Mirror Earth (5 page)

Still, by comparing the lines they saw in the Sun with the ones they were able to create in labs on Earth, scientists were able to figure out what the Sun is made of. (It's made mostly of the same elements Earth is made of, although in very different proportions.) So that was useful. But spectral lines turned out to have another crucial property. They're exquisitely sensitive motion detectors. You can think of light as waves of electromagnetic energy. In the visible spectrum, the more tightly packed the waves are—which is to say, the shorter the distance between them, or the shorter the wavelength—the closer a color is to the violet end of the rainbow. If the waves are more loosely packed, they're closer to the red end. The same goes when you go beyond the visible part of the spectrum: Ultraviolet light is more tightly packed than violet; gamma rays are more tightly packed than ultraviolet, and X-rays even more. The piano analogy works here too: a note sounds higher pitched because sound waves—which are simply
pressure waves in the air—are more tightly packed together. Looser packing makes for a lower note.

All of this is pretty straightforward when the thing you're looking at or listening to is sitting still. But now imagine something that's making a lot of noise while it's coming toward you—a train speeding toward you with its horn blaring, for example, as you stand close to the track. As the train comes toward you, its motion squeezes the sound waves together, so the pitch sounds higher to you than it really is. When the train passes and starts moving away the squeezing stops abruptly, and suddenly the sound waves are being stretched instead. The pitch drops instantly (in old movies, train whistles are doing this all the time; these days, you mostly hear the effect with police and ambulance sirens).

This change in pitch works for things that are moving toward or away from you; if they're moving from side to side, there's no squeezing and no stretching. So you need to be right next to the railroad track to hear the switch from high pitch to low.

Exactly the same thing happens with light. If a shining object is moving toward you, its light looks slightly higher-pitched than it really is. That is to say, it looks bluer. If it's moving away, it looks redder. If a star happens to be moving toward you, it isn't just the light, but also the dark lines that interrupt it, that shift in the blue direction. This trick of light is what let Edwin Hubble discover the expanding universe back in the 1920s. When he broke apart the light from galaxies beyond the Milky Way, he thought he would see shifts in
the locations of dark spectral absorption lines to show that some of the galaxies were moving away from us and some moving toward us. Instead, the lines showed that they were all speeding away. This meant either that the Milky Way was somehow very repulsive, or, more reasonably, that the whole universe was expanding, with every galaxy speeding away from every other galaxy.

During graduate school and on into his postdoctoral fellowship, Geoff Marcy had gotten very good at finding and measuring spectral lines. So he went with his strength and decided to look for planets this way. With astrometry, you have to be looking down on an alien solar system from above to see the star moving from side to side (“down” and “above” don't really have any meaning in space—you could just as easily say “up from below,” and you'd be equally inaccurate, but it's such an instinctive way to describe it that astronomers talk this way anyway, and nonscientists understand instantly what they're talking about).

Marcy wanted to look for that same motion, but from an edge-on perspective. He wanted to catch planets tugging their stars toward Earth (just the tiniest bit), then away. Since the red-shifting and blue-shifting of spectral lines betrays that sort of motion, that's what he proposed to look for.

The only problem with this idea, Marcy soon learned, was that it was nearly impossible. When cosmologists use shifting spectral lines to measure the speed of galaxies racing apart as the universe expands, they're looking at objects moving at many thousands of miles per hour. They take a spectrum from
a galaxy's collective starlight, lay it next to a reference spectrum from a motionless object—the Sun, for example, or a laboratory reference lamp that generates an artificial spectrum—and see how far a given line has shifted. There's some imprecision in the process, but if they're off by a couple of thousand mph or so, that's plenty accurate enough. They don't have any need to improve their precision.

But the back-and-forth motion Jupiter causes in the Sun is a piddling 28 mph or so. The greatest expert in measuring cosmological redshifts would fail utterly to detect it. So Marcy tried tightening up the procedure in every way he could think of to make it more precise. He succeeded up to a point: He managed to get his accuracy down to about 450 mph. But since he was trying to find distant Jupiters, this wasn't nearly good enough. No matter how careful he was, the act of moving the telescope from star to reference lamp changed his measurement system enough to make the measurement unreliable. Imagine you wanted to measure the length of two different objects—two bricks, say—with high precision. You'd be smart not to use two different rulers, since one of them might be just a little bit off. But if you wanted to be
really
precise, it's a problem even if you use just one ruler—moving the ruler from one brick to the other could change things. The second brick could be in a slightly warmer place, so the ruler might expand just an infinitesimal amount. Or you might hold the ruler in a slightly different way, so it would sag under gravity differently, distorting its shape. These are absurdly small changes, but if you really needed absolute accuracy, they could make a
difference. The best way to make absolutely sure you're measuring things exactly the same way is to measure them at exactly the same time.

Marcy decided to do just that. He'd measure a star's spectrum and a reference spectrum all at once, so he didn't have to move anything. He might have figured out a way to do it, but it turned out that he didn't have to. A Canadian postdoc named Bruce Campbell, at the University of British Columbia, had come up with a solution a half decade or so earlier. Working with a colleague named Gordon Walker, Campbell had realized that you could take a gas whose spectral absorption lines were thoroughly understood, put it in a glass container, and let starlight pass through the gas on its way to the spectrometer. The reference spectrum and the real spectrum would be measured at exactly the same time with exactly the same instrument.

It worked. Campbell and Walker were able to measure the wobbling of stars to an accuracy of plus or minus 30 mph or so. That still wasn't precise enough to find an alien Jupiter, however, since the measurement error was as big as the signal you'd be looking for. Beyond that, Campbell and Walker had settled on hydrogen fluoride gas for their reference—well understood, but also corrosive, explosive, and horribly toxic. Marcy needed a better gas, and he also needed a collaborator for this project, which was growing increasingly complicated.

By now, Marcy was on the faculty at San Francisco State University. After asking around a bit, he learned of a recent San Francisco State graduate who had joint degrees in physics and chemistry, and who also had a strong interest in astronomy.
Paul Butler was still at the university, working on a master's in physics and looking for a research topic. When Marcy approached him, Butler was intrigued. Like Marcy, Butler was drawn to research with long odds but a potentially huge payoff. And he loved the challenge of trying to make measurements more precise than anyone had ever been able to pull off.

The only downside was that Paul Butler had a somewhat rough personality. He divided the world into good guys and bad guys, and the bad guys included some of the world's most eminent scientists. These would eventually include, once he signed on with Geoff Marcy, many of their professional colleagues. He thought nothing of describing senior astronomers at places like Caltech and Cornell and especially Harvard as evil, or cowardly, or even mentally ill. He would say these things openly, and later on, when Marcy and Butler finally began making discoveries and getting some public recognition, he would sometimes even say them to reporters.

Nevertheless, Butler was very good at his job. He spent more than a year hanging out with chemists, trying out one element or compound after another, looking for the ideal reference gas to use for finding planets. Ultimately, he settled on iodine. It was not only safe, but it also had an enormous number of spectral absorption lines that spanned the visible spectrum all the way from red to violet. That would give each measurement plenty of cross-checks. The lines created by hydrogen fluoride, by contrast, were not only fewer in number, but they also bunched up in a small part of the visible-light spectrum.

So Marcy and Butler built what they called an “iodine cell” to attach to the Hamilton Spectrograph at Lick Observatory near San Jose, and began using a small telescope to take data on relatively bright, nearby Sun-like stars, looking for wobbles. They didn't have the software yet to analyze their observations; the spectrum of iodine was so horribly messy that they couldn't disentangle its spectrum on their images from the spectra of the stars. But Butler was also a talented software writer, so while they continued to take unreadable measurements, he worked on code that might someday make sense of them.

In the end, it took him six years. “It's my Rembrandt,” Butler told me in 1996. “It's as close to great art as I'll ever get.” Even then, however, Marcy and Butler could get to a precision of only twelve meters per second—still not good enough to find an alien Jupiter. The Hamilton Spectrograph, built by Steven Vogt, Geoff Marcy's thesis adviser in grad school at Santa Cruz, was now the limiting factor. Vogt had to upgrade the device, and then Butler had to rewrite his software to account for the upgrade. Finally, Geoff Marcy, Paul Butler, and Steven Vogt, their new collaborator, were able to measure the wobbles of stars to within an astonishing and unprecedented three meters per second. They could find a planet like Jupiter.

After Geoff Marcy and Paul Butler had all the kinks worked out in their hardware and software, the only thing standing in their way was time. It takes Jupiter eleven years to orbit the Sun just once. If you were an alien astronomer looking toward our solar system using an iodine cell and the Hamilton Spectrograph, it would take you eleven years to watch the Sun move toward you, then away, then back in a single orbital cycle. And if you were a really careful alien astronomer, who didn't want to risk the embarrassment of making a discovery that turned out to be wrong, you'd want to see not just one, but at least two or three cycles to convince yourself you really were seeing a planet and not, say, some weird pulsation of the star itself.

Geoff Marcy knew very well that astronomers had fooled themselves about planets before. The best-known example was the “discovery” of planets around a nearby star known as Barnard's Star, by Swarthmore College astronomer Peter van de Kamp in the 1960s. What van de Kamp thought was a
side-to-side motion in the star, caused by a planet, was actually a change in his telescope—a minuscule repositioning of a lens when the telescope was refurbished. The slight change in focal length made Barnard's Star appear to move, and van de Kamp interpreted the motion as the tug of a planet. When the mistake was discovered in the 1970s, van de Kamp compounded the problem by being slow to acknowledge it (in fact, it's not clear that he ever did).

To be sure they could claim a planet discovery with confidence, Marcy, Butler, and Vogt would have to wait a long time to confirm that any wobble they spotted matched the signature of a planet. They'd also have to convince other astronomers that their complicated apparatus and their complicated software—which may have been Butler's Rembrandt, but which was so complicated that it would take years to analyze observations of a single star—really was capable of doing what they claimed. Some of their colleagues at other universities literally laughed at them when they heard a description of the research. Many of the laughers ended up on Butler's growing enemies list.

Near the top of that list was an astronomer named David Latham. Latham has been at Harvard since the late 1960s; he got his Ph.D. there in 1970. When I arrived as a freshman in the fall of 1971, he was the deputy instructor of a hugely popular undergraduate course called The Astronomical Perspective, taught by the astronomer and historian of science Owen Gingerich. I remember looking down from the back row of a steeply angled lecture hall and seeing Latham, a young, skinny, nondescript guy, standing off to the side as Gingerich lectured.
Forty years later, I sat in his office at the Harvard-Smithsonian Center for Astrophysics—the CfA, if you want to sound like an insider—listening to an older guy, less skinny, with grayer hair, wearing a jacket and tie (he hadn't worn them as a teaching assistant in 1971, and most astronomers don't wear them now unless they're getting a major award). He was still teaching the course I'd taken so many years earlier, but it now had a different name, and he was now the lead instructor, since Gingerich had retired. Latham was still pretty nondescript, but now it was in an avuncular way—low key, companionable, easy to get along with. You wouldn't immediately guess, nor would you have guessed in 1971, that he was and is a competitive motocross racer (a photo of Latham kicking up dust on his dirt bike appears on his Harvard homepage) and a hockey player. It's less surprising to learn that he's a wine connoisseur.

About midway between his appearance at the front of that lecture hall in 1971 and our conversation in 2011, Dave Latham had become a planet hunter too. Originally, he was largely interested in cosmology, the study of the origin and evolution of the universe. The Big Bang had become the dominant theory of the universe in the mid-1960s, after lingering around the edges of science for many decades, but there were all sorts of profound unanswered questions remaining. When did the universe begin? What happened during the first moments after the Big Bang? What is the universe made of? What will its ultimate fate be?