Stars are huge - thousands, millions, if not billions of times the size of Earth - so it's come as quite a shock to astronomers who've discovered dozens of them are missing.
At least 100 that appeared on charts and graphs in decades past aren't there anymore, according to a new paper.
"Finding an actually vanishing star - or a star that appears out of nowhere - would be a precious discovery and certainly would include new astrophysics beyond the one we know of today," says Beatriz Villarroel of Stockholm University.
After noticing in 2016 a star in the constellation of Lupus had vanished, Dr Villarroel led an international team that looked at military sky maps and other public sources dating back to the 1950s to see what else might be missing.
The Vanishing and Appearing Sources during a Century of Observations (VASCO) project found about 100 "red transients" - stars that appear and disappear very quickly on an astronomical timescale.
When stars vanish, they usually do so in spectacular fashion - in a supernova - or very slowly, just burning out over millions of years before turning into a white dwarf.
Other possibilities do exist, but they're either exceedingly rare or have never been witnessed before, including a failed supernova - "when a very massive star collapses into a black hole without any visible explosion" or a Dyson sphere.
"A Dyson sphere is a hypothetical giant structure surrounding a star to harness its energy," the VASCO team said in a statement, somewhat doubtfully.
"We are clear that none of these events have shown any direct signs of being extra-terrestrial intelligence," said co-author Martin López Corredoira of the tituto de Astrofísica de Canarias in Spain.
"We believe that they are natural, if somewhat extreme, astrophysical sources."
A Dyson sphere was suggested as a possible cause of a bizarre anomaly surrounding a star in the constellation of Cygnus known as Tabby's star, named for its finder Tabetha Boyajian. It was later deemed to probably be just dust.
The VASCO astronomers have only analysed about 15 percent of the 150,000 possible red transients in their data, so are looking at setting up a crowdsourced analysis project.
Their findings to date were published in The Astronomical Journal earlier this month.
