Dinner in Andromeda, and the Dust Is Still Doing the Dishes
A Nature study says a small star spiralled into a swollen supergiant in the neighbouring galaxy and the two merged, observed by the James Webb Space Telescope. A rival model says the pair may still be circling each other.
On October 19, 2025, a reporting group called MASTER logged a new object in the Transient Name Server, the registry where astronomers record things that suddenly appear in the sky. It sat in M31, better known as the Andromeda galaxy. It was filed as AT 2025abao [5]. Astronomers now call it Bao.
A study in Nature, announced by NASA on November 18, 2026, says Bao was the death of a partnership. A small star spiralled into the outer layers of a much larger companion, and the two ended up as one [1][4]. NASA's own headline calls it a first-of-its-kind stellar merger.
The evidence is unusually rich. Astronomers had observations of the system from decades before the blow-up, records of it brightening in the run-up, and Webb data taken 47 days after the eruption peaked, the earliest follow-up of this kind of event described in NASA's account [1]. Stars do this often across the universe. Almost never does anyone get a seat for it.
The story is still moving. A separate model argues the two objects may not have merged at all. More on that below.
A giant and a smaller star
Before the eruption, the main star was a bloated giant. The researchers estimate it held about seven times the Sun's mass, stretched to 350 times the Sun's radius and shone with 18,000 times the Sun's brightness [1]. Set where the Sun is, a star that size would reach more than one and a half times the distance from the Sun to Earth. Earth would be inside it.
Its companion was small by comparison, estimated at about 0.7 solar masses, a tenth of the giant's weight [1]. A separate peer-reviewed study in Astronomy & Astrophysics found that the pre-eruption light was consistent with an M giant or an asymptotic giant branch star, two names for an old, swollen, cool star near the end of its life [2].
NASA says the two had orbited one another for millennia. Then the smaller star began interacting with the giant's surface, and the final transformation took about seven years, or seven orbits [1]. The system began brightening unusually in 2023 and erupted in 2025 [1].
How a star eats a star
The mechanism has a name that sounds like a cold-weather garment: the common envelope. The giant's outer layers are thin and puffy, and when a companion dips into them, the two stars end up sharing one cloud of gas [1][4].
The physics is close to dragging a spoon through honey. The spoon slows. In the star's case, the companion loses orbital energy to drag as it moves through the shared gas, and so it sinks inward [1][4]. The lost energy has to go somewhere. It goes into the gas, which is stripped from the giant's outer layers [1].
That phase is short and hard to catch. It is also important, because theory says it is one of the routes by which pairs of stars end up so close together that they can later become compact binaries, including pairs involving neutron stars or black holes. A well-observed event can test how orbital energy is handed to gas and how much material escapes.
Bao is the kind of event astronomers call a luminous red nova. The A&A study traces its light rising to a peak and then holding in a red-band plateau for about 70 days. It is the fourth such nova recorded in M31 [2]. The Transient Name Server also lists it under that classification [5].
What the Webb saw
The merger conclusion is an inference, and the reasoning is worth following. The remnant cannot be seen directly, because dust currently hides it [4]. Webb observed the dust itself in the mid-infrared, the wavelengths that warm dust glows in [1].
What the team could do was add up the gas. Science News reports that the amount of outer gas estimated to have escaped was small, and that the energy of the fast-moving gas fits the picture of the companion reaching the giant's core [4]. NASA's account describes the stars as having merged, and describes the result as a potentially rejuvenated single star [1].
That word, rejuvenated, is the source of the headline-friendly idea of a star reborn. A merger can in principle give an ageing star a fresh supply of material to burn. It is a possibility NASA states with care, and it is the part of the story that cannot yet be checked.
The rival reading
The cleanest picture has a loose end. In an arXiv preprint dated October 5, 2026, a different group models another outcome. It argues that the smaller main-sequence companion and the giant's core could have survived in an eccentric orbit for at least 400 days after the common-envelope phase began, without merging [3]. The preprint has not been through peer review. Its authors say direct observational evidence for a companion is lacking [3].
That is a fair admission, and it cuts both ways. No one has seen a surviving companion. No one has seen a fused core either. Both readings are built from the same sky, one with a peer-reviewed paper and Webb data behind it, the other with a model.
The merger case has the stronger standing today. A rival model is not a refutation. But the open question is a real one: did the cores fully merge, or did they survive in a tight orbit [1][3][4]?
What to watch
The dust is the gatekeeper. As it thins, the remnant should come into view, and the researchers plan to keep monitoring the system while it does [4]. They also want to find other systems in the slow-brightening phase, the 2023-style warning that Bao gave, so that the next common-envelope event is caught before the eruption rather than after [1][4].
Other questions are waiting in the same light. How much mass left the system, and how much is still bound to it? Do the Nature interpretation and the preprint's model make different predictions that later observations can test?
Bao is far enough away that nothing here threatens anyone, and near enough, as these things go, that astronomers could watch it unfold. They now wait for the dust to settle and show them what was left.
Our call
At least one new paper on AT 2025abao, other than the Nature study, the Astronomy & Astrophysics paper and the October 2026 preprint, will be posted to arXiv between November 25, 2026 and February 12, 2027.
Bao is the closest well-observed luminous red nova caught with early Webb data, a preprint already challenges the Nature reading, and the researchers say they will keep monitoring the system as dust clears.
Elena Marsh
Elena MarshSenior writerWrites the long pieces that make people read to the end. Believes every story is a history story if you go back far enough, and usually goes back just far enough.Model: Claude Sonnet 5.5 says
80%. Jev says 32%.
By February 12, 2027. How we will check: Search arXiv for 'AT 2025abao' or 'AT2025abao' in titles and abstracts. The call is true if any paper first submitted between November 25, 2026 and February 12, 2027 presents new observations or new modelling of the object and is not one of the three papers named above.
Open · Every call we made
Every edition in brief, three times a day, on our Telegram channel, on Bluesky and on Threads.
Spotted an error? Tell the editors
- NASA Data Clinches First-of-Its-Kind Stellar Merger - NASA Science NASA
- AT 2025abao: The fourth luminous red nova in M 31 arxiv.org
- Unravelling the Nature of AT 2025abao: From Precursor to Final Fate arxiv.org
- Two stars in the Andromeda galaxy merged before astronomers' eyes sciencenews.org
- 2025abao | Transient Name Server wis-tns.org



