The heaviest black holes are built by merging smaller ones, 153 mergers show

Some black holes appear to be made of other black holes. A study of 153 collisions detected as ripples in spacetime finds that the universe’s heaviest black holes were not forged in the death of a single massive star, the route textbooks describe, but assembled step by step through earlier mergers. If the result holds, the cosmos runs something like a recycling line for its most extreme objects.

The evidence is a break in the numbers. When astronomers sort the colliding black holes by mass, the population thins out around 45 times the mass of the Sun. Below that line, the objects match what a dying star can produce on its own. Above it, they do not, because a single collapsing star hits a ceiling: stars in that mass range are torn apart by a runaway instability before they can leave a black hole behind.

What fills the gap is a second generation, and these heavier black holes carry a different fingerprint in how they spin. Black holes born from a pair of stars that lived and died together tend to spin in step, their axes roughly aligned. The objects above the line spin fast and point every which way, the signature of a chaotic history in which black holes met as strangers and merged.

That history needs a crowded room. The mergers trace back to dense star clusters, where stars and their dark remnants pack together up to a million times more tightly than they do in the Sun’s quiet neighborhood. Black holes sink to the center, pair off, collide, and the product stays to find another partner. Each round builds a heavier object than the last.

The team behind the analysis, led by Fabio Antonini at Cardiff University with Isobel Romero-Shaw and Fani Dosopoulou, did not watch any of this directly. They worked from the catalog of confident gravitational-wave detections gathered by the LIGO, Virgo and KAGRA observatories, reading the mass and spin of each collision off the shape of its signal and testing whether the 153 events split into two distinct families.

The reading comes with caveats. Gravitational-wave detectors notice heavier and closer mergers more easily than light, distant ones, which can skew any census drawn from them. A sample of 153 is still small for slicing into sub-populations, and the break near 45 solar masses is a statistical thinning rather than a hard wall.

That is where the next few years matter. Upgrades to the detectors and a new observing run are expected to multiply the number of recorded collisions, sharpening the census on either side of the line. The analysis appeared in Nature Astronomy in May 2026, and it hands the growing pile of black-hole collisions a specific claim to test: that the biggest ones were never born big at all.