Ten planets that appear to be drifting in interstellar space have been spotted by an international team of astronomers. The planets are so far from any host stars that they may not orbit a star at all, and could be drifting unbound through space. The team believes that such rogue planets could outnumber normal stars almost 2:1 and their existence could confirm computer simulations of solar-system formation.

More than 550 planets have so far been found beyond our solar system. The vast majority of these extrasolar planets – or exoplanets – have revealed themselves by their gravitational influence on their host star, or by the dip in brightness that they cause as they pass in front of their star. However, a clutch of 12 worlds had previously been found by gravitational micro-lensing.

This technique relies on the object of interest passing directly between the observer and a more distant background object. The mass of the foreground object acts like a lens and magnifies the light from the object beyond. If the foreground object is a star, then any orbiting planet leaves its own tell-tale fingerprint in the shape of the magnification. However, due to the need for an exact alignment, fewer than one in a million stars in the central part of the Milky Way are micro-lensed at any given time. This is why the number of exoplanets detected this way is low.

Sifting through 50 million stars

In an attempt to get around this problem the Microlensing Observations in Astrophysics (MOA) collaboration observes many stars at once. The new rogue planets were found in MOA observations of 50 million stars within the Milky Way between 2006 and 2007. "Over all the stars observed we are very confident that we witnessed 474 definite lensing events," lead-author of the study Takahiro Sumi, of Osaka University, Japan, told physicsworld.com. Of these 474 events, 10 lasted for less than two days. Seven of these 10 events were later confirmed by data from the Optical Gravitational Lensing Experiment (OGLE) collaboration.

The more fleeting the duration of the event, the less massive the lensing object; a duration of less than two days implies the mass of the foreground object to be much less than that of a star. In fact, Sumi believes the culprits to be planets roughly the mass of Jupiter. What is more, no stars were observed within 10 astronomical units of the lensing objects – one astronomical unit is the distance between the Sun and the Earth and Saturn orbits at about 9 astronomical units. "There is a possibility that these planets do have a host star. However, direct imaging of exoplanets by other teams suggests that such distant planets are very rare," Sumi explains. "This led us to conclude that the lensing objects are freely floating planets, unbound from any star," he adds.

Because they are short-lived events, and the result of chance alignments, Sumi didn't expect to uncover such a high yield of planet-lensing events with MOA. From statistical analysis of his data he was able to extrapolate a figure for how common these free-floating planets might be. "We found that unbound planets, with roughly the mass of Jupiter, should be 1.8 times more common than the stars we observed," Sumi explains.

Scattered into space

The existence of rogue planets isn't completely unexpected: they have been predicted from computer models of solar-system formation. "We think they are formed in the same way as other planets but get scattered from the system by gravitational interactions between them," says Sumi. Joachim Wambsganss, of the University of Heidelberg, Germany, who was not involved in the work, says that this research quantifies this process for the first time. "We just didn't know how often this happened," he said. "This research gives us an idea," he adds.

Wambsganss went on to describe the research as using a "clear and solid method", however he thinks some people may not believe the claims of the rogue planets' abundance. "They used a very extensive statistical analysis, using several different factors, but others may argue with the numbers they used," he explains. One way of strengthening the research's claims will be to use the next stage of data from the MOA experiments. "There are three more years of data for 2008–2010 that they can work through in the same way. They should find more of these events and this will provide an even stronger statistical basis for their claims," he says.

The planets are described in Nature 473 349.