By Geoff Brumfiel (MA '01)
Opening Photo by Mike Ciesielski

Mapping the stars has never been easy. The Greek astronomer Ptolemy counted one-sixth of the 5,000 stars visible to the naked eye over his lifetime. The renowned 16th-century Danish astronomer Tycho Brahe cataloged just over 700 in the course of his career. More recently, the Lick Survey, conducted at the Lick Observatory in Santa Cruz, California, during the 1950s, accurately recorded the positions of a million galaxies in only a decade. Now, the Sloan Digital Sky Survey housed at Apache Point observatory in Sunspot, New Mexico, has changed mapping forever. "With Sloan," says Alex Szalay, a Hopkins astronomer and Sloan collaborator, "we can do a million galaxies in a night."

Sloan's telescope is able to record this huge amount of information because, unlike previous surveys, Sloan uses digital imaging and high-speed computing to record images of the sky. Building the database has required the cooperation of 11 universities and research institutions from around the world. Since 1992, Johns Hopkins University has been a major contributor to the collaboration, offering both financial support and the expertise of astronomers like Szalay. When the survey is completed in 2005, it will have created a chart of 100 million galaxies, one-quarter of the sky. Szalay and other astrophysicists will use these charts to probe the very beginnings of the cosmos.

The Sloan will do more than answer questions about the origins of our universe--it will forever change the way astronomy is done. When the survey is completed, it will contain information on over 200 million celestial objects. This data, enough to fill 60,000 sets of encyclopedias, will be accessible to researchers worldwide via the Internet. No longer will an astronomer need a costly telescope to look at the stars--all that will be required is a high-speed Internet connection. Astronomy, according to Szalay, is entering the age of the "dial-up sky."

Electrons buzzed quietly about their nuclei. Inert hydrogen drifted aimlessly. The waves of the primordial sea became frozen in time. In the darkness, gravity began to act--pulling the remnants of the waves into crisscrossing tendrils of hydrogen that stretched for billions of light-years in every direction. The tendrils collapsed into galaxy-sized clouds of gas, and thermonuclear reactions began to ignite. One by one, stars were born. After another 8 billion years, the frozen sea of the early universe became a glittering cobweb of hundreds of millions of galaxies that can still be seen today. This cobweb is a snapshot of the turbulent sea of particles that existed nearly 14 billion years ago. By capturing that snapshot, Szalay and others hope to learn about the ratios of those fundamental particles, how the universe came to look the way it does, its eventual fate, and much more.

It is only within the past 100 years that cosmologists have drawn this theoretical picture of the universe's first moments. At the turn of the 1900s, many physicists, including Albert Einstein, believed that the heavens were composed of a homogeneous cloud of stars. But by 1917, a handful theorized that we lived within an "island universe," and that other such "island universes," or galaxies as they are now called, might exist outside our own. During the mid-1920s, astronomer Edwin Hubble looked deep into the Southern California sky and surmised that the Milky Way was just one of many millions of galaxies; in effect, he invented the idea of the universe. What's more, he offered the first evidence for the Big Bang Theory by showing that the cosmos is actually expanding.

Unable to look deeply enough into space to see the great cobwebs, Hubble, who died in 1953, concluded that the galaxies were uniformly distributed throughout the universe. It was not until the mid-1950s that astronomers first saw galactic patterns in the night sky. At the Lick Observatory in 1947, Donald Shane and his graduate student Carl Wirtnanen undertook an ambitious project to chart a million galaxies. Wirtnanen spent seven years carefully scrutinizing over 1,200 photographic plates, identifying, by eye, each galaxy and recording its position in a ledger. When he had finished, his map revealed a sprawling web over 1.5 billion light-years wide. It was the first observational evidence of large-scale galactic structure.

While observational astronomers like Wirtnanen and Shane worked to chart these massive structures, cosmologists were developing theories about how they had been created. One of the leading theories of the time came from the preeminent Soviet physicist Yakov Boris Zel'dovich. During the 1960s, Zel'dovich devoted most of his energy to building the H-bomb for the USSR, but toward the end of the decade he turned his attention to cosmological large-scale structure.

Zel'dovich theorized that large clouds of gas scattered throughout the young universe collapsed into very thin "pancake" structures of galaxies. "He translated it to 'pancakes,' but he actually meant 'crepes,'" says Szalay. "They were really thin, not like the thick American pancakes." This was a "top-down" approach to galaxy formation--the thin crepes of gas formed first, and within them galaxies formed. To Zel'dovich, these crepes held a promise: If he could learn how the sea of particles led to these gigantic structures, then he could use them to probe the first moments after the Big Bang.

It was the beginning of a fruitful collaboration between Zel'dovich and Szalay. "I loved working with him," says Szalay. "He gave me interesting challenges, challenges that were just a little beyond what I could do." Zel'dovich assigned Szalay the task of understanding how the swirling of fundamental particles in the early universe affected the shape and size of these crepes. Szalay spent the next few years working on this problem, and it eventually became the topic of a series of short papers.

"Hungary was a unique place at that time in the '70s," Szalay explains. It was part of the Eastern Bloc, so Soviet physicists like Zel'dovich could travel there without restriction, but its close ties to Western Europe meant that Western physicists were still willing to visit. At conferences, Szalay was exposed to both Eastern and Western theories of the universe's large-scale structure. While most Eastern Bloc physicists like Szalay and Zel'dovich believed in "top down" theories of large-scale structure, Western cosmologists theorized that the universe formed from the "bottom up"--that is, galaxies formed first and then clustered into the structures we see today. "There was bitter animosity for a while between the two groups," says Szalay, "and of course it was played out East against West."

Between 1980 and 1984, cosmologists found a way to reconcile the two theories by extending fundamental particle computations, including those Szalay had completed under Zel'dovich's guidance. These computations showed that the crepes and individual galaxies formed almost simultaneously, proving both "top-down" and "bottom-up" theories correct. Furthermore, the computations helped explain how the roaring sea of the early universe had created patterns in galactic structure. "It opened up the connection between early particle physics and the beautiful universe that we see today," says Szalay. Because Szalay's papers contained some of the only available calculations at the time, he found himself in demand at conferences worldwide. In December 1980, he traveled to Baltimore to attend a conference on cosmology. He spent the next spring touring laboratories and attending meetings across the United States. It was the beginning of a new collaboration for Szalay, between East and West.

For the next decade, Szalay divided his time between Eštvšs University in Budapest and astronomy departments throughout the United States. In 1981, Hopkins professor Gabor Domokos asked Szalay to give a series of talks on particle astrophysics. They struck up a friendship, and throughout the '80s, Szalay's trips to the U.S. would invariably include a talk at the Hopkins Physics and Astronomy Department. Attracted to the many research opportunities at Hopkins, Szalay accepted the invitation to become a full-time professor in 1989.

If Zel'dovich, Szalay, and their Western counterparts were correct, then the large-scale galactic structures seen today were directly related to the Big Bang. Cosmologists believed they could deduce important ratios of fundamental particles, the amount of matter in the cosmos, and even the rate at which the universe expands, from an image of these structures. Creating this picture in the mid- 1980s became the goal of three astronomers--Donald York and Richard Kron of the University of Chicago, and Jim Gunn of Princeton University. Constructing an image of large-scale structure was in itself no small task, but there was another caveat. In order to accurately picture the galactic foam, the image had to be three-dimensional. That meant not only being able to tell the precise position of each galaxy, but its distance from our own galaxy as well.

To determine each galaxy's distance from ours, the survey would also have to measure its spectra, or range of colors. This is because the further a galaxy is from our own, the redder it appears (this is known as the red shift). To take these measurements, the Sloan would need a spectrograph, a special instrument that, like a prism, separates the light from a distant star or galaxy into its individual colors. Most spectrographs could only look at one celestial object at a time; at that rate, to find the spectra of 100 million galaxies would take over 200 years. The only solution was to invent a device that could take numerous spectra simultaneously. Gunn came up with a design for a new type of spectrograph, which used fiber optics to take hundreds of spectra at once. Using two of these spectrographs in tandem, the Sloan would be able to take 640 spectra simultaneously. Like the camera, the spectrographs would use CCDs and contribute to the digital database.

Gunn showed his plans for the spectrographs to Hopkins astronomers Paul Feldman and Alan Uomoto. The design was bold. Each one would need to take 320 spectra simultaneously--more than any other spectrograph in the world. Uomoto was skeptical. "You think, 'maybe we can do that,'" he says, "but it's sort of on the edge."

Uomoto began with a sketch of the spectrograph taken from a grant proposal Gunn had written. "Basically the drawing was all we had to start with, that and the requirements for what the spectrograph should do," he says. From these, Uomoto and Feldman drafted a blueprint of the complex instrument that would combine the ideas of a diffraction grating and a prism, creating a "grism." The goal: to be able to collect light from hundreds of different galaxies and quasars and turn it into spectra at the same time. Fiber optic cables would be used to collect the light from each galaxy. To align the fibers with their galaxies, holes were drilled in an aluminum "plug-plate," each of which corresponded to a galaxy's location at the back of the telescope. These holes were then stuffed with fiber optics, and the plate placed on the base of the telescope. As light from the galaxies came through the telescope, the fiber optics would feed it to the spectrograph and then be photographed by a digital camera. There was little room for error; a mistake as small as one two-thousandth of an inch could ruin the results.

But there were problems of another kind. Over 100 astronomers, most of whom had previously worked in collaborations of only three or four, were expected to coordinate their efforts to make Sloan a reality. "I remember one of our first meetings," Uomoto confides. "We sat down, had some really wonderful arguments, and shortly after we left I realized we had accomplished nothing toward the progress of the project. It was clear to me, at that point, that management was going to be the big problem." To complicate things even further, the astronomers were dispersed at nearly a dozen institutions across the country. While Hopkins was building the spectrographs, the CCD camera was being built at Princeton, the telescope constructed at the University of Washington, and the software to bring it all together written at the University of Chicago. "It was very difficult to get this consortium of universities to work in any coherent way," says Gunn. "There were turf wars; people were determined not to let [go of] their projects."

Throughout the early '90s, the Sloan fell further and further behind schedule and the cost began to grow. It was then that the collaborators realized they had another problem on their hands. It would be impossible to construct the Sloan on its initial meager budget of $25 million. "There were a number of unrealistic estimations," says Szalay. "One was that we would be able to build it with faculty time, so that we only really had to pay for the parts. The other major misunderstanding was the nature of the software."

The software needed to operate the Sloan was unlike anything ground-based astronomy had ever seen before. With its 30 CCDs, and two fiber spectrographs, the Sloan generated enough data to fill 1,000 sets of encyclopedias each day it ran at full capacity. When the project is completed in 2005, it will contain 40 terabytes of data, enough to fill the Library of Congress several times. Organizing and storing this much data required powerful computing software. Ground-based astronomers usually took pictures of only a few celestial objects a night. They looked at the pictures themselves, differentiated between stars and galaxies, and filed them accordingly. "Once you have 40 terabytes of data, you cannot do that anymore," says Szalay, who was assigned the job of building the digital archives. "You have to have it all automated." Sloan programmers ended up creating the largest software program in the history of ground-based astronomy. "We wrote over a million lines of code," he says.

By 1996, things were starting to slip. Hopkins astronomer Tim Heckman, who was then serving as the chairman of the Sloan's science advisory board, realized that action needed to be taken. "It was sort of like trying to pull an airplane out of a power dive," Heckman says. "Apart from trying to solve things technically and politically, we were also in a very precarious financial situation." In an effort to turn the Sloan around, the advisory board hired expert project manager Jim Crocker. Crocker is a tall southerner with a master's degree from Hopkins (MS '89) in management of technical and scientific programs. He had helped coordinate numerous large astronomy projects including Skylab, the Very Large Telescope in Chile, and COSTAR (the optics package that fixed the mirrors on the Hubble Space Telescope).

Crocker's first step was to bring in experts to perform a cost-to-complete review. "I didn't want to take the project on if we didn't have an honest assessment of what it was going to cost," he says. When the estimates were done, the projected price of the Sloan was well over twice the original figure. "The amount of effort to do this project had been significantly underestimated," Crocker says. "People equate the [cost] of the project to the size of the telescope, and this is a modest telescope. But the instrumentation is phenomenal, and the quantity of data is enormous, and the processing required is incredible." After revising the cost estimate, Crocker made sure that the consortium's institutions and universities would back the project to completion.

Once Crocker had completed his cost estimate, he went to work trying to find the best way to allocate the Sloan's resources. One of the first things he focused on was the Sloan's monitor telescope, designed to provide invaluable information on atmospheric conditions such as cloud cover, wind, and temperature fluctuations. The Sloan, which sits 9,150 feet above sea level, needed this information to correct its own images, so that images taken on two different nights, under two different sets of conditions, could be made to agree. However, the monitor telescope was in poor shape. Its mirrors were cracked, and it was out of focus. Without it, the main telescope was useless, but because the project to replace it posed little scientific challenge, no one was interested in fixing it.

"I think I'd been on the project two weeks when I realized that the monitor telescope was the critical path," Crocker says. He took Hopkins's Uomoto off the spectrograph project and set him to work moving a small 20-inch telescope from the top of Hopkins's Bloomberg Center for Physics and Astronomy to the Arizona desert, where it became the new monitor telescope.

Under Crocker's management, the Sloan quickly got back on track. "He reorganized the project and made it really professional," Szalay says. "I think that's when things really started moving along." Crocker stayed on with the Sloan for 18 months, providing the structure that was so desperately needed. Today, under the care of a new project manager, collaborators are willing to send projects to other universities, and things are running smoothly, according to Gunn. "I think that in the end it has worked out extremely well," he says.

Already, the Sloan's online archives have contributed several major discoveries to astronomy. Last spring, Hopkins astronomers Wei Zheng and Zlatan Tsvetanov discovered the oldest, most distant object ever seen in the universe--a giant black hole surrounded by a fiery disc of stellar dust, known as a quasar. "The brightest quasars are hundreds of times brighter than our galaxy," says Zheng. But because they lie on the very edge of the known universe, they are difficult to see. "Finding a quasar hiding in the sky," says Zheng, "is more difficult than finding a needle in the hay. The odds are one in millions." By sifting through the massive Sloan database, Zheng and Tsvetanov were able to break the 10-year-old record for the most distant object in the universe.

Their fame, however, was to be short-lived. Within two weeks, a group at Princeton using the same database had found an even more distant quasar.

While the Princeton group was searching for these faint quasars, it stumbled across another stellar oddity, a small, dim star known as a brown dwarf. Brown dwarfs are objects that are something between a star and a planet. At around 80 times the mass of Jupiter, they are large enough to produce heat but too small to support the thermonuclear reactions created by most stars. As a result, brown dwarfs spend their short lifetimes glowing, like embers pulled from a fire.

Because they fade so quickly after they are born, they are very difficult to find. Few people know more about brown dwarfs than David Golimowski, an astronomer at Hopkins who has been searching for these dim stars since 1992. In November 1994, Golimowski and a team of researchers set out to find a brown dwarf that had been predicted by theory, but was as of yet unseen by astronomers. Golimowski methodically plodded through more than 300 stars, looking at each one to see if it had a brown dwarf orbiting it. He and his colleagues finally isolated a star only a few light-years from Earth that seemed to have a brown dwarf companion. After nearly a year of careful observation, his team announced it had discovered the first confirmed brown dwarf.

But these dim stars proved elusive, and five years later astronomers had yet to find a second one. That is, until the Princeton team announced that they found a likely brown dwarf candidate. "I had to admit, I was taken by surprise," says Golimowski. "It happened so early in the data taking." Since that time, Golimowski and others have used the Sloan's database to identify more than 50 brown dwarfs. Using the incredible wealth of Sloan data on these objects, Golimowski and others hope to better understand the continuum between stars and planets, and in doing so learn more about how our galaxy evolved.

These two discoveries are just a small sample of what is to come. "This thing is really big," says Uomoto. "People will continue to get PhD theses out of this for the next 20 to 30 years."

"I don't think I have enough vision to know just how deep this is going to go," says Gunn. "I also don't have enough hubris to think the Sloan will be the final word, but it is the beginning."

And what of the three-dimensional map of the universe? By the time this article is published, cosmologists will have already begun checking their theories against the Sloan's data-- confirming the relationship between the stellar cobweb we see today and the interactions of particles shortly after the Big Bang. "This is something I've been working on for 25 years," says Szalay. "We should know the answer in six to eight months."

Geoff Brumfiel is a master's degree candidate in the science writing track of The Writing Seminars at Hopkins. For more information about the Sloan Digital Sky Survey, visit the SDSS website at http://www.sdss.org/.

Return to April 2001 Table of Contents