A brief letter in Nature was John Debes's inspiration. The 1999 piece, by David J. Stevenson (Caltech), proposed that planets with liquid water oceans — and even life — could exist in the cold, dark depths of interstellar space far from any star. Based on the knowledge that some fraction of planets must get gravitationally ejected from their systems during the systems' formation, the paper theorized that some of these ejected planets could, with enough internal heat, keep their atmospheres and stay warm enough to support liquid water below a thick frozen crust.

What might happen if such an outcast had a big moon? To find out, Debes (at the Carnegie Institution of Washington) ran 2,700 computer simulations based on an Earth-mass planet and a lunar-mass companion.

"The ejection process can be very intense," says Debes. "It wasn't clear to us if any bound systems would actually survive." But in 123 of the cases, or between 4 and 5 percent of the time, the "Earth-Moon" system did survive ejection from its solar system intact.

"Anytime something happens in astronomy a few percent of the time, it is interesting to us because on the grand scale of things, it means it's happening a lot and people should probably know about it," says Debes.

And these pairs have a better chance to harbor life, since the dissipation of tidal energy between the moon and the spinning planet causes the interior of the planet to warm. Debes' models predict that this heating would match what happened in Earth's case more than 4 billion years ago, when the young Moon was much closer and Earth was rotating faster.

In the October 20th Astrophysical Journal Letters, Debes and Steinn Sigurdsson write that the heating would likely be localized in hot spots of volcanism or other geothermal processes. Biologists are finding many examples on Earth of life surviving on these energy sources, such as at mid-ocean ridges. Could such "extremophiles" be the most common form of life in the universe?

The team's best-case scenario found that an ejected Earth-Moon system can sustain its heat for up to 250 million years — long enough for life to arise. But is it long enough for that life to adapt to the eventual decreasing temperatures?

And if dark, free-floating Earths exist, when will astronomers be able to detect one? It's not as impossible as you might think. Debes estimates the next generation of space-based microlensing surveys will have around a 2% chance.