A Third Galaxy on the NGC 1052 Trail Appears to Be Missing Dark Matter

In the NGC 1052 field, a faint trail of galaxies stretches across the sky. Two of them, NGC 1052-DF2 and NGC 1052-DF4, have spent years as cosmic anomalies: their stars move so slowly that astronomers cannot detect the gravitational tug of a dark matter halo. Now a third galaxy on the same trail, NGC 1052-DF9, has been measured. It, too, appears to be missing dark matter.

Dark matter is the invisible mass that dominates the gravity of most galaxies. Astronomers infer it from motion: if stars and gas move faster than the visible mass can explain, something unseen must be providing the extra pull. Dwarf galaxies are normally the most dark-matter-dominated systems of all. DF2 and DF4 broke that rule, and their shared home in a tight line of faint galaxies raised an obvious question: did they form together in some rare event that stripped away their dark matter?

A Trail That Moves Together

The trail was identified earlier as a dozen low-luminosity galaxies aligned so tightly that random chance seems unlikely. Follow-up spectroscopy showed that their radial velocities -- motion toward or away from us -- change linearly with position along the trail. That kinematic link suggests the galaxies are not just neighbors but relatives, born together in the same event.

One proposed origin is a 'bullet dwarf' collision. In this scenario, a high-speed encounter between gas-rich dwarf galaxies separates gas from dark matter. The gas, now free of the dark matter that once anchored it, could cool and form new galaxies. Computer simulations of such collisions have reproduced both the trail's geometry and the unusual properties of DF2 and DF4. The scenario makes a strong prediction: other galaxies on the trail should also lack dark matter.

Reading a Faint Spectrum for the Tug of Gravity

DF9 was the best candidate to test that prediction. It is the trail galaxy most similar to DF2 and DF4 in luminosity, size, and bright star clusters. But it is still faint and diffuse, so measuring the internal motion of its stars required deep spectroscopy.

Michael A. Keim of Yale University and colleagues observed DF9 with the Keck Cosmic Web Imager (KCWI) on the Keck II telescope in Hawaii. KCWI spreads light into detailed spectra across a small field of view. Over two nights in October 2024, the team collected 24 on-target exposures of 750 seconds each -- about 5 hours on DF9 -- plus matching sky exposures, for a total combined exposure time of 10.83 hours. The spectra targeted absorption lines from calcium in the combined starlight.

If all stars in DF9 moved at exactly the same speed, those lines would be sharp. Their actual width reveals the stellar velocity dispersion -- the spread of stellar speeds along the line of sight. By comparing the observed spectrum with synthetic stellar population models, the team fitted that spread. They found a raw dispersion of about 8.7 kilometers per second, then corrected for effects that can broaden lines without reflecting the galaxy's gravity, such as internal stellar rotation, macroturbulence in stellar atmospheres, and stars orbiting unseen binary companions. The corrected gravitational dispersion is 6.4 kilometers per second, with an uncertainty range from about 2 to 10 kilometers per second.

The Difference Between 6 and 27 Kilometers per Second

That slow motion is the key. From the stellar mass enclosed within DF9's effective radius -- about 0.71 x 10^8 solar masses, half of its total stellar mass of about 1.4 x 10^8 solar masses -- astronomers expect a velocity dispersion of 8.3 kilometers per second. The measured 6.4 is consistent with that, given the uncertainties. Within that same effective radius, the dynamical mass is about 0.42 x 10^8 solar masses. The uncertainties on the dynamical mass and on the stellar mass overlap.

If DF9 instead sat in a normal dark matter halo, following the observed relationship between stellar mass and halo mass for low-mass galaxies, its stars should move at about 27 kilometers per second, with an uncertainty of 3. That is far above the measurement. In other words, DF9's kinematics do not require dark matter. The stars alone can account for the motion.

An independent, weaker check comes from two star clusters previously measured in DF9. Their velocities are close to the galaxy's diffuse starlight, which is consistent with a low dispersion. With only two tracers, however, this check cannot rule out dark matter; it can only show that a large velocity difference, which would demand a massive halo, is not seen.

A Third Bead on a Strange String

The result places DF9 alongside DF2 and DF4 as outliers in the relationship between galaxy size, stellar mass, and velocity dispersion. In a broader comparison with dwarf galaxies in the Local Group and the Virgo Cluster, galaxies with similar stellar masses tend to have much larger dispersions and far more massive dark matter halos. If their low dynamical masses were matched by small 'normal' dark matter halos, those halos would lie below the universal share of baryons -- the ordinary matter that makes up stars, gas, and planets. That is unphysical in the standard picture, because a galaxy cannot convert more baryons into stars than its halo ever contained.

The more straightforward interpretation is that these galaxies formed without dark matter. The bullet-dwarf collision is one way to do that, but it is not the only proposed route. Quasar-driven outflows, tidal debris from a massive galaxy merger, and stripping from an infalling stream could also separate gas from dark matter. The paper notes that such top-down formation channels would likewise predict dark matter deficiency in DF9. The specific appeal of the bullet-dwarf model is that it anticipated the trail and its dark-matter-free members before DF9 was measured.

What the Measurement Does Not Prove

This is a preprint, and the conclusion is statistical rather than absolute. The velocity dispersion is low and the uncertainties are large. The corrections for binary stars and stellar broadening rely on estimates, and the integrated-light measurement cannot remove every binary star individually. The authors conclude that dark matter is not required to explain DF9's kinematics; they do not claim the galaxy has exactly zero dark matter. A small amount could still hide within the error bars.

Future observations could sharpen the picture. Measuring neutral or ionized gas in the trail could reveal whether leftover gas from the original collision is still present -- potential 'smoking gun' evidence for a bullet-dwarf event. Gas motions could also provide dynamical constraints for trail galaxies too faint for optical spectroscopy. Extending the measurement to a fourth or fifth galaxy would be valuable, but the paper notes it will be very difficult because those objects are even fainter.

For now, DF9 is a third bead on a strange string: a galaxy whose stars move as if dark matter were never there. Whether it is a clue to how dark matter behaves or a rare, violent accident of galaxy formation, the NGC 1052 trail has become one of the most unusual laboratories in the nearby universe.