Vera Rubin Observatory Maps 20 Billion Stars. Dark Matter Remains Invisible.

On a ridge 8,800 feet above the Atacama Desert, Claire Hebert stands in front of a computer screen displaying 189 galaxies that did not exist in any catalogue until 3:47 this morning. She is not excited. She scrolls past them without stopping. Behind her, through reinforced glass, the Vera Rubin Observatory's primary mirror—8.4 metres of polished glass that cost $28 million and took eleven years to grind to a tolerance of ten nanometres—tracks westward across a sky so dark and dry that astronomers chose this spot over 43 other candidate sites across three continents. The mirror feeds the largest digital camera ever built: 3,200 megapixels, the size of a small car, capable of photographing the entire visible sky every three nights. It has been doing this since September 2024. Hebert, 39, is the lead scientist on Target Selection Team Three, which searches the resulting data for gravitational anomalies that might reveal the location of dark matter. In eighteen months, her team has found 14,661 anomalies. None of them were dark matter.

She refreshes the screen. Another 112 galaxies appear. She scrolls past those too.

"People think this job is about discovery," she says, not looking away from the monitor. "Mostly it's about waiting for the universe to show you something it has been hiding for 13.8 billion years. And then waiting longer."

The Problem Fritz Zwicky Saw in 1933

The search for dark matter began with an accounting error that would not resolve. In 1933, Swiss astronomer Fritz Zwicky at Caltech studied the Coma Cluster, a collection of more than 1,000 galaxies 321 million light-years from Earth. He measured how fast the galaxies were moving. Then he calculated how much mass would be needed to generate enough gravity to hold them together at those speeds. The numbers did not match. The visible mass—stars, gas, dust, everything his telescopes could detect—accounted for less than 15 percent of the gravity required. Either Newton's laws were wrong, or 85 percent of the universe was missing.

For forty years, most astronomers assumed Zwicky had made a mistake. Then in the 1970s, Vera Rubin—for whom the Chilean observatory is named—measured the rotation curves of spiral galaxies and found the same problem: stars at the outer edges were moving far too fast to be held in orbit by the visible mass alone. They should have been flung into intergalactic space. They were not. Something invisible was holding them in place.

By the 1990s, observations of the cosmic microwave background—the afterglow of the Big Bang—confirmed the proportions with precision. Ordinary matter, the stuff of stars and planets and human bodies, makes up 4.9 percent of the universe. Dark energy, the force driving the universe's accelerating expansion, accounts for 68.3 percent. The remaining 26.8 percent is dark matter: matter that interacts gravitationally but emits no light, absorbs no light, and has never been directly detected by any instrument.

What the Camera Sees Every Night

Hebert's office at the Rubin Observatory base facility in La Serena, Chile, is a repurposed shipping container with airconditioning that struggles above 28 degrees Celsius. On the wall, she has taped a timeline of dark matter searches: the Cryogenic Dark Matter Search, begun in 2003; the Large Hadron Collider's search for supersymmetric particles, 2008 to present; the XENON experiment in Italy's Gran Sasso laboratory, which in 2020 reported a potential signal that turned out to be contamination from tritium; the LUX-ZEPLIN detector in South Dakota, operating since 2022, which has found nothing. Beside the timeline, in smaller print, she has added a quote from physicist Katherine Freese: "We have been looking in the same places for forty years. Maybe it is not there."

The Rubin Observatory takes a different approach. Instead of trying to detect dark matter particles directly, it maps the gravitational distortions they create. When light from a distant galaxy passes near a massive object, the object's gravity bends the light—a phenomenon called gravitational lensing. Dark matter, because it has mass, should create the same effect. If you photograph enough galaxies with enough precision, you can map the invisible scaffolding of dark matter on which visible matter hangs.

That is the theory. The 3,200-megapixel camera photographs 3.5 billion pixels of sky every 30 seconds. Each image covers an area 40 times the size of the full moon. Over ten years, the survey will produce 20 billion galaxies, 17 billion stars, and 6 million asteroids. The raw data stream is 20 terabytes per night. By 2034, the total archive will exceed 500 petabytes—more astronomical data than has been collected in the entire history of astronomy before 2020.

Hebert's job is to find dark matter in that avalanche. Her team's algorithm scans for weak gravitational lensing: tiny, systematic distortions in galaxy shapes that indicate mass where no visible matter exists. Since January 2025, the algorithm has flagged 87,439 candidates. Hebert and her colleagues have reviewed 31,226 of them. Most are instrumental artifacts—smudges on the sensor, cosmic rays, diffraction spikes from bright stars. Some are gravitational lensing, but from visible matter: galaxy clusters massive enough to appear in other catalogues. Fourteen are unexplained.

"Fourteen unexplained signals in eighteen months," Hebert says. "At first that sounds promising. Then you realize we are sampling point-zero-zero-zero-one percent of the observable universe. If dark matter is diffuse—if it is spread evenly like a fog rather than clumped like stars—we will never see it this way. And if it does not interact with light at all, we are aiming the wrong instrument at the wrong question."

The Candidates That Keep Failing

Physicists have proposed more than sixty candidates for what dark matter might be. The leading theory for three decades has been WIMPs: Weakly Interacting Massive Particles, hypothetical subatomic particles with mass between 10 and 1,000 times that of a proton. WIMPs would rarely interact with ordinary matter, but occasionally—perhaps once a year in a detector containing several tons of ultra-pure xenon or germanium—one should collide with an atomic nucleus, producing a flash of light or a pulse of ionisation.

No one has ever detected such a collision. The LUX-ZEPLIN experiment, which began operating in February 2022 at the Sanford Underground Research Facility in South Dakota, uses ten tonnes of liquid xenon shielded by 4,850 feet of rock to block cosmic rays. After two years of operation, it has recorded zero WIMP candidates. The XENON collaboration in Italy reported similar results in 2023. The PandaX experiment in China's Jinping Underground Laboratory—the deepest physics facility on Earth, 7,900 feet below the surface—published null results in November 2024.

Elena Aprile, spokesperson for the XENON collaboration, told a conference in Geneva in March 2026 that the experimental sensitivity has now reached levels where the original WIMP hypothesis should have produced dozens of detections. "We have looked where the theory told us to look," she said. "It is not there. Either the theory is wrong, or the parameter space is smaller than we thought, or we are looking for the wrong thing entirely."

Alternative candidates have fared no better. Axions—hypothetical particles with a mass a trillion times smaller than an electron—were proposed in 1977 to solve a separate problem in particle physics, but could also constitute dark matter if they exist in sufficient quantities. The Axion Dark Matter Experiment at the University of Washington has been searching since 2018 using a resonant microwave cavity tuned to detect the faint electromagnetic signal an axion would produce in a strong magnetic field. Five years in, it has found no axions. The ADMX-EFR detector, which began operations in January 2025, extends the search to higher frequencies. Preliminary results, released in April 2026, are negative.

The Heretic in Toronto

Stacy McGaugh, 54, is a professor of astronomy at Case Western Reserve University in Cleveland, Ohio. He does not believe dark matter exists. He is not a crank: his 2016 paper on galaxy rotation curves has been cited 1,847 times, and he serves on the editorial board of the Astrophysical Journal. But he argues that the dark matter hypothesis is a century-long mistake, and that the evidence points instead to a modification of gravity itself.

"We invented dark matter to save Newton," McGaugh says over a video call from his campus office, where the shelves are lined with three decades of observational data on low-surface-brightness galaxies. "Galaxies were not behaving as Newton predicted, so we added invisible mass until the equations balanced. But there is another possibility: Newton's laws are accurate for the solar system and laboratory scales, but break down at galactic scales. Modified Newtonian Dynamics—MOND—proposes exactly that."

MOND, first proposed by Israeli physicist Mordehai Milgrom in 1983, posits that at very low accelerations—such as those experienced by stars in the outer regions of galaxies—gravity behaves differently than Newton described. The theory has a single adjustable parameter, and with it, McGaugh has successfully predicted the rotation curves of more than 150 galaxies without invoking dark matter. In 2016, his team published predictions for the rotational behaviour of ultra-diffuse galaxies based on MOND. When those galaxies were observed in 2020 using the Dragonfly Telephoto Array, the predictions were accurate to within measurement error.

"Dark matter has had forty years to show itself," McGaugh says. "It has not. Meanwhile, MOND keeps making successful predictions. At some point, you have to ask whether the data is trying to tell you something."

The problem with MOND is that it struggles to explain phenomena beyond galaxy rotation. The cosmic microwave background, gravitational lensing around galaxy clusters, and the large-scale structure of the universe all fit dark matter models more naturally. Attempts to extend MOND into a full theory of relativistic gravity have produced frameworks such as TeVeS and AQUAL, but none has gained the predictive power or empirical support of General Relativity. Most cosmologists regard MOND as an interesting empirical rule that awaits a deeper theoretical foundation—or as a historical curiosity that will be abandoned when dark matter is finally detected.

What the Silence Means

In February 2026, the Particle Physics Project Prioritization Panel—a body convened every decade by the U.S. Department of Energy and the National Science Foundation to recommend the future of American particle physics—released its long-awaited report. It did not recommend a new dark matter detector. Instead, it proposed shifting resources toward neutrino experiments and precision measurements of the Higgs boson. The report acknowledged that "direct detection experiments have reached sensitivity thresholds where continued investment yields diminishing returns in the absence of a theoretical breakthrough."

The decision was not universally welcomed. At CERN, planning continues for a next-generation collider—the Future Circular Collider, a 91-kilometre ring that would cost an estimated €20 billion and could produce dark matter particles if they exist within a certain mass range. Proponents argue that the LHC's failure to find supersymmetric particles does not rule out dark matter; it only rules out certain models. Critics point out that the same argument was made about the Superconducting Super Collider, cancelled by the U.S. Congress in 1993 after $2 billion had been spent and 22 kilometres of tunnel excavated in Texas.

Hebert, in Chile, is agnostic about the politics. Her contract runs through 2029. The Rubin Observatory will continue photographing the sky regardless of whether dark matter is found, because the data serves dozens of other purposes: mapping asteroids, tracking supernovae, measuring cosmic expansion. But the original scientific justification—the Legacy Survey of Space and Time, designed explicitly to map dark matter through gravitational lensing—rests on an assumption that is looking increasingly fragile.

"I think about Vera Rubin," Hebert says, turning away from the screen of galaxies. "She spent fifty years measuring galaxy rotation curves. She proved something was wrong with our picture of the universe. But she died in 2016 without knowing what dark matter actually is. I think about whether I will die without knowing either. Whether anyone will."

The Ridge Above the Atacama

At 11:30 p.m., Hebert drives back up the ridge to the observatory. The road is unpaved and climbs 1,400 vertical metres through terrain so dry that NASA tests Mars rovers here. The dome, when it comes into view, is white and enormous and completely incongruous, like a cathedral in a desert where no one has ever lived. Inside, the telescope is already tracking. The camera is already recording. The data is already streaming to servers in California and Illinois and France.

She does not go inside. Instead, she walks 200 metres east to a flat outcrop where the Milky Way is visible as a luminous band from horizon to horizon, so bright it casts shadows. Somewhere in that light, or in the darkness between the light, is the answer. She has been looking for eighteen months. The telescope will keep looking for another eight years. After that, someone else will design a bigger telescope or a more sensitive detector or a new theory, and the search will continue.

Or the search will stop, and physicists will conclude that dark matter, as imagined for ninety-three years, does not exist. That the universe is stranger than anyone predicted. That the scaffolding is not invisible matter but something else entirely: a curvature of space, a failure of Einstein's equations at cosmic scales, a dimension we cannot perceive. That we have spent a century looking in the wrong place for the wrong thing.

Hebert stands on the ridge until the cold becomes uncomfortable. Then she walks back to the car, drives down the mountain, and returns to the office, where 221 new galaxies are waiting on the screen. She opens the first file. She begins again.