At the heart of every large galaxy sits a supermassive black hole — millions, sometimes billions of times heavier than the Sun. Astronomers have long treated these monsters as permanent fixtures, glued to the galactic center for eternity. But a new study suggests something far more dramatic: some of them get kicked out.
Analyzing data on around 100,000 quasars from the Sloan Digital Sky Survey, researchers at Eötvös Loránd University (ELTE) in Budapest have found statistical evidence that supermassive black holes can be ejected from their positions after forming. The study, led by Bence Bécsy (an ELTE graduate now at the University of Birmingham) and published in the Monthly Notices of the Royal Astronomical Society, tests a prediction made by the very same research group ten years ago. The result matters not only for our picture of galaxy evolution — it could determine how many black hole mergers the future LISA space observatory will be able to detect.
When Two Black Holes Become One — and the Winner Gets Kicked
To understand the new result, start with how galaxies grow. Galaxies collide and merge all the time, and when two large galaxies become one, their central black holes sink toward each other, form a binary pair, and eventually merge into a single, even larger black hole.
That final merger is one of the most violent events in the universe. In the last moments before the two horizons join, the system blasts out gravitational waves — ripples in spacetime itself — carrying away enormous amounts of energy. And here is the key: if those waves are not emitted equally in all directions, the leftover black hole receives a "kick," just like the recoil of a gun when a bullet leaves the barrel. Physicists call this gravitational recoil, and general relativity says it must happen.
The kick can be extreme — fast enough to displace the black hole from the center of its galaxy entirely. But the ejected black hole does not simply vanish. If the kick is not strong enough to escape the galaxy altogether, gravity eventually hauls it back. It falls through the galactic center, speeds through, slows on the other side, and turns around — oscillating around the nucleus for a very long time, like a pendulum. And just as a pendulum moves fastest at the lowest point of its swing, the displaced black hole moves fastest as it passes through the densest part of the galactic center.
How Do You Measure the Speed of an Invisible Object?
Before 2016, the problem was observational. Individual "offset quasar" candidates had been spotted — quasars whose bright cores did not line up with their host galaxy centers — but every single case had an alternative explanation. Proving a black hole had been kicked looked almost impossible.
So Peter Raffai, Zoltán Haiman, and Zsolt Frei took a different path. In 2016, they built computer models of what a population of oscillating, ejected black holes would look like in real survey data, and asked: what would we actually see?
Their models pointed to a clever two-part signature. The first part is velocity. Some quasars are "active" — their black holes are greedily consuming surrounding gas and blazing with radiation. Gas very close to the black hole travels with it, while gas farther out stays put. Light from the two regions is therefore shifted slightly relative to each other — the same Doppler effect that makes an approaching ambulance siren sound higher-pitched. That small shift reveals how fast the black hole is moving relative to its host galaxy.
The second part is dust. The galactic center is the dustiest region of a galaxy, and dust reddens light — the same reason the setting sun looks red through a thick layer of atmosphere. If a quasar looks redder than comparable objects, more dust is obscuring it.
Put the two together, and the 2016 team made a specific prediction: the kicked black hole moves fastest exactly where the dust is thickest — at the galactic center. So quasars hosting faster-moving black holes should, on average, be redder and more dust-obscured than quasars with slow-moving ones.
100,000 Quasars Say: The Prediction Holds
That is the prediction Bécsy and colleagues have now tested. They selected roughly 100,000 quasars from the Sloan Digital Sky Survey with high-quality data, measured each black hole's velocity and each quasar's redness, and checked whether the two were related.
The relationship emerged clearly. Quasars containing faster-moving black holes are, on average, redder — more dust obscures them. Natural variation between individual quasars makes the effect nearly impossible to spot in any single object, but across this enormous sample the signal is statistically very strong. The most striking difference appears among the dustiest quasars: the share of heavily dust-obscured objects is three times higher among quasars with the fastest-moving black holes than among those with the slowest.
The team also ran validation tests that make the result harder to dismiss. The correlation held under different ways of selecting and measuring the data — but it vanished when they compared only light from gas clouds that do not follow the black hole's motion. That is exactly what you expect if the velocity measurement is really tracking the black hole itself.
As Bécsy put it, proving that a single quasar's black hole was ejected is very difficult, because other phenomena can produce similar signatures. What is new here is that the team did not hunt for one unusual object — they searched statistically for the traces of an entire population.
Why Ejected Black Holes Matter for Galaxy Evolution
If this interpretation holds up, the implications reach far beyond one survey. The strength of the signal suggests that in roughly half of the quasars studied, two supermassive black holes merged relatively recently — although the authors stress that this estimate is uncertain and other physical processes cannot be completely ruled out.
That frequency matters for how galaxies grow. Supermassive black holes and their host galaxies grow together in a tight relationship: bigger galaxies host bigger central black holes, and the black hole's energy output regulates star formation across the galaxy. But what happens when the black hole is suddenly not at the center anymore?
A wandering black hole changes the story. A kicked black hole carries its accretion disk with it, so it can keep shining as a quasar while displaced — an "offset quasar" wandering through its own galaxy. A galaxy that recently lost its central black hole may grow an unusually large, diffuse core, because the oscillating black hole scours stars out of the center as it swings back and forth. Some galaxies observed with mysteriously empty or oversized cores may be the aftermath of ancient kicks.
And the biggest kicks can be final: a black hole ejected from its galaxy entirely becomes a rogue, drifting through intergalactic space with nothing but its fading accretion glow to mark it. The new result hints that the universe may be littered with such wanderers.
This is why large sky surveys have become such powerful tools for answering deep questions. Modern catalogs let astronomers test ideas statistically across hundreds of thousands of objects — the same approach that is reshaping everything from cosmology, as in our coverage of CHIME's detection of hydrogen glow from the early universe, to planetary science, as with Euclid's first comet discovery.
What LISA Will Settle
The most exciting consequence concerns the future. In 2035, the European Space Agency plans to launch the LISA space observatory: three spacecraft flying in formation 2.5 million kilometers apart, detecting gravitational waves by measuring tiny changes in the distances between them. LISA is designed to hear precisely these events — the mergers of supermassive black holes.
Here is the connection. If the new signal really is caused by ejected black holes, then supermassive black hole mergers are common — which directly determines how many such events LISA will be able to detect. A universe full of frequent mergers means a rich harvest of signals; a universe of rare mergers means a quieter sky. The ELTE team has been involved in LISA's scientific preparations since 2006, and their waveform-modeling work will be essential for extracting black hole masses, distances, and fates from LISA's future signals.
Professor Zsolt Frei, head of the HUN-REN–ELTE Extragalactic Astrophysics Research Group, noted that few research groups get to maintain this kind of continuity: the recoil idea was put forward at ELTE a decade ago, and now — led by Bécsy, who graduated from the same university — the group has tested it against real data themselves.
How Certain Is It?
Honest science reporting requires the caveats. This is statistical evidence, not a smoking gun. No single quasar in the sample can be pointed to as a confirmed kicked black hole. The estimate that mergers happened recently in about half the sample is, in the authors' own words, uncertain. And the correlation between velocity and dust, while predicted by the recoil model, is the kind of signature that demands independent confirmation — ideally from other surveys and more detailed models.
That is exactly what comes next. The authors emphasize that further observations and better models are needed. If confirmed, the method becomes a general tool: measuring how often supermassive black holes merge, how fast they are ejected, and what environments they carry with them. And within a decade, LISA will listen to the mergers themselves, testing the whole picture from the other end — the gravitational waves, not just their dusty fingerprints.
Either way, one idea is now harder to ignore: the monsters at the centers of galaxies are not always sitting still. Sometimes, the most violent event in the universe gives them a kick — and the evidence of that kick may have been hiding in quasar light all along. It is a reminder that even our most settled assumptions deserve a second look, much like JWST's discovery of a lopsided exoplanet atmosphere challenged assumptions about distant worlds.
Source: Bence Bécsy et al., "Statistical evidence for massive black hole recoils in active galactic nuclei," Monthly Notices of the Royal Astronomical Society (2026), DOI: 10.1093/mnras/stag1367. Based on the October 22, 2026 report from Eötvös Loránd University.