A trip to see friends and give an informal talk about his graduate work turned out to be much more fruitful than Rasmus Voss had anticipated. The visit, to Radboud University in Nijmegen, the Netherlands, led to work that takes astronomers a step closer to establishing how one type of supernova forms.

At Radboud, Voss, now a fellow at the Max Planck Institute for Extraterrestrial Physics in Garching, Germany, spoke about his work analysing X-ray telescope data. Gijs Nelemans, a theoretical astronomer at Radboud, and the adviser of one of Voss's friends, attended the talk. He realized that with his knowledge of theoretical models and Voss's expertise, the two of them could tackle a model which suggested that the presence of X-ray signals could precede a type Ia supernova — an exploding star astronomers use to measure the expansion of the Universe.

At the time, two schools of thought prevailed about the origins of these supernovae. One theory was that supernovae result from the merging of two white dwarfs, which are small, dense stars. Such a merger would not be expected to be preceded by any X-ray data. The other possibility was that a white dwarf slowly accretes material from a companion star, producing X-ray signals in the process. If Voss and Nelemans found X-ray signals in positions at which type Ia supernovae later occurred, this would lend credence to the accretion model.

The origins of supernovae are difficult to study, because by the time researchers detect an explosion, the object that led to it has gone. So Voss and Nelemans decided to hit the archives in search of optical telescope data that showed supernova activity. Voss then trawled through earlier archival data from the Chandra X-ray Observatory, which can detect the most distant X-ray sources, to look for X-ray signals in locations at which supernovae had occurred. Their attempts, which focused on three supernovae that had occurred in 2002, 2004 and 2006, did not yield conclusive results.

The two decided to put the project on the back burner and wait for a new supernova to occur. “Not many of these events happen within a distance where they can actually be seen,” says Voss. “So we basically sat for half a year. Then a nearby supernova exploded.” But Voss didn't immediately scramble to the Chandra X-ray data archives, thinking that his chances of finding X-ray data to validate the accretion model were slim. “I had been working on another project,” he says, “So I delayed looking at the data.”

But when he did go back to the archives, he found the X-ray signatures that had eluded him on the other three supernovae. His reaction, he says, was a mixture of being “very embarrassed” for not immediately jumping on the data and “very excited” because the presence of X-ray signals preceding the supernova demonstrated the feasibility of their approach.

Their findings (see page 802), Voss says, do not exclude the merger model, because type Ia supernovae may occur by multiple means. But a likely outcome of the work is that more supernova specialists will start looking at X-ray data in their studies. And, as far as Voss is concerned, the next time he sees any supernova activity, he'll attack the X-ray data immediately.