Mirror Earth (22 page)

“There was a little section in the paper,” he recalled, “saying if there was another planet that had these kinds of properties and if you observed transit times, they would vary by a few seconds.” The gravitational tug from the invisible planet wouldn't just keep the visible planet's orbit elongated; it would also make the transits vary from perfect clockwork, beginning a few seconds earlier or later than you'd normally expect. “I thought, ‘Hey, that's kind of neat,'” said Holman. “There's this little dynamical thing.” It surprised him that the variation was as long as a few seconds. “My intuition said it should be much smaller,” he said, “so I set up my own numerical integration to check—and they were right. So I got to thinking about, okay, how big would that transit-timing variation be for a range of different types of planets with different orbits and different masses.”

This was just the kind of problem that intrigued him. Unlike Dave Charbonneau, his colleague down the hall, Holman didn't become an exoplaneteer to discover life in the universe. “That's not really what motivates me,” he said. Instead, he explained, “I'm always motivated by precision. What's always excited me about planetary dynamics is that you can make very careful, detailed predictions and detailed measurements and you can write down the equations of motion and I like
that.” In the 1980s, the field of chaos theory was just emerging, and along with a handful of other planetary scientists at the time, Holman had tried to figure out whether the solar system is stable over the very long term. (The answer is no, but planets are stable enough for long enough that we don't have anything to worry about—unless an unstable body like a comet or asteroid changes direction and heads right for us. Which has happened before, and will almost certainly happen again.)

“What was really exciting about extrasolar planets,” he continued, “was that suddenly there were a lot more systems to study.” Thanks to Kepler, moreover, there were going to be lots of systems where the timing of transits could be clocked right down to the second. If there were any variations in that timing, it would allow scientists to deduce the mass of the planets—an entirely different way of doing so than through radial-velocity wobbles, and one that didn't depend on the star being especially bright.

Fortunately for Holman, the Kepler team had created a category of semi-insiders called Participating Scientists. They weren't formal members, but they were allowed to propose supplementary projects that would make use of the Kepler data. “It's really a grant program,” said Holman, “where not only do you get the money to do the work, but you get the opportunity to be part of the Kepler team. I actually don't know whether that's done commonly or not.” Sara Seager is a Kepler Participating Scientist (she's looking for reflected light from Kepler planets to try to characterize their atmospheres). So is Dave Charbonneau (using the infrared-sensitive Spitzer
Space Telescope to weed out false positives among Kepler Objects of Interest), Eric Ford (characterizing the eccentricities of Kepler planets), and several others.

So Holman wrote a proposal to look for transit-timing variations in the Kepler data, and it was accepted, and he and a group of colleagues began looking through the Kepler Objects of Interest as they came out of the pipeline. “I'm not looking at 150,000 or 160,000 things,” he told me. “I'm looking at the ones where they have already said, ‘We think that there are planets here and we've gone through the exercise of ruling out all the obvious other things they could be.'” The “they” in this case was mostly Jason Rowe, a postdoc on the Kepler team who was leading the vetting process. “He's the one that finds the objects of interest. He actually did the preliminary time measurements so we could see that there's more than one planet there and one looks like it's speeding up, one looks like it's slowing down. Once the transit-timing variation group saw that, we went, ‘Okay, that's the one. We're going to really focus on that.' ”

The system they focused on was originally known as KOI-377; by the time the paper was published in
Science
, it had been rechristened Kepler-9, signifying that it was no longer just a star with planet candidates, but with actual planets. (A word about the Kepler numbering system: The first five stars where Kepler discovered planets, announced months earlier at the AAS meeting in Washington, were called Kepler-4, -5, -6, -7, and -8. There was no Kepler-1, -2, or -3, however, since Kepler numbers go only to planets actually discovered by the Kepler Mission. The first three stars where Kepler spotted transiting
planets had already been discovered by ground-based telescopes. Borucki and his team had decided to ease into the project with three detections that should be absurdly easy. If Kepler couldn't see planets everyone already knew were there, it would have been a very bad sign.)

In the case of Kepler-9, there were two transiting planets, one with a nineteen-day orbit and the other at thirty-nine. But the timing of those transits wasn't like clockwork: They varied by four minutes and thirty-nine minutes, respectively. Holman was the lead author on the
Science
paper that announced the new result, but he had more than three dozen coauthors. They were all listed by name, but they simply have been identified as the Exoplaneteer All-Stars, since they included Bill Borucki, Natalie Batalha, Dave Charbonneau, Eric Ford, Geoff Marcy, Dave Latham, Debra Fischer, and others.

This discovery was a very big deal. Not only was it conceptually elegant, but it also allowed scientists to calculate, using the laws of gravity, exactly how massive each of the planets in the system was. Until now, the only way to get a clue about a transiting planet's mass and density was to use the radial-velocity-wobble technique to see how hard it tugged on its star. For most of the stars in the Kepler catalog, this wasn't possible; the stars were too faint. Transit-timing variations were a second way into the problem, and the brightness of the star was irrelevant. You did need more than one transiting planet, obviously.

But the Kepler insiders knew something they hadn't yet revealed to the world: There were far more systems with multiple transiting planets than anyone had suspected. “It was a big surprise,” Dimitar Sasselov told me confidentially a month or
so after Holman's paper came out. Sasselov, who got his Ph.D. in Communist Bulgaria before immigrating first to Toronto and then on that well-worn path from Toronto to Cambridge, Massachusetts, was a full-fledged Kepler co-investigator, along with Geoff Marcy and Dave Latham and—until she was promoted to deputy principal investigator—Natalie Batalha. “We didn't expect it,” he said. “We kind of had the feeling, ‘Who ordered this?' You know? We expected maybe 5 systems with multiple transiting planets. We've found 170.” He noticed a look of astonishment on my face. “Yeah, 170 multiple-transiting systems, many of them with double transits, but there are plenty of triple transits. One has six. One has five, and there are a few with four.”

This was really more than a nonphysicist's mind can grasp all at once, but the second author on Matt Holman's
Science
paper, a Santa Cruz postdoc named Daniel Fabrycky, came up with a rather brilliant way to illustrate it. It happened by accident, really. When the six-planet system (it was dubbed Kepler-11) was finally announced early in 2011, Fabrycky, who would co-author that system's discovery paper as well, wanted to visualize what was just a bunch of numbers only a scientist could love.

“Someone had figured out,” he explained, “that three of the planets would sometimes cross in front of the star at the same time, and I wanted to figure out where the other three were when they did.” So he wrote a computer routine that would show where all the planets were at a given time. To advance the simulation one notch forward in time you'd press on the computer's F key. Knowing where all the planets were
allowed NASA to create a vivid and accurate artist's rendering of what the system might look like if we could travel there and see it up close. The editors of
Nature
would end up putting the illustration based on Fabrycky's simulation on the magazine's cover.

But Fabrycky's son and daughter, ages six and four at the time, respectively, figured out something else. If you kept hitting the F key over and over, you'd turn the simulation into a crude movie of the planets going around and around and around. It was pretty entertaining. Fabrycky, who has since joined the faculty at the University of Chicago, also realized that his program would work equally well for all 170 or so of the multiple-planet systems as it did for Kepler-11. So he put together an animation showing all of these systems, as though seen from above, on a single screen, and set them in motion. Fabrycky titled his creation the Kepler Orrery, after the gorgeous mechanical solar systems built in the 1700s to illustrate the motions of the planets.

Then he added one perfect final touch. Since the animation was virtually buzzing with a swarm of planets, he created a sound track with Rimsky-Korsakov's “Flight of the Bumblebee,” played on a xylophone. When he (or anyone else) flashes it on the screen during a talk, there's a moment of stunned silence while the audience takes it in. Then the room invariably erupts into delighted laughter.

When Geoff Marcy and Michel Mayor began finding planets in the mid-1990s, their discoveries triggered a wave of excitement in the astronomical community that was unlike anything the field had seen, probably ever. It was also immediately clear to everyone that the ultimate goal would be to detect life beyond Earth. But it was also clear that existing telescopes weren't nearly powerful enough to do that. As Bill Borucki had discovered back when he first started thinking about Kepler, NASA had been playing with ideas for finding planets, and even finding life, for decades. The agency had produced a number of reports and white papers on the topic, but hadn't done much more than that.

All that planning, however, wasn't entirely in vain. At the same meeting where Geoff Marcy announced his first two planets back in 1996, Daniel Goldin, then the NASA administrator, was able, thanks to those years of study, to step up to a microphone the day after Marcy's talk and lay out a fully formed, step-by-step strategy for identifying and then studying a Mirror Earth. The key technology would be something
called interferometry, a technique for combining the light from two widely spaced telescopes to simulate a single, gigantic scope with extremely high resolution—the ability to take super-sharp images.

The first step, called the Space Interferometry Mission, or SIM, would use that sharp resolution to do astrometry—to measure the side-to-side wobbles a planet imposes on a star rather than the forward-and-back wobbles used in radial-velocity searches. Astrometry was so hard to do that both Bill Borucki and Geoff Marcy rejected it when conceiving their own projects. Nevertheless, said Goldin, SIM would do astrometry with such precision that it would be able to detect the wobbles induced by Mirror Earths as they orbited around Sun-like stars. (Goldin was also the NASA chief who declared when he took office that henceforth, the agency would do everything “better, faster, and cheaper.” Scientists and engineers generally agreed that they could do any two of the three at one time, but not all of them.)

The next grand step would be the Next Generation Space Telescope, a successor to the Hubble. The NGST, which has since been renamed the James Webb Space Telescope, should be in orbit by 2007, said Goldin (the current best estimate is 2018). While the NGST wouldn't be designed just for planet-hunting, it might be able to take images of giant planets, as long as they were far enough away from their stars that they would be lost in the glare. Then, by 2020 or thereabouts, he said, the crown jewel of NASA's planet-searching program should be ready for launch. Called the Terrestrial Planet Finder, or TPF, it would be an interferometer like SIM, but with four
huge space telescopes instead of two small ones. These four telescopes would have to be so widely separated that they couldn't sit on a single structure. They would have to fly in formation, out in the general neighborhood of Jupiter, maintaining their separation to within a fraction of an inch as they sailed through interplanetary space.

If it all worked out, the Terrestrial Planet Finder would be able to do something remarkable. By adjusting the spacing of the telescopes just slightly, interferometry would cause the light to cancel out in parts of the image. In principle, you could blank out the star, making it much easier to see an Earth-size planet, and even to probe its atmosphere for the chemical signature of life. This would be incredibly difficult, technically, but hadn't NASA landed men on the Moon, and set cameras down on Mars, and detected the faint afterglow of the Big Bang with the COBE satellite?

Despite the fanfare and the promise, however, this grand scheme didn't play out quite as Goldin had portrayed it. “In 1999,” Marcy recalled, speaking more than a decade later, “the Space Interferometry Mission was approved, the budget was roughly $50, $60, $70 million per year. We met three, four times a year, the science team did—an enormous effort.” Eric Ford's graduate thesis at Princeton was in support of the SIM program. NASA ultimately spent $600 million on the project without even starting to build the hardware, and then canceled it for budgetary reasons. The more ambitious Terrestrial Planet Finder hasn't been canceled, but it's been put on hold. If TPF launches by 2030, astronomers will be very surprised.

The problem with TPF isn't just that it's expensive and
technically difficult, but also that astronomers can't even agree on what the instrument should look like. The original concept involved those four space telescopes flying in formation, but in the early 2000s, designers came up with two simpler (though less powerful) alternatives. The first was to build a large, single space telescope and fit it with a coronagraph, a device that blots out the light of the central star to let a planet shine through. The second was like the first, but would involve the launch of two separate pieces of hardware: the telescope itself and a device called an occulter. The occulter would fly thousands of miles away from the telescope and position itself in just the right way to blot out a star. Each version had its proponents who argued that theirs was the right one and everyone else's was wrong. In the end, NASA threw up its hands, put TPF on the back burner, and cut the project's budget drastically.