NASA's Kepler space telescope was launched in 2009 and stopped taking data last year after a mechanical failure. Yet in its relatively short lifetime, it has offered up a wealth of discovery. In February scientists announced a new harvest that brought Kepler's tally of discovered planets to nearly 1,700. “This is the biggest haul ever,” says Jason Rowe of the nasa Ames Research Center, who co-led the research. The scientists studied more than 1,200 planetary systems and validated 715 planets. All the new worlds are members of multiplanet systems—stars with more than one orbiting satellite.

Researchers used a new method for weeding out false signals. Kepler searched for planets by measuring dips in a star's brightness, which occur when a planet passes in front of it. This technique, called the transiting method, is very accurate, but sometimes a nonplanet can fool the telescope. One of the most common reasons for a “false positive” is an eclipsing binary—a pair of orbiting stars that sometimes cross in front of each other from our perspective.

Stars with a single planet can be hard to distinguish from eclipsing binaries. But multiplanet systems are far less likely to be frauds. “It happens, but it's unlikely that you have two eclipsing binaries in the background of the same star,” says Francois Fressin of the Harvard-Smithsonian Center for Astrophysics, who was not involved in the study. It is also possible, albeit extremely unlikely, to have an eclipsing binary and a star with a planet lying right on top of each other.

Rowe and his colleagues tried to weed out false signals by examining light from the candidate planets. They looked for a particular signature known as a moving centroid: an off-center point of light that could be created only by an eclipsing binary, not by a planet.

What remained among the trove of ample discoveries: a potentially rocky world; an odd binary star system where each star has planets of its own; and cramped systems where the multiple planets are each gravitationally tugging one another around. “Of course, we have every type of planetary system in our validated set that people can think of except the perfect Earth analogue,” Rowe says. For now that remains Kepler's holy grail.