A Breakthrough Listen survey was thought to have covered about 288,000 stars catalogued by Gaia. But a new simulation of the Milky Way estimates that more than 6.18 million stars actually fell within its radio-telescope beams—millions of potential worlds too faint or distant for Gaia to catalogue reliably, yet still within range if they were transmitting strongly enough.

A radio beam is wider than its target

A large radio telescope may be pointed at one named star, but its receiver accepts radiation from a patch of sky around that target. Every foreground and background star within the useful beam is therefore observed incidentally. SETI researchers call this extra population stellar bycatch.

An optical telescope can separate points of light into a catalogue. A single-dish radio observation instead combines incoming power across its beam, with the strongest response at the centre and declining sensitivity toward the edge. A sufficiently strong narrowband signal from an untargeted star could still enter the recorded data.

The new MNRAS analysis argues that earlier counts missed most of this bycatch. Louisa Mason, Michael Garrett and Andrew Siemion simulated 6,182,364 stellar objects across 1,229 usable Breakthrough Listen pointings, extending the population to 25 kiloparsecs from Earth.

The result does not mean six million stars were newly discovered, nor that six million planetary systems were searched to equal depth. It means a statistical Milky Way model places that many stellar objects inside the beam footprints. Detectability still depends strongly on distance, position and transmitter power.

The search being recounted

The calculation revisits observations made with the 100-metre Green Bank Telescope in West Virginia and the 64-metre Parkes telescope in Australia. Breakthrough Listen’s radio search programme aimed these large single dishes at nearby stars and recorded fine-frequency data across multiple receiver bands.

Search software looked for narrowband radio emission, a useful technosignature candidate because natural astrophysical sources rarely concentrate power into extremely fine frequency channels. It also examined frequency drift. Relative motion between a transmitter and Earth should Doppler-shift a narrow signal during an observation.

A 1,327-star survey covered 1.10 to 3.45 gigahertz with Green Bank and Parkes. Its named targets were nearby, but each beam included additional sightlines through the Milky Way. Those unplanned stars were present in the same data even when researchers lacked reliable catalogue distances for them.

The original null result applied to the frequencies, drift rates, observing times and signal strengths tested by the pipeline. It did not imply that every kind of transmission from every star in the beam would have been detectable. The new paper changes the statistical accounting, not the recorded observations.

Gaia raised the count first

A 2020 Gaia-based extension counted 288,315 stellar objects within the full width at half maximum of the telescope beams. At that boundary, beam sensitivity has fallen to half its peak. The result was already more than 200 times the central target list.

That study showed how a targeted survey could support a broader census. Gaia parallaxes supplied distances to real foreground and background sources, allowing each star’s required transmitter power to be estimated. Yet the method could count only objects that Gaia detected and measured well enough.

The Gaia mission has transformed Galactic astronomy, but it remains an optical survey with selection limits. Faint stars fall below its magnitude threshold, crowded fields confuse sources, and parallax uncertainties grow important at large distances.

Those omissions matter along long paths through the Galactic disc, where enormous numbers of dim red stars should exist behind dust and brighter foreground sources. A catalogue is strongest when identifying actual objects. It is not automatically a complete inventory of everything geometrically inside a radio beam.

Building the unseen population statistically

Mason and colleagues used the official Besançon Galaxy Model, a population-synthesis framework representing the Milky Way’s thin and thick discs, bulge and halo. It is constrained by multiple surveys and by the Galaxy’s gravitational structure.

The model combines stellar density, star-formation history and an assumed initial mass function. Evolutionary tracks then assign luminosity, surface gravity and temperature. Atmosphere and three-dimensional extinction models translate those physical properties into colours and apparent magnitudes along each line of sight.

Each synthetic entry can carry position, distance, magnitude, colour, age, mass, temperature and spectral class. The researchers simulated a rectangular sky region for each pointing, then filtered it to the circular full-width-at-half-maximum footprint used for the radio observation.

These are simulated stellar objects, not six million new catalogue detections. Repeated population draws provide statistical uncertainty, but no synthetic entry becomes a star with a measured coordinate or a confirmed planet. The model estimates what the Milky Way should contain where optical completeness fails.

Why 6.18 million is not one sensitivity

The exact total, 6,182,364, came from 1,229 unique pointings from the Price survey. Two exceptionally dense fields toward the Galactic Centre were excluded because they took too long to simulate through the model’s web portal. The reported total applies to the successfully modelled set.

The synthetic population extends to 25 kiloparsecs, or about 81,500 light-years. That reaches through much of the Milky Way, but merely occupying the beam does not establish useful sensitivity. A transmitter’s received flux falls with the square of distance, while beam response also weakens away from the centre.

The practical quantity is effective isotropic radiated power. It expresses how powerful an equally bright transmitter radiating in all directions would need to be. For each star, the threshold depends on distance, telescope sensitivity, spectral resolution, integration time and the signal-to-noise cut.

Frequency drift creates another penalty. If a signal moves across channels faster than the search can correct, its power smears and sensitivity falls. The team’s calculator includes a dechirping efficiency to estimate this loss. Faraway stars therefore contribute only to limits on extraordinarily powerful transmitters.

What the null result can actually say

The strongest quoted population limit is not based on all 6.18 million objects. Within 2.5 kiloparsecs, the model produced 301,472 plus or minus 549 stars. At that distance the observations could test a defined, still exceptionally powerful class of transmitters.

Using the standard 95 percent Poisson upper bound after zero detections, the paper estimates that no more than 0.000995 plus or minus 0.000002 percent of those systems host continuous, high-duty-cycle transmitters above roughly 5 times 10¹⁶ watts at near-zero drift rates.

That is about one system in 100,000, but only under the stated assumptions. An intermittent beacon could be silent during the short observations. A civilization might transmit at another frequency, use broadband or pulsed signals, beam elsewhere, or sit below the power threshold.

The Royal Astronomical Society summary makes the boundary clear: no candidate in a narrow search does not mean intelligent life is absent. It means no signal matching the tested form appeared above the relevant sensitivity.

The Galaxy model has blind spots too

The Besançon model improves completeness at large distances, but it is not uniformly superior to direct observation. Within roughly 100 parsecs, it simulated fewer stars along these pointings than Gaia had actually measured. On the smallest scales, a smooth population model can miss local structure.

The model reaches down to stars above about 0.12 solar masses but omits brown dwarfs. Its low-mass initial mass functions remain uncertain, asymptotic giant stars are added empirically, and extinction maps are less complete along very dense lines of sight and spiral-arm tangents.

Crowding near the Galactic Centre creates difficulties for both approaches. It hides objects from optical catalogues and makes population simulations computationally expensive and sensitive to uncertain extinction. A large synthetic count in such a direction should not be mistaken for millions of individually verified systems.

Catalogues and simulations solve different problems. Gaia anchors the nearby, measured population. The model supplies a statistical estimate where direct completeness fails. A careful SETI analysis can use the stronger source in each regime rather than treating either one as a perfect Galactic inventory.

The same data now cover more systems

SpaceDaily’s earlier account of the Breakthrough Listen survey described detailed observations of nearby targets and the release of its radio data. The new work does not alter a recorded byte, add telescope time or retroactively observe new frequencies.

It changes the denominator used to interpret a null result. If a powerful transmitter anywhere inside the primary beam would have registered in the searched spectrum, then every sufficiently sensitive star in that footprint belongs in the statistical sample, whether or not it was the intended target.

That larger denominator tightens limits only after stars are grouped by the power detectable at their distances and beam positions. Counting all 6.18 million as equally searched would exaggerate the result. Counting only the named targets would discard genuine sensitivity to bright transmissions elsewhere.

The broader population also includes more spectral and luminosity classes than a nearby target list chosen around familiar stars. That makes the search less dependent on assumptions about which stellar environments a technological species would prefer, while leaving signal-form assumptions firmly in place.

A reusable accounting method

The team built a calculator that accepts a pointing direction, telescope field of view and instrument sensitivity, then filters Besançon simulations to the beam. It estimates transmitter thresholds and displays the expected spectral classes and luminosity types within the incidental sample.

The method can extend beyond traditional single dishes. The authors describe applications to incoherent beamforming surveys such as COSMIC on the Very Large Array and BLUSE on MeerKAT, as well as interferometric SETI observations, provided each instrument’s response and sensitivity are handled correctly.

Future gains can come from better Galactic models, deeper catalogues and more complete accounting, but also from widening the radio search itself. Additional frequencies, repeated visits, broader drift-rate coverage and different signal morphologies explore dimensions that a larger star count alone cannot touch.

The finding is an audit of search coverage, not a discovery of worlds or signals. Millions of statistically expected stars occupied sky already examined, but only powerful, persistent transmitters would have been visible at great distances. Better bookkeeping makes a null result more informative while preserving exactly what it cannot answer.