Until recently, most astronomers believed that the universe had entered a very boring middle age. According to this paradigm, the early history of the universe--that is, until about six billion years after the big bang--was an era of cosmic fireworks: galaxies collided and merged, powerful black holes sucked in huge whirlpools of gas, and stars were born in unrivaled profusion. In the following eight billion years, in contrast, galactic mergers became much less common, the gargantuan black holes went dormant, and star formation slowed to a flicker. Many astronomers were convinced that they were witnessing the end of cosmic history and that the future held nothing but the relentless expansion of a becalmed and senescent universe.

In the past few years, however, new observations have made it clear that the reports of the universe's demise have been greatly exaggerated. With the advent of new space observatories and new instruments on ground-based telescopes, astronomers have detected violent activity occurring in nearby galaxies during the recent past. (The light from more distant galaxies takes longer to reach us, so we observe these structures in an earlier stage of development.) By examining the x-rays emitted by the cores of these relatively close galaxies, researchers have discovered many tremendously massive black holes still devouring the surrounding gas and dust. Furthermore, a more thorough study of the light emitted by galaxies of different ages has shown that the star formation rate has not declined as steeply as once believed.

The emerging consensus is that the early universe was dominated by a small number of giant galaxies containing colossal black holes and prodigious bursts of star formation, whereas the present universe has a more dispersed nature--the creation of stars and the accretion of material into black holes are now occurring in a large number of medium-size and small galaxies. Essentially, we are in the midst of a vast downsizing that is redistributing cosmic activity.

Deep-Field Images

TO PIECE TOGETHER the history of the cosmos, astronomers must first make sense of the astounding multitude of objects they observe. Our most sensitive optical views of the universe come from the Hubble Space Telescope. In the Hubble Deep Field studies--10-day exposures of two tiny regions of the sky observed through four different wavelength filters--researchers have found thousands of distant galaxies, with the oldest dating back to about one billion years after the big bang. A more recent study, called the Hubble Ultra Deep Field, has revealed even older galaxies.

Obtaining these deep-field images is only the beginning, however. Astronomers are seeking to understand how the oldest and most distant objects evolved into present-day galaxies. It is somewhat like learning how a human baby grows to be an adult. Connecting the present with the past has become one of the dominant themes of modern astronomy.

A major step in this direction is to determine the cosmic stratigraphy--which objects are in front and which are more distant--among the thousands of galaxies in a typical deep-field image. The standard way to perform this task is to obtain a spectrum of each galaxy in the image and measure its redshift. Because of the universe's expansion, the light from distant sources has been stretched, shifting its wavelength toward the red end of the spectrum. The more the light is shifted to the red, the farther away the source is and thus the older it is. For example, a redshift of one means that the wavelength has been stretched by 100 percent--that is, to twice its original size. Light from an object with this redshift was emitted about six billion years after the big bang, which is less than half the current age of the universe. In fact, astronomers usually talk in terms of redshift rather than years, because redshift is what we measure directly.

Obtaining redshifts is a practically foolproof technique for reconstructing cosmic history, but in the deepest of the deep-field images it is almost impossible to measure redshifts for all the galaxies. One reason is the sheer number of galaxies in the image, but a more fundamental problem is the intrinsic faintness of some of the galaxies. The light from these dim objects arrives at a trickle of only one photon per minute in each square centimeter. And when observers take a spectrum of the galaxy, the diffraction grating of the spectrograph disperses the light over a large area on the detector, rendering the signal even fainter at each wavelength.

In the late 1980s a team led by Lennox L. Cowie of the University of Hawaii Institute for Astronomy and Simon J. Lilly, now at the Swiss Federal Institute of Technology in Zurich, developed a novel approach to avoid the need for laborious redshift observations. The researchers observed regions of the sky with filters that selected narrow wavebands in the ultraviolet, green and red parts of the spectrum and then measured how bright the galaxies were in each of the wavebands [see box on page 63]. A nearby star-forming galaxy is equally bright in all three wavebands. The intrinsic light from a star-forming galaxy has a sharp cutoff just beyond the ultraviolet waveband, at a wavelength of about 912 angstroms. (The cutoff appears because the neutral hydrogen gas in and around the galaxy absorbs radiation with shorter wavelengths.) Because the light from distant galaxies is shifted to the red, the cutoff moves to longer wavelengths; if the redshift is great enough, the galaxy's light will not appear in the ultraviolet waveband, and if the redshift is greater still, the galaxy will not be visible in the green waveband either.

Thus, Cowie and Lilly could separate star-forming galaxies into broad redshift intervals that roughly indicated their ages. In 1996 Charles C. Steidel of the California Institute of Technology and his collaborators used this technique to isolate hundreds of ancient star-forming galaxies with redshifts of about three, dating from about two billion years after the big bang. The researchers confirmed many of the estimated redshifts by obtaining very deep spectra of the galaxies with the powerful 10-meter Keck telescope on Mauna Kea in Hawaii.

Once the redshifts of the galaxies have been measured, we can begin to reconstruct the history of star formation. We know from observations of nearby galaxies that a small number of high-mass stars and a larger number of low-mass stars usually form at the same time. For every 20 sunlike stars that are born, only one 10-solar-mass star (that is, a star with a mass 10 times as great as the sun's) is created. High-mass stars emit ultraviolet and blue light, whereas low-mass stars emit yellow and red light. If the redshift of a distant galaxy is known, astronomers can determine the galaxy's intrinsic spectrum (also called the rest-frame spectrum). Then, by measuring the total amount of rest-frame ultraviolet light, researchers can estimate the number of high-mass stars in the galaxy.

Because high-mass stars live for only a few tens of millions of years--a short time by galactic standards--their number closely tracks variations in the galaxy's overall star formation rate. As the pace of star creation slows, the number of high-mass stars declines soon afterward because they die so quickly after they are born. In our own Milky Way, which is quite typical of nearby, massive spiral galaxies, the number of observed high-mass stars indicates that stars are forming at a rate of a few solar masses a year. In high-redshift galaxies, however, the rate of star formation is 10 times as great.

When Cowie and Lilly calculated the star formation rates in all the galaxies they observed, they came to the remarkable conclusion that the universe underwent a veritable baby boom at a redshift of about one. In 1996 Piero Madau, now at the University of California at Santa Cruz, put the technique to work on the Hubble Deep Field North data, which were ideal for this approach because of the very precise intensity measurements in four wavebands. Madau combined his results with those from existing lower-redshift optical observations to refine the estimates of the star formation history of the universe. He inferred that the rate of star formation must have peaked when the universe was about four billion to six billion years old. This result led many astronomers to conclude that the universe's best days were far behind it.

An Absorbing Tale

ALTHOUGH MADAU'S ANALYSIS of star formation history was an important milestone, it was only a small part of the story. Galaxy surveys using optical telescopes cannot detect every source in the early universe. The more distant a galaxy is, the more it suffers from cosmological redshifting, and at high enough redshifts, the galaxy's rest-frame ultraviolet and optical emissions will be stretched into the infrared part of the spectrum. Furthermore, stars tend to reside in very dusty environments because of the detritus from supernova explosions and other processes. The starlight heats up the dust grains, which then reradiate this energy at far-infrared wavelengths. For very distant sources, the light that is absorbed by dust and reradiated into the far-infrared is shifted by the expansion of the universe to submillimeter wavelengths. Therefore, a bright source of submillimeter light is often a sign of intense star formation.

Until recently, astronomers found it difficult to make submillimeter observations with ground-based telescopes, partly because water vapor in the atmosphere absorbs signals of that wavelength. But those difficulties were eased with the introduction of the Submillimeter Common-User Bolometer Array (SCUBA), a camera that was installed on the James Clerk Maxwell telescope on Mauna Kea in 1997. (Located at a height of four kilometers above sea level, the observatory is above 97 percent of the water in the atmosphere.) Several teams of researchers, one of which I led, used SCUBA to directly image regions of the sky with sufficient sensitivity and area coverage to discover distant, exceptionally luminous dust-obscured sources. Because the resolution is fairly coarse, the galaxies have a bloblike appearance [see illustration above]. They are also relatively rare--even after many hours of exposure, few sources appeared on each SCUBA image--but they are among the most luminous galaxies in the universe. It is sobering to realize that before SCUBA became available, we did not even know that these powerful, distant systems existed! Their star formation rates are hundreds of times greater than those of present-day galaxies, another indication that the universe used to be much more exciting than it is now.

Finding all this previously hidden star formation was revolutionary, but might the universe be covering up other violent activity? For example, gas and dust within galaxies could also be obscuring the radiation emitted by the disks of material whirling around supermassive black holes (those weighing as much as billions of suns). These disks are believed to be the power sources of quasars, the prodigiously luminous objects found at high redshifts, as well as the active nuclei at the centers of many nearby galaxies. Optical studies in the 1980s suggested that there were far more quasars several billion years after the big bang than there are active galactic nuclei in the present-day universe. Because the supermassive black holes that powered the distant quasar activity cannot be destroyed, astronomers presumed that many nearby galaxies must contain dead quasars--black holes that have exhausted their fuel supply.

These dormant supermassive black holes have indeed been detected through their gravitational influence. Stars and gas continue to orbit around the holes even though little material is swirling into them. In fact, a nearly dormant black hole resides at the center of the Milky Way. Together these results led scientists to develop a scenario: most supermassive black holes formed during the quasar era, consumed all the material surrounding them in a violent fit of growth and then disappeared from optical observations once their fuel supply ran out. In short, quasar activity, like star formation, was more vigorous in the distant past, a third sign that we live in relatively boring times.

This scenario, however, is incomplete. By combining x-ray and visible-light observations, astronomers are now revisiting the conclusion that the vast majority of quasars died out long ago. X-rays are important because, unlike visible light, they can pass through the gas and dust surrounding hidden black holes. But x-rays are blocked by the earth's atmosphere, so researchers must rely on space telescopes such as the Chandra and XMM-Newton X-ray observatories to detect black hole activity. In 2000 a team consisting of Cowie, Richard F. Mushotzky of the NASA Goddard Space Flight Center, Eric A. Richards, then at Arizona State University, and I used the Subaru telescope at Mauna Kea to identify optical counterparts to 20 x-ray sources found by Chandra in a survey field. We then employed the 10-meter Keck telescope to obtain the spectra of these objects.

Our result was quite unexpected: many of the active supermassive black holes detected by Chandra reside in relatively nearby, luminous galaxies. Modelers of the cosmic x-ray background had predicted the existence of a large population of obscured supermassive black holes, but they had not expected them to be so close at hand! Moreover, the optical spectra of many of these galaxies showed absolutely no evidence of black hole activity; without the x-ray observations, astronomers could never have discovered the supermassive black holes lurking in their cores.

This research suggests that not all supermassive black holes were formed in the quasar era. These mighty objects have apparently been assembling from the earliest times until the present. The supermassive black holes that are still active, however, do not exhibit the same behavioral patterns as the distant quasars. Quasars are voracious consumers, greedily gobbling up the material around them at an enormous rate. In contrast, most of the nearby sources that Chandra detected are more moderate eaters and thus radiate less intensely. Scientists have not yet determined what mechanism is responsible for this vastly different behavior. One possibility is that the present-day black holes have less gas to consume. Nearby galaxies undergo fewer collisions than the distant, ancient galaxies did, and such collisions could drive material into the supermassive black holes at the galactic centers.

Chandra had yet another secret to reveal: although the moderate x-ray sources were much less luminous than the quasars--generating as little as 1 percent of the radiation emitted by their older counterparts--when we added up the light produced by all the moderate sources in recent times, we found the amount to be about one tenth of that produced by the quasars in early times. The only way this result could arise is if there are many more moderate black holes active now than there were quasars active in the past. In other words, the contents of the universe have transitioned from a small number of bright objects to a large number of dimmer ones. Even though supermassive black holes are now being built smaller and cheaper, their combined effect is still potent.

Star-forming galaxies have also undergone a cosmic downsizing. Although some nearby galaxies are just as extravagant in their star-forming habits as the extremely luminous, dust-obscured galaxies found in the SCUBA images, the density of ultraluminous galaxies in the present-day universe is more than 400 times lower than their density in the distant universe. Again, however, smaller galaxies have taken up some of the slack. A team consisting of Cowie, Gillian Wilson, now at the California Institute of Technology's Spitzer Science Center, Doug J. Burke, now at the Harvard-Smithsonian Center for Astrophysics, and I has refined the estimates of the universe's luminosity density by studying high-quality images produced with a wide range of filters and performing a complete spectroscopic follow-up. We found that the luminosity density of optical and ultraviolet light has not changed all that much with cosmic time. Although the overall star formation rate has dropped in the second half of the universe's lifetime because the monstrous dusty galaxies are no longer bursting with stars, the population of small, nearby star-forming galaxies is so numerous that the density of optical and ultraviolet light is declining rather gradually. This result gives us a much more optimistic outlook on the continuing health of the universe.

Middle-Aged Vigor

THE EMERGING PICTURE of continued vigor fits well with cosmological theory. New computer simulations suggest that the shift from a universe dominated by a few large and powerful galaxies to a universe filled with many smaller and meeker galaxies may be a direct consequence of cosmic expansion. As the universe expands, galaxies become more separated and mergers become rarer. Furthermore, as the gas surrounding galaxies grows more diffuse, it becomes easier to heat. Because hot gas is more energetic than cold gas, it does not gravitationally collapse as readily into the galaxy's potential well. Fabrizio Nicastro of the Harvard-Smithsonian Center for Astrophysics and his co-workers have recently detected a warm intergalactic fog through its absorption of ultraviolet light and x-rays from distant quasars and active galactic nuclei. This warm fog surrounds our galaxy in every direction and is part of the Local Group of galaxies, which includes the Milky Way, Andromeda and 30 smaller galaxies. Most likely this gaseous material was left over from the galaxy formation process but is too warm to permit further galaxy formation to take place.

Small galaxies may lie in cooler environments because they may not have heated their surrounding regions of gas to the same extent that the big galaxies did through supernova explosions and quasar energy. Also, the small galaxies may have consumed less of their surrounding material, allowing them to continue their more modest lifestyles to the present day. In contrast, the larger and more profligate galaxies have exhausted their resources and are no longer able to collect more from their environments. Ongoing observational studies of the gaseous properties of small, nearby galaxies may reveal how they interact with their environments and thus provide a key to understanding galactic evolution.

But a crucial part of the puzzle remains unsolved: How did the universe form monster quasars so early in its history? The Sloan Digital Sky Survey, a major astronomical project to map one quarter of the entire sky and measure distances to more than a million remote objects, has discovered quasars that existed when the universe was only one sixteenth of its present age, about 800 million years after the big bang. In 2003 Fabian Walter, then at the National Radio Astronomy Observatory, and his collaborators detected the presence of carbon monoxide in the emission from one of these quasars; because carbon and oxygen could have been created only from the thermonuclear reactions in stars, this discovery suggests that a significant amount of star formation occurred in the universe's first several hundred million years. Recent results from the Wilkinson Microwave Anisotropy Probe, a satellite that studies the cosmic background radiation, also indicate that star formation began just 400 million years after the big bang.

Furthermore, computer simulations have shown that the first stars were most likely hundreds of times as massive as the sun. Such stars would have burned so brightly that they would have run out of fuel in just a few tens of millions of years; then the heaviest stars would have collapsed to black holes, which could have formed the seeds of the supermassive black holes that powered the first quasars. This explanation for the early appearance of quasars may be bolstered by the further study of gamma-ray bursts, which are believed to result from the collapse of very massive stars into black holes. Because gamma-ray bursts are the most powerful explosions in the universe since the big bang, astronomers can detect them at very great distances. In November 2004 NASA launched the Swift Gamma-Ray Burst Mission, a 250-million satellite with three telescopes designed to observe the explosions and their afterglows in the gamma-ray, x-ray, ultraviolet and optical wavelengths. In the two years since its launch, Swift has identified a number of gamma-ray bursts. The most exciting of these was the September 2005 discovery of an explosion that took place only 900 million years after the big bang. The hope is that this is just the first of many such detections stretching to even greater distances, thereby providing scientists with a much better understanding of how collapsing stars could have started the growth of supermassive black holes in the early universe.

In comic books, Superman looked through walls with his x-ray vision. Astronomers have now acquired a similar ability with the Chandra and XMM-Newton observatories and are making good use of it to peer deep into the dust-enshrouded regions of the universe. What is being revealed is a dramatic transition from the mighty to the meek. The giant star-forming galaxies and voracious black holes of the universe's past are now moribund. A few billion years from now, the smaller galaxies that are active today will have consumed much of their fuel, and the total cosmic output of radiation will decline dramatically. Even our own Milky Way will someday face this same fate. As the cosmic downsizing continues, the dwarf galaxies--which hold only a few million stars each but are the most numerous type of galaxy in the universe--will become the primary hot spots of star formation. Inevitably, though, the universe will darken, and its only contents will be the fossils of galaxies from its glorious past. Old galaxies never die, they just fade away.

THE AUTHOR

AMY J. BARGER studies the evolution of the universe by observing some of its oldest objects. She is an associate professor of astronomy at the University of WisconsinMadison and also holds an affiliate graduate faculty appointment at the University of Hawaii at Manoa. Barger earned her Ph.D. in astronomy in 1997 at the University of Cambridge, then did postdoctoral research at the University of Hawaii Institute for Astronomy. An observational cosmologist, she has explored the high-redshift universe using the Chandra X-ray Observatory, the Hubble Space Telescope, and the telescopes on Kitt Peak in Arizona and on Mauna Kea in Hawaii.