thousands of measurements of electrons and positrons colliding into hadrons and fed those into the calculation. The dominant term — the hadronic vacuum polarization — was estimated this way. The result was a Standard Model prediction that sat firmly below the Fermilab and Brookhaven measurements. Hence the anomaly. Hence the excitement about a possible fifth force.
A ten-year calculation on a supercomputer
An international team led by Penn State’s Zoltan Fodor took the other route. Rather than borrow from experimental data, they built the strong-force contribution from first principles using lattice quantum chromodynamics — chopping space and time into a fine grid and solving the equations of QCD on it numerically.
The calculation ran, on and off, for more than a decade. The final answer, published in Nature in April 2026, put the Standard Model prediction almost exactly on top of the Fermilab measurement. The gap collapsed to about half a standard deviation. In the team’s summary of the result, according to the lattice QCD team’s published results, their new calculation method showed the theoretical prediction matched experimental measurements, eliminating the apparent discrepancy.
Half a sigma is nothing. It is noise. The anomaly, in this reading, was a shortcoming of the older data-driven method, not a crack in the theory.
The disagreement about the disagreement
Except the story is not that clean.
The lattice result agrees with the Fermilab measurement. It disagrees, sharply in places, with the older data-driven calculations. Several electron-positron collision experiments show varying degrees of tension with the lattice calculations, pointing to unresolved questions about pion production cross-sections.
So one of the two approaches to the strong-force contribution is wrong, or at least incomplete. As Science News reported, the anomaly may have moved rather than dissolved: from a mismatch between theory and experiment, to a mismatch between two different ways of doing the theory.
Independent lattice groups have been converging on Fodor’s answer. A team using the Jülich Supercomputing Centre in Germany, working in parallel, reached a compatible number. Their announcement in 2026 described the result as resolving the discrepancy in the muon’s magnetic moment on the theory side.
The disagreement now lives in the electron-positron collider data. Which experiment measured the pion production cross-section correctly? That is a much smaller, much more specific question than whether there is a fifth force, and it is exactly the kind of question physics knows how to close.
What Fermilab’s ring actually accomplished
The temptation, reading the 2026 headlines, is to say Fermilab’s decade in the muon business ended in a shrug. It did not. The Batavia experiment delivered a value so tight that any future theoretical calculation has to account for it, achieving a precision of 127 parts per billion. The Scientific American account of the saga called the anomaly’s apparent disappearance one of the most anticipated results in recent particle physics — precisely because the measurement itself is so sharp.
The experimental number is not going away. It is a fixed point now. Every future model of physics beyond the Standard Model has to reproduce it.
Fodor described the moment with unusual candor, expressing disappointment to interviewers about finding no evidence for new physics. Fodor explained that the team initially hoped their calculation would support evidence for a fifth force, but instead their precise results confirmed the Standard Model’s predictions. What the team found instead, in his words, was a very precise proof of the Standard Model and of quantum field theory itself, holding to 11 decimal places.
What comes next
Fresh electron-positron data is on the way from several experiments, which should reconcile — or sharpen — the tension between the two approaches to the strong-force contribution. The MUonE collaboration at CERN is trying a completely different technique: firing muons at electrons and reading the hadronic vacuum polarization out of the scattering pattern directly. In Japan, a J-PARC experiment is preparing to measure g−2 using ultra-cold muons rather than the relativistic beam Fermilab uses. Independent method, independent systematics, independent answer.
If J-PARC agrees with Fermilab, the experimental number is locked. If the lattice calculations continue to converge, the Standard Model prediction is locked. If the two match, the muon anomaly closes as a case study in how long it can take to compute a number.
The Muon g-2 experiment is a companion piece to other long-arc physics stories — LIGO’s decade-long push to confirm Stephen Hawking’s black hole area theorem, for instance, or the slow accumulation of data that turns a hint into a fact. The instrument sometimes outlives the question that built it.
The ring at Fermilab is quiet now. Its final data have been read out, its final papers submitted, its Breakthrough Prize awarded. Fifty feet across, superconducting, hauled by barge from Long Island in the summer of 2013, it spent twelve years spinning muons in Illinois and produced a number good to 127 parts per billion — and, in doing so, forced the theorists to spend a decade of supercomputer time proving that number was exactly what the Standard Model had been saying all along.
