Black holes pile up at 10 solar masses, a hint of failed supernovae

In 153 black-hole mergers from the LIGO–Virgo–KAGRA catalogue, a new model places about three in four in a narrow group near 10 solar masses and finds the merger rate effectively zero between 12 and 16 solar masses, a pattern the authors link to stars that collapse without exploding.

Peer reviewed · Statistical analysis of a public catalogue · Model-dependent interpretation

In most black-hole pairs that LIGO, Virgo and KAGRA catch merging, the heavier black hole weighs about 10 times the Sun, and ones of 12 to 16 solar masses are strikingly rare. Isaac Legred, Jacob Golomb and Katerina Chatziioannou, working at Caltech, argue in Physical Review D (published 10 November 2026) that the 10-solar-mass group could come from failed supernovae: massive stars whose cores collapse straight into a black hole without a successful explosion. Nobody has watched that happen here. The link rests on a statistical model, and the data favour a true gap without requiring one.

Evidence at a glance

Is this a big deal?

It is a solid step on one well-known puzzle, not a discovery of failed supernovae. The 10-solar-mass peak was already established. What is new is a model that lets the merger rate fall all the way to zero above the peak, and the finding that current data are comfortable with that. If later catalogues confirm the gap, it would be strong evidence that the most common merging black holes have their own birth route.

What scientists found

The team modelled the heavier black hole in each merging pair (the “primary”). In their main model, the peak sits at 9.95 solar masses (90% range 8.70 to 10.48) and is narrow, with a spread of about 0.7 solar masses. The model assigns 76% of merging binaries to this group (90% range 59% to 83%).

Above the peak the rate collapses. At the peak it is 7.36 mergers per solar mass per year per cubic gigaparsec, a cube about 3.26 billion light-years on each side. Between 13 and 15 solar masses the highest value is 0.24 in the same units, roughly 30 times lower, and its 90% range reaches zero. The rate is consistent with “effectively zero” from 12.0 to 16.1 solar masses, then climbs again where a second, heavier population switches on near 16.2 solar masses.

The two groups also spin differently. The effective spin of the 10-solar-mass black holes, a measure of how much their spins line up with the orbit, is tightly clustered (spread 0.07), while the heavier group’s is about twice as broad (0.13). Caltech’s release quotes Chatziioannou: “Our data are showing that this group of black holes may have a common origin.”

How they found it

Gravitational waves are ripples in space-time. When two black holes spiral together, the detectors record a signal whose shape reveals the masses. Each mass comes with uncertainty, so the team did not simply count events in mass bins. They fitted a population model, a mathematical description of how many black holes exist at each mass, to all events at once, and corrected for the fact that heavier mergers are easier to detect.

Their model has two parts. One is a narrow bell curve for possible failed-supernova black holes. The other describes every heavier black hole, with its own start mass. Earlier models forced that heavier population to extend down underneath the 10-solar-mass peak. This one does not, which is what allows a gap to appear if the data want one.

Think of a lost-property office full of size-42 shoes, almost no 44s or 45s, then a scattering of larger sizes. You would suspect the 42s came from one team. The comparison breaks down in one way: the counters here cannot read sizes exactly, and a few uncertain shoes may belong in the empty shelf.

That matters because no merger has its primary mass firmly inside 12 to 16 solar masses, yet under a no-gap version of the model, four or five events have about half their mass estimate there. When the model allows a gap, those events get nudged out of it.

What a failed supernova is

A massive star ends its life when its core collapses. Often the collapse drives a supernova that blows the outer star apart and leaves a neutron star. In a failed supernova, the explosion never gets going and the star falls inward to form a black hole. You can read how this fits the wider story in our guide to the life cycle of a star.

The paper draws on computer simulations of dying stars. In one set of models for stars with the Sun’s mix of heavy elements (solar metallicity), only stars born with 11 to 14 solar masses failed to explode and made black holes. Another simulation series found stars born near 12 solar masses were the only failures. Stars slightly lighter or heavier exploded instead. A narrow band of failing stars could therefore feed a narrow band of black holes, which is the pattern the catalogue shows.

Why it matters to you

Nothing here affects daily life. Its pull is wonder: it is a clue to how the commonest merging black holes are born, read from vibrations in space-time that were first detected only in 2015. Our explainer on how black holes form covers the basics.

What the evidence does not show

It does not show that any particular black hole came from a failed supernova. The authors call the gap consistent with the data, preferred but not required. Their more conservative reading is that the rate at 12 to 16 solar masses is orders of magnitude lower than at 10, with a lower limit that is poorly constrained.

The answer depends on modelling. Swapping the description of heavier black holes from “power law plus peak” to a “broken power law” shrinks the zero-rate range and lowers the chance of a separation from 0.88 to 0.61. The Bayes factor between the two models is about 2.6, which the authors call neither strong nor easy to interpret.

Two further caveats. The gap appears only in the primary mass, not the lighter partner’s mass, and sorting each pair into heavier and lighter can itself shape the distributions. Repeated mergers, in which a black hole made by one collision merges again, could also put objects into the gap.

The “effectively zero” label has a concrete meaning. The authors estimate current detectors have surveyed about 20 cubic gigaparsec-years for pairs of 15-solar-mass black holes. At their threshold of 0.001 per solar mass, across a 5-solar-mass interval, that predicts about 0.1 detections (0.001 × 5 × 20), so a rate that low would be invisible.

Show the maths: how steep is the drop, and why does “effectively zero” mean zero detections?

Peak rate ÷ highest rate at 13–15 solar masses: 7.36 ÷ 0.24 ≈ 31, so roughly 30 times lower (medians; the 90% range at 13–15 reaches zero).

Expected detections at the threshold: rate × mass interval × surveyed volume-time = 0.001 per solar mass per year per Gpc³ × 5 solar masses × 20 Gpc³·yr = 0.1 detections.

Threshold versus peak: 7.36 ÷ 0.001 ≈ 7,400, so “effectively zero” is thousands of times below the peak; the paper describes it as a drop of about 10⁴.

All inputs are medians or estimates quoted in the paper; the divisions are SciQuest’s.

How this fits what we already knew

The peak near 10 solar masses has shown up in many analyses. Earlier work with the GWTC-3.0 catalogue found the rate dropping by factors of roughly 10 to 100 between the peak and about 15 solar masses, but flexible models did not find a true gap. One earlier model built specifically to search for a dip at 12 to 15 solar masses found a nonzero rate throughout; the new paper attributes that to its assumption that both black holes in a pair share one mass distribution.

A simpler explanation for the peak, that some stellar cores are unusually compact and collapse easily, predicts an excess of only about two to three times. The observed peak is at least 50 times the rate nearby, according to the authors, so they argue another mechanism is needed.

Independent perspective

SciQuest found no on-record comment from an independent astrophysicist and has not interviewed one for this story. The APS Physics synopsis presents the work as a model that explains the distribution, not a direct detection, and notes that why the heavier population starts near 16 solar masses is an open question.

How other headlines framed it

• Caltech’s news release frames the result as “hints” of a star population that failed to explode, and says the 12-to-16 deficit was found “for the first time”. The paper itself cites earlier studies that saw a strongly suppressed rate in a similar range; the new element is a model in which that rate can reach zero.
• Phys.org republished the Caltech release under a headline about black holes “born from failed supernovae”, hedged with “hint”.
• APS Physics uses a firmer verb, saying the observations “implicate” failed supernovae, and attributes the failures to compact cores. The paper is more cautious: it says core compactness alone cannot explain the size of the peak.

What happens next

LIGO, Virgo and KAGRA keep adding events, and the gap test will tighten. With the 257 binaries of GWTC-5.0, released while the paper was under review, the separation probability was 0.85 and the zero-rate range was 12.3 to 15.7 solar masses, so the picture held. The authors suggest that modelling repeat mergers through their spins could sharpen the test, and that fully general-relativistic simulations of failed supernovae will be needed to predict the peak’s detailed shape. The Caltech release adds that searching historical records for massive stars that failed to explode is another route. For another way gravitational waves are being used as a measuring tool, see our report on the first Hubble constant from gravitational waves and galaxies.

In one sentence

Merging black holes crowd near 10 solar masses and are nearly absent from 12 to 16, and a new model suggests that pattern could come from a narrow band of stars that collapse without exploding, though the gap is favoured rather than proven.

Quick check: did it click?

1. Did anyone see a failed supernova in this study? No. The link comes from fitting a population model to gravitational-wave catalogue data.
2. In what mass range is the merger rate consistent with zero in the main model? About 12.0 to 16.1 solar masses, for the heavier black hole in each pair.
3. Why is the gap not yet confirmed? The data also allow a small nonzero rate, and a different model for heavier black holes weakens the case (0.61 instead of 0.88).

Read next

• Life cycle of a star: every stage, from gas cloud to remnant, for where supernovae and black holes fit.
• How is a black hole formed?, the background on stellar collapse.
• First Hubble constant from gravitational waves and galaxies, another use of the same detectors’ catalogue.

Sources and disclosure

1. Legred I., Golomb J., Chatziioannou K. “Low-mass failed supernovae and the 10 M_☉ peak in the merging black hole mass distribution.” Physical Review D 114, 103023 (2026). doi:10.1103/mnzq-w1b5. Full text read in arXiv:2604.01420v2 (first posted 1 April 2026; version 2, 9 November 2026).
2. Caltech news release, “Astronomers Find Hints of Star Population That Failed to Explode,” Whitney Clavin, 10 November 2026: caltech.edu.
3. Sophia Chen, “Gravitational-Wave Observations Implicate Failed Supernovae,” Physics 19, s149 (APS), 10 November 2026: physics.aps.org.

Disclosure: This article was drafted with AI assistance and checked against the sources above. The study was funded by the US National Science Foundation, the US Department of Energy and the Simons Foundation, according to the paper. Ratios (peak versus 13–15 solar masses, about 7.36 ÷ 0.24) and the 0.1-detection estimate were rechecked by SciQuest; the graphic is a SciQuest original with schematic curve shapes and the paper’s values. Spotted an error? See our corrections policy.