DANBURY, Conn. -- Errors detected throughout the manufacturing process of the Hubble Space Telescope's flawed main mirror were repeatedly discounted by officials of the company that built it, a scientist assigned to the project said yesterday.
Dr. Robin Laurance, a scientist with the European Space Agency attached to the NASA board investigating the telescope's mirror, said yesterday that from the beginning of the manufacturing process, there were indications in test results that something "went radically wrong."
But each time, officials of the manufacturer, Perkin-Elmer Corp., discounted the test results because of their confidence in what they considered a flawless test device known as a reflective null corrector. It was that device that showed the 94-inch primary mirror surface to be near-perfect, Dr. Laurance said.
Investigators have found that a 1.3-millimeter spacing error in the null corrector -- caused by a mispositioned cap on a measuring rod -- has resulted in blurry images being transmitted from the $1.5 billion telescope, greatly reducing the orbiting observatory's capabilities.
Dr. Laurance said that repeated tests with a device called an inverse null -- which took the place of the primary mirror in the test setup to check the alignment of the equipment -- showed problems that were later discounted.
The alignment check with the inverse null was performed after every test on the mirror's surface during the polishing phase of manufacture, he said.
But, he explained, so long as the test equipment was found to be aligned properly, evidence of the other problems was discounted.
Officials of the National Aeronautics and Space Administration already have said that on at least two other occasions during the mirror's manufacture a decade ago, yet another testing device, known as a refractive null lens, showed something amiss.
One of those tests with the refractive null -- near the completion of the project in 1981 -- showed the mirror to have a spherical aberration of the same magnitude that has crippled the Hubble. But those conflicting test results also were discounted, the NASA officials have said.
"The real error, in my opinion ... was that there was no requirement for an independent check," said Dr. Laurance, whose agency built the faint object camera, one of five instruments aboard the orbiting observatory. "The error should have been discovered by an independent method."
Each of the incidents in which conflicting test results appeared should have run up a red flag that something was wrong with the mirror's surface, he said.
Low-level officials of the company, now known as Hughes Danbury Optical Systems Inc., were aware of the conflicting test results but apparently did not report them to their supervisors, Dr. Laurance said. So far, investigators have not found evidence that NASA was aware of those results, he said.
The NASA investigative board, which met behind closed doors yesterday for the first day of a two-day session, has determined that the 1.3-millimeter spacing error in the reflective null was caused by a mispositioned cap on a measuring rod used to construct the reflective null, Dr. Laurance said.
The aluminum cap was supposed to fit snugly on the rounded top of the measuring rod like a bottle cap, but investigators have found it did not.
The extra space threw off the distance between a small, spherical mirror inside the reflective null and a "field lens" on the bottom of the assembly, causing the mirror to be polished to the wrong shape.
Because the measuring rod was longer than it should have been, company officials installed washers on a flange holding the field lens in place, thus allowing the measuring rod to fit between the small mirror and lens, Dr. Laurance said.
It was those washers that NASA investigators believed a month ago could have caused the spacing problem, but they have since discovered the problem was in the measuring rod cap, he said.
The shape of the mirror was determined solely by an interferometric test, which utilized the reflective null corrector, but with no double checks -- a test method NASA officials at the time approved, but since have identified as a weakness in the manufacturing procedures.
In the interferometer test, a laser beam is shot through the null corrector assembly at the mirror. In turn, the beam bounces back through the null assembly for a computer analysis in which that image is compared to an image of the desired shape.
Problems aboard the Hubble first were discovered in June, two months after the telescope was launched, when scientists realized the images being transmitted from space were blurry and showed indications of a classic mirror flaw known as spherical aberration.