by Lee Billings
“I took this picture from my front door using an off-the-shelf five-megapixel camera,” Laughlin said, pointing to what looked like a lump of pale Legos against a deep blue background. “That’s Venus, zoomed in so that the individual pixels are visible. It’s emblematic of the situation we’re now in with exoplanets, in that we can see there’s some structure there, it’s mysterious, and we want to know more. It’s also emblematic in that most of the things we’re trying to understand about worlds orbiting other stars, we’ve already been through with planets in our own solar system.”
Venus was also symbolic of Laughlin’s early scientific beginnings. His first brush with astronomy had come when, as an eight-year-old boy in the soybean country of downstate Illinois, he scraped together enough money to buy a small, simple telescope. He looked at stars, and the Moon. In one early evening’s twilight he turned his telescope toward Venus, low and sparkling in the sky. He had expected to see the same blurry dot he witnessed with his naked eyes, albeit magnified. Instead the telescope revealed a crescent, sharp and whitish-yellow like a nail clipping. It dawned on him that he was seeing both day and night on Venus, and that the demarcation between the two marked a zone of twilight, just like the one he was now in on Earth. The view from his backyard in Illinois seemed at once larger and smaller than ever before; something about seeing an alien planet’s hidden details revealed before his very eyes made his mind effervesce. The feeling faded, only to momentarily return over the years each time he uncovered something unexpected and beautiful. The more he learned, the more profundity he saw in the purity of numbers and equations, the more majesty he found in the lives of planets, stars, and galaxies. Laughlin didn’t know it at the time, staring at sunlight shining on the distant cloud tops of Venus, but the vista spread before him in his telescope would draw him deep into the frontiers of planet hunting.
“Though closer to the Sun, Venus is covered by so many clouds that it actually absorbs less sunlight than Earth does,” Laughlin was saying to his audience. “And so for many years it was perfectly possible to imagine the Venusian surface looking like this.” On the screen behind him, an aerial photograph appeared of waterfalls cascading through a mountainous forest shrouded in mist. “Then, in the late 1950s, astronomers found that Venus was just spewing out microwaves with an emission corresponding to a temperature on the order of six hundred degrees Celsius. It soon became clear that Venus was a runaway-greenhouse world, a truly terrible place.” Laughlin summoned an image of the true Venusian surface and let it linger silently on the screen—a lifeless, flattened landscape of shattered rock beamed back in 1982 by the Russian Venera 13 lander moments before the probe melted and imploded beneath hellish temperatures and crushing atmospheric pressures.
“During the 1950s, there was this brief, wild interval when you could realistically speculate that Venus not only had a habitable environment, but also that humans would soon visit it. The Apollo program wasn’t far away, and the ability to travel between planets was just within our grasp—in a way that it doesn’t seem to be any longer. Think what would have happened, how history would have changed, and what our world would look like today if we had found a habitable Earth-like planet literally right next door. It’s an odd, tragic coincidence that these possibilities disappeared for us right around the dawn of the Space Age. And as soon as Venus and then Mars changed from being candidates for full-blown economic colonization to being objects of mostly scientific interest, public interest really shifted to planets orbiting other stars.”
A few slides later, Laughlin showed a graph plotting all the known exoplanets, with masses on the vertical y-axis and years of discovery on the horizontal x-axis. A lone dot resided in 1995’s column on the plot’s older, sparsely populated left side, high up between the masses of Jupiter and Saturn. The dot represented a gas-giant world in a star-grazing 4.5-day orbit around the nearby star 51 Pegasi. It was 51 Pegasi b, a “hot Jupiter,” the first confirmed exoplanet around a Sun-like star, and a planetary system so bizarre that it spurred theorists to rewrite their models of planet formation. Sweeping forward in time, the dots proliferated across the chart and spread out into a thick wedge spanning a wide range of masses. In a decade’s span, the number of confirmed worlds beyond the solar system had soared into the hundreds, with no obvious end in sight. The field of exoplanetology was booming as never before.
Nearly all those worlds had been detected through a technique called radial-velocity (RV) spectroscopy, which watched for stars that wobbled from the gravitational to-and-fro tug of orbiting planets. When a planet tugs its star toward an observer on Earth, the waves of light from the star are compressed toward the blue end of the spectrum; when a planet tugs its star away, the starlight is stretched out toward the red. The same effect can be distinguished in waves of sound rather than light, as when an ambulance’s siren rises in pitch as it roars toward you up a street, then falls in pitch as it speeds away. The frequency of a star’s wobble indicates a planet’s orbital period—its year. The wobble’s strength—whether it corresponds to a kilometer or a centimeter of motion per second, for instance—yields an estimate of a planet’s mass.
It is not easy to distill planetary signals from the motions of a million-kilometer-wide roiling ball of plasma we call a star, particularly when the planets are small and in more-distant orbits. To do so requires not only large telescopes, but also high-resolution ultra-stable spectrometers. A telescope’s mirror gathers and amplifies light from a target star, which is sent streaming into such a spectrometer. Within the spectrometer, the starlight passes through a labyrinth of mirrors, gratings, and prisms that shape, split, and sort the photons by their wavelengths before sending them to be captured and preserved in the memory of a charge-coupled device, a CCD akin to those in consumer digital cameras. A star’s raw spectrum looks like a stretched-out, chopped-up rainbow, its languorous red-to-blue continuum broken by thousands of black absorption lines. The lines come from particular atoms and molecules percolating at the star’s glowing surface, soaking up certain wavelengths of starlight before the photons can escape to space. A star’s wobble is discerned in these lines as they migrate redward, then blueward, in time with the star’s reflexive motion from an orbiting planet’s gravitational pull. To track the motions of the lines, astronomers project reference marks onto the spectrum like tiny ticks on a ruler. For small planets, the offset in a line’s position may only be a fraction of a single pixel in the CCD detector—cooling the detector to cryogenic temperatures helps minimize stray electronic noise in the pixels, allowing such faint gradations to be seen. Stray electrical currents or minor changes in air pressure and temperature can also create noise that masks or mimics planetary signals. Complex statistical reductions and analyses of all the gathered data introduce further opportunities for error. Teasing faint RV signals from the noise is part straightforward science, part arcane art, and anyone who has the physical resources and mental capacity to do it belongs to an exclusive club with at most a few dozen members worldwide.
The problems of instrumental stability and data calibration were not new to planet hunting. Indeed, they had been behind an earlier, nearly forgotten era of false-alarm planets. Beginning in the early 1940s and extending into the early 1970s, several astronomers claimed detections of worlds orbiting nearby stars, all of which would ultimately prove illusory. The most prominent planet hunter of that bygone era was Peter van de Kamp, a Dutch-American astronomer at Swarthmore College. In photographic plates taken over a period of decades at the college’s 24-inch Sproul Telescope, van de Kamp thought he spied planetary wobbles in the motions of Barnard’s Star, a dim red dwarf and the next nearest star to our Sun after Alpha Centauri. His claims of two gas-giant planets around Barnard’s Star were initially reported—and endorsed—in scientific journals and in such popular publications as Time magazine and the New York Times, but competitors could find no evidence of those worlds in their own observations. Subsequent investigations suggested that van de Kamp’s
wobbles had been due to aberrations caused by periodic cleaning and upgrades of the Sproul Telescope, as well as to faulty analytic techniques. Years of close surveillance never found further evidence of his planets. Van de Kamp died in 1995, a few months before the detection of 51 Pegasi b, never having forgiven his critics and unwavering in his certainty that his worlds were genuine. Astronomers today use his story as a powerful admonition against cavalier claims of planetary discoveries based on indirect evidence and weak statistics.
With the 2009 launch of NASA’s $600 million Kepler Space Telescope, another more direct technique besides RV had become popular. Instead of looking for stellar wobbles, Kepler looked for transits, when a planet crosses the face of its star and blocks a small portion of the star’s light as recorded on a CCD. The frequency of a transit’s recurrence yields the transiting planet’s year, and astronomers can estimate the transiting planet’s size based on how much it dims a star’s light. Unlike RV, which could over time conceivably detect most planets around most stars, transits rely on random geometric alignments. Only planets with orbits that, by chance, align approximately edge-on with the line of sight from Earth will transit. This means the vast majority of exoplanets are invisible to the technique. The gamble, however, was worth the jackpot. Surveying a single patch of sky containing more than 165,000 stars, at the time of Laughlin’s talk the Kepler mission had already found in excess of 1,200 candidate transiting planets. “Candidate” was used until each planet could be confirmed or validated by other techniques, though many of Kepler’s stars were too faint for robust follow-up measurements. By early 2013, the Kepler team had announced the discovery of more than a hundred confirmed worlds, and nearly 3,000 candidates.
Compressed and plotted onto the rightmost edge of Laughlin’s chart, the announced Kepler candidate planets formed an unbroken line of dots. In comparison to the relative sparseness of all prior years, the Kepler data looked like a solid wall, one extending from several times the mass of Jupiter down to a few hundredths of Jupiter’s mass—the mass of Earth. “Obviously, what this shows is that we’re finding more and more planets at lower and lower masses,” Laughlin said, gesturing at Kepler’s wall of worlds on his chart. “Only a few years ago, almost all of this was terra incognita. Just this year, right now, we have finally begun detecting planets of one Earth mass. We’ve reached the point where we can actually credibly talk about planets around other stars that are the same size and mass as Earth, and we’re starting to get a much better sense of how most planetary systems are arranged. We’re taking what I like to call the ‘galactic planetary census,’ and what we’ve found is reminding us again that, like with Venus and 51 Pegasi b, making simple extrapolations based on the Earth or our solar system can be dangerous.”
Laughlin played an integral part in the ongoing galactic planetary census, not as an observer at the telescope, but in analyzing the data the observers delivered. One of his specialties was the trial-and-error statistical process of piecing together a star’s planetary system solely from RV wobbles. If a star’s wobble was due to a lone orbiting planet, plotting the back-and-forth oscillation over time would yield a classic sine wave, with smooth and regular crests and troughs that repeated with the period of the planet’s orbit, looking like the single pure note of a vigorously plucked violin string. Spotting that pattern in the data was simple. In multiplanet systems, however, each world imparts its own subtle distinct pull to the star, creating a more complicated pattern of wobbles. Teasing apart the system’s architecture from those wobbles was rather like determining the layout and composition of an entire orchestra as each of its instruments played different notes all at once. If a planet hunter was too focused on a handful of isolated sweet tones in the music of the spheres, he or she could miss other planets hiding in the sour notes and residual noise. The smaller the world, the weaker its signal, and the more astronomers would strain to detect its presence in washes of stellar static. Laughlin had guided the development of a piece of software, the Systemic Console, that could discern planetary signals out of such complicated data sets. It had rapidly become one of the field’s standard tools. Pulling up Systemic on his laptop, he gave his audience a taste of what real planet hunting was like. A black-and-white grid popped up onscreen, and hundreds of dots blossomed all over it.
“This is RV data for 61 Virginis, a star about twenty-eight light-years away, provided by my colleagues Paul Butler and Steve Vogt from two telescopes over the past several years,” Laughlin explained. He clicked a button, and a sine wave superimposed itself over the data, intersecting many but not all of the dots. “Here’s what would be the signal of a planet with a period of several hundred days and roughly a quarter of Jupiter’s mass, but you can see it’s nowhere near a perfect fit.” He tweaked the model planet’s orbit and mass for a few moments, but its sine wave stubbornly resisted aligning with the data points. “Now we’ll just run the automated fitting routine, which tries out different stable planetary configurations, runs through optimizations, and comes up with its best fit.” Another click of a button, and within seconds three distinct curves materialized out of the dots, running through far more of them than the sine waves from Laughlin’s previous attempts.
“You can see the software’s solution was three planets, although there are still residuals here—61 Virginis could easily have more planets, even in its habitable zone,” he said. “This is a result my collaborators and I have published. What’s really interesting is that out of the nearest few hundred stars in our local neighborhood, 61 Virginis is one of the most similar to our own. It has almost the exact same mass, radius, and chemical composition as the Sun, and it’s of a similar age, and yet its planetary system is completely different. The masses of these planets, nearest to farthest, are something like 5, 18, and 23 Earth masses, and they’re all closely packed roughly within what, in our own solar system, would be the orbit of Mercury. Our solar system has nothing interior of Mercury’s orbit, but 61 Virginis has three whole planets packed down there! These sorts of architectures are turning up everywhere we look, yet they could not have been extrapolated from our own solar system. They were unexpected.”
From the audience, someone asked how many stars might lack planets entirely.
“It’s very hard to show that any star is devoid of planets,” Laughlin replied. “So it’s more useful to ask what percentage of stars have planetary systems like ours. Data from RV surveys and from Kepler is now starting to show that stars with a Jupiter-mass planet in something like a ten-year orbit like our own system are fairly rare. This is something that I think at most only ten percent of surveyed stars can now support. At least as far as Jupiter is concerned, our solar system is somewhat unusual. Right now, the data are telling us that the archetypal planetary system is a Neptune-like planet in a warm, short-period orbit, but part of that is selection bias—those planets are easier to detect.”
He began to deliberately pace the floor, regaining the rhythm of his presentation by walking back and forth between the podium and a window. “We really don’t know much yet about the distribution of Earth-size planets in Earth-like orbits, but the expectation is that they will be abundant. Kepler’s going to tell us soon, I think—it’s easier to validate transits, even though they reveal small fractions of total populations. The small planets we’re detecting today create RV signals on their stars of the order of a meter per second. I’m walking a meter per second right now. Now, that’s an incredible accomplishment to detect such a small change in the motion of an entire star many light-years away, but it’s not quite enough: the Earth’s RV signal on the Sun is only about ten centimeters per second. Stars move more than that from interior pulsations and vibrations, and from material flowing around their surfaces—at any particular moment, a star is creating all that noise, introducing astrophysical jitter that contaminates the signal.”
There was an implicit warning within Laughlin’s carefully phrased remarks. The most tantalizing worlds—those that might rese
mble Earth and harbor life—were also some of the most difficult to find. A low-mass planet in a clement orbit about its star could often only be detected as a sliver of signal cresting above a sea of stellar noise. As the pressure mounted for RV surveys to turn up alien Earths orbiting nearby stars, it was also getting tougher to know what was actually real.
Faint RV planetary signals could be amplified, Laughlin told his audience, by lavishing a quiet star with attention, hammering away with hundreds or even many thousands of observations, all averaged over time to beat down the star’s already low stellar noise. But the approach came with risks—a planet-hunting team fortunate enough to secure observing time on a world-class telescope and spectrometer might chase a star’s tantalizing signals for months or even years, only to ultimately discover that its potential planets were illusory. Careers and reputations would be forged or shattered on the barest probabilistic whiffs of planets emerging from a statistical haze. “[The] push toward lower-mass planets is part of an ‘arms race’ among the different competing groups,” Laughlin explained near his presentation’s conclusion. “You find planets, and then the time-allocation committees give you more time to find more planets. If at any point you don’t find planets, that’s it, you’re out of the game.”
• • •
For more than a decade after astronomers began regularly detecting exoplanets in the mid-1990s, the RV “game” had been limited to a contest between two great planet-hunting dynasties, one American, the other European. The first began in Pasadena, California, in 1983, while a struggling twenty-eight-year-old astronomy postdoc named Geoff Marcy was taking a long morning shower. Marcy’s research into stellar magnetic fields wasn’t panning out, and had been roundly criticized by a few senior astronomers. He felt incompetent, depressed, and burnt-out. As the water streamed over his downturned head, he realized that up until then his career had been mostly a failure, and that if nothing changed, it might end before it had even properly begun. He thought back to what had set him down his disastrous path of astronomy, when he had been a young boy wondering whether all the stars in the sky had planets. Where had his passion gone? Suddenly, an epiphany: if he was destined for failure, he should fail spectacularly, pursuing a topic he loved but few others took seriously. By the time he finished showering, he had decided to spend the rest of his perhaps-brief career searching for exoplanets.
