Three times physicists have run this experiment. Three times the result has come back wrong. And each time they have returned with more data and sharper instruments, the wrongness has not corrected itself. It has deepened.
Findings published in Physical Review Letters record the most precise measurement ever taken of a rare subatomic decay at CERN’s LHCb detector, built from 8.4 inverse femtobarns of collision data collected across seven years. The gap between what the experiment measures and what physics says should happen now stands at 4 sigma. That number means there is a 1-in-16,000 chance the discrepancy is random noise. It is not a rounding error. It is not a calibration fault. Physicists have spent over a decade hunting for both and found neither.
The Standard Model is the closest thing physics has to a complete rulebook for reality. It describes every known particle and every known force except gravity, and it has passed every direct experimental test thrown at it for fifty years. The W boson, the Z boson, the top quark, the Higgs boson: the Standard Model predicted all of them before any detector confirmed them. It does not get things wrong. Except here, buried in a decay so rare it occurs roughly once in every million attempts, it keeps getting something wrong, and nobody can agree on what is causing it.
The decay in question involves a particle called the B meson, a short-lived subatomic object that disintegrates in less than a trillionth of a second. Most of the time that disintegration is unremarkable. But occasionally the beauty quark inside it converts directly into a strange quark while producing two muons, heavy cousins of the electron. That process is almost entirely forbidden by the Standard Model. It can only happen through an elaborate series of virtual quantum steps, particles borrowed briefly from empty space and returned before they can be directly seen. Because the process is so suppressed, its geometry is predicted with extraordinary precision. That is exactly why a deviation is so hard to dismiss: the prediction is tight enough that any shift registers clearly against it.
The shift physicists keep measuring centres on a quantity called P5-prime, an observable engineered specifically to strip out the theoretical uncertainties that make other measurements easier to argue away. It first deviated from the Standard Model prediction at 3.7 sigma in 2013. It held at 3.3 sigma in 2016 with a larger dataset. It now sits inside a global tension of 4 sigma with the largest and most carefully constructed dataset the experiment has ever produced. A second independent detector, the CMS experiment, published its own measurement earlier in 2025 and recorded the same directional deviation. Two separate machines, two separate teams, two separate analysis pipelines, pointing at the same gap in the same place.
What fills that gap is the question nobody has been able to answer. The most rigorous theoretical work done on Standard Model corrections, using multiple independent calculation frameworks, consistently produces numbers too small to close the distance. Something beyond the known particle catalogue appears to be participating in the decay at the quantum level, contributing to the process in a way that bends the geometry of the output away from the prediction. Physicists have two serious candidates. One is a heavier, undiscovered cousin of the Z boson, the particle that carries the weak nuclear force, a hypothetical called the Z-prime that appears in several extensions of the Standard Model and couples to both beauty quarks and muons in precisely the way needed to produce the observed shift. The other is a class of particles called leptoquarks, objects that would connect quarks and leptons, two families of matter that currently have no direct interaction in the Standard Model at all. Both fit. Neither has been found.
What makes the current situation different from every previous version of this story is the size of what comes next. Since the dataset used in this analysis was closed in 2018, the LHCb experiment has collected approximately three times as many collision events in its upgraded Run 3 configuration. That data is currently being processed. The 5-sigma threshold required for a confirmed discovery in particle physics corresponds to a 1-in-3.5-million probability. The current result is at 4 sigma. The Run 3 dataset carries enough statistical weight to push past that boundary if the signal is real, or to collapse it entirely if it is not. There is no middle ground left. The experiment has run long enough, and the data volume is now large enough, that the next result will be definitive.
No publication date for the Run 3 analysis has been formally announced. The work is active at CERN.
Source:
LHCb Collaboration. Angular analysis of the B0 → K*0(→K+π−)μ+μ− decay. Physical Review Letters, 2025. DOI: https://doi.org/10.1103/24g9-yn9d
The commercials in between the paragraphs makes me not want to believe what it is I’m reading.
We need to pay the bills somehow.