Not all planets have a home. For decades, astronomers and science fiction authors alike have speculated about orphaned orbs cast adrift from their home stars, endlessly wandering the boundless reaches of interstellar space. Most theorists hold that such ejections should be quite common during the chaotic tumult of a planetary system’s early days, when closely-packed worlds whirling around a star can scatter off each other like billiard balls in a break shot. Studying the properties of these far-flung planetary nomads—their numbers, masses and trajectories—could allow scientists to reconstruct these bodies’ murky origins and peer into a crucial formative stage of planetary systems that is otherwise largely hidden to us.

Hard evidence for this population of planetary nomads has proved elusive—floating cold and lightless in the void, these dark worlds cannot be directly observed by any conceivable telescope. Very rarely, however, one might pass in front of a far-distant background star, creating a detectable blip of light as the planet’s gravitational field acts as a magnifying lens. The duration and strength of such a “gravitational microlensing” event could reveal not only a rogue planet’s existence but also its mass, as bigger worlds tend to create longer, stronger amplifications of a background star’s light. A typical free-floating Jupiter-mass planet, for instance, is estimated to create an amplification lasting one to several days. A smaller, Earth-sized object might only amplify a star for a few hours.

It takes intensive calculations and a complicated series of assumptions to extract a rogue planet’s basic details from the deceptively simple brightening of faraway stars. But experts broadly agree that it can be done—so a handful of telescopic surveys around the world now monitor hundreds of millions of suns night after night to seek these objects, gradually taking a bulk census of the Milky Way’s loneliest worlds from the telltale twinkles of chance cosmic alignments.

The latest results of that census appeared on Monday in Nature, and come from Poland’s Optical Gravitational Lensing Experiment (OGLE), a 1.3-meter telescope in Chile. Based on a statistical analysis of more than 2,600 microlensing events, drawn from six years of observations on about 50 million stars, the OGLE team estimates that there is perhaps one Jupiter-mass rogue planet for every four stars in the galaxy. This result meshes well with leading planet-formation theories and more conventional observations of other planetary systems, and appears to refute the previous best guess from a competing team that estimated rogue Jupiters should be roughly twice as common as stars. (Neither result is anything for Earthlings to fret about—space is so huge that the chances of a rogue planet wandering close enough to our solar system to cause harm are, well, astronomically low.)

“Our new microlensing observations are in agreement with theoretical expectations on the frequency of free-floating Jupiters, and with infrared surveys for planetary-mass objects in star-forming regions,” says Przemek Mróz, lead author of the OGLE paper and an astronomer at the University of Warsaw Observatory in Poland. “We found that Jupiter-mass planets are at least 10 times less frequent than previously thought.”

Those earlier estimates of giant rogue planets being twice as common as stars sent shockwaves through the astronomical community when they emerged in a 2011 Nature paper from the Microlensing Observations in Astrophysics (MOA) survey, which uses a 1.8-meter telescope in New Zealand. “Very few of us in the microlensing field believed the original MOA results, simply because they were so difficult to reconcile with other observations and theory,” says Scott Gaudi, an astronomer at The Ohio State University. “But it was hard to know what was causing the apparent excess of events.”

The puzzling nature of the MOA results can be understood by imagining Jupiter-sized planets being a bit like sumo wrestlers, says Douglas Lin, a planet-formation theorist at the University of California, Santa Cruz. “A Jupiter can easily eject smaller planets, just like a sumo could throw a baby out of a ring. But a sumo has a hard time throwing another sumo out,” he says. “The trouble with the MOA result was that they claimed not only to see a lot of big sumos being thrown out, but that there were more sumos being thrown out than ones you see being left behind!”

To explain the MOA results, some theorists guessed that many of the purported rogue giant planets were actually free-floating failed stars called brown dwarfs—intermediate objects that straddle the hazy line between being a planet and a sun. Or, perhaps more likely, MOA’s free-floaters may have actually been run-of-the-mill giant planets bound to their stars, albeit in very wide orbits. Lin and others also began considering more exotic ideas, envisioning ways that binary star systems might somehow become factories for manufacturing giant planets and flinging them across the universe.

Others simply attributed the surprising results to sweeping statistical extrapolations being made from very small numbers of events, or to systematic weaknesses in the MOA survey’s observations and analyses that the 2011 paper’s authors had failed to account for. (Those authors, it should be noted, include some members of the OGLE team, who used a handful of data points from OGLE to bolster the MOA result at the time. The OGLE team says, however, that its survey’s early support for the MOA result gradually evaporated after upgrades boosted the quality and sensitivity of the full OGLE data set.)

David Bennett, a senior NASA research scientist who co-authored the MOA paper, stands by the 2011 results. He maintains that MOA’s discrepancy with OGLE is probably due to procedural differences in how the two surveys gather and analyze their observations—what OGLE might treat as a “brown dwarf” in its analysis, for instance, MOA might instead define as a “giant planet,” and each survey gathers data from a different swath of sky. “The data itself,” Bennett says, “may be more consistent than the interpretation.”

Such arguments over microlensing results are likely to grow in the immediate future, largely due to another new claim from OGLE that rivals the 2011 MOA result in potential for controversy. In its latest results the OGLE team reports not only seeing a relative dearth of free-floating Jupiters, but also indications of a vast population of smaller worlds. Of the more than 2,600 microlensing events the OGLE team observed and analyzed, six were “ultrashort,” lasting less than half a day—suggesting they were caused by objects somewhere between one and 10 times the Earth’s mass.

Astronomers have predicted such objects for decades. In Lin’s analogy they would be the “babies” ejected by the Jupiter-sized “sumos” in embryonic planetary systems. Some have even speculated that these tiny exiles could be habitable, harboring subsurface oceans kept warm and liquid by the decay of radioactive elements, shielded by crusts of ice or thick hydrogen-dominated atmospheres. Habitable or not, evidence for the existence of this class of free-floating world has to date remained circumstantial at best because their microlensing effects are so short and difficult to see. Experts—including members of the OGLE team—view the survey’s claimed detections of such low-mass microlensing worlds with measured skepticism. “Because our sensitivity to such short events was very low, free-floating Earths should be very common, perhaps more frequent than stars. But we are unable to provide a precise number owing to the small number of detections,” Mróz says. “We only very carefully proposed that the six [ultrashort] events may be microlenses caused by low-mass Earth-type planets. This is to encourage further studies of such objects rather than to make a strong claim.”

Results from other microlensing projects could corroborate or refute OGLE’s claim in coming years. The MOA survey is still ongoing, and according to Bennett the team is presently assembling a new analysis of its latest data that could be published as early as next year. Another survey, the Korea Microlensing Telescope Network (KMTNet), holds particular promise as it relies on three geographically distributed observatories rather than just one, allowing longer monitoring of target star fields that would otherwise be limited to no more than 10 hours due to the Earth’s rotation. Jennifer Yee, an astronomer at the Harvard–Smithsonian Center for Astrophysics and KMTNet team member, says that longer monitoring would make it easier to detect the signals of rogue Earths—and to distinguish them from confounding effects such as stellar flares, which can mimic ultrashort microlensing events.

Ultimately, the census of free-floating worlds—and lingering questions over what exactly lurks unseen in the outer dark—will be completed by observatories in space freed from mundane limitations like Earth’s rotation and weather. NASA’s Wide-Field Infrared Survey Telescope (WFIRST), a sort of super-Hubble with a panoramic view, is planned for launch in the mid-2020s with an intensive microlensing survey as one of its main science objectives. The telescope’s provenance is a complicated one, but one root of its origins lies in a proposal first made to NASA in the 2000s by MOA’s staunch defender, David Bennett.

“If you want to understand the possibility of life on other planets, it takes more than just finding one in the same size and orbit as Earth and trying to study it,” Bennett says. “There are many things that feed into a planet’s habitability—its atmosphere, its history, its water content—and those things can go all the way back to the details of its formation…. If we really want to look for life, we need to understand the processes involved in planet formation, some of which can eject planets and make them free-floating.” WFIRST’s microlensing studies promise to do just that.