The 4 Percent Universe
Page 6
Forty years earlier Edwin Hubble had arrived at the evidence for an expanding universe by studying the behavior of galaxies. By tracing that expansion backward, as if running a film of the outward-flying galaxies in reverse, Georges Lemaitre had arrived at the idea of a primeval atom. Peebles hadn't believed that the universe could be that simple, but now he was becoming one of the leading interpreters of a simple universe: homogeneous and isotropic—one that looked the same no matter where you were in it and no matter which way you looked. After Penzias and Wilson, as well as Peebles and his colleagues at Princeton, had found evidence for what those initial "Primeval Fireball"* conditions might actually have been, Peebles began using that knowledge to refine his understanding of the expansion itself. It was as if he were running the film of the history of the universe again, only instead of rewinding it to the beginning he would be running it forward to today.
Now that was a movie he wouldn't mind seeing.
The computer he would be using at Los Alamos—a CDC 3600—was many magnitudes more powerful than any he could have found on a university campus, and he wouldn't even have to tap the Princeton Physics Department's research funds. He could run the computer as long as he needed—all night, even all weekend. And Peebles could do so even though Los Alamos was in high Cold War mode and he wasn't a U.S. citizen. Peebles had emigrated from Manitoba only eleven years earlier. He was a Canadian citizen—officially an alien. Yet apparently the work he was doing seemed either so primitive or so esoteric—or his demeanor so unthreatening; his reputation so established; his computational skills so (relatively) undeveloped—that his entire security detail consisted of a secretary who sat to one side, knitting.
Peebles would be performing what scientists call an N-body simulation. Take a number—N—of points, program them to interact according to whatever properties you want, and see how the action unfolds. In this case, Peebles would be taking 300 points and treating each as if it were a galaxy in one particular part of the universe—the Coma Cluster, the closest and most-studied galaxy cluster. He would assign each galaxy a position and velocity based on rough observations of real galaxies in the cluster, and he would teach the computer the law of universal gravitation. And then he'd let the model do whatever galaxies interacting gravitationally in an expanding universe do over billions of years.
He'd already been thinking about how clusters develop, and he'd done some cluster calculations during his time at Caltech. Now he took that initial research and converted it into a computer program. Then he punched the holes in the 7%-by-3%-inch computer cards himself, stacked them in the metal feeder, and ran a simulation. At the end of the simulation the points had moved a bit. He transferred the image to a frame of 35-millimeter film, and then he ran the next simulation, using the galaxies' positions at the end of the previous simulation as a starting point. Again, over a period equivalent to millions of years, the galaxies shifted slightly. When Peebles had enough frames, he ran them together, loaded the film in a projector, and sat back.
The universe swirled to life. Galaxies moved outward, following the flow of the Hubble expansion. But then they didn't. They slowed, also moving under the influence of their mutual gravitational attractions, and they continued to slow, until they stopped following the flow of the expansion and began to fall back on themselves. Smaller galaxies clumped toward the nearest larger galaxies, and those growing clumps clumped with other clumps. The more the galaxies clustered, the more the galaxies clustered.
Simple, sort of.
In one form or another, the question of how simple the universe was had come to occupy Peebles ever since the discovery, four years earlier, of the cosmic microwave background at a temperature that roughly matched his prediction. Despite his earlier mistrust of cosmology, he now felt that the field had probably possessed the requisite makings of a science since the early 1930s. Edwin Hubble had acquired a set of measurements: the direct correlation between galaxies' distances and redshifts. Georges Lemaitre and Aleksandr Friedman had attached a theoretical interpretation to those observations: a universe expanding from a Big Bang. And there was no shortage of open issues to drive further research. The assumptions that the universe would look the same on the largest scale (that it was homogeneous) no matter which way you looked (that it was isotropic) were not assumptions that Peebles would have made. And maybe his prejudice against simplistic assumptions had blinded him to the scientific possibilities of cosmology. But his attitude had begun to soften even before he heard of Penzias and Wilson's 3 K detection. As soon as he had finished his calculations for the cosmic microwave background's temperature—a prediction you could actually put to a test—he understood that he was going to have to take cosmology seriously.
At once he and Dicke began collaborating on a major paper, "Gravitation and Space Science," which they sent to Space Science Reviews in early March 1965, right around the time of the fateful phone call from Penzias. (They added a note in proof about the Penzias and Wilson detection as well as the two upcoming papers in the Astrophysical Journal.) Dicke handled the physics section of the paper, Peebles the cosmology. As a sort of belated rejoinder to the Rochester professor who a quarter of a century earlier had told Dicke that physics and general relativity had nothing to do with each other, they wrote in the introduction: "While in a limited sense gravitation is of no great importance to a physicist, this is much too naive an interpretation." In the first paragraph of the "Cosmology" section Peebles expanded on that philosophy. For physicists, he wrote, cosmology doesn't satisfy just "the obvious interest" in the origins of the universe; "we need cosmology as a basis of any complete theory of the galaxies, or for that matter, of the solar system."
If you wanted to understand specific problems concerning the evolution and structure of the universe—the clustering of galaxies, for instance—then you had to abandon any residual "island universe" assumptions. You had to learn to think about the universe not only as a collection of individual galaxies but as the sum of its galaxies—a single unit, a whole. You had to keep in mind that while the whole—the universe—was expanding, its parts—the galaxies—were evolving. "The moral of this section," Peebles offered in conclusion, twelve pages later, "is the unity of the universe."
Even as he was collaborating with Dicke on "Gravitation and Space Science," Peebles had begun writing an investigation into how the primeval conditions of the universe might develop into galaxies. Like the paper he wrote with Dicke, it went out in the mail in early March; unlike the other paper, it would later receive major revisions to accommodate the 3 K discovery. By the time this paper ran in the Astrophysical Journal, in November 1965, the Princeton and Bell Labs articles had appeared in print, and many of Peebles's peers were rapidly moving past the Why-take-cosmology-seriously? and into the How-exactly-does-this-cosmology-thing-work? phase.
Peebles showed them how: Look at the math. Look at the prediction; then look at the observation. See how they match? During one presentation at an American Astronomical Society meeting, Peebles was pacing back and forth before a blackboard, taking long strides and waving his chalk, when an astrophysicist in the audience called out, "One can make any point at all with a little slapdash arithmetic!"
Peebles spun toward the voice, smiling broadly. "My arithmetic may seem slapdash," he said with a flourish, never breaking stride, "but I can assure you it is impeccable."*
"The radiation," Peebles explained in one of his many papers during this period, "performs the great service of defining the epoch at which the galaxies can start to form." That epoch occurred when the temperature of the primeval fireball fell below 4000 K. At that point the electrons and protons that had been ricocheting independently since the first instants of the universe recombined to form atoms of matter. This matter now took on a "life" of its own and decoupled from the radiation—the fossil radiation that survived today as the cosmic microwave background. And although that background seemed to be as uniform—as homogeneous—as theory had predicted, it couldn't be e
ntirely uniform. It had to contain the irregularities, or inhomogeneities, that identified the concentrations of matter that existed at the moment that matter and radiation decoupled and went their separate ways—the matter that through gravitational interactions would have grown into the large-scale "distributions of mass and size" that we see today: "galaxies, and clusters of galaxies, and the material within galaxies," including us.
The universe was simple. It just couldn't be perfectly simple. Yet to the radio telescopes at the time, the microwave background was absolutely uniform; it lacked the inhomogeneities that had to be there in order for us to be here. Eventually, astronomers or physicists wanting to test the Big Bang theory would have to develop instruments sensitive enough to detect those subtle irregularities in the background. In the meantime Peebles would proceed, as always, with caution.
Having already made the movie of the universe, Peebles now wrote the book. In the fall semester of 1969, after returning from Caltech to Princeton, Peebles taught a graduate course on cosmology. His colleague John Archibald Wheeler—legendary theorist, Princeton fixture since 1938, longtime collaborator of Einstein's—suggested that Peebles use the course as the basis for a text, but Peebles demurred. So Wheeler began showing up at Peebles's lectures. Peebles would be pacing at the blackboard, enthusing in his usual wingspan-enhanced fashion. Wheeler would be sitting in the back, taking notes. Thoroughly unnerving Peebles. After class, it only got worse. Wheeler would present Peebles with the notes, written in perfect penmanship. The "blackmail"—Peebles's word—worked.
Wheeler, he soon saw, was right. If cosmology was in fact in the process of graduating from speculation to science, from metaphysics to physics, then it deserved a textbook. It needed a textbook, if a new generation was going to investigate cosmology properly. The cosmology texts still in use were the ones that Peebles had consulted as a graduate student while preparing for his general examinations; they were cobwebbed with decades-old theory of the kind that valued the simplicity of the math over the relationship to observations, if only because, before the advent of radio astronomy, those observations couldn't be made.
Peebles conceived Physical Cosmology as the first full-tength examination of the physics of the early universe. Such was the state of knowledge that he could explain the entire field in 282 pages. "The great goal now," Peebles wrote in the introduction, "is to become more familiar with the Universe, to learn whether any of these pictures may be a reasonable approximation, and if so how the approximation may be improved." He kept in mind a conversation he'd had with the physicist Philip Morrison shortly after the discovery of the radiation. The two of them were standing in a crowded room. "You measure the level of noise in this room," Morrison said, "and convert that into a thermal temperature and you'll get an absurd answer. How do you know you aren't doing the same thing here?" Peebles's answer: He didn't know. Nobody knew. Although the discovery of the 3 K radiation had provided his career with a new trajectory and inspired the book, he discussed the microwave background only as a "candidate" for the primeval fireball.
Peebles wouldn't even venture into the area of research that had been occupying so much of his time recently—"the very broad topic of galaxy formation, and the presumably related task of understanding irregularities of all sorts in an expanding universe." He could describe the behavior of clusters of galaxies but not the behavior of galaxies themselves. Some papers he was seeing—for instance, Mort Roberts's radio observations of Andromeda—were even indicating that the rotation curves of individual galaxies might be flat. Even though he'd included galaxy formation in the lectures themselves, he mostly omitted the topic from his book. As provisional as much of the cosmological physics in the rest of the book might be, the knowledge of individual galaxy formation was so raw that he decided it wasn't yet worth committing to a bound volume.
So Peebles was hardly surprised when a Princeton colleague, the astronomer Jeremiah Ostriker, stopped by his office to say that he couldn't make sense of the behavior of the Milky Way. Ostriker had seen the N-body simulations that Peebles loved to demonstrate, and he'd consulted with Peebles on the resulting paper. (With the help of a graduate student, Peebles had gotten the N up to 2,000.) Ostriker had been working on rotating celestial objects since he was a graduate student at Cambridge; he had written his thesis on rotating stars. Scientists had known since the nineteenth century that if you rotated an initially spherical liquid drop it would become oblate, increasingly so, and eventually compress into a bar shape. Ostriker had treated stars as liquid drops—as compressible objects—and found that they, too, would become oblate over time. Recently, he told Peebles, he had looked at a rendering of the Milky Way—a flat disk like the other spiral galaxies that astronomers had been collecting by the thousands. He could see at a glance that it should have become bar-shaped or broken up into two galaxies after one rotation. Yet by now the Milky Way was old enough to have completed a dozen rotations.
"There's something wrong here," Ostriker said.
Peebles agreed. He began work on creating an N-body simulation for the Milky Way. He laid the points into a spiral shape and set it rotating. Sure enough, it wobbled catastrophically during its first 200-million-year rotation. He and Ostriker would need something else to stabilize it: a surrounding mass of something to hold it together gravitationally. Something you couldn't see with a telescope—at least not yet—but something that had to be there. Something that Peebles and Ostriker would now add to the computer program.
For the first simulation of the rotation of the galaxy with this missing component, the amount of mass wouldn't matter; it would simply serve as a basis for further comparison. Ostriker and Peebles would surround the visible galaxy with this mass and see what happened. If they ran the simulation and the rotating disk stabilized, they'd shrink the halo and keep shrinking it until the disk destabilized. If they ran the simulation and the rotating disk didn't stabilize, they'd expand the halo and keep expanding it until the disk did.
Peebles shrugged. "Just choose a halo."
They did—a large one, one that swallowed a huge portion of the visible galaxy. Once again the Milky Way wobbled and gyrated. So they tried a larger halo. Another unstable galaxy. Another halo, another unstable galaxy. And another, and another.
This result wasn't entirely surprising to Peebles. The problem of "missing mass" had been shadowing astronomy for decades, for almost as long as astronomers had known of the existence of galaxies. But the problem had always related to clusters of galaxies. In 1933 the Swiss-born astrophysicist Fritz Zwicky, working at Caltech, studied eight galaxies in the Coma Cluster, comparing the mass he derived from their velocities relative to one another with the mass he expected just judging from appearances. His conclusion was that the density of mass had to be four hundred times as large as what the luminosity alone suggested.* If astronomers couldn't resolve this discrepancy, he wrote in a Swiss journal, "we would arrive at the astonishing conclusion" that the density of luminous matter in Coma must be minuscule compared with the density of some sort of dunkle —or dark —Materie—matter. Three years later, the astronomer Sinclair Smith published an article in the Astrophysical Journal about a similar pattern he'd noticed in the Virgo Cluster, suggesting the presence of "a great mass of internebular material within the cluster." That same year, Edwin Hubble addressed the problem in his landmark book The Realm of the Nebulae: "The discrepancy seems to be real and is important."
"This discrepancy is so great," Zwicky wrote in 1937, "that a further analysis of the problem is in order."
Further analyses came, but only sporadically and inconclusively. In science, progress often follows a self-fUlfilling logic: You work on the problems that either have the best chance of yielding conclusions or are most in need of them. Astronomy in the post-Hubble, galaxies-aplenty era was lousy with such problems. The motions of poorly understood objects (galaxies) in possibly coincidental formations (clusters) was not one of them. Peebles himself had regarded the missing-mass pr
oblem as one of those topics you discussed only if you were shooting the breeze during a coffee break, like the question of what came before the universe.
The rise of cosmology as a real science in the late 1960s, however, suddenly made the missing-mass problem more pressing. If you considered the evolution of the universe on the largest scales—as Peebles had, in Physical Cosmology—you couldn't exactly ignore the behavior of the largest structures in the universe, galaxy clusters. His conclusion there, however, could only echo Zwicky and Hubble's plea more than three decades earlier: "We urgently need comparable data."*
Now, though, he was seeing a similar problem in his simulations of individual galaxies, and he had begun seeing the same pattern in papers claiming flat rotation curves in galaxy after galaxy. What if missing mass wasn't a problem just in galaxy clusters? What if it was a problem in individual galaxies, too? What if it was the same problem?
Peebles and Ostriker kept giving their galaxy a larger and larger halo. Only when they had simulated an invisible halo with roughly the same mass as the visible parts of the galaxy—the disk of stars and gas, of central bulge and spiral arms—did the system stabilize. In 1973 Ostriker and Peebles published a paper arguing that "the halo masses of our Galaxy and of other spiral galaxies exterior to the observed disks may be extremely large."
Over the following year, working with a postdoc, they wrote a second paper on the subject. Instead of running hypothetical models on their computers, they analyzed observations that astronomers had already made. They examined the data on individual galaxies. They examined the data from binary galaxies—pairs of galaxies where each galaxy closely interacts gravitationally with the other. They examined the data from satellites of galaxies—dwarf elliptical galaxies that orbit large spiral galaxies. And when they were done, they collected all the data into one comprehensive table, compiled their analysis, and made, in the opening sentence of a second paper, a deceptively simple declaration: "There are reasons, increasing in number and quality, to believe that the masses of ordinary galaxies may have been underestimated by a factor of 10 or more."