In the 1980s, some astronomers started batting around the idea that the sun had a long-lost twin, circling undetected in the edges of the solar system. They suggested that the existence of a companion star to our own might explain some cataclysmic events on Earth, like the mass extinction of the dinosaurs. Perhaps the orbit of this star, they said, was capable of disrupting the Oort cloud, a massive region of icy objects beyond Neptune’s orbit. Its gravitational forces could dislodge comets and send them hurtling toward Earth. The astronomers named the hypothetical star Nemesis, after the Greek goddess of retribution.

Today, the theory of Nemesis, of an evil twin lurking in the night sky and messing with comets, has fallen out of circulation. The companion star has never been found. But that doesn’t mean it never existed.

Most sun-like stars in the universe—stars with masses similar to our own—exist in pairs. Astronomers don’t yet understand exactly how these pairs, known as binaries, form. But as they dig deeper into the mysteries of star formation, they’re finding some clues. The latest is a new study of a distant cluster of young stars in the Milky Way that suggests nearly all sun-like stars are born in pairs, bolstering the claim that our sun has a twin.

A pair of researchers from the Harvard’s Smithsonian Astrophysical Observatory and the University of California, Berkeley, used radio telescope observations to study the distribution and orientation of stars in Perseus, a massive, very cold gas-filled cloud located about 600 light-years from Earth. Perseus is home to dozens of sun-like stars less than a million years old, the baby versions of our 4.5 billion-year-old yellow orb in the sky. Using statistical models, the researchers turned back the clock and investigated how the star population within the cloud unfolded over time. They found that in order to explain what they were seeing, they had to assume that the stars inside Perseus first formed and lived in binaries, before other forces may have started pushing them around.

So, yes, the sun may have had a twin, in the very beginning—but not for long.

To understand why, let’s look at Perseus. Perseus, like other similar clouds, is home to the birthplace of stars: egg-shaped cocoons of molecular gas, known as dense cores. The researchers observed several kinds of systems across Perseus. There are wide binaries, which contain two young stars orbiting each other, separated by more than 500 astronomical units, or AU. (One AU is the average distance between the sun and Earth, about 93 million miles). These young stars appear to be aligned with the long axis of a dense core, which suggests they may have formed together. There are tight binaries, containing slightly older stars orbiting less than 500 astronomical units apart and showing no alignment to dense cores. And there are single-star systems, like our sun.

The researchers, based on their simulations, determined that all the stars started off as wide binaries, aligned with their egg-shaped cocoons. About 60 percent of the pairs split up a few million years later, flying apart. The rest spiraled closer together, forming tight binaries.

If our sun had a twin, it would have orbited at least 17 times farther from the sun than Neptune. After a few million years together, it drifted away for good, into interstellar space, mingling with the other stars of the Milky Way.

But why?

“We don’t know exactly how we lost it,” said Steven Stahler, a theoretical physicist at the University of California, Berkeley, and one of the researchers. Stahler said friction between the young stars and surrounding gas can cause them to shrink their orbits, moving closer and closer together. Mergers are rare, since young stars emit winds powerful enough to drive off clouds of gas and reduce friction. But the reasons for breaking apart remain more of a mystery. One star may be ejected from its dense core, or the core itself could splinter, leaving two halves with two stars to float apart.

Younger stars are trickier to observe than their older counterparts, thanks to their dusty, dense cores, which block out their light, said Sarah Sadavoy, the lead author and an astronomer at the Smithsonian Astrophysical Observatory at Harvard. But clusters of young stars are the best places to investigate the mysteries of star formation, she said.“There’s been less time for things to get disrupted,” she said.

Perseus is just one cloud in the cosmos, of course. Despite its location, the makeup of the cluster may not be universal. “All of the observations that are being conducted always give only a little snapshot of some phase in the process, and every observation that is being taken gives a different picture,” said Anna Frebel, an MIT astrophysicist who studies the early stars of the universe, and who was not involved in the study. “Today, 13 billion years later, the universe is a lot more complicated, and every different gas cloud where you might have stars has undergone its very specific, billion-year evolution.”

So, is there any hope of finding a sibling to our sun? Astronomers could study their velocity and motions and then work backwards, spinning time in reverse to see whether one once had the same origin. Or they could examine the absorption spectra of stars, to figure out their chemical composition. If astronomers spot a star whose composition matches that of our sun, they can speculate they both came from the same cloud billions of years ago.

But these searches only work well if the star recently moved away from the solar system. Billions of years have passed. Nemesis is probably thousands of light-years away. It could even be on the opposite side of the center of the galaxy.

“Ultimately, there’s no way to track this down,” Sadavoy said. “It’s lost in the cosmos at this point.”