The space shuttle's huge external fuel tank will probably have to undergo several design improvements to further minimize the dangerous shedding of insulating foam that occurred during the launch of Discovery last month, NASA officials announced yesterday.

That preliminary assessment of a need for multiple fixes reflects NASA investigators' inability so far to find any single "root cause" for the foam losses, space station manager William H. Gerstenmaier told reporters in a conference call.

Gerstenmaier did not speculate about how long it might take or how much it might cost to make the changes, which he characterized as "minor engineering modifications." But he said he does not expect the shuttle Atlantis to get off the ground before its scheduled September launch window closes.

The next launch opportunity will come Nov. 7.

"We didn't find any immediate, easy fixes," Gerstenmaier said, referring to efforts by NASA's "tiger team" of shuttle investigators. That team is working with five "program teams," each focusing on one of the five areas of the tank that lost notable amounts of foam.

Despite the setback, which officials emphasized will be subject to reevaluation as more is learned, officials presented a generally upbeat perspective on the initial findings.

"I was surprised by how well the tank actually performed," Gerstenmaier said, noting that a total of only 1.2 pounds of foam came off the tank, out of 4,192 pounds sprayed onto the giant missile.

"In terms of the aggregate, that's a pretty small number," Gerstenmaier said. "It's still not the performance we'd really like but a pretty good performance overall."

Gerstenmaier also emphasized that the single largest incident of foam shedding, a 0.9-pound loss from the "PAL" ramp, occurred late enough in the launch that the thin air rendered the foam no threat to the orbiter.

In fact, he said, the agency had not decided how many pieces of foam would ultimately be classified as having exceeded the agency's "allowable" limits for foam debris. If issues of timing are counted, he said -- a parameter not previously written into the agency's definition of "allowable" -- then the number might end up as small as "one."

This positive view "is not spin control," Gerstenmaier said in response to a question. "I just don't think it's as dire as it appeared at first."

John Logsdon, director of George Washington University's Space Policy Institute, said it is not unreasonable to change definitions of "allowable" to accommodate new factors such as timing, which can affect how dangerous a piece of debris is.

The difficulty, he and Gerstenmaier noted, is gaining assurance that a foam loss that happened late enough in a flight to pose no risk will not occur earlier in the next flight.

Although the fact-finding phase of the investigation is expected to go on for at least two more weeks, Gerstenmaier said that possible solutions have already started to come into focus. One might involve removing the PAL ramps from tanks, reapplying foam in that section using methods that may enhance adhesion and then replacing the ramps.

There is also a strong suspicion that the foam that flew from a "bipod assembly" may have come loose because of the presence of a wire that can probably be moved or redesigned, Gerstenmaier said.

He also expected minor changes in ice-deflecting vanes near the PAL ramp, where three small pieces of foam flew off.

In the future, he said, NASA may determine that the PAL ramps are unnecessary.