When two black holes spiral together and merge, they send ripples through spacetime that detectors on Earth can pick up. After years of collecting these signals, gravitational-wave scientists now have enough data to see something striking: the merging black holes do not all look alike. A statistical analysis drawing on the largest catalog of gravitational-wave events to date finds that binary black hole mergers sort into at least three distinct groups, separated by sharp mass boundaries near 27.7 and 40.2 times the mass of our Sun.
The result suggests that different astrophysical processes are producing these collisions. Some mergers likely originate from pairs of massive stars that evolved together and collapsed. Others may have been assembled through gravitational encounters inside dense star clusters. A third channel could involve stars born in metal-poor environments. If the finding holds up under peer review and further observation, it would mark a turning point in how researchers model stellar death and the environments that forge the universe’s heaviest objects.
What the data show
The foundation for this work is GWTC-4, the latest gravitational-wave transient catalog from the LIGO-Virgo-KAGRA (LVK) collaboration. LIGO operates two detectors in the United States (in Louisiana and Washington state), Virgo is based in Italy, and KAGRA operates in Japan. Together, they form the most sensitive network ever built for detecting spacetime ripples from colliding compact objects.
GWTC-4 incorporates data from the first segment of the collaboration’s fourth observing run (called O4a) along with all prior runs. It added 128 new compact-binary merger candidates with an astrophysical probability of at least 0.5. Of those, 86 carry a false alarm rate below one per year, meaning the signals are very unlikely to be noise. Combined with earlier detections, the dataset used for population-level studies now contains 158 binary black hole mergers.
That number matters. With dozens of events, earlier catalogs could identify broad trends in the mass distribution. With 158, researchers can start testing whether the distribution contains internal structure: peaks, valleys, and boundaries that separate one group of mergers from another.
Three subpopulations, not one smooth curve
A separate study, which is not an official LVK collaboration paper but was built on the publicly available GWTC-4 data, tested whether multiple subpopulations fit the observations better than a single smooth distribution. The authors found that the data favor at least three subpopulations with sharp primary-mass transition boundaries, at 90 percent credibility.
Black holes below about 28 solar masses behave as one statistical group. Those between roughly 28 and 40 solar masses form a second. And those above 40 solar masses constitute a third.
The LVK collaboration’s own population properties analysis, working from the same 158 mergers, independently identified persistent features that align with this picture. Over-densities cluster near 10 and 35 solar masses, with a possible third feature near 20 solar masses. Gaps sit between those peaks. Notably, the mass peaks near 10 and 35 solar masses were already visible in analyses of the earlier GWTC-3 catalog; the expanded GWTC-4 dataset reinforces those features with greater statistical weight.
Spin measurements from the same dataset show the merging black holes have moderate spins, ruling out scenarios in which they approach their theoretical rotational limits before colliding.
Why three groups matter
Different formation channels predict different mass signatures, which is what makes the three-group result so significant.
Isolated binary evolution occurs when two massive stars in a close orbit both collapse into black holes and eventually merge. This channel tends to produce black holes in a particular mass range dictated by how stellar winds strip material from the stars before they die. Supernova physics also plays a role: certain explosion mechanisms can prevent black holes from forming in specific mass windows, creating gaps in the distribution.
Dynamical assembly happens inside globular clusters or nuclear star clusters, where gravitational interactions can pair up black holes that were not born together. Because these environments allow repeated mergers, where the product of one collision finds a new partner and merges again, they can build black holes to higher masses than isolated evolution typically allows.
A third pathway involves stars with very low metallicity (few elements heavier than hydrogen and helium). These stars lose less mass to stellar winds over their lifetimes, so they can leave behind heavier black hole remnants. Such stars were more common in the early universe, but pockets of low-metallicity star formation persist even in the present day.
The three-subpopulation result is the strongest statistical signal yet that more than one of these channels is actively producing the mergers LVK detects. Earlier analyses hinted at structure in the mass distribution, but the expanded GWTC-4 dataset sharpens those hints into a statistically meaningful pattern.
What remains uncertain
The data favor three groups, but the analysis does not yet pin each group to a specific formation pathway. The transition masses at 27.7 and 40.2 solar masses are inferred boundaries, not hard physical thresholds. Their 90 percent credibility intervals leave room for the true boundaries to sit several solar masses higher or lower, and additional detections during the remainder of the O4 run could shift them.
Spin distributions offer one promising way to break the degeneracy. Dynamically assembled binaries in dense clusters tend to have randomly oriented spins, while isolated binary evolution can produce spins aligned with the orbital plane. The LVK population properties paper provides general spin constraints across the full sample but does not yet break those constraints down by subpopulation. Until spin measurements are mapped onto each mass group separately, the link between a given group and a specific astrophysical environment remains indirect.
The three-subpopulation paper is not an official LVK collaboration publication. It is hosted on the arXiv preprint server and has not yet appeared in a peer-reviewed journal. Preprint circulation is standard practice in gravitational-wave astronomy, where large collaboration papers routinely go through community review before formal journal acceptance. But the absence of peer review means the statistical methods and model choices have not yet passed independent editorial scrutiny.
There is also the question of the possible mass feature near 20 solar masses. It sits at lower statistical confidence than the peaks near 10 and 35 solar masses. If it fades with additional data, the case for three cleanly separated groups could narrow to two well-established populations plus a transitional zone between them.
What the remaining O4 run and future detections will test
The O4 observing run is still underway as of mid-2026, and each new detection feeds into tighter population statistics. More mergers will reduce noise in the mass distribution, clarify whether the 20-solar-mass feature is real, and narrow the credibility intervals around the transition boundaries. Full O4 results, expected after the run concludes and the collaboration completes its analysis, will provide the next major update. Peer review of the three-subpopulation paper, if it proceeds on a typical timeline, could conclude within the next year.
The community will also be watching for independent checks on the statistical framework. Alternative analyses using different parameterizations of the mass function, or nonparametric methods that let the data speak without assuming sharp transitions, can test whether the same broad structure appears regardless of modeling choices. Agreement across methods would significantly bolster confidence that the observed groupings reflect genuine features of nature.
If spin orientations can be reliably measured for a large subset of events and correlated with mass groups, that would provide the clearest test yet of which formation channels dominate in each mass range. Misaligned spins concentrated in the heaviest group, for instance, would strengthen the case for dynamical assembly in dense star clusters.
For now, 158 black hole mergers have sketched a mass distribution with real internal structure, and the simplest reading of that structure points to at least three distinct populations. Each additional ripple in spacetime that reaches Earth’s detectors will either reinforce those boundaries or force scientists to redraw them.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
