In just three years, the cloud of dust surrounding a young star has disappeared, indicating that there's something wrong with our current ideas of planet formation.
It may mean that planets can form much more quickly than previously thought or, alternatively, that stars harboring planets could be far more numerous.
"The most commonly accepted time scale for the removal of this much dust is in the hundreds of thousands of years, sometimes millions," says Inseok Song, assistant professor of physics and astronomy at the University of Georgia.
"What we saw was far more rapid, and has never been observed or even predicted. It tells us that we have a lot more to learn about planet formation."
The scientists examined data from the Infrared Astronomical Satellite, or IRAS, which surveyed more than 96 percent of the sky in 1983. The star, known as TYC 8241 2652 1, is located in the Scorpius-Centaurus stellar nursery, and was originally surrounded by a cloud of dust that was identifiable by its distinctive radiation of infrared energy.
Examination in 2008 using a mid-infrared imager at the Gemini South Observatory in Chile showed the same pattern. But when observations were repeated a year later, the team discovered that infrared emission had dropped by nearly two-thirds.
And when NASA's Wide-field Infrared Survey Explorer, or WISE, took a look in 2010, the dust had mostly disappeared.
"It's as if you took a conventional picture of the planet Saturn today and then came back two years later and found that its rings had disappeared," says Ben Zuckerman of UC Los Angeles.
The researchers have several different explanations - but not one fits with conventional thinking about planet formation.
One possibility is that the process of accretion, whereby minute particles of dust left over after a star forms clump onto each other, happens more quickly than thought. If so, though, it's a lot more quickly, given that models show the process taking hundreds of thoudsands of years, rather than just, er, three.
"If what we observed is related to runaway growth, then our finding suggests that planet formation is very fast and very efficient," says Song. "The implication is that if the conditions are right around a star, planet formation can be nearly instantaneous from an astronomical perspective."
Unfortunately, it's not yet possible to check, as the planet's distance - 450 light years - means it's too far away for a planet to be observed.
Alternatively, the star may have absorbed the dust itself, implying that planet formation is much less likely than previously thought; or the dust may have been expelled from the sun's orbit altogether.
The fact that such clouds of dust can be so very temporary implies that there may be many more planets out there than believed.
"People often calculate the percentage of stars that have a large amount of dust to get a reasonable estimate of the percentage of stars with planetary systems, but if the dust avalanche model is correct, we cannot do that anymore," says Song. "Many stars without any detectable dust may have mature planetary systems that are simply undetectable."