Science · 6 min read
Why the Muon Magnetic Moment Remains a Test of Theory
Why a settled precision measurement still awaits reconciliation among strong-interaction calculations before it can test new physics.

Why the Muon Magnetic Moment Remains a Test of Theory
A small anomaly measured with increasing precision
A charged particle with spin behaves in a magnetic field as if it carries a tiny magnet. For a pointlike spin-one-half particle, the simplest relativistic quantum equation gives a magnetic factor, called g, equal to 2. Interactions with quantum fields shift that value slightly. Physicists express the shift as the anomalous magnetic moment, aμ = (g−2)/2, for the muon.
The word "anomalous" is historical terminology, not a claim that the measurement conflicts with the Standard Model. Quantum electrodynamics predicts the leading correction, while weak interactions and processes involving quarks also contribute. Hypothetical particles could add further corrections. Because the total can be measured and calculated precisely, comparison is a sensitive consistency test.
Fermilab's Muon g−2 collaboration announced its final experimental result in June 2025 after analyzing six running periods. Its published value reached a precision of 127 parts per billion and agreed with the collaboration's earlier results and with the older Brookhaven E821 measurement. That convergence establishes the experimental target more firmly. It does not, by itself, establish a discrepancy with theory, because the theoretical reference has become the central unsettled part of the comparison.
The storage ring turns spin into a frequency comparison
The experiment stores polarized positive muons in a magnetic ring. Muons are unstable, with a mean lifetime of about 2.2 microseconds at rest, but relativistic motion extends their lifetime in the laboratory frame. As the particles circulate, their spins rotate relative to their momentum. The rate of that relative rotation depends on the anomalous magnetic moment and the magnetic field.
The muons decay into positrons and neutrinos. Because the weak interaction violates parity, high-energy decay positrons are emitted preferentially in relation to the muon's spin. Detectors around the ring record the changing rate and energy distribution of positrons. The oscillation in those counts yields the anomalous precession frequency.
The magnetic field is measured independently using proton nuclear magnetic resonance probes. The collaboration then forms a ratio between the muon precession frequency and a field-calibration frequency, adding measured corrections for beam motion, electric fields, detector effects, and other features of the stored ensemble. The final aμ value therefore does not come from reading a single instrument. It combines frequency estimation, magnetic mapping, beam dynamics, calibration constants, and blinded analysis procedures.
Fermilab reused the 14-meter storage ring from Brookhaven but built a more intense muon beam and improved detectors, field uniformity, and analysis. Agreement across run periods is an important check because operating conditions and systematic contributions changed. Yet shared apparatus and methods mean that agreement among runs is not wholly independent replication. The earlier Brookhaven result supplies a more separate comparison, though at lower precision.
Experiment and theory have different uncertainty budgets
On the experimental side, uncertainty can be divided into statistical and systematic components. More detected decays reduce statistical uncertainty. Systematic work includes knowing the average magnetic field sampled by the muons, modeling their distribution in the ring, controlling detector gain changes, and accounting for muons lost before decay. The final Fermilab result is precise enough that future improvement would require both a new experimental strategy and substantial control of systematics.
The Standard Model calculation is assembled from several physical contributions. Quantum electrodynamics supplies the largest correction and is calculated through very high perturbative orders. Electroweak effects from W, Z, and Higgs processes are smaller but sufficiently controlled. The dominant uncertainty comes from strong-interaction, or hadronic, effects that cannot be calculated with ordinary low-energy perturbation theory.
Two hadronic terms are especially relevant: hadronic vacuum polarization and hadronic light-by-light scattering. The light-by-light contribution has benefited from lattice quantum chromodynamics and data-informed methods that broadly support one another within uncertainties. Vacuum polarization is larger and is now the source of a methodological split.
This separation matters when headlines present one number called "the Standard Model prediction." A benchmark is a synthesis of chosen inputs and methods. Its uncertainty represents the authors' accounting within that framework; it is not a direct measurement made by the same apparatus as the experiment.
Hadronic calculations no longer give one uncontested benchmark
The Muon g−2 Theory Initiative's 2020 white paper used a data-driven evaluation of hadronic vacuum polarization based mainly on measured cross sections for electron-positron annihilation into hadrons. Compared with the experimental average available after Fermilab's first result, that benchmark produced a difference widely reported as about 4.2 standard deviations. If all uncertainties and inputs were correct, such a difference would be unlikely as random fluctuation.
In 2021, the BMW collaboration published a lattice-QCD calculation of the leading hadronic vacuum-polarization contribution that was higher than the data-driven value. Using it moved the full Standard Model prediction closer to experiment. Subsequent lattice calculations improved precision and studied the relevant distance ranges, while new measurements of electron-positron processes did not all agree with one another. The CMD-3 experiment's cross sections for the dominant two-pion channel, for example, created tension with earlier datasets and radiative-return measurements.
The Theory Initiative's 2025 white paper consequently did not restore a single simple comparison. It synthesized advances and documented the unresolved difference among data-driven and lattice approaches. The final Fermilab paper likewise emphasized that interpretation awaits resolution of the theoretical situation. This is not a procedural delay hiding an agreed answer; it is the scientific question.
Lattice QCD approximates space-time on a finite grid and uses large numerical computations to evaluate strong-interaction effects from the underlying theory. Data-driven dispersion methods use measured hadron-production cross sections plus analyticity and unitarity relations. The methods have different dependencies. Lattice results must control finite-volume effects, lattice spacing, quark masses, and electromagnetic corrections. Data-driven results depend on experimental calibration, radiative corrections, channel combination, and coverage. Agreement would provide powerful cross-validation. Disagreement requires locating which uncertainties or inputs are incomplete.
The open question moved from measurement to comparison
Several outcomes remain possible. Experimental cross sections could converge after revised calibrations and combinations. Lattice groups could find common effects that shift their results or reinforce their current direction. A more refined Standard Model synthesis could still leave a residual difference from the measured value. Only then would models of new particles or forces face a stable numerical target.
Other observables provide constraints. Many proposed new-physics explanations also affect electron magnetic moments, particle decays, collider events, or dark-matter searches. A model that can shift muon g−2 is not thereby supported; it must remain consistent with those measurements. Conversely, absence of a discrepancy would not make the experiment unimportant. It would be a stringent confirmation of a calculation spanning quantum electrodynamics, weak interactions, and hadronic physics.
What is established is the experimental value and its documented uncertainty, reproduced across Fermilab runs and consistent with Brookhaven. It is also established that strong-interaction calculations have advanced and that leading approaches are not yet fully reconciled. What is contested is the correct combined Standard Model reference and therefore the statistical size of any discrepancy. What remains unknown is whether the final comparison will require new physics or improved understanding of known physics. Precision has narrowed the location of uncertainty without dictating its resolution.
Sources and review note
4 sources, none of them with a link, so treat them as reading, not proof.
- Muon g−2 Collaboration, "Measurement of the Positive Muon Anomalous Magnetic Moment to 127 ppb," final Fermilab result (2025), with the collaboration's 2026 final report consulted for methods.
- T. Aoyama et al., Muon g−2 Theory Initiative, The anomalous magnetic moment of the muon in the Standard Model, Physics Reports 887 (2020), and the initiative's 2025 white paper.
- S. Borsanyi et al. (BMW Collaboration), "Leading hadronic contribution to the muon magnetic moment from lattice QCD," Nature 593 (2021).
- Fermilab Muon g−2 technical publications on magnetic-field measurement, beam dynamics, detector calibration, and run-by-run analysis.
