A galaxy is much more than a radiant agglomeration of stars. To modern astrophysicists, galaxies are more notable for their dark sides: their hidden material that is only “seen” by its gravitational pull upon the shiny stuff it seems to vastly outweigh. So-called dark matter is as much a defining feature of galaxies as stars and gas, and is thought to provide the gravitational seeds from which galaxies assemble and grow.

A galaxy without dark matter—or without some bizarre effect of gravity that would mimic dark-matter behavior—would be a very weird thing indeed. Finding such a thing would be like finding smoke but no fire, effect without cause. Yet that is what Yale University astronomer Pieter van Dokkum and his colleagues have just found, they report in a study published in March in Nature.

The galaxy, called NGC 1052-DF2, is about 65 million light years away. It is almost as big as the Milky Way but is “ultra-diffuse,” meaning it contains just a vanishing fraction of the stars found in our galaxy—only 1 percent, in this case. That stellar sparseness means it does not look much like a typical spiral galaxy, but rather a loosely connected, ghostly blob of star-pocked gas and dust. If it contained an amount of dark matter typical for a galaxy of its size, the dark matter’s gravity would hasten the motions of several star clusters that surround it. Instead, van Dokkum’s team found those star clusters moving languidly around NGC 1052-DF2, a sign that there may well be very little or no dark matter within that galaxy at all. That suggests dark matter can be decoupled not only from regular, visible matter, but from entire galaxies—a phenomenon astronomers have never seen until now.

“If it’s true that a galaxy exists where there is hardly any dark matter, I think that’s a problem for all theories about galaxy formation,” says Erik Verlinde, a theoretical physicist at the University of Amsterdam who has proposed an alternative to dark-matter-dominated gravity and was not involved in the research. “Even if you believe in a modified theory of gravity, you would expect to see something different from what has been observed here. It’s a problem for almost every theory that’s out there.”

Dark matter has never been seen or measured directly because it does not emit light. Its presence is instead inferred through the gravitational pull it exerts on any normal matter around it. Astrophysicists think dark matter’s gravity is crucial for forming the universe’s large-scale structure of filaments and sheets of galaxy clusters, and scientists have even measured how clumps of it act as gravitational lenses, magnifying light from far-distant background galaxies. Some physicists have postulated that there is no such thing as dark matter, however, and that what we perceive as giant, starlight-bending clumps of heavy, invisible material is actually something else that is profoundly misunderstood.

Only discovered in 2015, ultra-diffuse galaxies are thought to be particularly useful cosmic laboratories for understanding dark matter. Surely, astronomers thought, dark matter must play a role in forming these objects so devoid of normal star stuff. That thinking led van Dokkum and his colleagues to build the Dragonfly Telephoto Array, a telescope in New Mexico created for the express purpose of scrutinizing ultra-diffuse galaxies. The researchers initially used Dragonfly to study a different galaxy, one appearing to possess an almost inconceivably gargantuan amount of dark matter, which was a weird result in and of itself. When van Dokkum and his team found NGC 1052-DF2, they expected to see something similar.

“Instead we saw the opposite, leading to this remarkable conclusion that there’s actually no room for dark matter at all in this thing,” van Dokkum says. “It’s not something we were looking for or expecting. At all. But you go in the directions the data take you, even if it’s in contradiction to what you’ve found before.”

In Dragonfly images, NGC 1052-DF2 looked like a standard ultra-diffuse galaxy. But when the team compared them to a better image from the Sloan Digital Sky Survey, they found a surprising mismatch. What had seemed to be dim basic galactic structures in Dragonfly’s view appeared as point-like sources in the Sloan image. To resolve the discrepancy, the team scrutinized the galaxy with the Hubble Space Telescope, the W.M. Keck Observatory and the Gemini Observatory, the latter two on Mauna Kea in Hawaii.

The point sources proved to be 10 globular clusters—compact and spherical groupings of stars orbiting the galaxy’s core. The researchers then set about measuring the movements of the clusters as a way to estimate the galaxy’s total mass. Simply put, the velocity at which clusters orbit a galaxy is related to the amount of matter—normal or dark—that a galaxy contains. Using information from the Keck telescopes, the team found the globular clusters were moving much more slowly than expected.

Adding up the clusters’ motions and NGC 1052-DF2’s mass, they realized there might be no dark matter behind this particular galactic curtain. All the clusters’ movements could be explained solely by the mass of the galaxy’s observed stars.

It is possible that these measurements are imperfect, however, says Michael Boylan-Kolchin, an astronomer at the University of Texas at Austin who was not involved in the new study. “The alternative possibility is that the globular clusters, or the objects they think are globular clusters, aren’t really measuring the total mass in the way they think,” he says. “The key thing is to see whether the globular clusters really are tracing the mass of the galaxy as a whole.”

Assuming the results are right, there are a few theories to explain how galaxies like NGC 1052-DF2 could come together untouched by dark matter’s hidden hand. It could be that NGC 1052-DF2 was once a placid mass of gas and has been recently perturbed by another unseen galaxy nearby, sparking star formation. Or, van Dokkum speculates, perhaps this ultra-diffuse, dark-matter-free galaxy arose from two streams of gas that collided and compressed to form a scattering of stars. Another idea, first proposed more than two decades ago by Yale astronomer Priyamvada Natarajan, then at the University of Cambridge in England, suggests that galaxies like NGC 1052-DF2 may form from galaxy-sized gobs of gas clumping together in jets ejected by feasting supermassive black holes. NGC 1052-DF2 does reside in a region where such things could conceivably occur, lying near a giant elliptical galaxy with a supermassive black hole at its heart.

Alternatively, in the absence of any direct dark-matter detections, some theorists have suggested its existence is illusory—and that something else drives galaxy evolution and gravitational lensing. Some have proposed tweaking the laws of motion first codified by Isaac Newton, developing a class of competing theories called Modified Newtonian Dynamics, or MOND. More recently, Verlinde suggested an alternative called “emergent gravity,” in which gravity is a byproduct of quantum fluctuations and dark energy (another scarcely understood phenomenon, one that seems to be causing the universe’s expansion to accelerate).

In a paradoxical twist of the sort that makes astrophysics special, NGC 1052-DF2’s lack of dark matter is potentially a good thing for the theory as a whole. That is because even if dark matter is not real, the observations that hint at its existence are. If the large-scale universe is dominated by subtle alterations to the force of gravity, or to a fluctuating force field of dark energy, these effects would not discriminate and would manifest in all galaxies—NGC 1052-DF2 included. Yet in this strange galaxy, the projected signatures of these exotic effects are not seen.

Here, “the absence of dark matter is evidence of its existence,” van Dokkum says. “There is no way around it. It can be in a galaxy or not in a galaxy, but it is not a field, or some alternative thing that manifests itself rather than being a substance.”

Verlinde disagrees that the new galaxy puts his ideas or MOND to rest. “I indeed believe there is no dark matter that needs to be added to explain these observations. There might be a change in the ways that gravity works, compared to what Einstein and Newton would have said, but let’s not immediately draw conclusions,” he says. “I think it’s too early to say.”

Follow-up observations, including more detailed views of the globular clusters, should help astronomers settle the question and better understand these clusters’ relationship to the galaxy. Future observatories under construction, such as the European Extremely Large Telescope in Chile’s Atacama Desert, or NASA’s James Webb Space Telescope, should be able to take such measurements.

“How did we get a gas cloud of that size that was able to form a galaxy, that was able to have enough gravity?” says Risa Wechsler, an astronomer at Stanford University who was not involved in the new research. “There’s no question it’s interesting. It’s totally weird.”