Mirror Earth (16 page)

So that's one way to rule out a background eclipsing binary star. Another is to see if the center of brightness in the image shifts at all. If all the brightness is coming from a single star, its position won't seem to change when the star dims. If it's coming from the combined light of three or more stars, the center
will shift subtly. Yet another is to look in a wavelength of light other than the visible—infrared, say. When a planet passes in front of a star, the star's color doesn't change much. It just gets dimmer. But when one star eclipses another, the overall color does change, since it would be very unusual for two stars in a binary to have precisely the same color. An infrared telescope, which already sees a different mix of colors than a visible-light telescope, will see a dimming similar to what Kepler sees if there's a transiting planet. If there isn't—if it's an eclipsing binary—the infrared telescope and Kepler will see very different types of dimming.

Yet another way to rule out eclipsing binaries is to do it statistically. Based on what astronomers actually know about how many eclipsing binaries there are in the Milky Way, you calculate how many are likely to be in Kepler's field of view. “Then,” explained Borucki, “we say for this tiny area around this star, here's the probability that there could be an eclipsing binary that would imitate it. Generally, we try to keep the uncertainties much less than 1 percent.”

Even when they had ruled out false positives to the best of their ability, however—if a star had dimmed three times on a repeating schedule (ruling out one-time glitches) and had passed all of the false-positive tests the Kepler team threw at it—it could still be something other than a planet. To paraphrase Donald Rumsfeld talking about the unexpected problems that turned up after the invasion of Iraq, you can rule out the known unknowns, but that still leaves you with unknown unknowns. One more test, however, could transform a “planet candidate” into a bona fide, rock-solid planet detection. If you
could detect not just a periodic dimming but also the radial-velocity tug an orbiting object imposes on its star, you could be a lot more confident there was a planet there. It would be the mirror image of what had happened when Tim Brown and Dave Charbonneau on one hand and Greg Henry and Geoff Marcy on the other had found the very first transiting planet, HD 209458 b, and quieted the last of the planet skeptics.

Ideally, the Kepler team would confirm all their planets this way. In practice, it wasn't going to happen. For one thing, it takes a significant amount of telescope time to get a radial-velocity profile of even a single star, and there's only so much telescope time available. For another, most of the stars Kepler is watching are far away—deliberately, so the satellite can look at a lot of stars at once. But that means they're generally dim, so radial-velocity confirmations require time at a very big telescope, with maximum light-gathering power. That's in even shorter supply than time on a smaller telescope, like the Shane at Lick Observatory. “We did a calculation at one point,” said Batalha, “to come up with the number of hours on a telescope like Keck [in Hawaii] you would need to confirm every one of those planets and it was astronomical, literally. It was ridiculous, the number.”

Finally, Kepler is ultimately looking for Earth-size planets in their stars' habitable zones. For a star like the Sun the habitable zone is about where Earth orbits, ninety-three million miles out or so. “Right now,” Batalha told me, “an Earth-size planet in a one-year period around a Sun-like star causes a radial velocity that's too small for us to detect. So that type of signal we will not be able to confirm.” To the untrained ear,
this suggests that Batalha and Borucki and the other charter members of the Kepler team knew from the beginning that the mission could never actually achieve its goal. It could never point to a star and say with certainty, “There's an Earth-size planet orbiting in the habitable zone of a Sun-like star.”

But that was never actually Kepler's goal. The original name for the project, remember, was FRESIP, for FRequency of Earth-Size Inner Planets. Kepler was designed to find the
percentage
of stars with Earths in the habitable zone. For that, you didn't have to prove that any one star had such a planet. “We'll never be able to eliminate all of our false positives,” explained Batalha. “So we have to pick and choose which planet candidates to follow up on. What we really want to do is understand our false-positive rate in a statistical sense. So we have picked a subsample of stars that we will just throw every trick in the bag at—less than one hundred, maybe between fifty and one hundred. Seventy, maybe. And we'll take what we learn from that exercise and apply it to the rest of the catalog in a statistical sense. If we say to the public, here are a hundred candidates, and our false-positive rate is 10 percent, you know ninety of them are right and ten of them are wrong. You just don't know which ten. That is Kepler's objective, to determine if the Earth-size planets are abundant, not to figure out which ones they are necessarily.”

That may sound odd at first, but it actually makes sense. If your ultimate goal is to find life on other worlds, it's obviously best to search for planets around the nearest stars, since they're the easiest to observe. But if you make them too near, you might not have enough stars close by to give you a fair chance
of finding anything. How far away “near” is depends on how common Earth-size planets are. The best way of figuring that out is to do a broad survey of a huge number of stars first, which is just what Kepler is trying to do. “That's always been the plan,” Batalha told me. “Figure out the fraction of stars that harbor likely planets, then design a mission to find the ones closest to us. Do I have to look out fifty parsecs [about 160 light-years] or two hundred [about 640]?” That's what Kepler is meant to do. If it determines, say, that 30 percent of Sun-like stars have Earth-size planets in the habitable zone, you can get away with surveying a couple hundred of the nearest stars to look for life. If the percentage is only 1 percent, you need to cast your net a lot wider.

The people at NASA headquarters understood this rationale, and agreed with it wholeheartedly. They had right from the beginning. The reasons they kept shooting down Bill Borucki's proposals were largely practical ones: He hadn't figured out all of the technology yet, or they weren't convinced he was being realistic about cost. They weren't necessarily right, but they had the final say. Borucki had four chances to give up and move on to something else, and didn't. I suggested to him during one of our conversations that I could imagine that a longtime NASA employee might become philosophical about getting shot down so many times.

“Ah. Okay,” he said. “You can imagine that. Good for you.” He didn't say it in a mean way. He was just amused.

So what does it really feel like, I asked.

“Not like that at all, of course.” Natalie Batalha had told me rejection just rolled off Borucki's back, but that might have
been an oversimplification. “There's certainly discouragement,” he continued. “If we were going to go propose again, that's a lot of work and we had failed a number of times. Do you really want to be a consistent failure or just an occasional failure? See, you think about things like that. One of our team members was Carl Sagan and we would submit and fail, submit and fail, submit and fail, and then he came down with cancer. I asked him, Did he really want to be a member of the team when we proposed again? So, he wrote me a nice letter that said, Yes, he felt the mission was an excellent mission, we were going to accomplish great things and he was so enthusiastic about it. He wanted to participate and he was expecting to feel better in the coming months and he would work harder with us.”

Sagan died a few months later. “I still have that letter,” said Borucki. “So, it's something that we have to go through. We have to overcome our anger at people being so stupid as not to immediately fund the mission. We have to work with them to look at the suggestions from the review panel and ask, How can we incorporate those? Where can we get the money to do the new studies that are required? That means we go to headquarters, we go to Ames, people at different universities have to find funds to come up with a new and better proposal. So, yes, there's discouragement. There's even a little bit being pissed—a little bit. But if you've got an idea that you think is so valuable, so important to mankind's future, then you're willing to accept that. You're willing to say, Okay, I put up with all this crap, this discouragement, and I move ahead.”

Not everyone reacts the same way. There was a Venus
mission that went head-to-head with Kepler the last time Borucki went into the competition, in 2000, and a Jupiter mission as well. “Everybody knew that the Venus mission was good,” Borucki said. “Everybody knew that the Jupiter mission was superb. But when headquarters chose our mission, these people just stopped. They didn't propose on the next round. I think some of them are proposing again now but there were a couple of opportunities that they did not compete for. I don't know whether they changed teams or changed people or what's happened, but I can't blame them.”

Writing a proposal for a mission costs several hundred thousand dollars, he said, and it's not a lot of fun. When you get downselected—in NASA jargon, that means you've made the final round—they send you a bunch of questions. “You have twenty-four hours to answer them, okay? But one of the questions that they sent us was itself a twenty-part question on the communication system. What coding would we use for x-band, a-band, things like that. What's the reception rate? What antennas were we going to use? Twenty parts to one question, twenty-four hours, that's typical. It's an enormous effort to write a successful proposal for these things.”

Was there actual champagne involved when Kepler was finally selected? I asked.

“Yes, there was. We were all delighted. Where's my bottle? There's a glass up on the shelf, but the champagne bottle is here as well if someone hasn't run off with it. In any case, it was Moët.”

Kepler was approved in 2000, but it wouldn't launch until 2009. It was clear to everyone in the exo-planet business that the satellite would revolutionize the science of planet-hunting—eventually. It would if Kepler was actually completed, that is, and launched without blowing up, and reached its intended orbit, and worked properly when it got there. None of those things is ever guaranteed with space probes. Even if they were, it would have been preposterous for hundreds of scientists to sit idly by waiting for Kepler for the better part of a decade when there was so much they could do in the meantime to chip away at the question of how many stars have planets, and what sort of planets they are.

So while Bill Borucki, Natalie Batalha, and the rest of the Kepler team began the long, careful process of putting together a space mission, Geoff Marcy, Paul Butler, Steve Vogt, Debra Fischer, and their collaborators and postdocs and graduate students kept measuring radial velocities. Paul Butler took the job he'd been offered at the Anglo-Australian
Observatory, and then moved on to the Carnegie Institution of Washington, but he'd remained part of the team. (Like all the astronomers at the Carnegie, Butler joined the Department of Terrestrial Magnetism, whose delightfully archaic name became obsolete in 1929, when the department completed its charter task of mapping the Earth's magnetic field.)

Michel Mayor and his group in Geneva kept at it as well, taking radial-velocity measurements, trying to squeeze down their uncertainties to find smaller and smaller planets. Sara Seager kept theorizing about planetary atmospheres and about what you'd be able to tell about an exoplanet—not just the atmosphere, but also the surface, and even the vegetation—if you could ever take a direct image if it, with something like the Terrestrial Planet Finder. For a few years, she did it at the Carnegie; then she moved on to take a faculty job at MIT. Around the world, astronomers worked harder than ever to add new worlds to the exoplanetary tote board. They wanted to be able to say something meaningful about what was out there, and they had no intention of waiting for Kepler if they didn't have to.

And there were a lot more exoplaneteers now than there had been just a few years earlier. (They weren't calling themselves that yet, although the term would eventually become almost universal. Dave Charbonneau is pretty sure he was the first to use it, sometime in the late 2000s.) When Marcy and Mayor announced their first exoplanets, they were operating way outside the mainstream of astronomical research. But by the time Debra Fischer got into the planet business in the late 1990s, she told me, “it felt like there was a tidal wave coming.”
The approval of Kepler might have been the crest of the wave, but the swells that came before it were considerable. It wasn't just a bigger effort from existing groups; funding requests for new searches were also starting to appear in grant-makers' in-boxes.

Many of these new proposals were for transit searches, inspired in part by the discovery of HD 209458 b, by Geoff Marcy and Greg Henry, neck and neck with Tim Brown and Dave Charbonneau. The tidal wave was also influenced by the community's growing awareness of the Kepler project, generated through Bill Borucki's quiet evangelism at conferences and in proposals through the nineties. One of the new searches, for example, was established by a young Harvard astrophysicist named Gaspar Bakos. Bakos had arrived from Hungary to take up a predoctoral fellowship in 2001. He brought along a project he'd begun working on as an undergraduate student in Budapest. It was and still is called the Hungarian-made Automated Telescope Network, or HAT-Net, and it was in essence a more elaborate version of the system Dave Charbonneau and Tim Brown had used to find the very first transiting planet, HD 209458 b.

Bakos wanted to create a network of small telescopes only a little over four inches in diameter—in the end he ended up using off-the-shelf telephoto camera lenses—that would automatically look for transits in a selection of stars. “I remember looking at his proposal as an external reviewer,” Fischer told me, “and I said, Yeah, we should give this guy money, this is brilliant.” They gave the guy money, and Bakos, working with Harvard's Bob Noyes—the same astronomer who inspired
Dave Charbonneau and Sara Seager and a generation of young astronomers to go looking for planets—and Dimitar Sasselov, Seager's thesis adviser, began putting the project together. Ultimately, it would have telescopes in Arizona, Hawaii, Israel, Australia, Namibia, and Chile.