Einstein predicted it in 1915 and doubted anyone would ever observe it: a massive object — a star, a galaxy cluster, a concentration of dark matter — would bend the path of light passing near it, just as it bends the paths of planets and photons alike. The amount of bending depends on the mass of the deflector and the geometry of the alignment. If the geometry is perfect — source, lens, and observer aligned along the same axis — the deflected image wraps into a complete ring around the foreground object. Eddington's 1919 solar eclipse expedition confirmed gravitational lensing with the Sun. Sixty-three years later, in 1988, the first extragalactic Einstein ring was confirmed. Today, gravitational lensing has become one of the most powerful tools in cosmology, and it works equally well whether the mass doing the lensing is visible or not.
That last property is what makes lensing so important. Dark matter does not emit or absorb electromagnetic radiation. You cannot image it, spectroscopically analyze it, or directly detect it with any conventional telescope. But it does have mass, and mass curves spacetime. Wherever dark matter accumulates — in the halos surrounding galaxies, in the filamentary structure of the cosmic web, in the massive clusters where thousands of galaxies congregate — it bends the light of background sources, and that bending is measurable. Gravitational lensing is, in a meaningful sense, the only direct way to map dark matter's distribution.
Strong lensing: rings and arcs
Strong gravitational lensing occurs when the alignment between source, lens, and observer is tight enough to produce visible distortion — arcs, partial rings, or multiple images of the same source appearing around a foreground object. Galaxy clusters are the most powerful strong lenses because their combined mass, including both baryonic matter and dark matter halos, can reach 10¹⁵ solar masses. In Hubble images of clusters like Abell 2744, the arcs are unmistakable: blue streaks curving around the yellow cluster galaxies, each one a distorted and magnified image of a galaxy that lies far behind the cluster.
These images contain information. The radius of the arc, its shape, and the positions of multiple images of the same source all depend sensitively on the mass distribution of the lens. Astronomers construct detailed mass models — effectively 2D maps of the lens mass — by fitting these observations. The models consistently show that the bulk of the mass in galaxy clusters is not in the galaxies themselves but in a diffuse halo extending far beyond the optical extent of the cluster. And that halo's mass distribution follows the predictions of cold dark matter models rather than modified gravity alternatives — at least in most cases.
Weak lensing: the statistical signal
Most mass concentrations in the universe are not massive enough to produce visible arcs. But even a relatively modest galaxy in the foreground will slightly distort the shapes of background galaxies — stretching them tangentially relative to the lens direction by fractions of a percent. This weak lensing signal is invisible in any individual galaxy image, because galaxies are intrinsically elliptical and you cannot tell a lensed shape from a naturally elongated galaxy. But if you measure the shapes of millions of background galaxies and look for coherent patterns of alignment relative to foreground structures, the statistical signal emerges from the noise.
This is exactly what surveys like the Kilo-Degree Survey, the Dark Energy Survey, and now ESA's Euclid mission are designed to do. Euclid, launched in 2023, is conducting a weak lensing survey of roughly 1.5 billion galaxies over a third of the sky, measuring galaxy shapes with the precision needed to detect coherent alignments at the sub-percent level. The survey data will reconstruct the 3D distribution of matter — dark and baryonic — across the past 10 billion years of cosmic history.
The S8 tension and what lensing says about it
One of the most discussed problems in contemporary cosmology is a discrepancy between two measurements of how clumpy matter is on large scales, quantified by a parameter called S8. Measurements derived from the cosmic microwave background predict a value of S8 that is systematically higher than what weak lensing surveys measure from the late-time distribution of galaxies. This tension, at roughly 2 to 3 sigma significance, is either a systematic error in one or both measurements, or evidence that our standard cosmological model is missing something about how structure grows.
Euclid's first cosmological data releases, combined with ground-based surveys, will either strengthen the tension — making it harder to dismiss as systematic error — or reveal it as an artifact of calibration issues in galaxy shape measurements or photometric redshift estimates. Either outcome is scientifically valuable.
The arc around a galaxy cluster is beautiful in isolation. But what makes gravitational lensing scientifically powerful is the realization that the universe's dark matter skeleton is encoded in the shapes of billions of background galaxies, waiting to be decoded. Einstein's 1915 prediction has become a precision cosmological instrument — not something he could have anticipated, and not something most astronomical observations ever achieve.
