One flash, a mile under South Dakota

On 1 September 2026 the LUX–ZEPLIN collaboration did something its discipline is trained not to do. It published an event it cannot explain.

The event is a single nuclear recoil, recorded on one day in June 2023 and then held under scrutiny for more than two years. It carries an energy of 248 keV, with an uncertainty of 23 keV from statistics and another 23 keV from systematics. Every known source of background radiation in the detector — radon daughters, neutrons from the rock, stray gammas, the faint radioactivity of the detector's own steel — was modelled, and none of them wants to put an event there. A profile-likelihood test puts the tension with the background-only hypothesis at 2.6 sigma globally: about a one-in-two-hundred chance that ordinary physics produced it.

That is not a discovery. Particle physics reserves the word for five sigma, and the collaboration is careful to say so.

“With only one event, we don't want to get ahead of ourselves. We are not claiming to have seen dark matter. But we have seen something interesting that we want to share with the scientific community.”

Rick Gaitskell, LZ spokesperson, Brown University

The quietest room we have

LZ is seven tonnes of liquid xenon held at the 4,850-foot level of the Sanford Underground Research Facility in Lead, South Dakota — the old Homestake gold mine, roughly a mile of rock between the experiment and the cosmic-ray sky. The detector is managed by Lawrence Berkeley National Laboratory and exists for one purpose: to be so empty of ordinary events that an extraordinary one cannot hide.

When something scatters off a xenon nucleus, the detector sees it twice. There is a prompt flash of scintillation light, and then a cloud of ionisation electrons that drifts upward through the liquid under an electric field and produces a second, delayed flash in the gas above. The ratio of those two signals says whether the recoil was against a nucleus or against an electron — and dark matter, if it is there, should hit nuclei. The timing between them, combined with the pattern across the photomultiplier array, reconstructs the position to within centimetres, so events near the walls can be thrown away. It is the cleanest discrimination anyone has built at this scale.

This result comes from 220 live days of data taken between March 2023 and April 2024. The novelty is not just the exposure; it is the search window. Most WIMP hunts have concentrated on low-energy recoils, because that is where a particle of ten or a hundred proton masses deposits its energy. LZ analysed a broader band than previous searches — and the anomaly turned up at 248 keV, well above the region the field has spent two decades combing. If it is a dark-matter particle at all, it is a heavy one: at least 200 GeV, more than two hundred times the mass of a proton.

Why this one is harder to dismiss

Most anomalies die because someone finds the background that made them. LZ's collaboration spent the better part of two years trying and failing to do exactly that, then published the failure. That is an unusual and rather admirable way to lose an argument with your own data, and it is the reason the result is worth attention that a 2.6-sigma bump would not normally earn.

There is also a structural reason to care. WIMPs in the hundreds-of-GeV range are not a fringe idea; they are close to the mass scale that supersymmetric models kept pointing at before the LHC declined to find superpartners. That region has been squeezed but not closed, and the experimental sensitivity there is thinner than at lower masses simply because fewer analyses have looked.

What settles it

Only more xenon-days. LZ's planned exposure runs to roughly 1,000 live days; the 220 analysed here are a fifth of it. If the effect is real, events should accumulate at a predictable rate and the significance should climb steadily. If it was a fluctuation, the significance will sag as the exposure grows — the characteristic death of a statistical ghost. Meanwhile XENONnT in Italy and PandaX-4T in China operate at comparable sensitivity with different rock, different steel and different systematics. An independent event in the same energy band from either would change the conversation in a week.

Nothing needs to be believed yet. But for the first time in a long while, the dark-matter search has something specific to argue about.