At least one of the two mirrors in the $1.6 billion Hubble Space Telescope was built in the wrong shape and will not work properly, NASA officials said yesterday. The defect will virtually eliminate the use of a key camera and halt about half of all the telescope's planned scientific work until a shuttle crew can go back to the telescope and make repairs in 1993.

NASA has ordered an investigation of the mirrors' construction records to find out who made the mistake. Disappointed scientists and engineers are scrambling to find ways to minimize the impact of the problem aboard the orbiting observatory which, with its ground facilities in Greenbelt and Baltimore, has cost taxpayers $2.6 billion so far.

"Where we suspect the problem happened is in the technique used to measure and polish the mirrors," said Jean Olivier, deputy project manager for the telescope at Marshall Space Flight Center. "Somewhere in this complicated chain of events, there was a mistake made" in one of the two mirrors. "It was done carefully and it was done to the wrong figure," or curvature.

The built-in flaw gives the Hubble a serious case of blurred vision. Two onboard cameras -- the workhorse Wide Field and Planetary Camera, and the European-built Faint Objects Camera -- designed to take pictures in visible light, are the most seriously affected. Four other instruments that study the heavens by gathering ultraviolet light or that for other reasons do not depend heavily on crisp, clean visible images will suffer less, officials said.

A second generation of instruments was already being built for the Hubble and engineers say they can be outfitted with the equivalent of prescription glasses to compensate for the defect, the same way glasses correct nearsightedness, said NASA's chief Hubble astronomer, Ed Weiler. Hubble's "glasses" will consist of adjustments in the shapes of small mirrors used to route the telescope's light rays into cameras.

Those instruments are scheduled to be installed by shuttle astronauts on missions now scheduled for 1993, 1996 and 1997. Officials said they are looking into the possibility of advancing the schedule.

"We are not losing science," Weiler said. "We are deferring science."

In addition, officials at the Space Telescope Science Institute in Baltimore are searching the backlog of scientists' proposals for experiments that can replace those already approved but which now must wait for the new instruments to be launched. Ten times as many proposals were submitted as could be accommodated in the first round of research, officials said, and many of them could use the telescope as is.

Of the proposals already approved, "at least 50 percent of the science . . . is extremely viable," said Peter Stockman, the institute's deputy director.

Engineers also plan to use actuators, or mechanical devices on the primary mirror, to change its shape slightly, but they said that will only counteract "a small fraction" of the defect.

For the short term, Weiler said, "the important question is, 'Can we still do unique and important science?' The answer is an emphatic yes."

But he acknowledged that Hubble scientists, many of whom have spent 12 or more years developing instruments for the project, are frustrated and unhappy.

Specifications had called for the mirrors to focus 70 percent of a star's light at a precise point within the telescope. "We are now getting about 20 to 25 percent, so we're a factor of 3 1/2 off," Weiler said. That is about the equivalent of what ground-based telescopes can do.

The Hubble has a troubled management history, involving some 350 organizations and numerous delays and cost overruns. But its launch in April had stirred optimism that it would soon fulfill much of its original promise to provide a major leap in human understanding of the universe.

A primary claim for the Hubble was that it would provide unprecedented clarity in its images of the heavens. Its heart is its 94.5-inch primary mirror, often described as the most optically precise large mirror ever built. It and the 12-inch secondary mirror were built by Perkin-Elmer Corp., now Hughes Danbury Optical Systems Inc.

The new problem emerged in recent days just as the solutions to vibrations and other vexations that had plagued the telescope since its launch were in sight, engineers said.

Two weeks ago, engineers at the Hubble's control center at Goddard Space Flight Center in Greenbelt sent commands to move the secondary mirror along the barrel of the telescope in a way that should have significantly improved the focus of the image, but "it did not," said engineer Douglas Broome, who manages Hubble for NASA headquarters.

Over subsequent days, tests indicated that one of the mirrors -- no one knows which -- has what astronomers call "spherical aberration," which means that all the light rays striking the mirror do not focus at precisely the same point.

To confirm this, engineers moved the mirror "from one end to the other," including positions deliberately well out of focus, said Olivier. The result was a signature of spherical aberration so "textbook perfect" and "radially symmetrical" that "it had to have happened on the ground" and could not be a sag in the mirror caused by launch or other stresses.

Olivier said the mirrors were not tested together before launch because that would have required a facility costing hundreds of millions of dollars. But he and others indicated that tests of the mirrors individually should have found the error.

NASA chief scientist Lennard Fisk said that, despite its major flaw, Hubble still will produce "many fundamental discoveries" in the next few years. He also cautioned against "condemning everyone in sight" before the investigating panel "figures out what happened."