If you’re looking for Star Wars' forest moon of Endor or Avatar's Pandora among exoplanets, prospects may not be promising: such Earth-like moons may be rare, and the ones that exist could be stranger than anyone thought.

Using data from the Kepler space telescope, a pair of astronomers from Columbia University may have found the first example of an “exomoon”: a Neptune-size world circling a planet 10 times Jupiter's mass that is about between 80 and 90 percent as close to its sun as Earth is to ours. The exoplanet, dubbed Kepler 1625 b, is 4,000 light-years distant in the constellation Cygnus. Its moon (if it is indeed one) orbits at a distance between 17.2 to 21.2 times the planetary radius, or 747,000 to 922,000 miles. From the cloud tops of Kepler 1625 b the exomoon would be several times the diameter of a full moon as seen from Earth.

Finding an Earth-size, habitable moon would have been a coup, but it is possible that Neptune-size moons might be more common, according to the new study, led by Columbia astrophysicists Alex Teachey and David Kipping, which was posted to the preprint server arXiv and has been submitted to The Astrophysical Journal. Their models showed that if you visited any star with a planet orbiting from the same distance as Earth down to one tenth that, there is about a 38 percent chance (and likely less) that you would run into a planet and moon system similar to Jupiter's four Galilean satellites (Io, Europa, Ganymede and Callisto), with similar ratios of moon to planetary diameters and orbital to planetary radii.

Estimates of the odds of a planet in general vary; some studies suggest sunlike stars have about a one in 10 chance of hosting an Earth-like planet, for example, whereas others say it’s possible nearly every main-sequence star has at least one planet of some type orbiting it.

Although it seems Earth-like moons are in the minority, that might not be the whole story. “I’d hesitate to write [moons like ours] off,” Teachey says. “We looked at 284 planets, but in the galaxy we’re talking about hundreds of billions of planets. Our results don’t rule these worlds out; they are just going to be more rare than we hoped and expected.”

To find evidence of these satellite worlds, Teachey and Kipping first combed through 284 objects in Kepler's catalogue of candidates and confirmed exoplanets. They eliminated those with orbital radii less than one tenth that of Earth’s, because at that distance moon systems might not remain in stable orbits around their planets on billion-year timescales. They used the mean Earth–sun distance, also known as an astronomical unit, as an outer cutoff point. With population-level data they were able to calculate the odds that a moon (and its gas giant) would be in the habitable zone around a star—the region where liquid water can exist.

The next task was to track down an individual exomoon. Kepler watches for the slight change in the brightness of a star when an exoplanet passes in front of it, an event called a transit. Plotting the brightness of the star over time generates a U-shaped light curve. An exomoon will produce an additional, small dip on one or both sides of the U.

That was how Teachey and Kipping found what they think is Kepler 1625 b's moon. The telescope caught five exoplanet transits; three produced useful data—and the third showed something odd. “The shape of transit 3 really baffled me for a time," Teachey says. Analyzing the data brought out the small dips which, they thought, showed the moon they were seeking. “If [the moon is] real, maybe it shouldn’t be terribly surprising that we saw it, since it’s large enough to be detected as a planet in its own right, at least around some stars. It should be easier to spot the big [exomoons],” Teachey says. It’s a basic bias in transiting exoplanet surveys: Larger objects will produce larger changes in a star’s brightness, so Kepler is more likely to detect big planets or moons.Another bias is planets with shorter orbits. To be confirmed as a planet, an object needs to transit at least twice, but Kepler's primary mission only lasted four and a half years. In the solar system Jupiter takes 12 years to orbit the sun. An alien Kepler mission could not confirm the gas giant’s existence in that amount of time, even if it caught one transit.

The next question is how such a planetary system may have emerged. Moons can form in one of three ways: accretion around a developing planet, capture by a planet’s gravity or a giant impact from an asteroid or planet-size body that carves it out of a planet. The general consensus among planetary scientists is that accretion is extremely unlikely to produce moons more than a few ten-thousandths the mass of their host planet. A satellite that is bigger than that—like the moon—probably requires a giant impact or capture.

Impacts have been shown to produce giant moons in our solar system, including our own. But René Heller, an astrophysicist at the Max Planck Institute for Solar System Research who was not involved with the study, says he is not sure that applies in the case of the newly discovered exomoon candidate. A pair of icy or rocky protoplanets, forming in a kind of wide binary system in which they orbit each other far apart, seems more likely. “That said, I’ve never seen a study that investigated such a possibility,” he says. “So it might be impossible. Either way, if this ginormous exomoon were confirmed, it could require a completely new model for how moons can form.”

Whether the exomoon exists is still an open question. Teachey and Kipping plan to observe the system with the Hubble Space Telescope starting in October. Meanwhile astronomers not part of the work counsel caution: Eric Agol at the University of Washington notes that the team’s confidence in their results is based on assumptions that might not hold; stars' brightness can fluctuate slightly and create false positives. “The way to [confirm] is precisely what they have proposed to do: Follow up observations with other telescopes,” he says.

Tomer Holczer, a former exoplanet researcher at Tel Aviv University, says it is likely a wide variety of moons are out there. “I don’t think there is a question of [whether exomoons] exist,” but rather their rarity, their size distribution, the types of planets they orbit and so on, he says.

Teachey, for his part, does not want to speculate too much on the nature of a yet-unknown moon. “Let’s confirm this thing first, and then we can let the theorists figure out how it got there.”