EYE ON THE SKY: Perched on a 9,200-foot peak in the Sacramento Mountains of south-central New Mexico, far from any large cities, the Sloan Digital Sky Survey's 2.5-meter telescope peers into a night sky that is among the darkest in the United States. The atmosphere at the site is remarkably free of water vapor or pollutants that degrade celestial images.
When the most distant object in the known universe first appeared on a photograph taken through the Sloan Digital Sky Survey telescope at Apache Point, New Mexico, the astronomers on watch had no idea what they'd stumbled upon. The dim speck was barely visible under extreme magnification and looked utterly unremarkable amid several thousand points of light scattered across a swath of the northern sky. The treasure in the star field remained hidden until the digital photo was scrutinized at the Fermi National Accelerator Laboratory near Chicago by a cluster of powerful computers equipped with advanced image-processing software that notes the color and shape of each point of light.

Usually, a blob of light denotes a galaxy and a pinpoint of light means a star. But the computer indicated this pinpoint was too red to be an ordinary star. So the Sky Survey's astronomers took a closer look with the powerful Keck telescope atop Mauna Kea, Hawaii. They made a spectrogram that provided a more precise breakdown of the light's constituent colors, revealing what it is made of and where in space it is actually located. "As soon as the spectrum came up on the screen," recalls Princeton astrophysicist Michael Strauss, "it was obvious to everyone in the control room what we'd found." The speck of light was a distant quasar—a celestial powerhouse with a giant black hole at its center that sucks in primordial matter and heats it to an incandescence exceeding that of a trillion stars. This quasar's light began the 12-billion-year journey from the outermost edge of the universe to Earth less than 1 billion years after the Big Bang, making it the oldest heavenly object humans have ever seen. "We were," says Strauss, with a touch of nonchalance, "pretty excited."

No wonder. Finding record-breaking quasars—the Sky Survey has now snagged the four furthest away from Earth—is already giving astronomers invaluable information about the epoch when the first quasars and galaxies lit up and burned through the fog of opaque gas that filled the universe following the Big Bang. The project, run by a consortium of labs and universities that includes Fermilab, Princeton, the University of Chicago, and Johns Hopkins, has also discovered thousands of new asteroids in our own solar system and located more than 20 brown dwarfs—objects that are bigger than a planet but smaller than a star and which may be a key to understanding the process of planet formation.


COSMIC DISK: NGC 4437, a spectacular edge-on spiral galaxy 75 million light-years from Earth, is among the early batch of celestial objects included in the Sky Survey. The dark lanes are clouds of gas and dust in the galaxy's outer arms. The background objects are a mix of stars and other galaxies.
Photograph courtesy of Sloan Digital Sky Survey
And these discoveries are just the beginning. By the time the project is complete some four years from now, the Sky Survey's astronomers will have assembled by far the greatest map ever conceived. It will span one quarter of the known universe—everything not obscured by the bulk of Earth below or the broad, dust-clogged band of the Milky Way above—plotting the precise positions of a million galaxies, 100,000 quasars, and countless other celestial objects. More important, it will do so in three dimensions. Unlike the flat star charts astronomers have been constructing since the time of the ancient Mesopotamians, this map will not just show where on the sky a galaxy or a quasar lies but how far away it is. The Sky Survey will be a dynamic model that can provide a better sense of the overall structure of the universe, as well as the contours and topography of deep space. The model will contain some 15 terabytes of astronomical data, which in sheer volume will rival the amount of information stored in the entire Library of Congress.

This sprawling virtual universe will be available electronically. Astronomers will be able to sit at their desks and fly across billions of light-years of space. And because the Fermilab computers will keep track not only of each object's position but also its color, shape, and relationship to every other object, researchers will be able to look at any subset of objects they choose. For example, they could focus on bluish galaxies, where lots of new stars are forming, or on reddish ones, where the stars have grown old. Or they could look at clusters of both types of galaxies.

While dozens of scientists helped to create and refine the Sky Survey, the project is primarily the brainchild of Princeton astrophysicist Jim Gunn. In recent years, Gunn has gained a reputation as one of astronomy's great instrument makers: He led the team that built a digital camera for the 200-inch Hale Telescope on Palomar Mountain in southern California and was in charge of designing a wide-field camera for the Hubble Space Telescope. And he had long dreamed of using the most sophisticated digital technology to provide astronomers with a survey of the entire sky.


150,000 points of light: To make sense of the enormous amount of raw data collected by the Sky Survey telescope, scientists have created a visual model—a kind of three-dimensional planetarium—that allows them to fly through the universe and see the distribution of galaxies from any vantage. This particular image, generated by Mark SubbaRao of the University of Chicago, shows the view looking back toward the Milky Way from a location 500 million light-years away in the direction of the constellation Orion. The 150,000 galaxies shown here are depicted far larger than they would actually appear. Various shapes and colors represent 10 different types of galaxies, as defined by their spectral characteristics. The most distant galaxies appear as single, color-coded dots. With such modeling, astronomers can freely explore the large-scale structure of the cosmos for the first time.
Photograph courtesy of Mark SubbaRao/University of Chicago/Sloan Digital Sky Survey.

When his vision for the project began to take shape more than a decade ago, the best existing sky survey was the Palomar, a set of nondigital photographs taken in the 1950s. The Palomar survey covers the northern sky, and astronomers still use it as a standard reference. But while photographic plates were state of the art in the 1950s, technical weaknesses have since rendered them hopelessly out of date. For one thing, photographic plates trap at most about 1 percent of the light that falls on them. They're also inconsistent; one plate might be more or less sensitive than the next, and patches on the same plate can vary. A further drawback is that the information on the plates has to be processed by hand. To measure distances between galaxies or stars, an astronomer has to get out a ruler.


SKY MASKS: Aluminum disks by the score sit in racks awaiting their turn at the telescope. After an initial image of a swath of sky is analyzed by the computer, machinists drill 640 holes in a disk, with each hole corresponding to the position of a selected galaxy, quasar, or star. Eventually each individual swath of sky will be photographed again, with a corresponding disk clamped on the telescope to isolate the selected points of light for refined spectrographic analysis.
Gunn and his colleagues set out to update the Palomar survey using telescopic photographs taken with charge-coupled devices (CCDs), digitized light sensors that generate a celestial database, which can be prodded and probed with much greater precision. The team also decided they needed a telescope that could cover an unusually wide expanse of the heavens. Most telescopes have an extremely narrow field of view, taking in only a few arc minutes of sky at a time (for reference, the full moon is a half-degree, or 30 arc minutes, across). Photographing the entire sky would take decades at that rate, so the Sky Survey team designed a wide-field scope that is a modest 2.5 meters across (the most powerful telescopes span eight or 10 meters) but which can capture seven square degrees—36 full moons' worth—of sky at once.

The wide-field scope is equipped with a camera that boasts an array of 30 CCDs, each about two inches square, assembled mostly by Gunn himself in a clean room built in the basement of Peyton Hall, headquarters of Princeton's Astrophysical Sciences department. The job took five years and cost about $4 million. Once Gunn finished building the extremely delicate instrument, he faced the problem of getting it to the Apache Point observatory in one piece. "For a while, I considered driving it out there myself," he says. "But we finally found a moving company called Robert's White Glove Express, which specialized in transporting ultrafragile things. We insured it, of course, but that was an exercise in futility; we could recover the $4 million, but not the five years." The camera made it onto the mountain without incident, and the Sloan Digital Sky Survey went into formal operation about a year ago.

Gunn's precious camera—the heart of the Sky Survey Telescope—has since proven to be a light-gathering device of unprecedented sensitivity. Each of the 30 CCDs is made up of more than 4 million picture elements, which release electrons as light is absorbed. The electrons in turn are amplified into electronic signals that are digitized. Each eight-minute exposure taken by the camera produces an image twice as sharp as those taken for the Palomar sky survey; the camera also reveals celestial objects where the light is one tenth as bright.


ASSEMBLY LINE: On a good night, astronomers will install six or seven aluminum disks on the telescope in rapid succession. The plates are put in cartridges, with each of the 640 holes plugged with fiber-optic cables that connect to a spectrograph. The cartridge on the far right has been opened to show the disk (on top) and the cables.
What's more, the images taken by Gunn's camera are in color. The CCDs are arranged in five rows, each of which is equipped with a different colored filter. So the camera records the brightness of light in a palette of five basic colors: ultraviolet, green, red, near infrared, and far infrared. The colors of heavenly objects carry vital information about their temperature and composition. In general the hottest objects glow blue-violet; the coolest, red. That's why star-forming galaxies tend to appear blue; young stars tend to burn hot, while aging stars usually become progressively redder. Redness can also mean that an object's light is being partially obscured—by interstellar dust, for example, or, as in the case of record-breaking quasars, by clouds of intergalactic hydrogen gas.

These basic color images are useful for taking a census of different types of celestial objects, but they lack one essential element: depth. One star may look brighter than another, but that doesn't mean it's closer; it might just be brighter after all. The same goes for galaxies. Luckily, there is a convenient way to calculate such distances. In a generally expanding universe, the farther away an object is from our home galaxy, the Milky Way, the faster it appears to be receding from us.

And speed, it turns out, is easy to measure: The light from a receding object is stretched out as it travels away from us and thus looks redder than it actually is (an approaching object would look bluer). The effect is analogous to the way a police siren sounds higher in pitch when it's approaching, then drops abruptly when it passes by. It was by noting this redshifting of the light of distant galaxies, in fact, that Edwin Hubble discovered the expanding universe in the 1920s. But five-color images can only hint at such redshift patterns. Measuring redshift precisely requires spectroscopy, smearing the light from a particular celestial object into its full spectrum of colors.

In order to make such measurements, the Sky Survey switches into a second mode of operation. When a nightly run of the Apache telescope is complete, up to a dozen magnetic tapes bearing some 150 gigabytes' worth of data are sent to Fermilab and processed with custom-designed image-analysis software. If the object is starlike but falls outside the normal color range for stars—as happened with the record-setting quasar—the computer flags it for further study. In every field of view, the computer also picks out the 600 brightest celestial objects—which may include galaxies, stars, and the occasional quasar—and identifies them as targets for spectroscopy.







A RAINBOW OF ANSWERS: The objects in the column at left look similar, but their spectra at right—which indicate how bright they are at each wavelength, or color—unmask their wildly disparate identities. A brilliant quasar at the edge of the visible universe (top) displays an emission spike that has been drastically stretched and reddened by the expansion of the cosmos. A medium-distance galaxy (middle) emits a broad wash of light from myriad stars. A white dwarf star in our own galaxy (bottom) peaks at short, hot wavelengths, with dips of absorption from its hydrogen atmosphere.
Photographs and Spectra courtesy of Brian Wilhite, University of Chicago/SDSS collaboration.
Once the computers have finished analyzing a night's images, the locations of the 600 target objects, as well as 30 or 40 reference objects, are e-mailed to the University of Washington. There, machinists take 30-inch-diameter aluminum disks and drill holes in them that correspond to the location of the objects in the sky. Finally, the plates are shipped back to Apache Point, where they wait for nights that aren't quite crystal-clear—nights that aren't great for imaging with Gunn's CCD camera but which are fine for the more focused task of spectroscopy. On such nights, Gunn's camera is detached from the telescope and the plates are clamped on. Staff astronomers then plug fiber-optic cables into the predrilled holes, allowing the light from the target objects to flow into one of two spectrographs. On average, each exposure takes about 45 minutes, after which astronomers have to come down to the telescope from the control building, about 100 yards away, to change plates. "We've gotten pretty good at doing it quickly," says Scot Kleinman, the senior staff astronomer on site. "We use scooters to zip down to the telescope to save time. Our record—a world record—is seven plates in one night." Each object's distance from Earth, calculated by how much its light has been redshifted, is fed back into the Sky Survey database and linked with the object's celestial coordinates. The magic is that it turns what was a two-dimensional map into 3-D: a universe on every astronomer's desktop.

The completed Sky Survey will trace out the figurative hills and valleys of cosmic structures—the places where galaxies huddle together and the great voids, 100 million light-years across, that lie in between. The existence of these voids first became clear in the 1980s, in surveys of just a few thousand nearby galaxies; the Sky Survey will reveal the location in space of more than 1 million galaxies and let cosmologists see whether clusters and voids permeate the universe. It will also reveal whether there are even bigger structures out there. So far, with 90,000 galaxies in hand, it looks as though the webwork is similar everywhere you look, but astronomers are always prepared for major surprises.

Quasars, the celestial objects that lie farther away than anything humankind has ever seen before, will trace out cosmic structure at the very earliest times after the Big Bang. They'll also provide evidence of how galaxies came to exist in the first place. It clearly took a while after that primordial explosion for clouds of gas to congeal into a form dense enough for stars and quasars to ignite, and the Sky Survey is already prompting astronomers to question some of their assumptions about how that process unfolded. Even if most quasars formed, say, 3 billion years after the Big Bang, pure chance would suggest that you might find an early bloomer at 2 billion. But if you find two or three or four—as the Sky Survey team now has—it's more than just chance, and it gets very tough to explain. "If you'd asked me three years ago whether we'd find quasars this old," says Penn State astronomer Donald Schneider, chair of the Sky Survey's quasar group and the holder of the pre-Sky Survey distance record, "I'd have put the chances at less than 50 percent. Jim Gunn has built an amazing machine."


ASSEMBLY LINE: On a good night, astronomers will install six or seven aluminum disks on the telescope in rapid succession. The plates are put in cartridges, with each of the 640 holes plugged with fiber-optic cables that connect to a spectrograph. The cartridge on the far right has been opened to show the disk (on top) and the cables.
Astronomers can also use distant quasars as a probe of the matter lying in between the distant edges of the universe and Earth. For about a billion years following the Big Bang—an epoch astronomers have dubbed the Dark Ages—a dense fog of hydrogen gas filled the universe. When Gunn was a young graduate student at the California Institute of Technology during the 1960s, he and his colleague Bruce Peterson suggested that the light from any quasars that might have formed during the Dark Ages would have been partially obscured by that dense fog of hydrogen. Eventually, ultraviolet light from quasars and newly formed galaxies would have ionized the hydrogen, converting it into an electrically charged form that is transparent to light. But early on, most hydrogen would still be neutral and relatively opaque. Astronomers have been searching for this so-called Gunn-Peterson effect ever since—and now, based on follow-up observations of the Sky Survey's record-breaking quasar, they've finally seen it. Evidence indicates that the fog of hydrogen lifted during a transitional period that lasted tens or even hundreds of millions of years, during which there were opaque regions in the universe interspersed with bubbles of light and transparent gas.

The Sky Survey is also baring secrets that are much closer to Earth in time and space. Sorting some 2,000 newly found asteroids by color, the Sky Survey astronomers have learned that the asteroid belt between Mars and Jupiter is really two belts side by side, an inner one made mostly of rocky lumps and an outer belt with mostly carbon-rich asteroids. The Sky Survey's exquisitely sharp CCD camera has also revealed a streamer of stars that marks the area where the Milky Way swallowed a neighbor galaxy eons ago. "We're just getting started at looking at the properties of ordinary stars," says Michael Strauss, Gunn's colleague at Princeton. "We've begun finding comets. And we have a couple of objects that we don't really understand at all yet."

All these discoveries have occurred in just one year, with 80 percent of the Sky Survey mapping yet to be done. Once the Sky Survey database is complete, and astronomers begin systematically exploring every nook and cranny of this virtual universe, there is no telling what surprises may emerge.





The Sloan Digital Sky Survey Web site provides background information on the project, plus news releases relating to recent discoveries: www.sdss.org.

For more on how CCD cameras work, see "How Do CCDs Capture Images?" Jon Titus, Test & Measurement World (April 1999), www.tmworld.com/articles/04_1999_CCD.htm.