Reanalysis finds no evidence of planetary‑mass primordial black holes in Andromeda data

A new independent reanalysis says a recent claim of possible dark matter signals in the Andromeda Galaxy does not hold up. In a paper posted to arXiv, astronomers Przemek Mróz and Andrzej Udalski report that 12 supposed short microlensing events seen in Subaru telescope data are instead consistent with ordinary stellar variability, not planetary-mass primordial black holes.

The disputed claim came from a February 2026 preprint by Sunao Sugiyama, Masahiro Takada, Naoki Yasuda and Nozomu Tominaga. Using Subaru Hyper Suprime-Cam observations of M31, the Andromeda Galaxy, from 2014, 2017 and 2020, the team analyzed 39.3 hours of data taken at roughly two-minute cadence. From more than 25,000 variable candidates, they reported 12 microlensing candidates with timescales shorter than five hours. Under a primordial black hole interpretation, they said the objects would point to black holes with masses around 10^-7 to 10^-6 times the mass of the sun and a dark-matter fraction of about 0.1, or roughly 10%.

Mróz and Udalski revisited the same Subaru data with what they describe as an independent difference-image-analysis photometric pipeline, a separate method for measuring how sources brighten and fade in crowded images. Their paper, titled “Eppur non si trovano Vol. 2: No Planetary-mass Primordial Black Holes toward the Andromeda Galaxy,” is listed on arXiv as arXiv:2604.00111. The current version was revised May 4, and the arXiv record says it has been accepted for publication in Acta Astronomica.

Their conclusion is blunt. “We find no compelling evidence for short-timescale microlensing events among the candidates identified in the Subaru data set,” the authors write in the abstract.

According to the reanalysis, all 12 candidates show asymmetric light curves and-or variability on multiple nights, behavior that does not match the expected signature of short microlensing events. Instead, the authors write that “among them ten objects are RR Lyrae stars, one is an eclipsing binary, and one is an unclassified variable star.”

That distinction matters because microlensing searches are looking for a very specific kind of signal: a brief brightening caused when a compact object passes in front of a background star and bends its light. In a dense stellar field such as Andromeda, many stars vary naturally, and those changes can masquerade as lensing events if they are not filtered out correctly. Rejecting those contaminants is therefore central to any claim that a dark, compact object has been seen indirectly.

The February preprint stood out in part because Subaru Hyper Suprime-Cam observations of Andromeda had previously been used to place strong limits on primordial black holes rather than to suggest a detection. Primordial black holes are a long-discussed hypothetical dark matter candidate, so credible evidence for a planetary-mass population would have been a major result.

The new reanalysis also leaves the February interpretation in tension with prior survey results. In 2024, the OGLE survey — the long-running Optical Gravitational Lensing Experiment — reported a high-cadence search of the Magellanic Clouds that monitored about 35 million stars and found zero short-timescale microlensing events. That study concluded primordial black holes in the relevant planetary-mass range could make up at most about 1% of dark matter.

Taken together, the new paper weakens the case that the Subaru Andromeda data revealed a significant population of planetary-mass primordial black holes. More narrowly, it argues that the 12-event signal can be explained without invoking dark matter at all — a reminder that in astronomy, especially in crowded fields, independent reanalysis can be as important as the initial claim.