The Earth twin that is probably too light to be made of rock

To be built from rock like Earth, Kepler-22b would have to weigh about 15.6 Earths. Fifteen years after the headlines, the best measurements say it probably weighs less than 9.1, and that “probably” turns out to be the whole story.

In December 2011 NASA announced the first planet confirmed in the habitable zone of a star like the Sun. Kepler-22b circles a slightly smaller, slightly cooler star every 290 days, about 640 light-years away in Cygnus. Within hours it had a nickname, Earth’s twin, and a temperature: if its greenhouse effect matched ours, the press reported, the surface would average a pleasant 72°F.

“A major milestone on the road to finding Earth’s twin.”

That was Douglas Hudgins of NASA Headquarters on announcement day, and as a milestone it was fair. What happened next is less well known. Two teams, a decade apart, measured this planet as carefully as anyone could, and their numbers tell a quieter and more interesting story than the nickname.

A planet measured by its star

Kepler never saw Kepler-22b. It saw the star dim by 492 parts per million, three times, at 290-day intervals. That dip gives the ratio of the planet’s width to the star’s, and nothing more. To turn a ratio into a size you need the star’s size, and in 2011 that was 0.979 Suns. Hence the famous 2.4 Earth radii.

In 2023 a team led by Aldo Bonomo re-derived the star using a parallax distance from the Gaia satellite. It came out at 0.869 Suns, and the planet shrank with it to 2.10 Earth radii, about 30% less volume, without a single new transit. Nobody had made a mistake. The ruler had simply got better.

Where 72 degrees came from

The 2023 numbers put Kepler-22b’s sunlight within a percent of Earth’s. That part is genuinely Earth-like. The temperature is another matter. The discovery paper reported 262 kelvin, the figure you get if the planet reflects about 29% of its light, as Earth does. Add Earth’s 33 degrees of greenhouse warming and you reach 295 K, which is 71°F. Two of the three ingredients in the headline number were borrowed from our own planet. Strip the borrowed atmosphere away and the same calculation lands just below 0°F.

The scale

Size says little about what a planet is made of. Mass does, and mass comes from the star’s wobble: as the planet orbits, it tugs the star toward and away from us by a tiny amount. Nobody has detected Kepler-22b’s wobble. Every mass figure for it is a ceiling.

Planets built like Earth, with an iron core and rock mantle, get denser as they grow because their own weight squeezes them. A simple version of published interior models says a rocky world 2.10 Earths wide would weigh about 15.6 Earths. The 2011 ceiling was 124 Earth masses, five times higher, so it could not test the idea at all. The 2023 ceiling is 9.1. A rocky Kepler-22b would need 1.7 times more mass than the measurements allow.

What “at most” means

Here is the detail that never made the headlines. The 2023 paper defines its upper limits as the 68% point of a probability curve, a 1-sigma limit. Roughly one time in three, the true mass is above 9.1. The 2011 ceiling of 124 was a 3-sigma limit, a 99.7% statement. Dividing one by the other and calling it a 13.6-fold improvement compares two different kinds of claim. Put on the same footing, the measurement got about 4.5 times sharper, still a large gain but a different one.

If the possible masses follow a simple half-bell curve (a teaching model, not something either paper claims), a rocky Kepler-22b sits about 1.7 widths out, where roughly 9% of the probability lives. That is the honest summary. The evidence leans against rock without ruling it out.

How long until we know

Turn the question around. To rule out a rocky Kepler-22b with the 99.7% confidence physicists like, the curve has to narrow until its 3-sigma point drops below 15.6, which means a width of about 5.2 Earth masses instead of 9.1. If precision improves with the square root of the number of measurements, that takes roughly 215 measurements instead of 70. At the pace of 2012 to 2023, the extra 145 would take about 22 years. Better spectrographs or more telescope time would shorten that. A spectrum of the planet’s atmosphere, taken as it crosses its star, could settle the ocean-world-or-mini-Neptune question another way.

None of this makes Kepler-22b less remarkable. It still gets almost exactly Earth’s share of sunlight, and it still taught a generation of astronomers how to find such worlds. It just isn’t a twin we can stand on, as far as anyone can currently tell, and the most accurate headline for it includes the word “probably.”

Sources

Source of every measured figure: W. J. Borucki et al., “Kepler-22b: a 2.4 Earth-radius planet in the habitable zone of a Sun-like star,” The Astrophysical Journal 745, 120 (2012), for the transit depth, the 2011 star, the 124 Earth-mass 3-sigma limit and the 262 K temperature. A. S. Bonomo et al., “Cold Jupiters and improved masses in 38 Kepler and K2 small planet systems from 3661 HARPS-N radial velocities,” Astronomy & Astrophysics 677, A33 (2023), for the 2023 star, the 2.10 Earth-radius size, the 9.1 Earth-mass limit (defined in the paper as the 68.27% quantile), the 279 K zero-reflectivity temperature, the sunlight figure and the 70 measurements over 3,913 days. Distance from the Gaia DR3 parallax. The 72°F figure and the Hudgins quotation are from NASA’s December 2011 announcement and its press coverage.

Modelled, not measured: the rock line (mass = radius^3.7, simplified from L. Zeng et al. 2016), the half-bell chance curve and the 9% figure, and the observing-time estimate. They are teaching simplifications, not results from either paper.