Rohan and Gondor are two supermassive black holes — and we just found them

"The beacons were lit!" That's how Chiara Mingarelli, the Yale astrophysicist who led the new NANOGrav search, explains why two freshly-identified supermassive black hole binaries are now nicknamed Gondor and Rohan. The reference is Tolkien — but the finding is real, and it does what pulsar timing arrays have been trying to do for fifteen years. It picks out individual black hole mergers from inside the nanohertz gravitational wave background, the low-frequency hum that NANOGrav, the European Pulsar Timing Array, and the Parkes Pulsar Timing Array first confirmed in 2023. Two candidates have risen to the top. They are the first concrete benchmarks for a detection method that turns the background into a map.

What the new paper actually did

The paper was published on February 5, 2026, in The Astrophysical Journal Letters (DOI: 10.3847/2041-8213/ae3719), with Mingarelli as corresponding author and an international team spanning Yale, NANOGrav, and several European institutions. The setup is precise. Take NANOGrav's 15-year pulsar timing data set — the same one that produced the 2023 evidence for the gravitational wave background — and combine it with electromagnetic priors from 114 active galactic nuclei that have quasars at their centers. The team's working hypothesis, published separately in 2025, was that supermassive black hole binaries are about five times more likely to live in galaxies hosting quasars than in quiescent galaxies. The new study tests that hypothesis by searching for continuous gravitational waves from each of the 114 candidate hosts. (Source: Mingarelli et al. 2026, ApJL; NANOGrav news; Yale Physics)

Two targets rose to the top of the search. The first, SDSS J1536+0411, is now nicknamed Rohan — partly for the region of Middle-earth, partly for Rohan Shivakumar, the Yale undergraduate who first analyzed it. The second, SDSS J0729+4008, is Gondor — and the beacon imagery is Mingarelli's: in The Return of the King, the beacons of Gondor call the riders of Rohan to ride to war. Here, the quasars are the beacons. Pulsar timing is the rider.

The detection itself is not yet a confirmed continuous-wave signal — it is a set of gravitational-wave strain upper limits per source, plus the two candidates where the analysis framework returned the strongest motivation for follow-up. The paper's authors are careful about that distinction. What the two targets do establish is that the protocol works end-to-end: from a list of plausible hosts, through electromagnetic priors and gravitational-wave search, to a short list of priority candidates for the next data release. As Mingarelli put it: "Our work has laid out a roadmap for a systemic supermassive black hole binary detection framework. We carried out a systematic, targeted search, developed a rigorous protocol — and two targets rose to the top as examples motivating follow-up."

Why this is harder than it sounds

The challenge is one of signal-to-noise. The 2023 gravitational wave background result was a stochastic detection: a hum in the data, present in many pulsars at once, with the right cross-correlation signature (the Hellings-Downs curve) to distinguish it from local noise. That signal comes from the combined gravitational-wave output of thousands of supermassive black hole binaries orbiting each other slowly in galaxy merger remnants across the universe. Each individual binary contributes only a fraction of the total background — and each one's signal is, by design, hard to isolate.

Continuous gravitational waves from a single binary, by contrast, are quasi-monochromatic. A supermassive black hole binary inspiraling toward merger produces gravitational waves at a frequency set by its orbital period, which for these systems is months to years. That frequency is far below what LIGO can hear — LIGO's band is tens to thousands of Hz, while NANOGrav operates in nanohertz (periods of years). To detect a single source against the background of all the others, you need either (a) enough signal that it stands above the rest, or (b) enough independent information that you can correlate the timing residuals across multiple pulsars in the specific pattern the source should produce. The new paper takes the second route, and adds the prior that quasar-bearing galaxies are five times more likely to host a binary than average.

That's a meaningful prior. Without it, the search would have to cast an enormous net over thousands of potential hosts. With it, the team could focus on 114 candidates and apply the targeted continuous-wave search framework to each. The improvement in sensitivity is roughly a factor of two compared to an all-sky search, because the priors shrink the parameter space (sky location, distance, chirp mass, frequency) the analysis has to explore.

The two candidates that rose are not confirmed detections. They are the targets where the strain upper limits came in tightest relative to the prior expectation — the places where the next data release is most likely to either confirm or rule out a signal. This is a method paper with two test cases. That distinction is important for anyone reading the press coverage.

What changes for the field

Three things move forward with this paper.

1. The pulsar timing array program has a roadmap to individual sources. Until now, NANOGrav, EPTA, and PPTA have demonstrated the existence of the background but not a confirmed individual source. Individual sources matter because they let you measure the masses, sky locations, and orbital properties of specific black hole binaries — and a population of those measurements tests the astrophysical models of galaxy merger rates and binary evolution. The roadmap Mingarelli's team describes is the first concrete protocol to find them systematically.

2. The electromagnetic priors matter. The choice to focus on quasar-bearing galaxies — rather than all AGN, or all massive galaxies — is what made the targeted search tractable. It also means the next round of candidates is most likely to come from time-domain surveys that catch quasars in variable states. The Vera Rubin Observatory's Legacy Survey of Space and Time, now in its fifth year of operations, is the most likely upstream feed. Expect Rubin-identified quasar variability to start showing up as NANOGrav target lists over the next two years.

3. The next data release is the test. NANOGrav's 17-year data set is in active development; IPTA's combined second data release (DR3) is expected within the next eighteen months and will combine NANOGrav, EPTA, PPTA, InPTA, and Chinese Pulsar Timing Array data. If Rohan and Gondor survive that joint analysis with strain upper limits consistent with a binary, the field will treat them as confirmed candidates. If not, they fall back to "interesting upper limits" — still useful, still publishing, just not the first individual source detections the field has been waiting for.

Why this matters now

Pulsar timing array science is one of the quieter fields in astrophysics. The data are taken on decade timescales, the analysis pipelines take years to mature, and the headline results happen on roughly five-year cycles. The 2023 background detection was the first. The first confirmed individual source — if it comes — will be the second. Between those two events, the field's job is to develop the analysis tools and the target lists that make the second event possible.

What Mingarelli's group did is publish one of those tools. It is the first end-to-end pipeline that goes from a published gravitational-wave strain upper limit per source to a prioritized list of follow-up targets, using electromagnetic priors that are themselves a research product. The two test cases are not the destination. They are the proof of concept.

The other reason this matters now is the broader gravitational-wave landscape. LIGO is now in its fourth observing run, with the next-generation ground detectors (Cosmic Explorer, Einstein Telescope) in late-stage planning. LISA, the space-based detector led by ESA with NASA partnership, is expected to launch in the mid-2030s and will operate in the millihertz band — between NANOGrav's nanohertz regime and LIGO's audio band. NANOGrav is the only current facility operating in its band, and the techniques developed for it feed directly into LISA's source identification problem. A method that works for nanohertz continuous-wave searches is, with appropriate modifications, a method that works for millihertz ones. The pipeline Mingarelli's team published in February is one of the inputs that LISA teams are now reading.

Closing

The beacons are lit, and the riders are on their way. Whether Rohan and Gondor are confirmed in the next eighteen months or take another full data release cycle, the protocol that put them at the top of the list is now a public tool. That is the substantive news from this paper, more than the names themselves. The next round of pulsar timing array data will be the first real test.

For background on why the nanohertz band matters and what the 2023 detection actually showed, see our LEO explainer for the broader orbital context and our orbital shells primer for how gravitational-wave observatories fit into the wider sky-monitoring picture.

Sources

• Mingarelli et al. 2026, The Astrophysical Journal Letters — DOI: 10.3847/2041-8213/ae3719
• NANOGrav — "The beacons were lit!" A system to detect and map merging black holes (Feb 5, 2026)
• Yale Physics — Lighting the beacons to find merging black holes and design a new roadmap
• Phys.org — Detection system uses gravitational waves to map merging black holes (Feb 5, 2026)
• NASA SpaceNews — To map merging black holes: NANOGrav's new protocol (Feb 2026)
• Yale News — Astrophysicists present first evidence of gravitational wave background (June 28, 2023)
• NASA LISA mission page
• International Pulsar Timing Array
• NANOGrav 15-year data set: Search for anisotropy in the gravitational-wave background (IOPscience)

Mira Okafor

Senior Editor · Orbital Mechanics