TIGHAR was not one of those groups. Though it was formed in 1985 by aviation fanatics interested in investigating old missing-plane cases and, if possible, recovering the aircraft, Gillespie steered TIGHAR clear of the Earhart mystery. Earhart had run out of gas somewhere on a very large ocean, he figured, and her plane could be anywhere in it, miles under the water.

But in 1988, two of his members came to him with a proposal. What if Earhart didn't crash into the sea? What if she reached an uninhabited island?

"The key to it is her final message, where she says 'line of position 157 dash 337,'" Gillespie said. "That's a line that Noonan calculated from the sunrise, running 337 degrees to the northwest and 157 degrees to the southeast. And if you follow it far enough, there are two deserted islands on it, McKeon Island and Gardner Island."

It didn't take long for TIGHAR investigators to find that somebody else had already mentioned the possibility of Earhart landing on Gardner Island. In 1960, a 68-year-old ex-Marine named Floyd Kilts gave an interview to a San Diego newspaper recounting his visit to Gardner Island in 1946, when he was sent there to dismantle a navigational device installed there during World War II.

Kilts said a Micronesian tribesman living on Gardner told him that when the Micronesians moved onto the island in 1938, they found a partial human skeleton, along with a woman's shoe -- a sign that she was a foreigner, since the tribesmen all went barefoot. The remnants of a fire pit nearby contained burned bones of small birds and fish, which suggested the woman had lived there some time.

The bones had been given to a British colonial official, who thought they might be the remains of Earhart. The Micronesian didn't know what happened after that, and neither did Kilts.

That story sounds straight from the captured-by-the-Japanese template -- except in this case, British archives yielded a load of radio traffic about the discovery of the bones and detailed measurements by a British medical examiner. (The bones themselves had disappeared. The British doctor had concluded the bones belonged to a man of mixed Polynesian and European race, though forensic anthropologists who looked at the data in the 1990s thought it more likely they were those of a European woman.)

One other thing TIGHAR's research turned up: The Navy's belief that Micronesian tribesmen had recently been living on Gardner Island in 1937 when its pilots flew over it was wrong. The tribesmen arrived for the first time a year later. Those signs of habitation had been left by someone else.

Gillespie and his group made their first expedition to Gardner Island -- by now renamed Nikumaroro and part of the Republic of Kiribati -- in 1989. It was once again deserted; drought drove the population away in the mid-1960s. Some of their empty buildings, including a general store, survived. Otherwise, not much was found.

A second, better-funded expedition arrived in 1991. The past two years had been hard on the island; a major storm had knocked down what little remained of the Micronesian settlement. But as they poked through the rubble, investigators found a fascinating piece of junk: A scrap of aluminum, 19 inches wide by 23 inches long, with four precisely measured rows of rivet holes. It looked for all the world like the torn outer skin of an airplane.

Over the years, tests have shown that's exactly what it was. The scrap is made from an Alcoa Aluminum substance called 24ST Alclad, which was used in the manufacture of nearly all American planes manufactured in the 1930s -- including Earhart's Lockheed Electra.

But Gillespie got out well ahead of his forensic evidence in 1992 by holding a Washington, D.C., press conference where he declared that "every possibility has been checked, every alternative eliminated... There is only one possible conclusion: We found a piece of Amelia Earhart's aircraft."

In fact, as other Earhart-investigation groups (there are more of them than Justin Bieber fan clubs, and they can be just as temperamental) quickly pointed out, the rivet patterns on Gillespie's scrap were very, very different than those on Lockheed's Electra.

"It was soon apparent that the Earhart mystery was not solved," Gillespie admitted ruefully.

For years, the metal scrap was like a thorn in TIGHAR's paw.

"We knew it was significant, we knew it was a piece of a plane, but we just couldn't quite figure out where it fit," Gillespie said.

Three months ago, the group decided to come at the scrap from the opposite direction: If it wasn't from a Lockheed Electra, then what plane was it from?

Gillespie's investigators spent a day with the reconstruction team in Dayton, Ohio, at the U.S. Air Force Museum, which rebuilds World War II-era planes for a living. The team scoured its vast store of blueprints and technical drawings. It didn't fit anything.

"That's when one of our investigators said, look, we know there's one piece on that plane that wasn't built or installed by Lockheed -- the replacement for that missing window," Gillespie recalled. "So maybe that's the match."

TIGHAR began reviewing its massive archive of photos of Earhart's plane. But relatively few showed the right side of the aircraft, because photographers usually wanted to get Earhart herself in the shot, and her pilot's seat was on the left side. Only one shot offered a really good view of the patch: that 1937 photo from the Miami Herald.