NASA published a gallery of strong gravitational lenses observed by Hubble on November 12, 2026. The resolved systems can help researchers build lens models and prepare an automated search for similar arcs in Nancy Grace Roman Space Telescope images, NASA says. The gallery’s role is to prepare those analyses; Roman’s possible yields come from simulations, not observed counts. NASA’s gallery announcement

Gravitational lenses turn a galaxy’s warped light into a way to estimate mass that cannot be seen directly.

How a gravitational lens reveals hidden mass

Nine Hubble lens images arranged in a grid, with bright galaxy cores surrounded by curved or ring-shaped arcs.

A strong gravitational lens forms when the mass of a foreground galaxy bends and magnifies light from a more distant galaxy. The distortions—arcs, rings or multiple images—give astronomers a way to model the foreground system’s total mass, including its dark matter.

NASA’s nine-panel gallery includes an arc-counterarc pattern in panel A, an Einstein Ring in panel F, and quadruply lensed galaxies in panels C and I. In panel D, dust obscures part of a visible arc; infrared observations could help clarify whether it forms a more complete ring.

The systems in the gallery were first detected by the Dark Energy Spectroscopic Instrument (DESI) at Kitt Peak. Hubble then captured higher-resolution images, which give researchers detailed examples for modeling lens shapes and mass.

Hubble observes mainly visible and ultraviolet light, with some infrared coverage. Roman is designed to observe mostly in infrared, with some visible capability, over a much wider field. That makes the missions complementary: Roman can search a broader area for candidates, while Hubble’s detailed views can help researchers study individual lenses. NASA says the gallery can also support development of an automated search for similar arcs in Roman images.

What Roman’s lens forecasts count

A 2025 peer-reviewed study modeled possible yields for Roman’s High Latitude Wide Area Survey. Its fiducial simulation used 146-second exposures over a 1,700-square-degree survey and forecast about 160,000 detectable lenses. Of those simulated systems, about 500 met the study’s signal-to-noise ratio threshold of greater than 200 for characterizing substructure. The study’s Roman lens simulations

A separate scenario coadded four 146-second exposures and produced upper-limit forecasts. These figures describe different simulated conditions, not a confirmed Roman yield.

Simulation scenarioExposure assumptionDetectable lensesSystems for substructure characterizationCharacterization criterion
FiducialSingle 146-second exposures across a modeled 1,700-square-degree surveyAbout 160,000About 500Signal-to-noise ratio greater than 200
Four-exposure coadditionFour 146-second exposuresUp to 510,000Up to 5,600Upper-limit forecasts

The study cautions that effective exposure time does not increase linearly with the number of exposures. The four-exposure estimates are upper limits, not a promise that Roman will find those numbers of lenses.

Why Roman’s image quality matters across its field

Roman’s point-spread function (PSF)—the way an optical system spreads a point of light across an image—varies with position across its focal plane. The study finds that this variation can affect the small-scale image structure used to analyze dark-matter substructure. Its simulations therefore call for researchers to account for the PSF at each lens’s detector position.