Ten Billion Collisions and a Trick From 1990: China's BESIII Just Retested the Standard Model's Bookkeeping

The Standard Model of particle physics contains a small table of numbers called the CKM matrix, which sets the odds that one kind of quark will transform into another. The theory makes a hard demand of that table: the probabilities in each row have to add up to exactly one. Nothing may leak out. If they do not add up, there is something in the universe the theory does not know about.

Physicists at the BESIII experiment in Beijing have now checked one of those entries a new way, using pairs of entangled particles produced in electron-positron collisions. The quantity is called Vus, and it describes how readily a strange quark converts into an up quark. The measurement, published in Nature, draws on roughly 10 billion collision events and revives an experimental technique that had been proposed decades ago and then largely left alone.

The particles doing the work are lambda hyperons — unstable cousins of the neutron, containing a strange quark. When they are produced in pairs from a collision, the two are quantum mechanically entangled, and the correlation between them lets experimenters extract information about the decay that a single particle would not give up. That is the trick: entanglement turns a messy decay measurement into a cleaner one, without requiring a new accelerator.

The answer is that the bookkeeping holds. The BESIII value for Vus is consistent with the requirement that the row of the CKM matrix sums to one, which is another way of saying the Standard Model survived a test it did not have to survive. For anyone hoping a crack would open up here, this is a disappointment; for anyone who wants to know how well the theory is actually pinned down, it is a real gain in precision.

The interesting residue is a small disagreement that will not go away. Determinations of Vus based on kaon decays — the traditional route — sit in mild tension with other extractions, and the hyperon result does not resolve that tension so much as approach it from a different direction. Because the systematic uncertainties in a hyperon measurement are almost entirely unlike those in a kaon measurement, having both is worth more than having either one twice.

That is the argument for continuing. A discrepancy that survives two techniques with unrelated error budgets is a physics result. A discrepancy that shows up in only one is usually a mistake somebody has not found yet. BESIII has now put a marker on the board with 10 billion events behind it, and the kaon community has to explain why its number sits where it does.

The paper appears in Nature under DOI 10.1038/s41586-026-10818-8.