How do we know dark matter exists if we cannot see it?

How did astronomers first notice the missing mass?

The story starts with speed. The faster things move in a group, the more gravity it takes to hold the group together. So if you measure how fast things move, you can estimate how much mass must be there.

In 1933, the Swiss-born astronomer Fritz Zwicky did this for the Coma Cluster, a big group of galaxies. His estimate used only seven galaxies, but it was striking. To explain their speeds, he wrote that the cluster’s average density would have to be at least 400 times greater than the glowing matter suggested. He called the unseen material “dunkle Materie,” German for dark matter. A modern translator notes that Coma’s speed spread, measured today from more than 1,000 galaxies, is very close to what Zwicky found. Still, NASA says the idea stayed on the fringe for decades.

That changed in the 1970s. Vera Rubin and her colleague Kent Ford measured how fast stars and gas orbit in spiral galaxies. Their tool was a very sensitive spectrometer built by Ford, which measures the light an object gives off at different wavelengths, or colors. According to the Rubin Observatory, they studied more than 60 galaxies. Stars at the outer edges moved about as fast as stars nearer the center.

Why is that strange? Think of our solar system (an analogy, and it only goes so far). Almost all its mass sits in the Sun, so Neptune crawls along much more slowly than Earth. A galaxy’s visible mass is also packed toward its middle, so its outer stars should slow down in the same way. A review by the astronomers Benoit Famaey and Stacy McGaugh sums up the result: rotation curves of spiral galaxies stay roughly flat, again and again, out to the farthest points that can be measured. The simplest fix is a large, roughly round halo of unseen mass around each galaxy. The analogy breaks here: a galaxy is not a sun with planets, and its mass is spread out, but the logic of “more speed needs more pull” carries over.

How can bending light weigh something invisible?

Einstein’s theory of gravity says mass bends the path of light. A heavy galaxy cluster acts a bit like a lens. It stretches and shifts the images of galaxies far behind it. This is called gravitational lensing. By measuring how much those background galaxies are warped, astronomers can map where the mass is, whether it shines or not. NASA describes doing this for clusters such as Abell 209.

The most famous case is the Bullet Cluster, formally 1E 0657-56. Two galaxy clusters collided there. NASA says it lies about 3.8 billion light-years away. In a crash like this, the pieces behave differently. The hot gas between the galaxies, which holds most of the normal matter, slams together and slows down. X-ray telescopes like NASA’s Chandra can see that gas. The galaxies themselves mostly sail past each other.

In 2006, a team led by Douglas Clowe mapped the mass with lensing. They found that the mass did not follow the hot gas. It followed the galaxies instead, and the offset was measured at 8-sigma significance, which means it is very unlikely to be a fluke of the data. Their paper argues that simply changing the law of gravity cannot explain the offset, so most of the matter there is unseen. In the famous NASA image, the pink glow is hot gas and the blue shows where the mass is.

What does the oldest light in the universe add?

The cosmic microwave background is light released about 380,000 years after the Big Bang, according to ESA. It is almost the same in every direction, but it has tiny hot and cold spots. Those spots record sound waves that rippled through the young universe. The pattern of the ripples depends on what the universe contained.

Ordinary matter got pushed around by light in those early days. Dark matter did not, because it doesn’t interact with light. That difference leaves a mark in the spot sizes. In their review, Famaey and McGaugh note that a universe without this extra, non-glowing matter should show a third ripple peak smaller than the second. Instead, the third peak is nearly as tall as the second. The Planck satellite’s final 2018 results measure a dark matter density about five times the density of ordinary matter.

The growth of galaxies points the same way. NASA explains that computer models with slow-moving, “cold” dark matter build a universe of galaxies and clusters like the one we see. Models with fast-moving dark matter do not.

Could gravity itself be wrong?

This is the strongest objection, and it deserves a fair hearing. In 1983, the physicist Mordehai Milgrom asked whether the missing mass might really be a sign that Newton’s law of gravity changes at very low accelerations. His idea is called MOND, short for Modified Newtonian Dynamics. Instead of adding unseen mass, it changes the rule.

MOND has real successes. Famaey and McGaugh show that one simple rule predicts many patterns in galaxies, some before they were observed. That is impressive, and it is why some scientists still take it seriously.

But MOND has serious trouble elsewhere. The same review says that in galaxy clusters, MOND still needs two to three times more mass than we can see. So it needs some dark matter anyway. The review calls the Bullet Cluster an outstanding challenge for all MOND theories, though it adds that the Bullet’s high crash speed is also a puzzle for the standard dark matter model. The simple no-dark-matter version fails the test of the oldest light’s third peak. And in 2023, a study of 8,611 pairs of stars in wide orbits, using data from the Gaia space telescope, found that their motions fit Newton’s law far better than MOND. It excluded MOND at 16-sigma and concluded that MOND would need big changes at small scales to survive.

NASA says the existence of dark matter is now widely accepted, even though astronomers do not agree on what it is. The MOND debate still matters, because it keeps pointing at galaxy patterns that any full theory has to explain.

What do people often get wrong about dark matter?

“Dark” does not mean black. NASA explains that it is called dark because it doesn’t absorb, reflect or give off light. Invisible would be a better word.

It is not dim ordinary stuff. Could it be faint stars, dead stars or rogue planets? The EROS-2 survey watched millions of stars for 6.7 years for the brief brightening that such objects cause when they pass in front of a star. It found only one possible event, where about 39 were expected if the halo were made of such objects. For objects around 0.4 times the Sun’s mass, that means they make up less than 8% of our galaxy’s halo mass. Across a wide range of masses, up to 15 times the Sun’s, they are ruled out as the main ingredient. The oldest light also sets the total amount of ordinary matter, and it falls far short.

It is not dark energy. Dark energy is the name for whatever makes the universe’s expansion speed up. Dark matter pulls things together. They are two separate puzzles, as our story What Is Dark Energy? explains.

It has not been caught in a lab, yet. The LZ experiment, a detector holding 7 tonnes of liquid xenon at the Sanford Underground Research Facility in Lead, South Dakota, hunts for dark matter particles bumping into atoms. Its 2025 results found no signal above the expected background. In September 2026, the team reported one unusual event at a high energy. They could not identify a background that explains it, but the tension with background alone was 2.6-sigma after accounting for the many places they looked. They did not claim a discovery. One event is a reason to keep watching, not an answer.

How can you check this yourself?

• Look at the Bullet Cluster image on NASA’s Chandra site. Find the pink gas and the blue mass, and notice that they don’t line up.
• Read NASA’s plain-language page on dark matter. It lists the main candidates and says plainly that none has been confirmed.
• Skim the abstract of the 2006 Bullet Cluster paper by Clowe and colleagues. It is one paragraph long, and it states its claim and its evidence.
• For the other side, read the opening of the review by Famaey and McGaugh. It lays out where MOND works and where it struggles.
• Ask one question of any dark matter headline: is it about the gravity, which is well measured, or about the particle, which is still unknown?

For a shorter tour of the same ideas, see our story What Is Dark Matter?