An astronomer has reported discovering an immense string of galaxies -- 100 times more massive than any structure previously known to exist -- that stretches across 10 percent of the universe.

Linked like beads on an enormous celestial chain, the new "supercluster" appears to challenge one of cosmology's most fundamental assumptions: that galaxies are spread randomly but equally across the universe. Discovered by R. Brent Tully of the University of Hawaii, this "supercluster" is so large that theorists may have to develop new models to explain the formation of galaxies.

"If this is accurate, it will be very difficult to understand what physical process could have organized galaxies on such large scales," said Simon White, an astrophysicist at the University of Arizona. "We will really have to sit down and rework many of the basic notions we have."

Galaxies are clusters of billions of stars grouped together over a distance of hundreds of millions of light years. Until ten years ago, most astronomers believed that galaxies do not form clusters on a large scale, but are scattered randomly around space.

Over the past decade, as sophisticated computers and telescopes have become essential tools of the astronomer's trade, there has been increasing realization that galaxies seem to fall into coherent clumps and strings separated by vast voids.

The supercluster described by Tully, however, is larger than anything envisioned before, and many scientists yesterday said they would remain skeptical until they read his article describing the discovery, to be published in the Dec. 1 issue of Astrophysical Journal.

"The data is all new, and we have to be sure he was not seeing an accidental pattern, one that could have been random," said James Peebles, professor of physics at Princeton University. " . . . I look forward to seeing his work."

Tully named the structure the Pisces-Cetus Supercluster Complex after the constellations in which it was found. He said yesterday that he has preliminary evidence of similar superclusters in other regions of the universe.

"I am comfortable that this is no fluke," he said. "I'm not bold enough to suggest that all the theories are worthless. But I do think they will need some adjustment."

Creating hypothetical galaxies on computers, theorists are already scrambling to make sense of Tully's discovery. Calculating the motion of galaxies and plotting their distance from each other, astrophysicists will feed millions of bits of data into huge supercomputers, like the one used by Tully, to create a simulated universe that can accommodate the new supercluster.

How the universe was formed is the most basic question of astrophysics. To answer it, astronomers begin with the Big Bang, the cataclysmic explosion that is thought to have created all celestial objects.

The largest structures in today's universe are the results of infinitessimally small fluctuations set off by the Big Bang billions of years ago. As the universe expanded, the fluctuations grew and eventually reached the massive proportions of today's cosmos.

Tully has been mapping galaxies for more than a decade. Once he identified clusters, he compared their distance from each other with random spacings between clusters of galaxies.

By gauging the density and volume of galaxies, he then computed the statistical chances of their being in certain locations in relation to other neighboring galaxies. He concluded that the clusters are related to each other because the probability of their relationship occurring by chance is statistically slight.

"This is very exciting, but we have to take it for the pioneering new information that it is," said Michael Turner, a physicist at the University of Chicago.

Turner cautioned that the Tully discoveries need to be confirmed and that even now, only a small fraction of the universe has been mapped by scientists.