One Universe at a Time

Ghost Planet

One of the biggest challenges in the search for exoplanets is distinguishing real evidence of a planet from random fluctuations that give a false positive. In most cases we can’t observe an exoplanet directly, so we have to look at things such as a dip in a star’s brightness, or a star’s wobble from a planet’s gravitational tug. Just because some data looks like the effect of a planet, we can’t announce its discovery without being sure. This is why we distinguish between candidate planets and confirmed planets. But even when we’re careful there are times when a planet can be confirmed in error, such as the recent case of Alpha Centauri Bb.

The discovery of a planet orbiting Alpha Centauri B was announced about 2 years ago. It was found by observing the wobble motion of the star (what’s known as its radial velocity). The amount of radial motion was small, about the speed at which a person could walk. That’s a small signal to pull out of the noise, but the evidence seemed pretty reasonable, which is why the results were published. But further study placed doubts on the existence of a planet, and now a new paper demonstrates that the planet could be a false positive. The authors looked at simulated data of stellar fluctuations with no planet data, to see if results might look like a planet. What they found was that the rate at which data was taken could produce a false result.

The original observations were taken at La Silla observatory, which is a ground-based telescope. That means data could only be gathered when skies were clear, and the observing time had to be scheduled between other observations. It turns out that the observation sessions also had a periodicity to them, and this affected the analysis of the original data. It’s a subtle effect, and it’s not surprising that the original authors didn’t see it. This new work demonstrates that patterns in your sampling can affect a result as well as random fluctuations in your data. Particularly when the signal is weak.

So it’s very likely that Alpha Centauri Bb isn’t a planet after all. Of course this raises an interesting question. Technically, the name A Cen Bb should go back into play, so if a different (real) planet is discovered around the star it should be given that name. But since the name is already in the literature that would be confusing. It might be better to name any new planet A Cen Bc. If that’s the case, then Alpha Centauri Bb would become a ghost planet, being the name of a planet that doesn’t exist.

Paper: Vinesh Rajpaul, et al. Ghost in the time series: no planet for Alpha Cen BarXiv:1510.05598 [astro-ph.EP]