Physicists sank a neutrino telescope 1.5 miles under the sea, and Einstein’s relativity still held

A neutrino detector anchored about 1.5 miles under the Mediterranean has put one of Einstein’s central assumptions through a fresh test, and the assumption came through intact. Researchers with the KM3NeT/ORCA collaboration studied 5,828 atmospheric neutrinos recorded over about 17 months and found no sign that Lorentz invariance, the principle underneath relativity’s treatment of space and time, breaks down. The results were reported on Nov. 4 in the Journal of Cosmology and Astroparticle Physics.

Stay ahead on AI: Join 600,000+ professionals who read The Deep View, the daily newsletter that keeps you up to date with artificial intelligence.

Subscribe to The Deep View →

We may earn a commission if you sign up.

Lorentz invariance is the idea that the laws of nature look the same regardless of direction or speed. Relativity builds on it, which is why physicists keep probing it for cracks.

Six strings of light sensors off Toulon

The detector, called ORCA, sits in the Mediterranean off the coast of Toulon, France, about 2.4 kilometers below the surface, according to Live Science’s account of the study. Each detection unit is a line anchored to the seabed that carries 18 glass spheres, and each sphere holds 31 light sensors. When a neutrino collides with a particle in the water, the impact produces a faint blue flash that those sensors record. The wider KM3NeT research infrastructure pairs ORCA with a sister telescope, ARCA, built to catch high-energy neutrinos from distant cosmic sources. The 2026 analysis used only the first six units. Forty-four are in the water now, and the collaboration plans to activate 108 by 2030, so the dataset behind this result is a small slice of what the detector will eventually collect.

Alba Domi, a doctoral student who helped lower the first detection units to the seafloor, explained to Live Science why water matters. Water and ice scatter and absorb light differently, so a telescope in the sea reconstructs a neutrino’s energy and direction by a different recipe than one frozen into Antarctic ice. That difference gives the Mediterranean array a distinct way to cross-check results from rival detectors, including the ice-based experiments whose limits ORCA is now compared against. A technique that works in seawater and one that works in glacial ice share few of the same systematic worries, so agreement between them carries extra weight.

Neutrinos that crossed 12,700 kilometers of Earth

The 5,828 neutrinos in the dataset were born when cosmic rays slammed into the upper atmosphere. The team kept the ones arriving from below, having passed upward through the planet, because the rock and water of Earth filter out most other particles. Some of those neutrinos traveled about 12,700 kilometers, or 7,900 miles, through Earth before reaching the sensors.

Distance is the point of the exercise. Neutrinos change type as they travel, a behavior called oscillation, and a break in Lorentz invariance would shift the pattern of those changes, with the shift growing over longer paths. Lukas Hennig, a doctoral student at Erlangen-Nürnberg and a co-author of the paper, told Live Science that general relativity describes the universe well at cosmic scales and the Standard Model works at microscopic ones. He added that Lorentz invariance “might be violated” if some candidate theories of quantum gravity are right. The two theories have never been reconciled into one picture, and that gap is why a test like this one keeps getting repeated with new instruments. The full analysis is published under the Journal of Cosmology and Astroparticle Physics reference 2026/11/012.

That growth with distance is why atmospheric neutrinos suit the job. A neutrino produced on the far side of the planet has a long baseline in which a small deviation could build into a visible change in how often muon neutrinos turn into tau neutrinos. The ORCA team looked for such a distortion in the pattern of arrivals across energy and direction and found none.

Limits that rival Super-Kamiokande and IceCube

The oscillations in the ORCA data matched standard physics. Any violation of Lorentz invariance must therefore be weaker than the limits the team set, including the first direct limits on a violation that would treat muon and tau neutrinos differently, according to the Live Science report.

KM3NeT described the work on its own Lorentz invariance news page in March, when the paper appeared as a preprint. Its summary said “no violation was observed” and that the study yields the first upper limits on isotropic diagonal coefficients in the neutrino sector that do not assume the initial flavour ratios of astronomical neutrinos. The collaboration called that approach model-independent and said it makes the constraints more robust. The preprint, arXiv:2603.04264, reports 1.4 years of ORCA6 data and says the limits on a subset of coefficients are competitive.

The comparison points are large. Super-Kamiokande in Japan contributed about 12 years of data to its own limits, and IceCube in Antarctica used two years of atmospheric neutrinos from its full array. Hennig said he was surprised that six detection units could produce competitive results at all. He also noted what a null result does to the theory shelf: models predicting a stronger violation than the new limits are now ruled out, which narrows the room left for the most aggressive versions of quantum-gravity phenomenology.

The open question is how far the limits can tighten. With 44 units deployed and 108 planned, the ORCA collaboration will have far more neutrinos to sort, and any violation hiding just below today’s bounds would have to show itself in that larger sample. The preprint’s own framing is modest, saying the results complement and extend existing constraints rather than closing the question, and a null result of this kind can only ever lower the ceiling on how large a violation could be, never prove that none exists.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

Stay ahead on AI: Subscribe to The Deep View →

We may earn a commission if you sign up.