There should be billions of Earths out there. Why can’t we find them?

In 2009, the Kepler space telescope constantly watched over some 200,000 stars in our corner of the Milky Way. It was looking for where life might exist—by pinpointing small, rocky planets in the temperate zones of warm, yellow suns, and figuring out just how special Earth is in the grand scheme of things. While the mission revolutionized the study of exoplanets, those main objectives went largely unfulfilled. A mechanical failure cut short Kepler’s initial survey in 2013. Astronomers would later discover just a single Earthlike planet in its dataset.
A decade later, researchers are finally closing in on some of the answers to the questions Kepler raised. Earthlike planets are probably rare, but not exceedingly so. Roughly one in five yellow stars could have one, according to a new analysis of Kepler’s data published in May in The Astronomical Journal. If the researchers’ conclusions are correct, that would mean the Milky Way might be home to nearly 6 billion Earths. Yet of the 4,000 likely exoplanets we’ve spotted, just one looks anything like our home planet. So where are the rest?
“[Truly Earthlike planets] are not hiding per se, it’s just that the sensitivity of our telescopes is simply not good enough yet [to find them],” says Dirk Schulze-Makuch, an astrobiologist at the Technical University Berlin, Germany, who was not involved with the research.
If astronomers want to find Earth 2.0s, research calculating the frequency of such worlds will give future telescopes their best chances of success.
In the current context of exoplanets, the phrase “Earthlike” doesn’t necessarily imply a pale blue dot. From a telescope’s point of view, there’s no dot at all—just an occasional dimming of a star as a planet passes by, blocking some tiny fraction of its light. Yet from this flickering, researchers manage to extract a few key facts. Deep flickers indicate giant planets, for instance. And frequent dimming is the mark of a planet in a fast, tight orbit. A planet earns the moniker Earthlike if these characteristics place it in the star’s so-called “habitable zone,” a balmy band of orbits where back-of-the-envelope math suggests the star’s warmth will allow surface water to stay liquid.
Michelle Kunimoto, the exoplanet scientist who led the recent analysis, adopted one standard definition of what it takes to be an Earthlike planet: a world between three-quarters and 1.5 times as large across as ours, orbiting a sun-like (“G-type”) star, at between 0.99 and 1.7 times our orbital distance. Only Earth satisfies those criteria in our solar system, with Mars being too small and Venus orbiting too close for inclusion.
Worlds checking all three boxes are almost certainly out there, as Kunimoto’s work, which earned her a PhD from the University of British Columbia, suggests. But they’re hard to spot. Dimming from small planets is hard to see. Plus, they might transit in front of their star just once every few hundred days—and astronomers need at least three transits to confidently claim a detection. To make matters worse, yellow suns are rare to begin with, making up just 7 percent of the 400 billion stars in the Milky way. The vast majority of the galaxy’s stars are dim red dwarfs, which may bathe nearby planets in lethal flares.
Mission planners didn’t know it at launch, but Kepler had almost no chance of completing its initially intended search. To rack up three transits of slower planets orbiting at the outer edge of their suns’ habitable zones, the telescope would have needed to peer unwaveringly at the same patch of sky for more than seven years. But its pointing machinery broke down after four, long enough to find planets only in roughly the inner half of their stars’ temperate zones.
What’s more, Kepler was designed with our sun in mind. But our star turns out to be special in more ways than one. “The sun tends to be quite quiet,” Kunimoto says, while Kepler’s stars crackled more from their intrinsic burning. “Essentially, it’s a lot harder to find the Earthlike planets [than mission designers expected].”
Kepler delivered the scientific goods in the form of a huge haul of thousands of exoplanets, mostly massive giants hugging their host stars. But researchers have been trying to infer the less epic, more familiar worlds that Kepler couldn’t quite make out ever since. (Kepler 452b, which is 10% wider than Earth and has a year that’s only three weeks longer than ours, is one prominent Earthlike exception.)
The new work builds on a method developed by Danley Hsu, an astronomer at Penn State, in 2018. Previously, many researchers assumed there’d be an even spread of planet sizes and orbits, but as the population of exoplanets has grown, some kinds of worlds seem common than others. For planets with years shorter than 100 (Earth) days, for instance, many are 50% wider than Earth and many are 150% wider, but few have twice our planet’s girth. To accommodate these unexplained oddities, Hsu and Kunimoto both broke the Kepler data into many different categories of size and orbit and analyzed them all in a more independent way. Kunimoto went a step further and generated her own list of exoplanet candidates, not relying on the official catalogue.
In the end, Kunimoto found that an Earthlike planet may circle one in approximately every five sun like stars. She stresses that this figure represents an upper limit, however, and that the worlds could well be somewhat rarer. Her results represent an emerging consensus that the Earth-to-sun ratio of the local Milky Way should hover in the ballpark of 1:10. That figure remains a bit rough, Kunimoto acknowledges, but it’s tighter than the wide ranges published previously, which suggested between one earth per fifty suns, to two earths orbiting every single sun.
Schulze-Makuch calls the estimate “reasonable” and says that this kind of research gives us a valuable glimpse at the answers to otherwise unknowable questions, such as “whether our solar system is typical or kind of a freak system.”
He cautions, however, against letting one’s imagination run wild with images of a galaxy awash in billions of blue and green, cloud-studded orbs. The limited criteria of orbit, size, and star type say little about whether the planets have protective atmospheres and magnetic shielding, water, or the materials needed for life to emerge.
Estimates like Kunimoto’s may also shape future missions and give them more of a chance of finding more Earthlike planets than Kepler had. The more common these planets are, the more mission planners will be able to focus on designing instruments that scrutinize individual worlds, as opposed to wider sweeps.
Schulze-Makuch hopes, for instance, that the Keplers of the future will carry “star shades” that block out stars to capture exoplanets as single pixels, whose variations could betray the passage of seasons or the presence of ice caps. Such innovations could narrow researchers’ definitions of what it means to be an Earthlike planet, but he predicts a clear-cut discovery of a true Earth 2.0—one sculpted by life—remains a long way off.
“If we just use the technology we have right now,” he says, “it feels like we’re light years away.”
Diagnostic Tool