Galaxy DF2 challenges cosmology: can dark matter-free galaxies be real?

Full text

Out there beyond the Milky Way, there’s much more than a typical telescopic view reveals. While our eyes might be drawn to the enormous, massive, bright spirals and ellipticals that can be found all across the Universe, the reality is that for every Milky Way-like galaxy out there, there are dozens or even hundreds of small, low-mass, very faint galaxies out there: the dwarf galaxies of the Universe. Even though large, bright, massive galaxies contain the majority of our Universe’s stars, the majority of galaxies are faint, small, and low in mass. While some of those dwarf galaxies are compact and extremely rich in dark matter, there’s a small but now-significant population of extremely diffuse, puffy dwarf galaxies that may have little-to-no dark matter inside of them.

The first such galaxy ever spotted is simply known as DF2 : an ultra-diffuse galaxy on the outskirts of the NGC 1052 group that was spotted with the Dragonfly Telephoto Array back in 2018. In the time since, three more galaxies that appear to have little-to-no dark matter inside of them:

DF4 , discovered in 2019 and also around the NGC 1052 group,

FCC 224 , discovered in 2025 in a different part of the sky (near the Fornax cluster),

and now, most recently, DF9 , identified in 2026 and also around the NGC 1052 group.

Individually, any one of these galaxies wouldn’t make extremely compelling evidence for a dark matter-free galaxy. But taken all together, there’s a strong case to be made that these objects do represent a new class of objects: dwarf galaxies that are extremely dark matter-deficient, or possibly even dark matter-free. There’s just one nagging thought that causes some astronomers to remain skeptical. Here’s what we know today, and what it’ll take to settle the issue definitively.

This image shows the field of the NGC 1052 galaxy group, with main galaxy NGC 1052 being the elliptical at the far left, smaller disk galaxy NGC 1047 located above it, and the potentially much closer spiral NGC 1042 located at right. On smaller angular scales, more distant galaxies, like NGC 1048 at the lower-right, can also be seen, while nearby prominent galaxy NGC 1035 is just out of frame.

Credit : Adam Block/Mount Lemmon SkyCenter/University of Arizona

Back in 2018, most astronomers weren’t particularly expecting to find dark matter-free galaxies with the techniques that were available at the time. On large cosmic scales, dark matter and normal matter were already long-established to both be present, and in a roughly universal 5-to-1 ratio, with dark matter being more abundant than normal matter. This ratio was obtained from multiple lines of evidence: from the fluctuations in the cosmic microwave background, from galaxy clustering data within our Universe’s large-scale structure, from measurements of X-ray emitting galaxy clusters that exhibited gravitational lensing, and from tracking the motions of individual galaxies within galaxy clusters, as well as my measuring the rotations of large, massive, individual galaxies.

The major exceptions to that 5-to-1 rule came from just two places.

At lower masses, star-formation episodes expel large amount of gas from galaxies, as their gravitational potential wells are too shallow to hold onto their normal matter if there’s a large burst of star-formation. Therefore, on the low-mass end of the galactic spectrum, we’d see a strong enhancement of the dark matter to normal matter ratio.

And if you had a major collision, disruption, or interaction event between large galaxies, some of the internal normal matter could be stripped out of those galaxies, creating clumps of normal matter — rich in stars — that got separated from the initial dark matter-rich structures that contained them in the process.

That was it; those were the only primary ways we knew of at the time to make galaxies with more-or-less dark matter than the cosmic average.

Located within the Norma cluster of galaxies, ESO 137-001 speeds through the intracluster medium, where interactions between the matter in the space between galaxies and the rapidly-moving galaxy itself cause ram pressure-stripping, leading to a new population of tidal streams and intergalactic stars. Sustained interactions such as this can eventually remove all of the gas from within a galaxy, eliminating its ability to form new stars, while creating a wake of dark matter-free star clusters or mini-galaxies behind it. Phenomena such as this allow us to conclude that the galaxy, the cluster, and the gas within it are all made of matter, not antimatter, while the tidal streams of new stars will contain practically no dark matter at all.

Credit : NASA, ESA, CXC

However, there was an additional mechanism that was at least thought to be theoretically plausible for how you could make a galaxy that differed in its dark matter vs. normal matter contents: tidal forces. When galaxies form, they typically contain both dark and normal matter in that 5-to-1 ratio. Over time, the normal matter preferentially sinks towards the center, forming stars and other clumpy structures, while the dark matter remains in a more diffuse, larger-radius, “puffier” halo. However, if there’s a large mass that comes near to these low-mass dwarf galaxies, then there will be differential (tidal) forces acting on different components of that galaxy.

The innermost, most central components will be the most stable, and the most difficult for tidal forces to tear apart.

However, the outermore, more extended components will be the most susceptible to tidal stripping, gravitationally ejecting them from the system.

This results in the low-mass galaxy being stripped away from the outside-in: first the outer dark matter halo is stripped away, then the inner halo and any gaseous (baryonic) matter present there, and finally streams of stars begin to get pulled out of the galaxy itself, making a characteristic “S-shape” as they get pulled out.

This scenario was the one that was floated when NGC 1052-DF2 (DF2) was announced as the first candidate for a dark matter-free galaxy.

The first galaxy detected that supports its existence without dark matter, NGC 1052-DF2, is shown here as imaged by Hubble. The follow-up imaging was done in order to determine whether it was gravitationally connected to NGC 1052, at ~64 million light-years away, or NGC 1042, which could be much closer at 42 million light-years distant. The determination was 72 million light-years with an uncertainty of just a few percent, and hence, it ought not be connected to either. It cannot possess a typical amount of dark matter.

Credit : Z. Shen et al., ApJ, 2021

The interesting thing about DF2 is that it appears to be what astronomers call an ultra-diffuse galaxy: a galaxy where its stars aren’t particularly centrally concentrated, but rather are distributed over a large volume of space. Instead of having a disk, this galaxy’s stars are located in a spheroidal shape, but with a weak central concentration and a large radial extent.

When you have a galaxy like this, what you can measure are things like line widths, brightness, spectral energy distributions, individual very bright stars (depending on distance), globular clusters, and variations in light as a function of space and time. This allows you to infer things like:

how far away is the galaxy,

what is the velocity dispersion of the stars inside of it,

what is the total stellar mass (i.e., the mass in the form of stars) inside of this object,

and what is the total gravitational mass needed to provide it with the motions we infer.

Based on their measurements, the team that discovered DF2 initially concluded that this galaxy was located at a distance of around 20 Megaparsecs (around 65 million light-years), and that its low velocity dispersion implied that all or nearly all of its mass was accounted for by stars, alone. Hence, this was the first claim of a galaxy lacking dark matter entirely.

Many nearby galaxies, including all the galaxies of the Local Group (mostly clustered at the extreme left), display a relationship between their mass and velocity dispersion that indicates the presence of dark matter. The lower in mass a galaxy is, in general, the higher its dark matter-to-normal matter ratio. NGC 1052-DF2 is the first known galaxy that appears to be made of normal matter alone, and was later joined by DF4 in 2019, FCC 224 in 2025, and DF9 in 2026. Galaxies like Segue 1, however, are particularly dark matter-rich; there are a wide diversity of properties, and dark matter-free galaxies are only poorly understood, with many questioning their nature.

Credit : S. Danieli et al., ApJL, 2019

However, there are multiple different ways to measure the distance to a galaxy, and the values that one infers for its gravitational mass and stellar mass, as well as other properties, are highly dependent on what that distance actually is. A concern was raised early on , just months after the galaxy’s status as dark matter-deficient was announced, that if the distance were wildly wrong, and was at 13 Mpc (42 million light-years) instead of 20 Mpc (65 million light-years), then the inferred properties would be much more normal for the galaxy DF2.

At a much closer distance, the total mass of the galaxy would have been somewhat lower, the velocity dispersions would be somewhat greater, but the biggest chance would come in the estimated stellar mass of the galaxy, which would have been much lower. (Remember, brightness scales as the inverse of the distance squared, so a much smaller stellar mass at a closer distance can lead to the same observed brightness.) If this were the case, the galaxy would be made up primarily of dark matter, like nearly any other galaxy. The issue is confounded by the fact that in the vicinity of this ultra-diffuse dwarf galaxy, there are four large, massive galaxies nearby:

NGC 1052 , located 19-20 Mpc (62-65 million light-years) away,

NGC 1042 , with a widely disputed distance ranging from 42-55 million light-years away,

NGC 1047 , located at the same distance as NGC 1052,

and NGC 1035 , located near NGC 1042, and more like 15 Mpc (49 million light-years) away.

If the dwarf galaxy in question is nearer than the closer member, rather than to the farther member, of this grouping, then that would vastly change the story of its dark matter abundance.

In a series of papers published in 2020 and 2021, astronomers used ultra-deep imaging with multiple different telescopes, including the flagship Gemini telescope, to search for ultra-faint tidal tails, making a predicted S-shape, around the alleged dark matter-free dwarf galaxies DF2 (left) and DF4 (right). While such a tail was indeed spotted around DF4, DF2 instead appeared completely symmetric, lacking any signature that could be interpreted as a tidal tail down to an incredibly faint magnitude of +30.5.

Credits : M. Montes et al., Astrophysical Journal, 2020 and 2021

In addition, galaxy DF2 doesn’t exhibit any evidence of that tidal stripping , or that S-shaped stream of stars, that was initially proposed as its likely formation scenario. Independent measures of its distance, albeit all with large uncertainties, pointed to that closer distance , rather than the farther one.

However, the story swiftly became a lot richer and more complicated. In 2019, galaxy DF4, also on the outskirts of this same galaxy group, was announced, and was also taken to be dark matter-free. This time, detailed faint imaging did indeed reveal evidence of the expected tidal stripping along with the characteristic S-shaped stream of stars coming from both sides of the galaxy. In 2021, DF2 underwent follow-up imaging with Hubble, and individual red giant stars were measured, providing a low-uncertainty distance measurement that was even greater than what was expected : of 22 Mpc (72 million light-years), implying that there was no dark matter at all inside of this galaxy.

And then, in 2025, galaxy FCC 224 was found in a completely different region of the sky: the Fornax cluster. More recently, in June of 2026, a third dark matter-free galaxy candidate was found in that same NGC 1052 group: DF9, bringing the total known number of these ultra-diffuse, dark matter-deficient (or even dark matter-free) galaxies up to four.

This image shows the stars of the ultra-diffuse galaxy FCC 224, located far away in space from the other three dark matter-free candidate galaxies: DF2, DF4, and DF9. Instead of being near the NGC 1052 group, this galaxy is located in an entirely different part of the sky: the Fornax cluster. Identifying a dark matter-free candidate galaxy in an entirely different environment from all of the others bolsters, but does not prove, the case that these galaxies indeed lack dark matter.

Credit : ESA/Hubble & NASA

Despite the fact that some of these galaxies show evidence of tidal stripping while others don’t , there are a series of properties that all of them share.

They all are ultra-diffuse galaxies, extended in space,

with tens-to-hundreds of millions of stars inside,

with unusually bright globular clusters (where unusually bright-or-faint objects traditionally being a hint that we’ve got the distance wrong),

where the galaxy exhibits slow rotation about its short axis,

with low densities and no new evidence of recent star-formation,

as well as several other common properties concerning their stars and globular clusters.

The fact that three of them, including the most recent, DF9 , are all in the same galactic group raises some red flags, but the fact that one of them (FCC 224) is in an entirely different region of the sky helps assuage some of those concerns. The inferred velocity dispersions of all of these objects are so significantly much lower than what would be expected if they had the expected amount of dark matter means that even if their distance estimates were somewhat off, there would still be a substantial dark matter deficiency. It’s these multiple lines of evidence, and the fact that we’re not just using one, but a series of independent objects, that bolster the confidence we have that these galaxies really are dark matter-deficient, and that some of them may well be dark matter-free.

While the overwhelming majority of galaxies in the Universe that have very small velocity dispersions are extremely compact and low in stellar mass (lower left), or higher in mass with larger velocity dispersions (large outlined circles on the left graph), four (solid circle) galaxies stand out as outliers: DF2, DF4, DF9, and FCC 224. These high mass, ultra-diffuse, but low velocity dispersion galaxies are the first (and thus far, only) candidates for being extremely dark matter-deficient, or even dark matter-free.

Credit : M.A. Keim et al., Astrophysical Journal, 2026

However, there is still the risk of bias there, and that cannot be so easily ignored. One concern is that the measurements of the velocity dispersions of these objects all have large uncertainties: uncertainties at around the 3-5 km/sec level. While that may not seem like a big deal, as those numbers are relatively small, we’re talking about velocity dispersions that are only at the 6-8 km/sec level overall, so uncertainties that are 50% or more of the measured value. If the actual numbers turn out to be one or two standard deviations higher than the inferred values at present, the case for dark matter-deficiency drops substantially.

Another issue is that, for these four systems, the measurements of velocity dispersion were all conducted with the same instrument aboard the same telescope: the Keck Cosmic Web Imager . While KCWI is a truly great instrument, you really want independent measurements from separate instruments to confirm an assertion that’s this revolutionary.

There’s also the issue that the now-preferred mechanism of formation of these galaxies is the bullet dwarf scenario : a high-velocity collision between two initially gas-rich galaxies, leading to dark matter-free clumps of stars forming along a relatively linear trail. The evidence thus far is ambiguous, but potentially interesting , as the DF2, DF4, and DF9 galaxies all do seem to align.

This close-up image shows galaxy DF9 on the outskirts of the NGC 1052 group. While the distance uncertainties to DF9 are large, the authors suggest a “bullet dwarf” scenario, where a long-ago collision between two gas-rich galaxies created a trail of dwarf galaxies with no dark matter, and that DF2, DF4, and DF9 are all members. The other dwarf galaxies, shown in unlabeled boxes, either would also be dark matter deficient or would have formed via an alternate mechanism.

Credit : M.A. Keim et al., Astrophysical Journal, 2026

However, as you can see from the wide-field DECaLS image in the picture above, there are more than those three identified faint galaxies along that approximate line; there are more like ten such galaxies. If they were formed by the same process, why don’t the other seven exhibit the same properties? And why does DF4 exhibit the tidal stripping, while DF2 and DF9 don’t?

If the globular clusters inside of these galaxies are all truly overly luminous, then how did they form? As European Southern Observatory astronomer Maria Luísa Buzzo, who led the work on FCC 224 in 2025, related : “Globular clusters are often used as a proxy to estimate the amount of dark matter a galaxy contains. For some reason this galaxy (FCC 224) has an unusual number of luminous clusters and no dark matter, at least within its inner regions. No existing galaxy formation model within our standard cosmological paradigm can currently explain how this galaxy came to be.”

And finally, if these galaxies really are dark matter-free or dark matter-deficient, why haven’t we imaged them with our flagship space telescope, JWST, to truly pin down their distances and narrow the uncertainties on their velocity dispersions and stellar masses? The right set of observations could lead to far superior measurements and greatly reduced error bars on all of these properties.

While the assumed distance of DF9 is 20.6 Mpc (67 million light-years), based on DF2 at 21.7 Mpc and DF4 at 20.0 Mpc, the authors of the latest study show that the expected velocity dispersion only changes by approximately 11% if the distances are raised or lowered, assuming they all form a straight line. If those assumptions do not hold, however, and the distances are well below what’s currently inferred, much larger velocity dispersions are suddenly admissible, with potentially devastating consequences for the dark matter-deficient hypothesis.

Credit : M.A. Keim et al., Astrophysical Journal, 2026

The big “nightmare scenario” for these dark matter-deficient dwarfs, of course, is simply that they do have dark matter, and that we’ve made a mistake somewhere along the way when it comes to inferring properties of each one of these galaxies. Although unidentified rotational motion could be at play, the most worrisome culprit, despite Hubble data to measure the evolved stars on the tip of the red giant branch within them, is that we’ve incorrectly inferred the distance to these objects. While the authors of the most recent work were able to show (above) that if the distances to DF2, DF4, and DF9 were somewhat closer than currently presumed, they would still exhibit dark matter deficiencies , a significantly closer set of distances — like the 13 Mpc (42 million light-year) distance considered in 2019 — would indeed remove the need for a significant dark matter deficiency.

The way to remove our doubts, and to become certain of the distances to these four objects, is to acquire observations superior to what telescopes like Keck, Gemini, or even Hubble can obtain: by using JWST to measure the stars within it. While Hubble can only resolve individual stars out to distances of roughly 50-60 million light-years at most, JWST can resolve many more types of stars out to those same distances, with some types accessible at distances in excess of 100 million light-years. If we truly want to know whether these galaxies are dark matter-deficient or dark matter-free, we need unambiguous data that closes all of the potential loopholes for these galaxies to still possess substantial amounts of dark matter. While the presence of dark matter enriched low-mass galaxies is not in doubt, we must take extreme care not to fool ourselves, especially when our understanding of the very nature of “what makes up the Universe” is on the line.

This article Is it finally time to take dark matter-free galaxies seriously? is featured on Big Think .