Runaway supermassive black hole · z = 0.964 · seen as it was ~7.7 billion years ago

RBH-1

A black hole of roughly 20 million suns, kicked out of its galaxy, leaving a 200,000 light-year trail of newborn stars. Scrub the timeline to watch the wake form.

[O III] 500.7 nm: bow shock Hα 656.3 nm: cooling gas New stars, under 10 Myr Ageing stars, 30+ Myr
Time
—
BH speed
—
From host GX
—
New stars
—
Today
Wake length62 kpcAbout 200,000 light-years, projected on the sky. Twice the Milky Way's diameter.
Black hole speed954 km/s+110 / −126 km/s, from the JWST bow-shock velocity map. About 3.4 million km/h.
Black hole mass~2×10⁷ M☉At least 10⁷ M☉ from energy conservation. Not measured directly.
Time since kick~70 Myr2023 estimate was 39 Myr at 1,600 km/s. Revised with the slower JWST speed.
Stars formed~2×10⁸ M☉Born in situ from halo gas condensed behind the shock.
Inclination29° +6/−3Tip tilted out of the sky plane, toward us.

How the story changed

  1. A straight, thin line of stars

    Pieter van Dokkum's team found a 62 kpc linear feature pointing at a compact galaxy, ending in a bright [O III] knot. Star colours suggested ages increasing away from the tip. They proposed a three-body slingshot about 39 Myr earlier at ~1,600 km/s, with a binary black hole recoiling the other way to make a fainter counter-feature.

  2. Maybe just an edge-on galaxy?

    Sánchez Almeida and colleagues argued the feature could be a thin galaxy seen edge-on. The later deep data show the light dropping by over 40× right at the tip, which a galaxy disk does not do.

  3. The bow shock is resolved

    An integral-field map shows a ~600 km/s velocity jump across the tip, the signature of a supersonic bow shock. The derived speed is 954 km/s at 29° inclination. Deep HST/UVIS images show no counter-wake on the opposite side.

  4. Merger recoil now favoured

    The host galaxy, GX, has a stellar velocity dispersion of only ~60 km/s, which makes three-body ejections unlikely. With no second runaway in sight, the best fit is a gravitational-wave recoil from two black holes merging ~70 Myr earlier (mass ratio ~3, spinning, precessing).

Two scenarios, side by side

Toggle between them in the viewer. Each uses the numbers from its own paper. Neither is drawn to physical scale for the black hole or the gravitational waves.

Merger recoilThree-body slingshot
StatusFavoured 2025–26Proposed 2023
Black holesTwo merge; the merged remnant is RBH-1Three interact; the lightest is flung out, the binary survives
KickUneven gravitational-wave emission at mergerGravitational slingshot
Speed954 km/s now~1,600 km/s
Age~70 Myr~39 Myr
Counter-wakeNone expected, none seenExpected from the recoiling binary

How the stars ignite: the black hole compresses ~10⁶ K halo gas in a bow shock. Behind it, turbulent mixing with cooler gas makes the gas radiate and cool quickly. The cooled gas condenses into knots that form stars. The viewer models each knot igniting a few Myr after the black hole passes, so the youngest stars sit nearest the tip.

Open resource

Build the scene in Blender

One Python script builds the whole animation with no add-ons: the host galaxy, the black holes, the bow shock, the growing wake and about 400 star-forming knots that ignite on schedule. Choose either scenario at the top of the script. It is tested on Blender 5.1 and written to work on 4.x.

blender --python rbh1_scene.py -- --scenario gw
View rbh1_scene.py
The script is in the project's blender/ folder (rbh1_scene.py).

Sources

  1. van Dokkum et al. 2023, A Candidate Runaway Supermassive Black Hole Identified by Shocks and Star Formation in its Wake, ApJL.
  2. van Dokkum et al. 2025, JWST Confirmation of a Runaway Supermassive Black Hole via its Supersonic Bow Shock, submitted to ApJL.
  3. Progenitor of the recoiling super-massive black hole RBH-1 identified using HST/JWST imaging, 2026 preprint.

The kinematics in the viewer are illustrative. For the recoil scenario the speed eases from ~1,030 km/s to 954 km/s, so the path covers 71 kpc in 70 Myr and projects to 62 kpc at 29°. Knot ignition delays (3–15 Myr) and the stellar-mass counter are scaled to the published totals, not taken from a simulation.