The Milky Way may be lying on its side. Two research teams said so in 2026, working from completely different data and landing in the same place: the Milky Way disk flip, a slow tipping of our galaxy’s flat sheet of stars by more than 90 degrees since it formed, reorienting inside the invisible halo of dark matter that wraps around it.
Two teams, two methods, one strange conclusion. Here is how each got there, and why neither is ready to call the case closed.
The short answer: In 2026 two research teams reported that the Milky Way’s disk appears to have tipped over by more than 90 degrees across billions of years, reorienting inside its dark matter halo after ancient collisions with smaller galaxies. One team read the tilt from the shape of the invisible halo; the other read it from the unexpectedly slow spin of the stellar halo. Both reach the same conclusion, and both stress it is not yet settled.
What a galaxy flip actually means
A galaxy flip means its flat sheet of stars has changed orientation by more than 90 degrees, tipping over inside the larger structure of matter around it. Nothing crashes and nothing breaks. The disk stays whole; only the direction it faces changes.
Kirill Batrakov, the Durham University astronomer behind one of the two 2026 studies, is careful with the word. A flip, in his definition, is a slow lean that eventually passes the 90-degree mark. As Batrakov put it, the disk could “turn on its side, not necessarily upside down.”
The pace is the part that trips people up. This is not a maneuver you could ever watch happen.
Speed up the eons and view it from Earth, and the clearest sign would be the night sky rearranging itself: constellations drifting apart, new ones assembling as the stars slid into fresh positions, far too slowly for any single lifetime to catch.
The clue was in the slow-moving stars
The suspicion started with stars in the galaxy’s outskirts that move too slowly. Mapped by the Gaia satellite over the past decade, the ancient stars of the stellar halo — the sparse cloud of old suns wrapped around the bright disk — circle the center far more sluggishly than the Milky Way’s mass would predict.
That mismatch has an easy explanation and an interesting one. The easy one says earlier work overestimated how much the galaxy weighs. The interesting one says the halo really is spinning slowly, and something long ago knocked the whole system out of its first alignment.
Batrakov’s group went after the interesting explanation.
A dark matter halo standing on its side
A team led by the Chinese Academy of Sciences mapped the Milky Way’s dark matter out to roughly 50 kiloparsecs and found the halo’s flattened plane standing vertical to the disk, an orientation almost no one had predicted. The invisible cocoon around our galaxy is slightly squashed, like a beach ball someone sat on, and the squashed face points sideways relative to the star disk instead of lining up with it.
To get there, Ling Zhu and colleagues built a model that reads the gravitational field straight from the motions of 14,497 K-giant stars, using full six-dimensional positions and velocities from the LAMOST survey and the Gaia mission. Their fit puts the halo’s short axis at 0.92 of its long axis, close to round but measurably flattened, and pins the dark matter within 50 kiloparsecs at about 5.4 × 1011 solar masses.
Here is the oddity. In a settled galaxy the star disk lines up with its dark matter halo, and the small satellite galaxies orbit in the same plane, like plates stacked flat. In the Milky Way, the halo and the satellite swarm agree with each other, and both stand at right angles to the disk. The disk is the odd one out.
- 90 degrees or more — the orientation change that counts as a flip
- ~50 kiloparsecs — reach of the new dark matter map
- 3 of 105 — simulated Milky-Way analogues matching our exact geometry
- 14,497 — halo stars in the LAMOST–Gaia sample
What could cause a Milky Way disk flip?
Small galaxies falling in on backward orbits can slowly torque a disk over, and that gentle push is the mechanism both teams point to. When an intruder spirals in against the direction of the disk’s spin, its gravity drags on the disk for billions of years, tilting it a little more with each pass.
The force involved is tiny. In the closest simulated match to our galaxy, the tilt built up at around 13 degrees every billion years, a rate that works out to a mere 2 kilometers per second at a radius of 10 kiloparsecs. Over the long haul, that whisper of a nudge swung the simulated disk by about 80 degrees.
Put that rate in human terms. At 13 degrees per billion years, across the entire 300,000-year span of our species, the disk would have leaned by roughly four thousandths of a degree, far too little for any astronomer of any era to have charted.
Two suspects fit the Milky Way’s record. The Gaia Sausage, a galaxy that merged with ours some 10 billion years ago, is the prime candidate for starting the tilt. The Sagittarius dwarf, which has looped through the halo again and again, is a strong bet for keeping it going.
Two teams, two methods, one answer
They start from different data and meet in the middle: one study weighs the shape of the dark matter, the other clocks the speed of the stars. Set side by side, the overlap is what gives the flip idea its weight.
Table 1 — How the two 2026 studies reach the same conclusion (Zhu et al., Astronomy & Astrophysics; Batrakov et al., RAS NAM 2026).
| Study | Zhu and colleagues | Batrakov and colleagues |
|---|---|---|
| Lead institution | Chinese Academy of Sciences | Durham University |
| What they measured | 3D shape of the dark matter halo | Rotation speed of the stellar halo |
| Data and tools | K-giants from LAMOST and Gaia | Auriga simulations vs Gaia motions |
| Key result | Halo’s long axis sits vertical to the disk | Slow halo matches an early major merger |
| Their reading | Disk flipped inside a fixed halo | Disk tipped after the Gaia Sausage hit |
| Stated confidence | Geometry strong; the twist tentative | A likely story, not yet certain |
Do the simulations agree?
Up to a point, yes. When Zhu’s team searched 105 simulated galaxies chosen to resemble the Milky Way, only 3 reproduced its exact geometry, and every one of those three had flipped its disk to get there. Thirteen came within a looser statistical range; the rest looked nothing like us.
So the configuration is rare, yet not impossible, and it does not fall outside the standard model of how galaxies grow. Our galaxy comes out special rather than freakish.
Batrakov’s simulations attack the question from the star side. Running the Auriga models, his group found that galaxies which formed early and swallowed a large companion tend to end up with slowly rotating stellar halos, the exact sluggishness Gaia sees in the Milky Way today.
How sure can we be?
Not sure, and both teams say so without softening it. Zhu’s group calls the twist in the halo tentative and carries a wide error bar on one of its axis measurements. Batrakov is just as frank: a flip is a likely story, but too early to claim with full confidence, and astronomers should hunt for other fingerprints of past flips before calling the case closed.
One caution about the coverage. The phrase “flipped sideways” makes the event sound sudden and finished. The papers describe something slower and less certain: a lean still under way, and in Zhu’s case a halo geometry that is measured strongly while the flip invoked to explain it stays tentative.
A third line of evidence keeps the idea alive. Using DESI observations, a separate group mapped the stellar halo out to about 50 kiloparsecs and found it twisted, turning vertical to the disk in the outer regions, a result reached without any reference to the dark matter shape and landing in the same place.
We can’t tell from the public summaries whether the DESI map and Zhu’s halo were tied to the same distance scale. If they weren’t, the neat “three methods agree” headline is a little softer than it sounds, and that is exactly the cross-check we’d want to see next.
Three separate methods, all pointing the same way. That is the kind of agreement that makes astronomers sit up.
Where this leaves the Milky Way
The honest verdict is probably, not certainly. Three independent measurements — the squashed shape of the dark matter halo, the sluggish spin of the stellar halo, and DESI’s twisted map of the outer stars — point to a disk that has swung far from where it began, most likely shoved by the Gaia Sausage and kept turning by the Sagittarius dwarf. What would settle it is a fourth fingerprint, and none of the three groups has found one yet.
If the picture holds, the night sky stops being a fixed backdrop and becomes one frame of a galaxy still finding its footing. We’ll update this page when the next line of evidence arrives.
Gerd Dani, President of FreeAstroScience — Science and Cultural Group
Sources
- Zhu, L., Cai, R., Kang, X., Xue, X.-X., Yang, C., Zhang, L., Mao, S., and Liu, C. (2026). A vertically orientated dark matter halo marks a flip of the Galactic disc. Astronomy & Astrophysics, 706, A193. https://doi.org/10.1051/0004-6361/202557613
- Starr, M. (2026). The Milky Way May Have Flipped Sideways After a Collision With a Sausage-Shaped Galaxy. ScienceAlert. https://www.sciencealert.com/strange-clues-in-the-milky-way-suggest-our-galaxy-may-have-flipped-sideways
- Batrakov, K., and colleagues (2026). Stellar halo rotation and disk flips in Milky Way analogues, presented at the Royal Astronomical Society National Astronomy Meeting 2026, as reported by ScienceAlert.
