It was late Saturday afternoon, and submarine pilot Marshall Flake and rocket engineer Chuck Nighswonger were doing what they had been doing twice a day for nearly six weeks -- creeping along the murky ocean bottom in a tiny submarine, looking for space shuttle wreckage 40 miles off the Florida coast and 600 feet down.

They were stalking an object known then only as "sonar contact No. 131."

In the weeks between the sudden disintegration of the Challenger and that mid-April afternoon, 722 sonar contacts had been picked out on the ocean floor. There was no way of telling which were shuttle debris and which were coral heads or 55-gallon drums or old shipwrecks until divers, robots or submarines could go down and inspect each one.

Using sonar and moving at the sub's top speed of a single knot, Flake and Nighswonger took 10 minutes to find the jagged piece of curved metal. Twenty feet long and 11 feet across, nearly the size of the sub, it was stuck in the sand like a footbridge. Most of the white paint was scorched off. Along one edge, the piece looked as if it had been carved with a blowtorch.

"I think," Flake radioed to his mother ship, "we may have found our ticket home."

Sonar contact 131 was a hunk of the Challenger's right solid rocket booster, including part of the joint that sprang a leak on Jan. 28, ultimately causing the destruction of the shuttle and its crew of seven 72 seconds after launch. Its recovery, and the retrieval of the crew compartment, climaxed the largest underwater search and salvage operation in history.

At its height, it was costing NASA $1 million a week to find and bring back the pieces of its shattered spaceship. When the search formally ends this month, it will have cost $13 million and used 16 aircraft, two dozen ships, three submarines and 10,000 personnel from the Coast Guard, the Navy, the Air Force, NASA and a dozen private contractors.

The gargantuan undertaking covered an area of Atlantic Ocean bottom seven times the size of the District of Columbia, every square foot of which was sonared and plotted so thoroughly that, as the search leader put it, "We can go back now and find individual conch shells."

Finding Challenger debris meant investigating every significant blip the sonar revealed. After separating the relevant from the irrelevant, the debris had to be hauled to the surface -- a task simpler in the description than in the execution, for much of the important shuttle wreckage was in pieces that weighed tons, pieces that had wedged themselves six feet into the sand.

"The doability of it was never in question," said Air Force Col. Edward A. O'Connor, the man who directed the search and salvage operation, his first. "I think the fact that I didn't really realize what was involved probably helped."

The world of people skilled at underwater retrievals is, as one of the participants put it, "a small bathtub." Bound by the thrill of finding things others have lost, they travel from job to job and ocean to ocean, and gather for the biggest searches -- the Challenger, the KAL jet shot down by the Soviets, the Air India jet that exploded near Ireland -- as if for a reunion.

When asked what he does for a living, Bob Kutzleb, the sonar guru of the Challenger search, replied with a broad grin, "Anything that makes a nickel on the water."

He was hired by his son Mike, who owns Steadfast Oceaneering in Falls Church, the principal search contractor for the Navy and the principal search contractor for the Challenger. Mike hands out pencils engraved with the motto: "If we can't find it, it ain't lost."

Don Dean is a project manager for Eastport International, the principal salvage contractor for the Navy, which directed the retrieval of objects found by the Kutzlebs. Just before moving on from the Cape to the next salvage job, he proclaimed in a booming voice, "We're getting pretty good at this. We could pick it all up now!"

The Navy's supervisor of salvage, in charge of the day-to-day operation, is Capt. Charles (Black Bart) Bartholomew, a tall, blond, soft-spoken man who says he quit Navy flight school because it bored him. Ask the others how he was to work with and they all say, "Same as always." Which means he doesn't raise his voice, doesn't lose his cool and finds what he's looking for.

The operation was coordinated from another fixture of worldwide salvage operations, Bartholomew's command van -- the ratty trailer section of an ancient 18-wheeler, flown to wherever things fall into the ocean, done up inside with yellow Formica tables, brown naugahyde stools, a computer, a magnetic status board and a chalkboard. The van looked at home amid the palmetto scrub and sand of the Cape, with an American flag in front propped up with sandbags.

The command van was joined for the round-the-clock, 16-week search by a house trailer equipped with another computer, videotape machines and a staff of radio operators. It was christened "the other van."

Two tiny submarines, one of which would find contact 131, were brought in. The Harbor Branch Foundation, a nonprofit oceanographic research institute, supplied the Johnson-Sea-Link subs and their support ships and crews, which Bartholomew pronounced "the workhorses of this operation."

From Connecticut came the Navy's smallest nuclear submarine, the NR1, also used mostly for scientific research. Once Bartholomew and company figured out how to use it, the NR1 would save the salvagers precious time, working without surfacing for days on end.

The enormous oil work ship Stena Workhorse, with its helicopter deck, its cranes capable of lifting 200,000 pounds from the ocean floor and its price tag of $12,000 a day, arrived from Orange, Tex. Its Swedish crew of 30 was happy to be working. The Stena ordinarily does oil rig work in the Gulf of Mexico, but because there isn't much offshore oil work these days, it was in drydock when the shuttle crashed.

From Elizabeth City, N.C., came a unit of Coast Guard divers who called NASA and volunteered their services. To them fell one of the grimmer tasks -- recovering debris from the shuttle's crew compartment. More unsettling than the remains, which were not recognizable as human, were the arresting reminders of the humanity of those who died: a condiment rack from the shuttle's galley, including packets of mayonnaise and catsup; a jar of Jiffy peanut butter, unbroken; a pair of jockey shorts, from the locker of one of the astronauts.

Among the most unschooled in salvage were a handful of normally landlocked rocket engineers such as Chuck Nighswonger and Pat Patterson, who for weeks rode around in submarines, squinting out at debris on the ocean floor and "seeing things most people only get to see in National Geographic," as Patterson said.

NASA wanted the Challenger debris found immediately, salvage crew members said, and had ideas of its own about how to do that. It figured, for instance, that there had to be a way to use all the radar data, equations and computers to predict where parts of the shuttle had fallen without having to search the entire ocean for them.

"Patience was a problem," said Bartholomew. "NASA kept saying, 'We want you to go look here! We want you to go look there!' . . .

"When I got to the Cape, there were a lot of knee-jerky things going on," he said. "NASA tends to be a little knee-jerky. With all their technology -- well, the way they work is not the way we work. They do things faster, their engineering is much more precise, they are much more impatient."

Bartholomew said he told NASA, "We're going to do it slow and methodical, and we're going to cover the entire area. That's the way we work . . . You can be a couple hundred yards from the gold mine and not know it. If you do it haphazardly, you'll never find anything."

The story of sonar contact No. 131, the piece that not only solved the Challenger mystery but turned the salvage operation into an unqualified success, illustrates that maxim.

Bob Kutzleb, the sonar expert, first located it on a sonar trace March 1. His role in the search was to sit all day in "the other van" and read the sonar tracings. Fifteen hours a day, seven days a week for nine weeks, he scanned the maroon etchings a sonar recorder makes on an endless roll of eight-inch-wide paper.

Four ships searched the 420-square-mile area, in 420 straight lines, 200 yards wide, each line run at a patient three knots, north to south, trailing a sonar "fish" three-quarters of a mile behind the ship. Each day, a small boat collected the rolls of tracings and took them to Kutzleb.

Sonar does not reveal shapes particularly well. And the Challenger, which showered more than 3 million pounds of debris into the ocean, crashed in an area off the Florida coast that was already littered with garbage. Each day, Kutzleb noted with a black felt-tip pen the maroon marks on the tracings that he thought might be shuttle debris, as opposed to fish holes or depressions in the sand.

"People think I do it with mirrors," he said with a grin. "I shouldn't tell you this, but a lot of it is just gut feel."

Contact 131 looked like a nail clipping on the sonar tracing. Kutzleb gave it its number, punched its location into the computer and added it to the list of contacts for future investigation by submarine or robotic vehicle. Eventually, he would pick 722 likely contacts out of more than 5,000 miles of sonar mapping.

"There were a lot of things I didn't send them on," he said. "If we had come away and not found contact 131 , then I would have been scratching my nose."

"We chased a lot of junk," said Bartholomew.

The "junk" salvagers found included marine batteries, trash cans, paint cans, a Pershing missile, half a torpedo, a refrigerator, a file cabinet with the drawers sitting next to it, a kitchen sink (retrieved and presented to Bartholomew) and a toilet (which he wanted retrieved as a "counter-present").

Also, eight shipwrecks, including a small coastal freighter and a fishing trawler; a DC3, World War II vintage; the transom of a motorboat, the motor still firmly screwed in place; and the 20-foot-long rudder from a freighter.

Also, a Cuban beer bottle and, as Kutzleb put it, "more 55-gallon drums than you can shake a stick at."

Unquestionably the most valuable nonshuttle item recovered, hauled out of the water on April 7, was a floating duffel bag containing 25 kilograms of cocaine. It was turned over to the Drug Enforcement Administration, which estimated its worth on the street at $13 million -- just enough to cover the cost of the entire salvage operation.

Initially, Bartholomew and the Kutzlebs thought booster rocket debris, some of the most important from the Challenger, would be found in the narrow corridors beneath the sky where radar showed the boosters had passed. But a piece that fit contact 131 like a puzzle was found seven miles away from it.

On the sonar tracings, contact 131 didn't look particularly important, because it was well south of the radar path of the right-hand booster. It wasn't a high priority, and it languished, unexplored, until mid-April.

Pat Patterson, a NASA engineer assigned to the NR1 sub because he knows what booster parts look like, was the first to take a look at it.

The NR1 makes today's military submarines look like cruise ships. Most of its 137-foot length is nuclear reactor, and the crew compartment, which on this mission held 12 people, is about the size of six telephone booths. The captain of the sub sleeps on the floor, the crew sleeps in shifts because there are only enough bunks for half of them. The food consists of TV dinners and anything that comes out of a can. There is no shower.

The advantage the NR1 offered Bartholomew was that it can stay submerged for days. The Johnson-Sea-Links provided better visibility, but because they are battery powered, they can only dive twice a day for about four hours and must be hauled from the water onto the fantail of a ship after each dive.

On one spectacular cruise the NR1 remained underwater for 13 days, touring 281 contacts. The smaller subs, the robotic vehicles and the divers were each clearing two to five contacts a day.

Patterson was on that marathon cruise. Afterward, asked if he felt like hugging his wife and taking a shower, he replied, "Yeah, and eating a green salad!"

He peered at contact 131 through a five-inch porthole, but couldn't get a very good view. The viewport and video cameras mounted on the hull of the sub require it to hover 20 feet from the bottom.

"I thought, that part doesn't look like SRB [solid rocket booster]. It looks more like a part of some boilers [from old ships] we've seen," Patterson said. The NR1's 14 cameras, however, had a better view.

Those videotapes -- and tapes of all the contacts the NR1 examined -- were eventually shipped back to shore for scrutiny by NASA engineers.

On April 11, NASA told Bartholomew that contact 131 looked like something that might be of more than passing interest -- it showed signs of scorching.

Bartholomew ordered one of the small, nimble Johnson-Sea-Link subs to dive on it. On the afternoon of April 12, Flake and Nighswonger climbed into the bubble of the Johnson-Sea-Link 1, were hoisted over the side of the mother ship and dropped down onto contact 131.

They spent two hours videotaping the piece from all angles, and attaching a pair of metal cables with looped ends so that it could be hauled to the surface. They attached a sonar "pinger" as well, so the Stena Workhorse could zero in on it.

Even when the men returned to their ship, there remained some question whether contact 131 was The Piece. Flake radioed to "the other van" on shore, advising cryptically that they had found "a piece we recommend place a high priority on recovering." The oblique message was necessary, the salvagers believed, because all the major news organizations were listening to ship-to-shore radio communications.

The Stena Workhorse was diverted to bring contact 131 to the surface. One of the salvage contractors had outfitted the ship with a remote-control diving robot, the Gemini, with two arms and an array of video cameras. Gemini attached a cable to one of the loops Flake and Nighswonger had left, and one of the Stena's big cranes brought it up.

"Everybody knew this was the piece," said Roy Truman, who directed recovery operations on the Stena. No. 131 hit the stern deck at 4:51 a.m., and Truman remembers that nearly the entire crew of the ship was assembled on its enormous helicopter pad, atop the ship's bridge.

There was some applause, he said, but no celebratory beers. The ship carried none.

Later that day, April 13, two NASA engineers were sent to the Stena to look at the scorched hunk of metal.

"Were they excited?" Bartholomew said. "No, actually, they were quite green." Both engineers became horribly seasick, and were quickly returned to shore.

But they brought with them confirmation that sonar contact No. 131 was what had until then been "the apparent cause" of the Challenger disaster.