The Crowded Universe

Walker and Bruce Campbell, his close colleague at the University of British Columbia, were true pioneers in the field of searching for planets around other stars. In the late 1970s, they had developed an ingenious technique that would enable them to discover Jupiter-mass planets in orbit around other stars similar to the Sun. They had put their idea to good use, spending 12 years of their lives and many nights of precious telescope time on the 114-inch (3.6-meter) Canada-France-Hawaii Telescope (CFHT) on Mauna Kea, Hawaii. From 1980 to 1992, Walker, Campbell, or a member of their group at UBC had traveled to Hawaii to spend between six and twelve nights each year looking for the first hints of a planet outside the Solar System.

Even a “gas giant” planet such as Jupiter, with 318 times the mass of Earth and 11 times its diameter, is nearly impossible to find when it is in orbit around a star. The problem is not so much the faintness of the planet itself—Jupiter is no fainter than the distant galaxies that have been imaged by the Hubble Space Telescope in the Deep and Ultra Deep Field surveys. The problem is that when a planet is orbiting a far brighter object, it is exceedingly difficult to see the planet in the glare of the star’s light. Stars such as the Sun give out most of their light at visible wavelengths, where our eyes are best suited to seeing. At visible wavelengths, Jupiter is about a billion times fainter than the star it circles. Even with its incredibly powerful cameras, NASA’s Hubble is incapable of snapping a photograph of a planet orbiting a star—the star’s light would drown out the planet’s light many times over.

Campbell and Walker developed a completely different technique for detecting the presence of a seemingly invisible planet lurking in the glare of its star. Rather than trying to see the planet directly, they would infer its presence by the effects that it must have on its own star. Their scheme relied on the fact that something we are all taught in elementary school is not quite right—the planets of the Solar System do not orbit the Sun. Rather, all the planets and the Sun itself orbit a single point in space, the center of mass of the Solar System. This center of mass is the place where the entire Solar System could be balanced on a fulcrum if there were a teeter-totter large enough for all to join in. To balance a teeter-totter, an adult must sit much closer to the central fulcrum than a small child sitting on the other end of the teeter-totter. Similarly, because Jupiter is 1000 times less massive than the Sun, the balance point for the Jupiter-Sun system is 1000 times closer to the Sun than to Jupiter. As Jupiter orbits the Sun over a period of 12 years, the Sun orbits the center of mass, or barycenter, of the Solar System in a circle that is 1000 times smaller than Jupiter’s orbit. If astronomers living on a planet around a nearby star were willing to spend 12 years watching the Sun, they might be able to detect this periodic motion of the Sun. If they did, the only explanation possible would be that the Sun must be orbited by a planet 1000 times less massive. There is no other physical explanation for a star appearing to wobble back and forth across the sky this way.

Because of the presence of Jupiter, the Sun moves around the center of mass of the Jupiter-Sun system on a circle whose diameter is roughly equal to the diameter of the Sun itself. Detecting such a miniscule wobble of the Sun from the great distance of another star is not easy, but compared to trying to take a direct photograph of an extrasolar Jupiter, it is simple. Walker and Campbell decided to hunt for Jupiters by clocking the speed of stars as they move around the barycenters of their planetary systems. The Sun moves with an orbital speed of about 30 miles per hour, or 13 meters per second, around the Sun-Jupiter barycenter. A speed of 30 mph sounds like it should not be hard to detect; if you are traveling 30 mph over the speed limit, you can be sure that a police officer will have no trouble pointing a radar gun at you and writing you a speeding ticket that will stand up in court. The problem for stars is that the relevant standard of comparison is the speed of light, which is about 186,000 miles per second, or 670,000,000 mph. Thus 30 mph is practically stationary by comparison.

The star’s speed is compared this way because Walker and Campbell planned to find planets by measuring it through the Doppler effect. Christian Johann Doppler, an Austrian physicist, hypothesized in 1842 that light waves emitted by a moving star behave exactly the same way as sound waves emitted by a moving train. When a train blows its whistle as it is moving toward you, you hear the whistle at a higher pitch, or frequency, than when the whistle is sounded while the train is stopped. When the train is traveling away from you, the whistle sounds lower in pitch than when it is at rest. The distinctive change in pitch of the sound of an automobile engine as it passes by is familiar to NASCAR fans, who are dazed by the roar of the Doppler effects created by dozens of stock cars racing by at speeds of 180 mph or more.

The change in frequency of the sound produced by the Doppler effect depends on the ratio of the speed of the NASCAR race car to the speed of sound. The sound speed in air is about 750 mph, or 340 meters per second, so a race car can move at about one-quarter of the speed of sound. That means that the sounds heard during a NASCAR race can be shifted to higher or lower pitches by a similar fraction— a distinctive change equivalent to the difference in pitch between the musical notes A and D.

Doppler postulated that this frequency shift would occur for light in the same way it works for sound, because light and sound are both wave phenomena. He hoped the effect would explain the red and blue colors of different binary stars, systems where two stars are in orbit about their own center of mass, with red stars moving away from the observer, and hence emitting lower-frequency and longer-wavelength light, and blue stars moving toward the observer, producing light of higher frequency and shorter wavelength. Sadly, he was mistaken, because the frequency shift associated with the Doppler shift of light is small indeed compared to the intrinsic variations in the colors of stars, whether they are in binary systems or all on their own. Even binary stars moving with speeds of several miles per second exhibit a frequency shift of their light of only one part in 100,000. The French physicist Armand Hippolyte Louis Fizeau (1819-1896) independently predicted the Doppler effect for light in 1848, leading the French to describe it as the Doppler-Fizeau effect.

The zenith of Doppler’s career came in 1850, when he was appointed founding director of the Institute of Physics at Vienna’s Imperial University. As director, he was responsible for deciding which candidates would be admitted for study at the university. (One candidate he turned down was Johann Gregor Mendel, who later gained admission to the university through a different department and became a pioneer of genetics research. Mendel’s mathematical skills were not considered on a par with what Doppler required of a physicist, to the everlasting benefit of modern genetics.)

The Doppler effect that Walker and Campbell sought to measure was quite small. If the motion of the Sun induced by the presence of Jupiter is 30 mph, dividing that speed by the speed of light—670,000,000 mph—yields a Doppler shift of about one part in 20,000,000. Campbell and Walker devised a means of measuring such a small shift by including a glass-ended container filled with hydrogen fluoride gas in their telescopic instrumentation. The hydrogen fluoride was introduced to serve as a stable set of reference lines for measuring the expected tiny Doppler shift. The light from the target star would pass through the telescope, bounce off the telescope’s mirrors along the way, and then pass through the hydrogen fluoride cell before it could enter the telescope’s spectrograph. There, the star’s light would be split into different colors, just as a prism splits white light into the colors of the rainbow. The hydrogen fluoride gas absorbs some of the star’s light only for very specific, narrow bands of color (wavelengths), thereby superimposing a set of absorption lines for comparison with the emission and absorption lines in the star’s spectrum. The star’s spectral lines are produced by the ions of elements, such as calcium, sodium, and iron, and by molecules, such as titanium oxide, in the star’s outer atmosphere.

As the target star moves in orbit around the barycenter, alternating between moving toward Earth and moving away from Earth, the star’s spectral lines will vary in frequency (equivalently, in wavelength or color) according to the Doppler effect. Meanwhile, the hydrogen fluoride lines do not change in frequency at all, so they provide a stable reference for making precise measurements of the changes in frequency of the star’s spectral lines. Given that it would take 12 years of observations to follow the stellar wobble induced by a planet similar to Jupiter, the hydrogen fluoride cell provided the stable reference source that is crucial to carrying out such a long-term search. Hydrogen fluoride had the particular advantage of providing a number of widely spaced lines with just the right colors to optimize the search, but it also had the disadvantage of being poisonous. Accidentally breaking the cell that contained the hydrogen fluoride would halt the observing program—and possibly an astronomer or two. With the hydrogen fluoride cell in place, Walker was able to make Doppler measurements accurate to about 35 mph, or 15 meters per second, close enough to the expected wobble of 30 mph induced by a Jupiter-like planet to begin the search.

Walker’s planet search was granted a special status on the Mauna Kea telescope as the only long-term program during the years 1980- 1992. This meant that Walker and his group did not have to worry about producing new discoveries every year in order to be awarded their allotment of telescope time for the following year, as is usually the case for astronomers. (Most telescopes are oversubscribed, so the committees charged with assigning telescope time often have to be brutal and reject nonproductive ongoing projects.) Walker had the luxury of being able to search for hidden beasts that could be detected only by taking Doppler measurements for a decade or longer: gas giant planets similar in mass to Jupiter. With their hydrogen fluoride cell accuracy of 35 mph, in fact, Walker could hope to find planets only of Jupiter mass and above; there was no hope of searching for the much smaller Doppler shift that would be produced by an Earth-like planet. Still, judging on the basis of the only planetary system known at the time, our own, Jupiters appeared to be natural products of the planet formation process, so they should be out there, waiting to be discovered. A Jupiter-mass planet on a Jupiter-like (12-year) orbit around a Sun-like star would be expected to be a signpost for Earth-like planets in the same planetary system.

Bruce Campbell thought that they had found something earlier. In 1988 he and Walker published a paper in the Astrophysical Journal reporting their results to date. Of the 16 stars on their target list at that time, 7 showed long-term trends that were consistent with the wobbles induced by planets with masses in the range of 1 to 9 Jupiter-masses. They suggested that they had found “the tip of the planetary mass spectrum.” There seemed to be good evidence for a 1.7-Jupiter-mass planet on an orbit with a period of 2.7 years around one of the stars in a binary star system called Gamma Cephei, the third brightest star in the constellation Cepheus.

In 1992, however, Walker published a paper in the Astrophysical Journal that retracted the 1988 claim for a planet in the Gamma Cephei system. By then, Bruce Campbell had quit the planet search program and left astronomy. Four more years of data had changed the orbital period of the suspected planet from 2.7 years to 2.5 years, the same period at which they found that the star’s calcium emission line varied. The variation was thought to be caused by the rotation of a giant star in the Gamma Cephei system with a rotational period of 2.5 years. A rotating star can also produce a Doppler shift if the surface of the star is covered with spots and other irregularities that affect the amount of light coming from different regions, some of which are moving away from Earth, and some toward it. Gamma Cephei’s “planet” seemed to be nothing more than grumblings from a cantankerous old star.

The paper that Gordon Walker and his remaining colleagues published in 1995 stated that they had looked for Doppler wobbles in 21 stars over a period of 12 years and had found no firm evidence for any planets with masses greater than Jupiter on Jupiter-like orbits. Considering this outcome along with the results of other planet searches, they furthermore claimed that 45 nearby stars showed no evidence of Jupiter-like planets and stated that their results presented a “challenge to theories of planet formation.”

Wetherill had studied what would happen to the formation process of habitable planets if he varied some of the basic parameters, such as the mass of the central star, the mass of the planet-forming disk of gas and dust, and the effects of the giant planets. The calculations he ran were based on the Monte Carlo technique, an apt name for what was going on. The route through which rocky planets such as Earth are built up from a large population of smaller bodies is a random one.

This lengthy process entails the banging together of progressively larger and larger solid bodies as they orbit their central star. It is thought to begin with tiny dust grains too small to be seen by the human eye. Wetherill had taken a shortcut and was studying the final phase of the collisional accumulation process, when a swarm of several hundred lunar-mass bodies are in orbit, having taken about 100,000 years to grow that large. This final phase takes tens of millions of years to play out, because the lunar-sized bodies have to wait longer and longer to smack into each other and grow even larger. The Moon’s mass is 81 times smaller than that of Earth, so hundreds of fierce collisions between sizable bodies are needed to form a planet as large as Earth.

The lunar-sized bodies are spread out over an immense area, and the chances of random collisions occurring are small. The situation is like a Dodge-Em car ride at a State Fair, but one in which dozens of cars are allowed to roam over the entire state of Kansas, instead of being confined to an area the size of a basketball court. In the planet formation ride, when two bodies of equal mass collide, they are likely to stick together and form a new body with twice the mass, instead of just bouncing off each other. Also, lunar-mass bodies are large enough that their gravitational attractions for each other, though small, are strong enough to determine the ultimate outcome of the ride. Mutual gravitational interactions pull the careening bodies toward each other, resulting in near misses, sideswipes, head-on collisions, and rear-end impacts without the benefit of airbags.

The entire process is an unplanned, chaotic one, and changing the initial position of just one of the Dodge-Em cars at the start of the ride can result in planets too close or too far from the Sun to be habitable. As a result, Wetherill had to run many different versions of the same basic simulation of the rocky planet formation process in order to get a good statistical picture of what the most likely outcomes would be when he played with the various free parameters in the modeling effort. For the Icarus paper, Wetherill ran about 500 Monte Carlo models in order to study 20 different assumptions about the mass of the star, the mass of the planet-forming disk of gas and dust, and the effects of the giant planets.

What he found was tremendously reassuring for those who imagined that planetary systems like our own might be common. Essentially all of the variations he tried resulted at times in one or two habitable worlds—that is, roughly Earth-mass bodies orbiting at distances from their stars where it would be warm enough to have liquid water on their surfaces, but not so hot as to turn the water into steam. Stars with masses half as large as the Sun or half again as large as the Sun appeared to be likely to have Earth-like planets. And for stars with the same mass as the Sun, the formation of at least one habitable planet seemed inevitable. When the amount of mass in the disk was doubled or halved, habitable worlds still formed, but with twice the mass each or half the mass each, respectively.

Wetherill noted that the presence and location of the gas giant planets that were assumed already to have formed in these systems by the time he turned on the electricity for the Dodge-Em cars were crucial for determining the final outcome of the ride. The gravitational forces exerted by the gas giants determined how far out from the central star rocky planets could grow, just as they did in our Solar System. Were it not for the gravitational yanks from Jupiter and Saturn, Mars would probably be a larger planet than it is today, with about one-tenth the mass of Earth, and there would be another rocky planet beyond Mars, instead of the horde of frustrated “wannabe planets” in the asteroid belt. When Wetherill ran models without any gas giant planets at all, he found that the planets that formed were likely to be twice as massive as when Jupiter and Saturn were up to their dirty tricks, and that the planets extended all the way out to Jupiter’s orbital distance, five times Earth’s distance from the Sun (93,000,000 miles, or 150,000,000 kilometers, a distance that is called the astronomical unit, or AU).

Wetherill concluded that “abundant populations of habitable planets” could be produced for stars of any of the masses he considered. On purely theoretical grounds, Wetherill had shown that one could expect that habitable planets were abundant in the universe.

Earlier in the year, the January 20 issue of Science had included a paper in which I discussed the chances for detecting the first Jupiter-mass planets around nearby stars. I had argued that, on the basis of computer models I had calculated of how planet-forming disks would heat and cool, it was to be expected that gas giant planets would form at Jupiter-like distances even around stars with lower masses than the Sun. Such “red dwarf” stars are much more common in the Sun’s neighborhood of the Galaxy than are stars like the Sun or more massive than the Sun. Because of their relative closeness and abundance, red dwarf stars seemed like the first place to look for another planetary system similar to our own. My calculations implied that if red dwarfs had gas giant planets, they would be orbiting far enough away from their stars for there to be rocky terrestrial planets orbiting comfortably in the habitable zone of the star. Red dwarf stars, the most common type of star, could thus be home to the majority of the habitable planets in the Galaxy.

Gordon Walker’s search had concentrated primarily on stars like the Sun, rather than on the more numerous and nearby red dwarfs. We had all assumed that Sun-like stars would have Jupiter-like planets, based on the undeniable evidence of the one obvious example we had—our own Solar System. But maybe the Solar System was not the proper example to consider after all. George Wetherill had suggested as much several years before, during a talk on finding new planetary systems that he had delivered at a conference held at Caltech, in Pasadena, California, on December 7-10, 1992. Wetherill had shocked the audience of several hundred astronomers and planetary scientists by pointing out that the mere fact that the one known planetary system contained a Jupiter did not necessarily mean that Jupiters were common. Jupiter protects us from the comets that revolve in the Kuiper Belt beyond Neptune’s orbit. When a malevolent comet decides to break out of the Kuiper Belt and make a suicidal dash toward Earth, Jupiter plays the role of the batsman protecting the wicket in a cricket match. It swats the comet out of the Solar System, or forces it to smash harmlessly into the Sun, or takes it right in the face, as Jupiter did with the startling collision of Comet Shoemaker-Levy 9 just 2 years later, in 1994.

Without Jupiter, Wetherill noted, Earth would be whacked by comets roughly 1000 times more often than is the case with Jupiter at bat. Dinosaur-killing events such as the collision that marked the Cretaceous-Tertiary extinction event 65 million years ago would be occurring every 100,000 years, instead of every 100 million years or so. It is hard to imagine intelligent life evolving if the evolutionary clock were reset by that kind of catastrophe every 100,000 years. Wetherill made the point that we therefore did not have a single unbiased example of a planetary system to use to predict what would be found elsewhere in the Galaxy. The problem of interpreting the statistics of a sample composed of only a single solar system was eliminated, because even that sample was biased and had to be rejected. We really had not a single planetary system to use to predict how common Jupiters would be.