Corey S. Powell has been an editor at Discover since 1997. This article is adapted from his just-published book, God in the Equation. Copyright © 2002 by Corey S. Powell. Reprinted by permission of The Free Press, a division of Simon & Schuster, New York.
Saul Perlmutter darts around his modest office at Lawrence Berkeley National Laboratory, a cluster of drab buildings nestled in the hills above the University of California campus. With his edgy movements, shaggy hair, and Woody Allen-ish gestures, he could be mistaken for a computer programmer. But it's soon clear that these institutional-lab white walls and gray steel bookshelves— even the rolling landscape outside— are only a minuscule part of who he is. Riffling through a stack of journal reprints and computer printouts, Perlmutter fishes out an article titled "Measurements of Omega and Lambda from 42 High-Redshift Supernovae." During the past 10 years, working in step with a rival group of scientists centered at Harvard University, Perlmutter and his collaborators have peered to the far edge of what astronomer Edwin Hubble called "the dim boundary— the utmost limits of our telescopes." The results, summarized in this innocuous-sounding document, have rewritten the saga of the Big Bang. They offer both a new chronicle of how the universe has evolved and an unnerving prophecy of how it may end.
When he set out on his cosmic quest, Perlmutter was still in his twenties, full of improbable ambition. “It goes back to childhood,” he says. “I’ve always been interested in the most fundamental questions.” He began by studying subatomic particles, but by 1983 he was fed up with complicated physics experiments that took years to execute. He sought a different path to universal truth and found it in astrophysics.
Ever since 1929, when Hubble presented evidence that galaxies are flying apart from the Big Bang, cosmologists had known that the fate of the universe lay in two numbers: the rate of its expansion and the rate at which that expansion is slowing down. The best way to determine those numbers was to measure the distances to extremely remote galaxies and how much their light had been stretched over time. But it was a notoriously difficult task—Hubble’s followers were still bitterly debating the answers six decades later. Perlmutter decided to gamble on a relatively untried technique: He would reckon the distances by the light of supernovas.
German-born astronomer Walter Baade suggested the idea in 1938 as he worked at the Mount Wilson Observatory in California. Then as now, astronomers estimated the distances to galaxies by studying Cepheid variables, an unusual class of stars whose brightness rises and falls predictably: The longer the period of variation, the more luminous the star. But even the most powerful telescopes of the time could detect Cepheids only in a handful of nearby galaxies. Supernovas, in contrast, are so brilliant that they can be seen across the entire universe. Formed when a star self-destructs, supernovas exist for only a few weeks before fading away; but for those few weeks, they shine more brightly than a billion suns. If all supernovas are essentially the same, Baade reasoned, their light can be used as “standard candles” to reckon cosmic distances.
But the supernovas were not as standard as Baade hoped. He soon learned that some are much more luminous than others. If observers did not understand the nature of those variations, their distance measurements could be off by more than a factor of two. By the time Perlmutter began his quest, a number of researchers—among them supernova guru Robert Kirshner of Harvard—had identified that a class of exploding stars could light a path through such difficulties. Dubbed Type Ia, these supernovas form when middleweight stars like the sun grow old and burn out, leaving behind a white dwarf star. Normally, a white dwarf is stable. But if it has a companion star, it can grab material from its partner and keep growing more massive. Eventually, it hits a point at which gravity can no longer support its bulk. The star implodes, setting off a titanic thermonuclear blast.
Type Ia explosions have a distinctive light pattern, or spectrum, that makes them easy to identify. As luck would have it, they are also the most luminous supernovas. Perlmutter and his Berkeley Lab colleague Carl Pennypacker decided to see if these stars could, at last, provide the kinds of cosmological revelations that Cepheid variables could not. The two researchers persuaded a few graduate students and colleagues to help and in 1988 began the Supernova Cosmology Project. But they weren’t the only ones drawn to supernovas. Soon they would find themselves in a heated competition.
Supernovas are among the rarest of celestial events. The last one seen in our galaxy was recorded by Johannes Kepler in 1604, five years before Galileo turned his first telescope skyward. In any one galaxy, a Type Ia explosion lights up just once every 300 years or so. But on a cosmic scale, the numbers pile up quickly. There are so many galaxies in the universe—about 100 billion—that today’s largest telescopes could in principle detect supernovas every few seconds. The problem is where to look.
The detectors, known as charge-coupled devices, or CCDs, record every iota of light they receive as digital fields of ones and zeros. Perlmutter decided that if images were converted to digital data, they could be searched to find a single supernova in a field full of galaxies. He would begin by recording the light from a patch of sky. Then, a few weeks later, he would record it again and subtract the binary numbers in the first image from those in the second. If everything stayed the same, nothing but background noise would remain. But if something new appeared—if a star exploded and brightened—it would pop out immediately. That was the idea, anyway. In practice, nobody could make it work. Perlmutter spent long hours writing software to combine, clean up, and analyze the images. “A lot of time you think, ‘Boy, you’re spending your whole life on this stupid computer,’” he says, laughing.
The first big break came in 1992, when the Supernova Cosmology Project bagged its first distant Type Ia supernova using a new CCD detector on the two-meter (6 ½ foot) Isaac Newton Telescope at the La Palma Observatory in the Canary Islands. Over the next two years, Perlmutter recorded a succession of supernovas, proving that systematic searches were possible. Then a new uncertainty stripped the bloom off astronomers’ rosy optimism. Preliminary surveys of relatively nearby supernovas in the late 1980s and early 1990s showed that Type Ia supernovas are not identical after all. Some brighten and fade faster than others; some are inherently more luminous. Slowly, a team led by Mark Phillips at the Carnegie Institution of Washington’s Las Campanas Observatory in Chile uncovered a meaningful pattern within the chaos. Sluggish supernovas are consistently brighter at their peaks than fleeting ones. The correlation is so tight that the steepness of a supernova’s light curve—a plot of its changing brightness over time—accurately predicts its intrinsic brilliance.
Adam Riess of the Space Telescope Science Institute, one of Kirshner’s disciples, devised a statistical technique to extract that measurement. A little later, Perlmutter came up with his own, more geometric solution: Expand the light curves to correct for the supernovas’ differences. “I drew light curves stretched in time, and they were amazingly close,” he says. “They all fell on top of each other. It was clear there was some physics making that happen.” Both teams eventually claimed they could calculate intrinsic luminosity to within about 10 percent, an astonishing level of accuracy.
Still, not everyone agreed on how to interpret the results. From time to time, Perlmutter contacted other members of the tiny supernova community to answer a question or help interpret an observation. One of those who lent a hand was Brian Schmidt, a soft-spoken 25-year-old Harvard graduate student. Under Kirshner’s guidance, Schmidt had started out studying the mechanics of how supernovas detonate. “I liked them as physical objects,” he recalls. He also knew that exploding stars could illuminate the greatest mysteries of cosmology—but they could just as easily fool anyone who failed to apprehend their tremendous complexity. Schmidt and a few of his Harvard colleagues followed the progress of Perlmutter’s team, both at conferences and in person, and started to feel uneasy. “We were not terribly happy with the way they were analyzing the data at the time,” he says.
Schmidt conferred with Kirshner and suggested that they launch their own, independent supernova search. Kirshner was skeptical. People had found supernovas before, only to realize that they could not squeeze useful cosmological information from them. “Yes, we could do it better,” Kirshner said. “But could we do it?” Schmidt convinced him that they could. In 1994, together with a number of the other supernova experts in their circle, they formed the competing High-Z Supernova Search. (Z is the term astronomers use to denote how the light of distant objects is stretched by the expansion of the universe.)
Perlmutter had a huge lead in software development. Schmidt, for his part, had a group of colleagues intimately familiar with supernovas and the knowledge that the project seemed at least technically feasible. Drawing on his expertise with astronomical computation, Schmidt sat down and hammered away at the same programming problems that had bedeviled Perlmutter. “Saul’s group worked for six years on software,” Kirshner says, sounding like a proud father. “Brian said, ‘I could do that in a month.’ And he did.” The two groups were off and running.
Actually, it was more like they were chasing each other through knee-deep molasses. Hunting supernovas calls for a singular mix of frantic activity and limitless patience. It begins in a frenzy of administrative activity, securing time on a large telescope just after a new moon, when the sky is dark, and three weeks later when moonlight again is not a problem. Both teams booked time on the 4-meter (13-foot) telescope at the Cerro Tololo Inter-American Observatory in Chile, whose huge CCD detector could capture the light of 5,000 galaxies in 10 minutes. Once the researchers secured two images of the same area, they had to make sure the views were properly aligned. Then they had to account for changes in atmospheric clarity and eliminate the many flickering objects that were not supernovas. All told, they might look at more than a hundred thousand galaxies in one season.
If a blip of light looked promising, another round of work began. The scientists made a pilgrimage to the huge Keck Observatory atop Mauna Kea in Hawaii. Each of the twin Keck telescopes has 36 aluminized glass-ceramic hexagons that form a 33-foot-wide Cyclops eye, able to gather enough light from a suspected supernova to spread the beam into a spectrum. Once Perlmutter and Schmidt identified the telltale sign of a Type Ia, the real frenzy began. To get an accurate read, the scientists had to track each supernova for 40 to 60 days at observatories around the world. After that came data processing to correct for intergalactic dust and other possible sources of error. Final analysis could take a year or more, until the supernova had faded from view, when it was possible to get a clean view of the galaxy where it lived and died. All the while, each team felt the other breathing down its neck.
For Schmidt, terrestrial distances became nearly as vexing as celestial ones. In 1995 he took a position at Mount Stromlo and Siding Spring Observatories near Canberra, Australia—now called the Research School of Astronomy and Astrophysics—and ended up on the other side of the world from his colleagues. “I had just had a child, I had just written software that had never been used before, and I was attempting to look for supernovas and debug the software across 13 time zones between Chile and Australia,” he says. “It was nearly a disaster.” Thankfully, Schmidt soon found his first cosmologically significant supernova, proving that his efforts were not in vain.
Cosmologists investigate the history of the universe by looking at two aspects of supernovas, brightness and redshift, that relate in a very complicated but meaningful way. The brightness reveals how far the star’s light has traveled; the redshift shows how much it has been stretched by the expansion of the universe. Most scientists assumed that the expansion of the universe had been slowing ever since the Big Bang, as the gravitational attraction of 100 billion galaxies tried to pull the whole works back together. In that case, faraway supernovas (which we see as they were billions of years ago, when the growth was more rapid) would have accumulated redshift more quickly relative to their distance than nearby ones. Put another way, those distant objects would be nearer, and therefore brighter, than you would naively expect if you simply extrapolated back from the way the universe is expanding closer to home. The amount of additional brightness relative to the redshift tells you the rate of deceleration and hence the overall density of the universe.
The first supernova that Schmidt’s group fully investigated was actually a bit dimmer than expected, but the High-Z team needed more data to understand the meaning of that single odd result. Perlmutter, meanwhile, slogged through observations of 23 distant supernovas and analyzed seven of them by the end of 1996. The stars lay roughly 4 billion to 7 billion light-years from Earth, or as much as halfway to the visible edge of the cosmos. At first, the Supernova Cosmology Project members believed they saw what cosmologists had long expected. The light of the supernovas was a little brighter than it would have been if they were hurtling away at today’s speed; the expansion of the universe must have been faster in the past.
But as Perlmutter worked through more observations, the picture improbably reversed. One after another, the supernovas seemed to grow fainter and fainter relative to their redshifts. As the brightnesses dropped, so did the implied density of the universe. Eventually, the density fell into the minus zone, implying that the universe contains less than nothing. “I guess we’re not here,” Perlmutter joked nervously.
There was only one sensible way to interpret the finding: The universe is not slowing down at all—it is speeding up. Reluctantly, Perlmutter turned to an idea first raised by Albert Einstein in 1917. Einstein’s theory of general relativity had overturned Newton’s ideas about gravity and showed how to describe the structure of the universe with a simple set of equations. But it had also yielded a paradox: Gravity should make the cosmos curve inward and crash in on itself. The reason it didn’t, Einstein suggested, was that the universe was filled with a subtle long-distance repulsion that he called Lambda.
Lambda was a hypothesis—a prayer almost—for which there was no observational evidence. When the discovery that the universe is expanding was announced, it eliminated the original motivation for Lambda, and Einstein quickly denounced it. Yet cosmologists still invoked it from time to time, intrigued by its ability to make their models of the universe look more balanced. Now these supernovas suggested that Einstein had been on the right track after all. If the universe was accelerating, the vacuum between galaxies had to be full of some kind of energy that acted just like Lambda, pushing them apart. Perlmutter saw negative densities because some mysterious energy dominates the universe, overwhelming the gravitational pull of ordinary matter.