A graveyard of black holes, mapped in a new catalog

The universe is quieter than it used to look. On 26 August 2025 the LIGO, Virgo and KAGRA collaborations released the fourth version of their Gravitational Wave Transient Catalog, adding 128 new significant candidates detected during the first part of the fourth observing run, which began on 24 May 2023. Set against the electromagnetic sky, the number is modest. Set against the first detection a decade and a half ago, it describes a field that has turned event-catching into a census.

The catalog is best read as a map of the stellar graveyard. Each gravitational-wave event is the ringdown of two compact objects merging, and from the waveform the collaborations infer the masses and spins of the objects involved. Plot those on a graph and a population structure appears: a population of binary black holes distributed across tens of solar masses, a smaller scattering of neutron stars, and gaps that theorists now argue about. The new release widens the statistics enough that the shape of that distribution is starting to constrain how massive stars end their lives.

One candidate illustrates why the catalogue matters. GW231123, flagged in the release, is likely the highest-mass binary black hole merger observed to date. Very heavy mergers sit in a mass range where ordinary stellar evolution struggles to produce a clean pair, because the pair-instability of very massive stars tends to blow them apart rather than collapse. Finding one out at the edge sharpens the question of whether these giants form directly, through hierarchical mergers of smaller black holes, or through some channel the models have missed. A single event would be ambiguous; a population of such events starts to discriminate.

What the release does not yet contain is as telling as what it does. The companion papers being prepared for an Astrophysical Journal Letters special issue include a search for gravitational lensing signatures, and after four observing runs the lensed, repeated signals that cosmology would love to exploit remain elusive. Strong lensing would let the network measure Hubble tension and test general relativity in a new regime. The fact that the catalogue has grown this large without a confirmed lensed echo sets a real upper bound on how common those alignments are, which is itself a cosmological statement.

The scale of the operation is worth pausing on. Across four observing runs the network has issued more than three hundred public alerts in real time, so that astronomers anywhere can point telescopes at the patch of sky where a merger may have lit up. O4a was only the first segment of a run scheduled to continue through late 2025, and further releases will follow. Detection has shifted from a red-letter event to a steady stream, and the hard work is now the follow-up: population statistics, rate estimates, and the hunt for the odd event that breaks the model.

There is a humility built into the numbers. A 128-event catalog is a big jump from the first dozen, yet most events are still black-hole mergers, which emit no light and leave no electromagnetic counterpart to study in detail. Neutron-star mergers, the rare ones that also glow, remain precious. The graveyard is well mapped; the luminous side of the story is still a trickle.

For all that, the catalogue is the deliverable a decade of detector engineering was building toward. Each new event is another data point on how stars die, how black holes assemble, and how spacetime itself responds when the two meet. The census is far from complete, but it is finally dense enough to read.