In September 2014, when astronomer Iair Arcavi found a new supernova in the night sky, he didn’t think much of it. It looked like any other star that had just died and violently burst apart. The object had brightened some time ago, and now it was fading — a sign that the explosive event was coming to an end. Since the juiciest part seemed to be over, Arcavi abandoned the star in search of supernovae that had exploded more recently.

But robotic telescopes continued to monitor the star over the next couple of months, just in case anything interesting happened. Then, in early 2015, Arcavi asked a student to look through telescope data to see if any of the stars they had found might be acting strange. Sure enough, that “ordinary” supernova wasn’t so normal anymore. It was getting brighter — almost as if it had exploded again. “We’ve never seen a supernova do that before,” Arcavi, an observational astronomer at UC Santa Barbara and Las Cumbres Observatory, tells The Verge.

This immediately intrigued Arcavi, so he and his team started monitoring the star every couple days with the Las Cumbres Observatory’s telescopes located all over the world. Over the next two years, their data unveiled just how strange the supernova was: it stayed bright for a lengthy 600 days, instead of the typical 100 days supernovae usually shine before finally going dark. During this time, the star repeatedly grew brighter and fainter up to five times. It was like it was erupting over and over again, as if the star just refused to die.

This supernova, detailed today in the journal Nature, is unlike anything ever observed before. Now, scientists are scrambling to figure out what may have caused the strange fluctuations they saw. The leading idea is that this may have been something of an imposter — an event that looks like a supernova, but doesn’t ultimately lead to the destruction of a star. But even the best theories don’t quite match what astronomers observed, so the origins of this supernova still remain something of a mystery. “No existing model or theory of supernovae fully explains what we see here,” says Arcavi.

To learn more about the exploding star, the astronomers initially took its light and separated it into its component colors, known as taking spectra. “That allows us to get a fingerprint of the supernova,” says Arcavi. These colors can tell astronomers what materials a star is made of, how fast it’s moving, and how hot it is. The spectra showed that this supernova looked almost exactly like the most common supernova we know about.

But it was also incredibly different: not only did it fluctuate in brightness for a long time, but it also seemed to age in slow motion, according to Arcavi. Astronomers can tell how old a supernova is by measuring how fast the material is moving around it; usually the materials start to slow down a while after a supernova has exploded. With this star, however, the materials around the event remained at really high speeds. So after 600 days, the supernova looked as if it was only 60 days old.

Adding to the mystery is that this star may have even exploded before 2014 — actually, over half a century ago. When digging through archived telescope data, Arcavi and his team found another supernova that had exploded in essentially the same spot in the sky back in 1954. They’re pretty sure it’s the same star — or else a very unlikely coincidence. How did this star explode at least once in the ‘50s and then again three years ago?

So far, the best theory to explain what’s being seen is something known as pulsational pair-instability, or PPI. It’s an event that kind of masquerades as a supernova, occurring in stars around 100 times the mass of our Sun. When such a star reaches the end of its life, its core can get incredibly hot — upwards of billions of degrees — and become unstable. At this point, the oxygen inside the core ignites and the star will blow off an outer layer of material. The result is something that looks like a big explosion, but it ultimately leaves the core of the star intact. The core can then pulse like this again and again, shedding material until it eventually collapses into a black hole.

No one’s ever actually seen a PPI in action before, so this would be a big first. “It’s been in our arsenal of models for quite some time, but we’ve never had a real clincher before,” Stan Woosley, an astrophysicist who came up with the PPI theory and wrote an accompanying article in Nature, tells The Verge. However, even the PPI theory doesn’t exactly match what Arcavi and his team saw. The fluctuations from a PPI are thought to be much more varied in brightness, with more time in between bursts. (Astronomers also didn’t think a PPI could stay as bright as this event did for so long.) Plus, if this star did erupt back in 1954, it probably would have gotten rid of most of its hydrogen back then, but Arcavi says the team found the supernova to still be rich in hydrogen. “Maybe that model needs to be changed or revised, or this could be something completely new,” he says.

There are other less viable explanations for the zombie star, but nothing that fits quite right. After a long time shining, the supernova is now is fading to darkness. Arcavi says his team will continue to monitor the dimming event, and next month NASA’s Hubble Space Telescope will observe it to possibly find more clues — and that’s exactly what astronomers want.

“The bottom line is we don’t know what it is, but for scientists that’s the best news of all,” says Woosley. “We love a puzzle.”