For decades, astrophysicists have thought some sort of invisible dark matter must pervade the galaxies and hold them together, although its nature remains a mystery. Now, three physicists claim their observations of empty patches of sky rule out one possible explanation of the strange substance—that it is made out of unusual particles called sterile neutrinos. But others argue the data show no such thing.

“I think that for most of the people in the community this is the end of the story,” says study author Benjamin Safdi, an astroparticle physicist at the University of Michigan, Ann Arbor. But Kevork Abazajian, a theoretical physicist at the University of California, Irvine, says the new analysis is badly flawed. “To be honest, this is one of the worst cases of cherry picking the data that I’ve seen,” he says. In unpublished work, another group looked at similar patches of sky and saw the very same sign of sterile neutrinos that eluded Safdi.

Astrophysicists think each galaxy forms and resides within a vast clump, or “halo,” of dark matter, like the pit in a peach. The gravity of the invisible substance helps prevent the stars within from flying off into empty space. Theoretical physicists have dreamed up numerous hypothetical particles that might make up dark matter, among them cousins to nearly massless, barely detectable subatomic particles called neutrinos, which gush out of the Sun and nuclear reactors. The particles that make up dark matter would be hypothetical “sterile” neutrinos, heavier and even more elusive. An ordinary neutrino can interact with an atomic nucleus; sterile neutrinos would only interact with other neutrinos, arising when an ordinary neutrino morphs into a sterile one through a process called neutrino mixing.

The idea that sterile neutrinos might make up dark matter got a boost in 2014. Observations of nearby galaxies and the center of our own Milky Way revealed a faint glow of x-rays with a specific energy, 3.5 kilo-electron volts (keV). That glow would be expected if sterile neutrinos with a mass of 7 keV pervaded the galaxies. Very rarely, a sterile neutrino would decay into an ordinary neutrino and an x-ray, which would have an energy equal to half the sterile neutrino’s mass.

But a new analysis of astronomical observations shows the telltale glow cannot come from dark matter, Safdi and colleagues report today in Science. They looked at data not from distant galaxies, but from blank spans of sky between the stars in more than 4000 archival images snapped by XMM-Newton, an x-ray space telescope launched in 1999 by the European Space Agency. If our own galaxy lies within a vast cloud of sterile neutrinos, then the telescope must be peering through that cloud—and the sky between the stars should also faintly glow with 3.5-keV x-rays.

Safdi’s team found no sign of such a glow. The no-show suggests the glow in distant galaxies isn’t coming from dark matter, but from some more ordinary source such as hot gas, Safdi says.

Alexey Boyarsky, an astroparticle theorist at Leiden University, is unconvinced. “I think this paper is wrong,” he says. Boyarsky says he and his colleagues performed a similar, unpublished analysis in 2018, also using images from XMM-Newton, and did see a 3.5-keV glow from the empty sky, just expected from peering through a halo of sterile neutrinos.

How do two groups look at the same data and come to opposite conclusions? The difference lies in their methods, Boyarsky says. Because our galaxy is filled with a thin ionized gas, the sky emits x-rays, which can peak as specific energies even without a contribution from dark matter. The XMM-Newton telescope itself can also glow and emit x-rays at certain energies. And some x-rays come from beyond our galaxy, too. To see a 3.5-keV glow from dark matter, researchers must sift it from those background contributions.

To do that, Boyarsky and colleagues analyzed the entire spectrum of x-ray energies that XMM-Newton can detect, modeled the entire background, and subtracted it from the data. Crucially, Boyarsky says, his team removed known peaks at 3.3 keV and 3.7 keV to reveal the unexplained 3.5-keV peak. Safdi says his team took a different approach. Borrowing statistical techniques developed at atom smashers, they analyzed the spectrum from each image separately and analyzed data only over a much narrower range of energies.

However, that energy range isn’t much wider than the peak the team is looking for, Abazajian says. Boyarsky adds that because Sadfi and his team did not take out the two other background peaks, they may have mistaken a plateau created by the three overlapping peaks for a flat spectrum.

Not so, Safdi Says. His team found that subtracting the other peaks and widening the energy window doesn’t change the result. If a 3.5-keV peak exists, he says, his team’s more sophisticated technique would have revealed it.

Boyarsky says he is going to try to publish his blank-sky analysis. A physics journal turned it down, saying it wasn’t sufficiently “interesting,” he says. Now, he says he will submit it to Science. “I don’t care if it gets published, but I would like it to be peer reviewed,” he says. “They can’t say it isn’t interesting.”