After decades of studying dark matter scientists have repeatedly found evidence of what it cannot be but very few signs of what it is. That might have just changed. A study of four colliding galaxies for the first time suggests that the dark matter in them may be interacting with itself through some unknown force other than gravity that has no effect on ordinary matter. The finding could be a significant clue as to what comprises the invisible stuff that is thought to contribute 24 percent of the universe.

“This result, if confirmed, could upend our understanding of dark matter,” says physicist Don Lincoln of the Fermi National Accelerator Laboratory in Illinois, who was not involved in the research. So-called “self-interacting dark matter” has been suggested for some time but it has generally been considered unorthodox. The simplest model of dark matter portrays it as a single particle—one that happens to interact with others of its kind and normal matter very little or not at all. Physicists favor the most basic explanations that fit the bill and add extra complications only when necessary, so this scenario tends to be the most popular. For dark matter to interact with itself requires not only dark matter particles but also a dark force to govern their interactions and dark boson particles to carry this force. This more complex picture mirrors our understanding of normal matter particles, which interact through force-carrying particles. For example, protons interact through the electromagnetic force, which is carried by particles called photons (particles of light).

Now scientists led by Richard Massey at Durham University in England report in Monthly Notices of the Royal Astronomical Society the first signs that dark forces and dark bosons might really exist. Researchers used the MUSE (Multi Unit Spectroscopic Explorer) instrument on the Very Large Telescope in Chile, along with the Hubble Space Telescope to examine the Abell 3827 cluster, where four galaxies are colliding in a cosmic car wreck.

To determine where the invisible dark matter lies, astronomers took advantage of a natural phenomenon called gravitational lensing, predicted by Einstein’s general theory of relativity. Lensing occurs when mass warps spacetime, causing light traveling through this bent region to take a curved path. The dark matter in Abell 3827 is plentiful, so it warps the space around it significantly. When light from a distant object behind the cluster travels to Earth, it passes through this distorted area and produces telltale signs of lensing, such as arcs of light and double images, that astronomers used to “weigh” the unseen matter in the cluster. The scientists found that in at least one of the colliding galaxies the dark matter in the galaxy had become separated from its stars and other visible matter by about 5,000 light-years. One explanation is that the dark matter from this galaxy interacted with dark matter from one of the other galaxies flying by it, and these interactions slowed it down, causing it to separate and lag behind the normal matter.

The interactions would be similar to what happens when two protons pass near one another. Each releases a photon that is absorbed by the other, causing both particles to recoil. This repellent force happens between any two particles with the same electromagnetic charge and it could happen between any two dark matter particles as well. But because dark matter is not affected by the electromagnetic force, only a new “dark” force, carried by a so-called dark-photon, could produce the repulsion. It could also be that only some portion of dark matter interacts with itself whereas the bulk of it is a more traditional single-particle type. “It does seem pretty good for our kind of model, in which only a small fraction of the dark matter interacts,” says Harvard University physicist Lisa Randall, who has envisioned such a model.

The study leaders are cautious about interpreting their observations. “This is very, very exiting because it is the first potential detection of nongravitational interactions, but we have to see this in more of these objects,” says team member David Harvey of the École Polytechnique Fédérale de Lausanne in Switzerland. “It’s by no means confirmed.” It is possible, for example, that the pattern the telescopes observed was caused by extra dark matter outside the cluster but along Earth’s line of sight, rather than self-interacting dark matter. “This is one of the situations where it would be so incredibly exciting if it ended up being dark matter that everybody will need to be a little bit cautious in how they approach it just because of that simple fact,” says physicist Neal Weiner of New York University, who was not involved in the study.

Self-interacting dark matter with dark forces and dark photons may not be as simple as the single-particle explanation but it is just as reasonable an idea, Weiner says. “The strongest motivation for considering dark matter to have its own interactions is simply that when we look at the Standard Model,” which describes all the known particles and forces, “we see that it’s full of all sorts of different interactions. It seems quite natural that dark matter could have its own force.” This setup could also explain some small discrepancies between predictions of the single-particle model and what astronomers actually observe. For example, the single-particle model says that the centers of galaxies should be denser than they really are; if dark matter interacts with itself, however, it would tend to collide in galactic cores and push away.

So far, no signs of self-interacting dark matter have showed up in other galaxy collisions. Another famous crash site, the Bullet Cluster, was one of the first to provide strong evidence that dark matter exists, because gravitational lensing shows that most of the cluster’s mass resides in a different place than the visible matter. But the separation there is not large enough to suggest that the dark matter is interacting. “The result is not in conflict with the Bullet Cluster,” says Maruša Brada of the University of California, Davis, one of the leaders of the original 2006 Bullet Cluster study, because that example put only upper limits on how strong the interaction could be. And another more recent study, led by Harvey and published in March in Science, surveyed 72 collisions of galaxy clusters and also found no signs of self-interacting dark matter. But the targets of that study, as well as the Bullet Cluster, are colliding galaxy clusters—not individual galaxies crashing together as in Abell 3827. Lone galaxies collide at slower speeds than entire clusters do, and that might have given the dark matter particles more time to interact with one another and slow down. The true test will come when astronomers look at other individual galaxy smashups. “There are other clusters that we can look at that are in similar states but none are as perfect as this,” Harvey says. “But with future surveys we’ll hopefully be able to see these objects.”