It took forever to glue on the thermal tiles that shielded the space shuttle from the scorching heat of reentry -- nearly two man-years of work for every flight -- and the glue dried so fast that technicians had to mix a new batch after every couple of tiles. But they came up with a solution: spit in the glue so it took longer to harden.
The trouble was that spit weakened the adhesive bond between the tiles and the shuttle's aluminum shell, making the tiles more likely to fall off during the spectacular stresses of space flight. When NASA officials found out about this home remedy, they put an end to it.
Spitting in the glue -- a common practice in the shuttle's first decade -- is just one example of the troubled history of one of the shuttle's most crucial features -- the 24,000 heat-resistant, ceramic tiles that cloak the spacecraft's underside. A catastrophic failure of those tiles is a prime suspect in Saturday's disintegration of the shuttle Columbia and the deaths of the seven astronauts aboard it.
If it turns out the tiles failed -- NASA says it has ruled out nothing in its investigation -- Columbia's obliteration would be the most tragic outcome in a long history of glitches and casualties, minor and significant, in a system that never seemed precisely to answer the difficult problem it was called upon to solve.
From the first shuttle's first flight in 1981 -- when 16 tiles fell off and 148 were damaged, according to NASA documents -- tile problems plagued the shuttle program.
NASA worried not only about how to make them more quickly and maintain them more easily, but, more important, how to keep them from falling off, getting knocked off or breaking off.
The answers were never fully found. But NASA always knew that the loss of a single tile could by itself doom the spacecraft, even if it did not trigger a dreaded "zipper effect" -- in which a lone gap in the mosaic would allow superheated gas to peel off a whole row of tiles, like a tornado ripping shingles from a rooftop.
In the beginning, the tiles appeared to be the best answer that materials science could devise for a problem that seemed nearly intractable in 1972. That was when President Richard M. Nixon announced plans to develop a reusable spaceship to replace the throwaway rockets and capsules that first carried humans beyond Earth's atmosphere.
"In the old days, like with Apollo, you had a system that would actually melt away," said Paul S. Fischbeck, an engineering and public policy professor at Pittsburgh's Carnegie Mellon University, who studied the tile system for NASA. "But it could only be used once. Now they were trying to do a new system that could be used over and over. It's a hard problem."
Lockheed Missiles and Space Co. won the contract in 1973 with a lightweight silica formulation devised by materials engineer Robert Beasley. "Think of it as rigidized fiberglass -- pure silica fiberglass," said Lee Jachter, former lead manufacturing engineer at Lockheed's Sunnyvale, Calif., plant, which made the tiles. "He had invented it back in the 1960s, but it was a product looking for an application."
The space shuttle was the opportunity, but it was a tall order. The tiles would need to withstand the extreme vibration of liftoff and acceleration to 17,000 miles per hour. In space, temperatures could drop to 240 degrees below zero Fahrenheit. During reentry, the heat could build to 3,000 degrees.
"It couldn't have extreme thermal expansion like a metal; it couldn't go from an amorphous silica to a crystalline form, and it had to survive the temperature change without cracking," said Rutgers University's Stephen H. Garofalini, a ceramics scientist who worked for Lockheed on tile development from 1973 to 1980.
"Also, it had to be lightweight, and it had to inhibit thermal radiation," Garofalini continued. "You wanted it to be mostly air, and it needed a smooth, solid coating so water couldn't penetrate and you wouldn't have frictional heating."
The Beasley glass, finished with a black coating developed by NASA's Ames Research Center at Moffett Field, Calif., filled the bill.
"If you were to pick up a tile, it would feel like you were holding a piece of Styrofoam, and, in terms of how well it insulates, it's excellent," said Donald H. Emero, who worked in various shuttle program management positions from 1974 to 1997.
But there was also a pronounced downside. "In terms of its ability to resist damage -- its robustness -- it leaves a lot to be desired," Emero added.
Losing or damaging tiles during liftoff could exact a heavy price during reentry, when heat on a pitted or naked hot spot "could create a little blowtorch effect," Emero said. "The aluminum structure would disappear."
If the missing tile happens to lie over a fuel tank, "then that little blowtorch is focusing" on it, he continued. "The pressure goes up, the temperature goes up, and, bang, it explodes."
The tile mosaic is breathtakingly complicated, requiring that each tile be machined to different thicknesses, shapes and contours, and then painstakingly installed by hand. The process is remarkable for its labor-intensive nature and quirky problems.
Early in the shuttle's history, NASA discovered that the porous tiles tended to suck up moisture like a sponge, then crack in the freezing void of space.
"NASA developed a procedure to inject waterproofing through the tile coating with a syringe," said Jachter, the former Lockheed engineer. "It was dissolving the glue. One year we had 20,000 tiles to replace. It cost me my Christmas vacation in 1985 because I was machining tiles."
NASA then "went out and bought cases of Scotchgard, the commercial stuff, right off the shelf," said Carnegie Mellon's Fischbeck. "They Scotchgarded the bottom of the orbiter."
Every shuttle had lost or damaged tiles on a flight -- sometimes both, sometimes by the hundreds, Jachter said. A 1995 NASA document recorded that each shuttle required pre- and post-mission inspections for damage and re-waterproofing of more than 20,000 tiles. The process was so complicated that replacement and maintenance averaged 17,000 man-hours or more for each flight -- about 708 man-days, a 1999 NASA report said..
In 1989, a NASA news release announced that Rockwell International, which installed the tiles that Lockheed made, had moved part of its tile operation from Palmdale, Calif. to Cape Canaveral. That meant that it now took only three days -- not six -- to make each new tile and install it.
"The biggest problem was literally the detailed design itself. The fact that there are 22,000-plus of these things, and they have to sit very precisely, and have to have a line that's very smooth -- that's problem number one, said Thomas Moser, a NASA manager who was responsible for the tiles for many years. "The other one is the actual process of applying the tile. It depends on people doing their jobs right. A tile has to be bonded. The adhesive has to be mixed properly, applied properly, cured properly. It's not a machine that does it."The plan was first to cover the aluminum shuttle skin with a flexible material called Nomex, similar to the outside of a tennis ball, then glue the tiles to that. NASA got its first embarrassing look at problems to come in 1979, when the newly built Columbia flew through a thunderstorm on the back of a Boeing 747. Hundreds of tiles were damaged or lost.
"You have this kind of crunchy Styrofoam, and this felt surface, and it was very hard to get a good, solid bond," Fischbeck said. "It was a complete and total disaster."
More tests followed. Tiles were subjected to high wind tunnel pressures and hung from bungee cords to see how much weight they could hold. In the end, NASA developed a chemical treatment to make the bottoms of the tiles denser and harder, enabling them to set better in the adhesive.
The glue itself was virtually identical to the silicone adhesives used in residential bathrooms, but when a batch was mixed for application, it hardened so rapidly that technicians couldn't install more than two or three tiles at a time.
"So what they would do is actually spit into the glue so they could install more tiles per batch," Fischbeck said.
Fischbeck and colleague Elisabeth Patư-Cornell of Stanford University revealed the glue-spitting as part of a tile-hazard study they undertook for NASA in 1988. By this year, Columbia's tiles may have been especially prone to failure because of age -- some of them had been on the orbiter for 22 years, but tile damage, loss and replacements were chronic features of the shuttle program.
"I estimated we replaced a hundred or so in the early flights, then fewer as they got better at repairs," Jachter said. "The trouble was the tile had to be perfect, like jewelry, and [the installers] were using things like grouting tools to fix them."
Besides installation foul-ups, NASA realized it had another serious problem with debris -- ice, pieces of external fuel tank insulation, pieces of shuttle -- coming loose during the bone-rattling liftoff and smacking into the delicate tiles.
In 1998, part of a solid-fuel booster rocket flew off and skittered along the undercarriage of the shuttle Atlantis during liftoff, causing significant damage to 298 tiles. With the robotic arm and camera it carried, the crew was able to examine the damage in flight.
"It was significant," said Guy Gardner, a former astronaut and Atlantis pilot. "Mission Control thought we were okay, but frankly, there was nothing we could do if we weren't."
The Columbia was not rigged with a robotic arm on its last mission.
Using data from NASA's own tests and other sources. Fischbeck and Patư-Cornell calculated that the tiles contributed 10 percent of the risk of accident on any given shuttle flight. To improve the odds, NASA implemented a series of recommendations: It boosted installers' pay to attract better workers, reduced pressure on them to work too fast, and made the foam insulation on the external tank less likely to break away.
In the end, none of it may have been enough to save Columbia.
"It would have been nice to have looked at Columbia's damage so we could at least have a clue," said Gardner, adding that his own experience did not cause him undue anxiety. "There's an old proverb that says kneel to accept the things you cannot change."
Researcher Margot Williams contributed to this report.