For every star in the Milky Way, there may be about 20 planetary-mass objects travelling without a detectable sun. Applied across a galaxy with hundreds of billions of stars, the central estimate reaches into the trillions.
That number is not a tally of observed worlds. It comes from a 2023 paper in The Astronomical Journal that used six unusually short microlensing events to infer a much larger hidden population. The authors described the objects as free-floating planets or planets in very wide orbits, an important qualification that tends to disappear when the result is shortened to “rogue planets.”
This is a population estimate from one nine-year survey, not a direct galactic census. Its central value is striking, but its uncertainty is broad enough that the true number could be substantially lower or higher.
The result began with 3,535 flashes
The Microlensing Observations in Astrophysics collaboration, or MOA, monitored crowded fields toward the Milky Way’s centre from Mount John University Observatory in New Zealand between 2006 and 2014. Its statistical sample contained 3,535 high-quality single-lens events.
Most were caused by stars or other relatively massive objects. Six events with crossing times shorter than half a day were likely caused by planetary-mass lenses. After simulating the survey’s sensitivity, including the brief signals it would have missed, the team fitted a mass distribution that rises steeply toward lower masses.
The preferred model yielded 21 free-floating or very wide-orbit planets per star, but with a one-sigma range of about eight to 44. It covered objects from 0.33 Earth masses to 6,660 Earth masses, a range extending from sub-Earth worlds through bodies much more massive than Jupiter. The “20 times more” line is therefore the centre of a wide probability distribution, not a count accurate to the nearest planet.
Gravity reveals a planet that gives off almost no light
A cold, isolated planet is extremely difficult to see directly. Gravitational microlensing avoids that problem by detecting its effect on light from somewhere else. When a foreground object passes almost exactly between a telescope and a distant background star, its mass bends space-time and briefly magnifies the star.
The event’s duration carries information about the lens, although mass, distance and relative motion are partly entangled. Planetary lenses can produce signals lasting only hours. The shortest event in this sample, MOA-9y-5919, lasted 0.057 days, roughly 82 minutes. A companion analysis of the low-mass candidates gave it a central estimate of 0.75 Earth masses, with large uncertainties on either side.
These alignments do not repeat. Once the lens and background star move apart, that particular brightening is over. A survey must already be watching the right part of the sky at a fast enough cadence to catch it.
“Earth-sized” is more precisely Earth-mass
The light curve does not ordinarily disclose the lens’s radius. It also says nothing directly about a surface, atmosphere or chemical composition. For that reason, the title’s “size of Earth” should be read as a reference to mass, not a measured diameter.
NASA made the same boundary clear in its July 2023 account of the work: the candidate may share little with Earth beyond a similar mass. It could be rocky or icy, but microlensing alone cannot decide. Nor does an Earth-like mass imply Earth-like conditions on a world that receives no sunlight from a nearby star.
Some may have a star too distant to register
“Rogue” suggests an object definitively travelling alone. The observations are less categorical. If no host star appears in a short microlensing event, the planet may be unbound, or it may orbit a star at such a great distance that the star produces no clear lensing signature.
That ambiguity is why the paper repeatedly used the longer phrase “free-floating or very wide-orbit planets.” Later high-resolution observations can sometimes reveal a possible host. For many events, however, the classification remains statistical.
The inferred preference for low masses is consistent with ejection from young planetary systems. During formation, planets exchange energy through close gravitational encounters, and the lighter bodies are easier to throw onto distant or unbound trajectories. Some isolated planetary-mass objects may instead form directly from collapsing gas and dust. A reliable mass distribution could show how much each route contributes.
The estimate revised an older picture
An earlier MOA analysis had suggested a large population of Jupiter-mass rogues. The nine-year study did not recover that excess. Its much larger total instead came from extrapolating a rising abundance of lower-mass objects, based on a handful of short events and a new treatment of detection efficiency.
That shift matters. The result is not simply “more planets than expected.” It replaces a proposed population dominated by giants with one in which small worlds are far more numerous, while retaining substantial statistical uncertainty at the least observed end.
Space Daily’s earlier account of the same survey set out the six-event basis and the eight-to-44 range. The remaining question is whether a far larger experiment will preserve that steep low-mass distribution.
Roman is the test the estimate has been waiting for
NASA’s Nancy Grace Roman Space Telescope is built to turn such extrapolations into a much stronger census. Its wide-field infrared camera will monitor dense star fields in the galactic bulge, where microlensing events are comparatively frequent. NASA’s current microlensing overview says Roman should detect free-floating bodies from about Mars mass upward.
The original NASA report said Roman could find about 400 Earth-mass rogues. That number was a forecast derived from the MOA model, not a mission guarantee. More importantly, a large, consistently selected Roman sample would reveal the shape of the mass distribution and test whether the six short events represented a truly enormous population.
The mission’s schedule has also changed since the 2023 announcement. NASA is now targeting 30 August 2026 for Roman’s launch on a Falcon Heavy, rather than the older “by May 2027” date. The telescope will travel to the Sun-Earth L2 region about a million miles from Earth.
Twenty per star is a claim designed to be tested
Observing the same microlensing event from Roman and Earth creates a long baseline. The difference between the two views can provide microlensing parallax, helping separate mass from distance and motion. NASA’s updated rogue-planet guide presents this two-location measurement as a route to much more accurate masses.
Until those observations arrive, “trillions” remains the galactic implication of a model, not a catalogue. The evidence points to many more low-mass wanderers than earlier surveys could see, but the estimate deliberately carries a large range and includes planets whose distant stars may have escaped detection.
The stronger conclusion is methodological. Astronomers have learned how to infer an unseen population from brightenings that last less than a working day. Roman should soon show whether the Milky Way really contains far more planets between its stars than around them.