Galaxy DF9 Moves at 6.5 km/s, Where Dark Matter Predicts 24

A faint galaxy called DF9 appears to contain little or no dark matter, astronomers report, making it the third such oddity in one trail and a possible test for theories that try to do without dark matter.

The study, by Michael A. Keim, Pieter van Dokkum, Zili Shen, Shany Danieli and Imad Pasha, was published in The Astrophysical Journal on June 16, 2026. ^[2][3] It adds DF9 to DF2 and DF4, two galaxies earlier reported to have little or no dark matter. All three sit in a narrow line of faint galaxies in the field of NGC 1052. ^[1][3][4]

Why a missing ingredient matters

Dark matter is the invisible stuff that astronomers say supplies most of the gravity holding galaxies together. Nobody has caught a particle of it. Its existence is inferred from how galaxies move, which is why a galaxy without it is so odd. Ordinary galaxies seem to come wrapped in it.

The rival camp holds that there is no such stuff and that gravity itself behaves differently than Newton said. The best known version is modified Newtonian dynamics, or MOND, proposed by the physicist Mordehai Milgrom in 1983. It changes how gravity acts at very low accelerations, to explain galactic observations without invoking dark matter. ^[1]

A galaxy built from the same ordinary matter as the others but missing the extra pull is awkward for that picture. That is why three of them in a row draw attention.

What the Keck telescope saw

The team pointed the Keck Cosmic Web Imager, an instrument on the W. M. Keck Observatory's telescope, at DF9. They observed over two nights in October 2024, for 10.83 hours of combined exposure. ^[4] DF9 is so faint that this was a hard measurement.

The method rests on a simple idea. The more mass a galaxy holds, the faster its stars and star clusters swirl around inside it. The spread in those speeds, called the velocity dispersion, therefore works as a scale for weighing the galaxy.

The measured dispersion in DF9 is 6.5 kilometres per second, with an uncertainty of plus 3.9 and minus 4.3. The stars alone should produce 8.3 km/s, give or take about one. ^[2][4] The two agree.

A typical dark-matter halo would change the picture sharply. DF9's stars add up to about 140 million solar masses. Give it a halo of 14 billion solar masses, a hundred times heavier, and the authors' comparison predicts 24 km/s, give or take three. ^[2][4] That is nearly four times what Keck saw.

The authors conclude that dark matter is not required to explain DF9's motions. ^[2] That is a statement about weight, not a sighting. The result is evidence for a low mass, not a direct measurement of dark matter's absence.

The error bars are wide, and the paper says only two of DF9's star clusters had velocity measurements. The authors argue the gap between the observed dispersion and the halo prediction is large enough to make the check meaningful anyway. ^[4] They also say the results do not depend on the assumed distance to the trail. ^[4]

A trail, and a possible crash

DF2 and DF4 are part of a statistically significant straight-line alignment of about a dozen faint galaxies, and the paper cites earlier work finding a relation between their positions and speeds. ^[4] DF9 is the galaxy on that trail most like the other two in brightness, size and bright cluster population. ^[4]

The authors propose that the three formed together, out of gas pulled apart from dark matter in a high-speed collision they call a "bullet dwarf" event. ^[1][2][3] No one has seen that collision. It is a proposed origin.

It does make a prediction. If one event made the trail, other galaxies along it should also lack dark matter, and DF9 was a test of that idea. ^[4] Future observations could also hunt for gas left behind by the collision and pin down the galaxies' gas content, the authors say. ^[3]

Keim, the lead author, told Quanta Magazine that the finding shows dark matter is "not just a modification of gravity." ^[1]

His point is that if dark matter can be separated from a galaxy, it is a thing and not a feature of how gravity works. ^[1] Galaxies made from separated gas would be hard to explain if gravity were modified in place of dark matter. ^[1][4]

Daniel Whiteson, a physicist quoted by Quanta, called the collision explanation "very convincing." ^[1]

Not everyone is convinced

The origin story is where the argument now lives. Kinwah Wu suggested the galaxies could instead have lost their dark matter while passing through a large group of massive galaxies. ^[1] That would strip them without any head-on collision between dwarfs.

Stacy McGaugh, a proponent of modified-gravity theories, told Quanta that the collision is a plausible origin if it happened. He said he was not convinced it was the only one, and did not think the galaxies establish the nature of dark matter without a firm origin story. ^[1]

That is a fair line to draw. The stellar motions in DF9 are a measurement, and the paper establishes them.

The collision is a story built on top, and a story is only as good as its next test. A measurement in hand and an explanation still on trial is how most good science looks in its middle chapters.

The test would also be narrow. It bears on MOND, not on every theory that tries to do without dark matter. ^[1][4]

Three galaxies in a line are harder to dismiss as a fluke than one, and a line makes a promise. If the shared-origin idea is right, more galaxies along the trail should turn out light on dark matter. If they do not, the collision story gets much harder to tell.

The next steps are plain. Astronomers can search the trail for gas left over from the proposed crash and measure more of DF9's star clusters to firm up its mass. ^[3][4] The gas, if it is there, would be the clearest evidence yet that something violent happened.