Bye Bye to a Lovely Planet

The planet Gliese 581g might be a chimera.

This intriguing object has dominated science news for the last few weeks because it was the first world found in deep space that might sport an environment comparable to our own. Gliese 581g could be wrapped in oceans, a thick atmosphere and -- who knows? -- some biology. After discovering nearly 500 planets around other stars, it appeared that astronomers had finally tripped across one that might approximate the Earth.

Well, buck up and stand down. A new analysis by astronomer Michel Mayor and his Swiss team suggests that Gliese 581g is an apparition -- a planet conjured into existence by other researchers' faulty interpretation of noisy data. It now seems you can stop fantasizing about oddball Gliesians 20 light-years from your doorstep.

Disappointing, sure. But there's nothing either novel or disturbing in this. Astronomy is largely an exploratory science, heavily dominated by observation. Astronomers have mapped out the cosmos by using their telescopes first and their imaginations second. And since the really exciting discoveries are perforce made at the hairy frontier of telescope performance, mistakes happen.

The discovery of extrasolar planets -- one of the hardest things to do in astronomy -- has been especially replete with false alarms.

You probably don't remember (unless you're a connoisseur of arcane astronomy), but in the 1960s, Peter van de Kamp, using a telescope at the Sproul Observatory in Philadelphia, claimed to have uncovered two giant planets around Barnard's Star, a mere six light-years' away. That was a major, major first. Van de Kamp examined stellar photos made over the course of months and years, and detected slight changes in this star's position. He saw side-to-side wobbles, which he understandably ascribed to orbiting planets.

Alas, the wobbling was actually due to tiny changes in the optics of his telescope, caused by routine maintenance of the lenses. An honest, if disheartening, error. Van de Kamp's planets were vaporware.

Then there was the case of British researchers Andrew Lyne and Matthew Bailes, who in 1991 reported seeing slight variations in the repetition rate of pulsar 1829-10 which they said were caused by planet-induced wobbles. Again, this would be dramatic, as Lyne and Bailes would be the first to discover planets around someone else's star (admittedly, in the case of a pulsar, a dead star). But the scientists had incorrectly compensated for Earth's own motions, and stoically retracted their claim only months later.

The take-home lesson is clear: If you're hunting for planets, you'd best be prepared to occasionally shoot and miss. As any researcher knows, science is not a ratchet, a mechanism that relentlessly increases our knowledge. It's a deep and wide migration towards the truth, beset by wrong turns and cul-de-sacs. But that's OK, because mistakes are not an indictment of science, but an essential part of the process. Bummer.

Of course it's discouraging that Gliese 581g has seemingly evaporated. But there's also some good news for those eager to find worlds that could be cousins of Earth. Andrew Howard and his team at the University of California at Berkeley have just announced an analysis of a large data set taken with the mammoth Keck Telescope in Hawaii, and concluded that nearly one in four Sun-sized stars probably hosts terrestrial-size planets. If so, there are tens of billions of Earth analogs in the Milky Way. And that's just our galaxy.

So we may have lost one, but gained billions. Not a bad scorecard for the week.