Gravity is the quiet architect of the Universe. Over billions of years, it drew matter together, building structure step by step: stars formed into galaxies, galaxies merged into groups, and these grew into enormous clusters containing thousands of galaxies and up to a million billion times the mass of the Sun. Together, they now trace a vast cosmic web that spans the observable Universe.
What we see, however, is only part of the story. Most of the mass that creates gravity to shape this web is dark matter —an invisible component that cannot be observed directly, but reveals itself through its gravitational influence. One of the most powerful ways to map this hidden mass is weak gravitational lensing: the subtle distortion of distant galaxy images caused by matter along the line of sight.
In this featured Nature Astronomy paper, we present an ultra-detailed wide-area weak lensing map constructed from high-resolution imaging with the James Webb Space Telescope (JWST) as part of the COSMOS-Web survey. The map covers a contiguous area of 0.54 deg², nearly three times the size of the full Moon, and reaches an angular resolution of 1.00±0.01 arcmin—more than twice as sharp as any previous space-based mass maps over comparably large areas. But the path to this result began long before any JWST data existed.
Waiting for JWST
I first learned that I would be working with JWST data in February 2022, months before the telescope had even launched. At that point, JWST was still a promise rather than a reality. We had an approved observing program, but no photons yet, and everything we planned was built on expectations about how well this new observatory would perform.
When JWST finally launched and began delivering data, that promise became real. I started my postdoc in early 2023, just as the first COSMOS-Web observations were arriving. Suddenly, we were looking at the deepest and sharpest views of the Universe ever taken, while also confronting the challenge of turning raw images into precise weak lensing measurements.
Some time later, collaborators including Gavin Leroy and David Harvey joined the project, bringing complementary expertise in mass-map reconstruction and lensing systematics. Together, we began converting these extraordinary images into a map of the invisible matter behind them.
Returning to COSMOS
The COSMOS field has a special place in the history of weak lensing. In 2007, a pioneering mass map derived from Hubble Space Telescope (HST) imaging—produced by Richard Massey and Jason Rhodes—provided the first wide-area view of the dark matter distribution in a representative region of the Universe.
Revisiting the same field nearly twenty years later with JWST naturally carries a sense of continuity: returning to one of the earliest wide-field lensing studies with a new generation of people and observational capabilities. Our high-resolution near-infrared images now reveal a much richer and more distant view of the same sky.
In this work, we measure the shapes of roughly 250,000 distant galaxies, corresponding to 129 galaxies per square arcmin—almost double what was achievable with Hubble. These galaxies serve as background markers whose shapes are subtly distorted by foreground mass, including invisible dark matter. By averaging over these distortions across many galaxies, we can reconstruct a map of the underlying mass distribution, in direct analogy to the original COSMOS lensing map. Measuring galaxy shapes in two filters (F115W and F150W) further improves the precision of these measurements, increasing the statistical power of the dataset and producing a sharper and more sensitive mass map.
The moment it became real
There was one moment in particular when the significance of this project crystallized. In May 2025, I presented a preliminary version of the JWST mass map at a COSMOS collaboration meeting, showing it side by side with the original 2007 HST COSMOS mass map.
As the slide appeared, people started clapping with excitement. It wasn’t polite applause; it was recognition. The same cosmic structures were there, but now resolved into finer filaments, sharper peaks, and richer detail. That moment captured what this project was really about: not just improving a measurement, but opening a new window on the dark Universe.
Comparison of weak lensing mass maps in the same region of sky derived from Hubble Space Telescope data (2007, left) and James Webb Space Telescope observations (2026, right). The increased resolution of JWST reveals finer structure in the projected dark matter distribution and previously unresolved structures. Credit: NASA/STScI/A. Pagan
What the new mass map reveals
The resulting weak lensing map traces the projected distribution of total matter—both dark and luminous—across galaxy clusters, filaments, and underdense regions. Prominent peaks correspond to massive clusters, many independently confirmed through X-ray emission from hot intracluster gas. Beyond these peaks, the map reveals faint structures connecting overdense regions: the strands of the cosmic web itself.
JWST’s depth also allows us to push these measurements much farther back in time. The map traces mass structures out to redshifts of about two, when galaxies were forming most rapidly. By comparing the mass map with X-ray and galaxy density maps in the COSMOS field, we see where dark matter, hot gas, and galaxies trace the same structures—and where they do not.
A zoom-in view of the COSMOS-Web field combining weak lensing mass, galaxy density, X-ray emission, and JWST imaging, highlighting a high-density region where dark matter, hot gas, and galaxies intersect. Credit: NASA/STScI/A. Pagan
A new regime for mapping the dark Universe
The most striking aspect of this result is not any single feature, but the observational regime it opens. The combination of wide area, high source density, and arcminute-scale resolution allows us to move beyond simply identifying the most massive objects and instead map the connective tissue of the cosmic web at earlier times.
Long used as a reference field for studies of galaxy evolution, it now provides a high-resolution benchmark against which future maps of the dark Universe can be calibrated and compared. While this work delivers an ultra–high-resolution view over a relatively small region of sky, it is best seen as a foundation. The next step is to extend this cartography across much larger volumes of the Universe using missions such as ESA’s Euclid and NASA’s Nancy Grace Roman Space Telescope, while moving beyond two-dimensional projections toward three-dimensional reconstructions of the cosmic web. The JWST COSMOS-Web data will serve as a key calibration point for that transition.
Returning to the COSMOS field with JWST—nearly two decades after the first space-based weak lensing mass map—highlights how advances in observational capability continue to reshape our view of the cosmos. With sharper eyes on the sky, the dark Universe becomes not only visible, but richly structured.