A short gravitational microlensing event observed from Earth and the Gaia spacecraft has given astronomers a direct mass estimate for an object with no detected host star. The lens, designated KMT-2024-BLG-0792/OGLE-2024-BLG-0516, has a mass of about 0.22 Jupiter masses, or roughly 70 Earth masses. That makes it slightly less massive than Saturn.
Subo Dong and 41 colleagues reported the measurement in a January 2026 paper in Science. Their analysis places the object about 3.05 kiloparsecs, nearly 10,000 light-years, from Earth. The event showed no lensing signature from a host star, although the data cannot exclude a star on an orbit so wide that it made no detectable contribution.
This is one study, not settled consensus.
The claim that these unseen worlds may outnumber stars by about 20 to one comes from a different paper and a different kind of evidence. It is a rounded population-model centre, not a count and not an upper limit.
The brightening lasted less than a day
The Korea Microlensing Telescope Network and the Optical Gravitational Lensing Experiment independently detected the event on 3 May 2024. Telescopes in Chile, South Africa and Australia watched a background red giant brighten as the gravity of an unseen foreground object bent and magnified its light.
The fitted Einstein crossing time was 0.842 days. Short events tend to indicate low-mass lenses because the duration scales with the square root of lens mass, but duration alone is ambiguous. Mass, distance and relative motion can trade off against one another to produce a similar light curve.
This event had a rounded peak caused by a finite-source effect. The lens passed across the apparent disc of the background star rather than treating it as a point, allowing the team to derive an angular Einstein radius of 18.6 microarcseconds. That removed one unknown, but a direct mass still required a microlensing parallax measurement.
Gaia supplied a second viewpoint
Gaia happened to scan the same position six times over a 16-hour period from its orbit near the Sun-Earth L2 point. It was an unusually favourable coincidence for a survey spacecraft that could not be redirected towards an alert. The event peaked about 1.9 hours later in Gaia’s data than it did in the ground-based light curves.
The delay arose because Gaia and Earth viewed the alignment from separated positions. Fitting that offset produced a microlensing parallax detection at about seven-sigma significance. Combining the parallax with the angular Einstein radius broke the usual mass-distance degeneracy.
The resulting lens mass was 0.219 Jupiter masses, with uncertainty of plus 0.075 and minus 0.046 Jupiter masses. Its inferred distance was 3.05 kiloparsecs, with uncertainty of plus 0.58 and minus 0.43 kiloparsecs.
That is a gravitational mass measurement, not a photograph of the object.
“Apparently drifting without a star” is deliberate wording
The event was consistent with one lens passing in front of one source. No host star appeared through an additional lensing feature. The Science paper therefore describes the object as either gravitationally unbound or on a very wide orbit, which is the usual observational boundary for microlensing candidates of this kind.
Calling it a rogue or free-floating planet is reasonable shorthand, but not every distant host can be ruled out from one brief event. Future high-resolution imaging may look for a faint companion once the foreground lens and background source have moved farther apart on the sky.
Saturn-mass does not mean Saturn-sized. Microlensing measures the gravitational influence of a lens; it does not reveal its radius, rings, atmosphere or surface temperature. The object could not be seen directly in these data.
Dong’s team argues that the mass is low enough, and the event’s position in the wider microlensing distribution is suitable, for formation in a protoplanetary disc followed by dynamical ejection. That is more likely in their analysis than star-like collapse into a very low-mass brown dwarf. The formation history is inferred rather than observed.
The 20-to-one estimate rests on a population model
The abundance figure comes from Takahiro Sumi and colleagues’ 2023 paper in The Astronomical Journal. The team analysed nine years of MOA-II observations towards the Galactic bulge, collected between 2006 and 2014. Its statistical sample contained 3,535 high-quality single-lens events.
Six events had Einstein crossing times shorter than half a day and were judged likely to have planetary-mass lenses. The researchers ran image-level simulations to estimate how detection efficiency changed with event duration and angular Einstein radius, then fitted a power-law mass function to the short-event population.
When integrated over objects from 0.33 Earth masses to 6,660 Earth masses, about 21 Jupiter masses at the upper end, the preferred model produced 21 free-floating or very wide-orbit objects per star, with uncertainty of plus 23 and minus 13. In ordinary numbers, its one-sigma interval runs from roughly eight to 44.
Twenty to one is therefore the rounded centre of a broad distribution. It is not the greatest value allowed by the study, and it is not 20 Saturn-mass planets around every star. The count is dominated by the much more numerous low-mass objects implied when the fitted power law is extended towards its 0.33-Earth-mass cutoff.
Why the galactic total remains unsettled
Microlensing surveys are least sensitive to the shortest events, so the inferred abundance depends on careful corrections for signals that occurred while a telescope was not observing, fell below its threshold, or were confused with variability and instrumental effects. A small number of detections can imply a large hidden population, but with correspondingly large uncertainty.
The same light curve can also come from a genuinely unbound planet or a planet in an extremely wide orbit whose star leaves no obvious trace. That is why the 2023 paper consistently grouped “free-floating or very wide-orbit planets.” The distinction matters when turning an event rate into a claim about worlds travelling entirely alone.
There is a further disagreement over the mass function. A 2025 study by Gavin Coleman and William DeRocco, available as an open manuscript and published in Monthly Notices of the Royal Astronomical Society, modelled planet formation and ejection across single-star, close-binary and wide-binary systems. It predicted 1.07 free-floating planets per star over nearly the same 0.33-to-6,600-Earth-mass interval, far below Sumi’s central value of 21.
The simulation does not overturn the observations. It instead predicts that a single power law fitted mainly above Earth mass will overstate the unseen population below that scale. Coleman and DeRocco found reasonable agreement with the observed slope above eight Earth masses, then predicted a turnover and fewer terrestrial objects. Space Daily’s recent examination of the 20-per-star result gives the survey calculation and its uncertainty more detail.
One weighed object does not settle the census
The Saturn-mass event answers a narrower question that earlier population work could not. It shows that at least one short, apparently isolated microlensing event was produced by an extrasolar object securely in the planetary-mass range. A single event cannot determine whether the Milky Way contains one such world per star, 20, or some number between or beyond those estimates.
NASA’s Nancy Grace Roman Space Telescope microlensing programme is intended to build a much larger sample, with sensitivity extending to lower-mass lenses. Event durations will set statistical constraints, while finite-source effects, parallax from separated viewpoints and later imaging can provide individual masses or test for host stars in favourable cases.
The next decisive comparison is the shape of the mass distribution: whether it continues rising steeply below Earth mass, as the 2023 fit implies, or turns over in the terrestrial regime, as some formation simulations predict.