Separating fact from speculation about Kepler-20's Earth-sized planets

There's an exciting story circulating today: A large team of researchers has announced in a Nature article the discovery of not one, but two, Earth-sized planets orbiting a star named Kepler-20. Previously, the smallest-reported exoplanet was 1.42 times the diameter of Earth (and therefore 2.9 times its volume). Today's announcement is about two planets having diameters of approximately 103% and 95% of Earth's, give or take 10% -- practically the same size as Earth and Venus. But they orbit so close to their stars that it's probably hot enough to melt glass on the surface of the smaller one, so these are definitely not Earth's twins.

I was happy today to be invited to participate in a Google+ Hangout with Fraser Cain and Nancy Atkinson of Universe Today, as well as Phil Plait (the Bad Astronomer) and MSNBC space journalist Alan Boyle. I had technical difficulties but finally managed to join about 40 minutes into the show. It was fun, and I hope to do it again!

As usual, I'm far from the first to comment on this story, but I'll try to compensate by being thorough and actually reading the paper rather than just the press release. In all scientific papers you have data (observations) and lots of detailed interpretations and often a concluding paragraph of speculation. With these exoplanet detections, there is very good data on their diameters and their orbital parameters. But when it comes to actually describing what these planets are like, there is an awful long chain of interpretations from interpretations from interpretations. I thought it'd be useful for you readers if I separated the observational facts from the quite-likely-to-be-true inferences from the downstream speculations.

First: the facts.

• The data are from 670 days of observations on the star Kepler-20, also known as KOI 070, KIC 6850504, and 2MASS J19104752 + 4220194. Kepler-20 is close to the same size as our Sun, but slightly less massive and slightly less luminous. It's roughly 300 parsecs (nearly 1000 light-years) away.
• Five "periodic transit-like signals" have been detected in the Kepler data.
• Three of these have been reported previously as resulting from three planets (Kepler-20b, Kepler-20c, and Kepler-20d) with diameters 191%, 307%, and 275% of Earth's, and orbital periods of 3.7 days, 10.9 days, and 77.6 days, respectively.
• The two new "transit-like signals," named Kepler-20e and Kepler-20f, recur with periods of 6.1 and 19.6 days and result in "flux decrements" (dimming of the star) of 82 and 101 parts per million, respectively.
• These flux decrements are very small. They could result from additional planets with diameters of 87% (plus or minus about 8%) and 103% (plus or minus about 10%) of Earth's, but in theory they could also result from a background eclipsing binary that falls within the same aperture as Kepler-20.

These facts are nicely summarized in a set of diagrams provided by the Kepler mission (many thanks to the Bad Astronomer for pointing these out to me):

• Kepler-20b
• Kepler-20c
• Kepler-20d
• Kepler-20e
• Kepler-20f

OK, now let's move on to the inferences we can be somewhat to reasonably confident about.

• Nearly all of the text of the paper concerns a sophisticated analysis performed to test the hypothesis that these two signals are "false positives," resulting from background eclipsing binaries or other things that are not planets. The results of the analysis are: "the hypothesis of an Earth-size planet for Kepler-20e is 3400 times more likely than that of a false positive, and 1370 times for Kepler-20f. Both of these odds ratios are sufficiently large to validate these objects with very high confidence as Earth-size exoplanets."
• Assuming that these are planets, measurements of the star's radial velocity from the Keck telescope provide an upper limit to their masses of about 3 and about 14 times that of Earth. These numbers are large because they basically result from the non-detection of a wobble in the position of the star in Keck data. They could definitely be a lot smaller, but they can't be any larger.
• From estimates of their masses (see below), they should produce motions of the star that could possibly be detectable from Earth "in the next few years."

So we can be pretty certain that these two signals in the Kepler data do result from the presence of two planets with diameters very close to Earth. Here's an artist's concept of the sizes of these two putative planets compared to Earth and Venus. Remember that the diameters reported here are uncertain by 8-10%.

Now it's time for the speculation, which is found only the very last paragraph of the paper and in a table summarizing the predictions of the planets' physical properties:

• Because they are close to their star, Kepler-20e and Kepler-20f are probably very hot, about 1040±22 and 705±16 Kelvins, respectively.
• Assuming those temperatures are roughly correct, because of their relatively small diameter and their proximity to their star, it is very unlikely that either Kepler-20e or Kepler-20f has a gassy envelope of hydrogen and helium; that should have been blown away by the stellar wind.
• If they don't have much gas, we can assume their composition to be a mix of silicate rock and iron-rich core. That assumption yields estimates of their masses of 39% to 167% of Earth's for Kepler-20e and 66% to 304% of Earth's for Kepler-20f.
• Both Kepler-20e and Kepler-20f could have initially accreted water with its rock and iron, but only if they initially formed beyond the "snow line" in the planetary nebula and then migrated inward to their current positions. It's very likely that they did migrate inward, given their currently very close position to their star, but it's less certain where they formed in the first place, so therefore it's impossible to know whether they started out with a significant amount of water.
• if Kepler-20f did initially form far enough from its star to have accreted water, it could retain its outer envelope of water for several billion years in its current orbit. Kepler-20e, which is both smaller and closer to the star, could not; it would lose all water to "extreme-ultraviolet-driven escape" within a few hundred million years of its arrival in its current orbit.

I love the speculation part of these papers, because they serve as nutritious food for the imagination. Exoplanet science is especially delicious; there are so very many kinds of alien worlds out there, and it's fun to imagine traveling to them or to imagine what sort of life might exist on any of these crazy planets or their moons. But the speculative parts of any scientific paper are hazardous because they're usually the most fun to write about, yet are the most likely sections of a paper to turn out to be wrong. Keep that in mind when you read stories about these Earth-sized planets!