The Sun's noise, measured
The measurement is a 3.3 sigma detection of coherent elastic neutrino-nucleus scattering from solar boron-8 neutrinos, collected over 6.77 tonne-years in liquid xenon under the Gran Sasso mountain in Italy. From it the collaboration infers a solar boron-8 flux of 5 million per square centimetre per second, on an interval running from 3 million to 8 million, consistent with previous measurements. The paper appeared in Physical Review Letters on 28 August, and the collaboration presented it in a webinar.[1]
Read as solar physics, that interval is wide, and the flux is already known far more tightly by experiments built for the job. The interest lies elsewhere. This number was produced by a detector designed to find dark matter, in the one channel where a neutrino and a light dark matter particle do the same thing: both make a whole xenon nucleus recoil, and the detector sees a recoil, not the thing that caused it. An instrument that can measure the Sun this way has, by the same fact, lost the ability to tell the Sun apart from what it is hunting.[1]
Ninety-three against ten
The paper puts a price on that loss. A 93 per cent increase in exposure over the previous search improved the median sensitivity to a 5 GeV/c² spin-independent dark matter particle by 10 per cent, and the authors describe this as diminishing returns. The likely mechanism is a change in how the limit scales: when the background is negligible, sensitivity improves roughly in proportion to exposure, and once an irreducible background grows with exposure alongside any signal, it improves roughly with the square root instead. A second explanation deserves its hearing: part of that shortfall could sit in analysis thresholds, efficiency and systematic terms specific to this dataset rather than in the fog itself, and the paper's single comparison between two searches cannot separate the two.[1]
An earlier column here tied the strength of a dark matter bound to re-asking an old census rather than to a fresh detection. The same shape returns in a harder form: the quantity a collaboration can most easily add, running time, is the quantity that has stopped paying. What would separate the two explanations is a third data release from the same detector. If the next XENONnT search reports a further large exposure increase and again gains only single-digit per cent at 5 GeV/c², the flattening belongs to the fog; a return to proportional gains would point at the analysis instead. The same dataset already carries the alternative use, a weak mixing angle measured at about 0.02 GeV/c momentum transfer, which is a number the fog helps rather than hinders.[1], [2]
Measuring the fog
For twenty years the neutrino floor was a projection: a curve drawn from solar models and cross sections, marking where a xenon experiment would eventually stop improving. It is now a measured background in the detector it constrains, with an error bar of its own. That changes the search space and leaves the particle untouched. The paper finds no evidence for light dark matter, and the honest reading of this result is that a low-mass dark matter search cannot be won by running longer. Directional sensitivity, annual modulation, a different target nucleus: the next argument has to be about telling a recoil's origin apart, because counting recoils has run out of room.[1]