PRIVATE DETECTIVES ARE still snooping. Experts are still analyzing paper and paint. But, most important, everyone involved in the meteoric rise or precipitous fall of the watercolors known as the "Canyon Suite" is still blushing. And losing sleep.

Billed as a stunning discovery 12 years ago--a surprise glimpse into the early Texas years of iconic American artist Georgia O'Keeffe--the whole episode suddenly morphed last fall into the most embarrassing scandal to hit the National Gallery of Art in years.

Whether a grand deception or just a garage-sale dream gone wrong, it never should have happened. The warning signs were there from the start, but they were swept away by a tsunami of money and wishful thinking. It crashed back to earth last fall, when it was revealed that the 28 watercolors found in an Amarillo, Tex., garage in 1987, coveted by the National Gallery and, finally, purchased for $5 million for the Kemper Museum of Contemporary Art in Kansas City, Mo., were, at best, not by O'Keeffe--and, at worst, were fakes.

That jaw-dropping condemnation came from a committee of scholars who'd spent six years compiling the Georgia O'Keeffe catalogue raisonne--the definitive, chronological, two-volume, 20-pound book--listing all her accepted works and the history of their ownership from the time they left the artist's hands to the present. It is the art market's bible on the artist, and any piece omitted, in essence, is rendered worthless. A joint project of the National Gallery and the Georgia O'Keeffe Foundation, the catalogue has also produced a monumental irony: The bevy of scholars who ultimately trashed the "Canyon Suite" by leaving it out are the very same scholars who had praised it--even begged for it--after its discovery in 1987, a year after the 98-year-old O'Keeffe died.

Their embarrassment was well earned: The art world runs on expertise. But for the power of expert opinion to bestow authenticity and, as happened here, manufacture rarity and value, the "Canyon Suite" fiasco never would have happened.

But it did, with help from an overheated '80s art market, overheated imaginations and, let's face it, greed. With O'Keeffe dead only a year, dealers and scholars were swarming over New Mexico, where O'Keeffe had spent the last half of her life, trying to wrest every farthing and fact from her legendary career. Fortunes were being made, most visibly by Santa Fe dealer Gerald Peters, 53, who'd become the primary outlet for O'Keeffe's estate.

After a year of negotiations, Peters bought the watercolors of the "Canyon Suite" from Terry Caballero of Amarillo for $1 million. Six years later, after a masterly packaging job, aided and abetted by the enthusiasm of the National Gallery, Peters sold them to Kansas City banker and museum founder R. Crosby Kemper at a 400 percent markup.

It all went up in flames last fall, when Kemper, without even a courtesy phone call, received a letter from the National Gallery saying that the catalogue raisonne would soon be published, but without the "Canyon Suite." The letter said their scholars had changed their minds, and no longer believed the paintings to be the work of Georgia O'Keeffe. No further explanation was given.

Kemper went public with his fury in the Wall Street Journal. Fearing a lawsuit, the National Gallery clammed up. In January, a delegation from the gallery, including catalogue raisonne author Barbara Buhler Lynes, curator Ruth E. Fine and paper conservator Judith C. Walsh, flew to Kansas City to explain their conclusions. Halfway through, Kemper excused himself and called his lawyers. He wanted his money back.

Just 'a Lark'

Since then, having refunded Kemper's $5 million and then some--he added a $2 million O'Keeffe painting from his own collection by way of apology--Gerry Peters has been staring bleary-eyed into the ruins, trying to figure out what happened and where he went wrong. The answer may be gallingly simple.

"We did it as a lark," Terry Caballero said by phone from her home in Amarillo. "We just found these watercolors in a box in my father-in-law's garage when we were helping him move. And we were astonished because they looked so much like O'Keeffe. We never said they were by O'Keeffe. . . . They were nothing but pieces of paper until Juan Hamilton [O'Keeffe's assistant and heir] said they were [by O'Keeffe]. I knew how these things worked--I used to be an antiques dealer. But we weren't trying to deceive anyone."

They didn't have to: Peters now concedes he did a fine job of deceiving himself.

Through a friend and agent in Santa Fe, Caballero's "find" was moved from her father-in-law's garage to Santa Fe, where Hamilton saw them and said he thought they were by O'Keeffe. That was enough to bring dealers and auction houses scrambling to make bids. It took a year of legal machinations, but Peters finally signed an unusual deal in which he promised to purchase the 29 watercolors that had been sitting in a garage for $1 million, if Hamilton would reiterate his endorsement.

He would, but not in writing. "He was unequivocal," says Peters, a characterization with which others concur. Hamilton, who now lives in Hawaii, did not return a phone call to his home.

"Juan was the man," Peters says, explaining why even a verbal confirmation from Hamilton carried so much weight. "He'd virtually lived with O'Keeffe for 14 years, and his knowledge of her work exceeded anyone's. The fact that he was fooled is my tough luck."

Hamilton may have been the first expert to be taken in by the paintings, but he wasn't the last. Soon every known O'Keeffe scholar had signed on to them, including then-National Gallery director J. Carter Brown, his successor Rusty Powell and gallery curators Fine and Jack Cowart, not to mention Lynes.

Charles C. Eldredge, Hall Distinguished Professor of American Art at the University of Kansas and former director of the Smithsonian's American Art Museum, was commissioned by Peters to write a 20-page essay extolling the "Canyon Suite's" importance in helping us "better appreciate the artist's working methods at a crucial point early in her career."

Lest someone else acquire the watercolors first, National Gallery officials invited Peters to leave them in the museum's storage vaults while they trolled for a donor. When billionaire art philanthropist Kemper was on the hook, they offered to carve his name in marble if he would present this "treasure" to the nation.

In 1994, Kemper bought the "Canyon Suite," but decided to keep the paintings for his new Kemper Museum of Contemporary Art. And thus it was Kemper who got the bad news last fall. He has since wondered publicly whether the committee would have worked so hard to disprove O'Keeffe's authorship if the works had gone to the National Gallery, a contention dismissed by the gallery as "absurd."

The Power of Imagination

While it lasted, the "Canyon Suite" had a sensational ride. During the 12 years between its rise and fall, it toured seven museums and inspired four books and an hour-long PBS special in which the same parade of experts stood before the pictures, music swelling in the background, waxing poetic about this "astonishing" new find. They spoke of how much these newly discovered works had helped them understand the origins of O'Keeffe's nature-based abstractions, which originated between 1916 and 1918 when she was teaching art at what was then called West Texas State Normal College, a training school for teachers in the Panhandle town of Canyon. They mattered because it was here, in this desolate landscape with its awesome canyons and endless sky, that O'Keeffe first fell in love with the Southwest and created the watercolors that some still see as the most beautiful and innovative works of her career.

In a 1916 letter, she described the power of the canyon in a way that also now applies to the "Canyon Suite" itself: a place where "imagination makes you see all sorts of things."

Cowboy Ted

The excitement over the discovery of the "Canyon Suite" was fanned by the tale of its origins, which provided a frisson of romance. It was told in an affidavit given to Peters at the time of his purchase, and signed by Emilio Caballero, one of O'Keeffe's successors as chairman of the West Texas State art department. Caballero claimed the source of the watercolors was Ted Reid, a name that could not fail to get the attention of O'Keeffe experts.

In 1916 Reid was 21, a handsome cowboy and football player studying at Canyon. There he met the 29-year-old O'Keeffe and began an intense relationship that apparently scandalized the town. In a 1917 letter to a friend, written in her poetic, elliptical style, O'Keeffe wrote: "There is no hesitancy in what I feel for him--We meet equally--on equal ground--frankly freely giving--both . . ."

But Reid would soon marry Ruby Fowler, a fellow student at the college. Even so, he and O'Keeffe remained lifelong friends.

Seven decades later, Emilio Caballero wrote an affidavit explaining how he'd received the works: "In the winter of 1975, Ted Reid visited me in my office at West Texas State University. . . . He was carrying a package under his arm which he gave me. He stated that the items contained within would be of interest to me and that he wanted me to have them. I hastily noted on the wrapper the source of the package and added it to [a box of] material I was taking home."

Caballero said he then put the box in his garage and forgot about it until he moved to a new house in 1987. "While cleaning out these lockers in our preparation to move," he continues in the affidavit, "I rediscovered the parcel Ted had given me 12 years before. Since my son had married Ted's granddaughter [Terry Caballero] in 1984, I felt that these things that had belonged to her [late] grandfather should remain in her possession."

Reid had died in 1983, at the age of 88. For years there had been rumors in his family that O'Keeffe had given him some of her work, though it had never surfaced. To this day, all three of Reid's adult children recall family tales of how their grandmother tried to burn some O'Keeffe works she'd found in a trunk when Ted was off at war, but that his sister Capitola stopped her, saying "you can't do that--they're Ted's."

"My grandmother used to call O'Keeffe 'that hussy,' " recalls one Reid daughter who now lives in Alaska. Like her siblings, she says she never saw any O'Keeffes around the house. But she still remembers being embarrassed as a child about the rumors that her father and O'Keeffe had had an affair.

Both she and her sister say they believe it is possible that Reid did have O'Keeffe works, and may well have given some to Emilio Caballero, a close friend. But Reid's son, J.W. Reid, a farmer, World War II bomber pilot and executor of his parents' estate, doesn't buy it: "I don't believe that story one bit," he says. "If he had that many works by O'Keeffe, he would have left them to his children. Or he would have given at least one of them to the school."

One Woman's Trash

According to various documents written by Caballero and his wife, Reid wasn't the first friend to hand him an O'Keeffe original. Twenty years before Caballero says Reid thrust the "Canyon Suite" at him, he and his wife, Mary, had gathered together at least 30 more works on paper, some of which apparently came from files O'Keeffe had accidentally left behind. This windfall came after O'Keeffe's successor, Isabel Robinson, had retired in 1955. In a letter to a dealer in 1978, Mary Caballero described the circumstances: "The storage room of things Isabel had collected for many years . . . was cleaned out. . . . I collected a group of drawings and paintings that appealed to me from the things to be thrown out. I did not know at that time that among the things . . . would be some of Miss O'Keeffe's early work."

What the Caballeros gathered from the files was unsigned, and it might have been the work of any of the many students who had studied there. The Caballeros learned that three of them were O'Keeffes years later, when a reporter for the Amarillo Globe News spied some of the rescued works on their walls, and asked if she could show them to O'Keeffe. Lo and behold, O'Keeffe recognized them as her own. But she refused to sign them, saying they were "no good."

Eventually, after Peters and another dealer bought them for $2,500, O'Keeffe would agree to officially sanction only one, a linoleum print of a redheaded woman.

Several other works from that forray into old files were subsequently purchased from the Caballeros by other Santa Fe dealers--on the theory that if the schoolroom trash had already yielded one authentic O'Keeffe, there might well be others.

One of the dealers who had a look at their trove was Nathaniel Owings, who vividly remembers going through them on the floor of his vacation home in Montana. "They sent me reams of stuff and Isabel Robinson files that had never been opened. And in it, I found five watercolors in folders stuck in this pile of stuff that were clearly by O'Keeffe, one of which I still have. Stylistically, there's no question," says Owings, who had bought and sold dozens of documented O'Keeffes.

It may have been clear to Owings, but none of these works made it into the catalogue raisonne.

"They threw out the baby with the bath water," said an angry Owings shortly after he learned of the omissions. "They simply eliminated everything with a Caballero provenance."

Seeing Red

So what happened? One day the "Canyon Suite" is worth a fortune; the next, it seems, it could have gone into the next yard sale.

"It's not true to say that they were worth $5 million one day and nothing the next," says catalogue raisonne author Lynes, now curator and chief scholar at the O'Keeffe Museum in Santa Fe. "There were 6 1/2 years in between [the discovery and the debunking]. There's a huge difference between what I know now and what I knew when I started. Nobody suddenly changed their mind."

What she knew then was this: A package of watercolors appeared in a place where O'Keeffe had lived and taught. The work resembled other known early watercolors by O'Keeffe. It was linked by oral history to a former lover of O'Keeffe's. O'Keeffe's assistant and closest friend at the time of her death confirmed their authenticity. The leading O'Keeffe scholar in the world had not only ratified that judgment, but amplified it in a 20-page essay. The National Gallery of Art wanted them.

There was one red flag among all the green lights: Wasn't it extraordinary that the same family that had been lucky enough to dig presumed O'Keeffes out of the trash would later have someone walk up and for no clear reason hand over a package of yet more O'Keeffes?

If that odd coincidence raised any concerns, they were ignored. "Everyone wanted to believe that these were real," says dealer Peters.

What finally made that impossible was not the "Canyon Suite's" soap opera provenance, but the very paper it was painted on.

In 1992, six years after O'Keeffe's death and five years after the "Canyon Suite" came to light, the Georgia O'Keeffe Foundation and the National Gallery began the massive project of cataloguing all the art produced by the artist in her remarkable 83-year career. O'Keeffe experts traveled the country, considering every one of more than 2,000 objects claimed to be O'Keeffe originals. Of these, they disallowed more than 250 for various reasons, 25 percent of them from the Caballeros.

More than 1,100 of the catalogue listings consisted of works on paper, like the "Canyon Suite." This huge sample allowed Walsh, the National Gallery's paper conservator, to sort out which papers O'Keeffe had preferred to use over her lifetime--and she was famously picky about it. Walsh--comparing O'Keeffe's work to a data base she had compiled of thousands of samples of old art paper--determined that fully 95 percent of O'Keefe's works on paper from the teens were on a student-grade paper identified as Bockingford cartridge paper, "cream color, of medium thickness and machine made, with a slightly fuzzy surface."

Not a single piece in the "Canyon Suite" was done on cartridge paper. In fact, some are on construction paper, which Walsh has determined that O'Keeffe never used.

But worse than that, Walsh had early on discovered that three "Canyon Suite" watercolors were on paper that wasn't even available until the 1930s--destroying the theory that O'Keeffe had painted these when she taught at Canyon from 1916 to 1918.

Of course, since Ted Reid's relationship with her had not ended in the teens, but lasted a lifetime, O'Keeffe might have given him works over a period of decades. But the stylistic similarities of all the Canyon works to each other, and to that 1916-18 period in O'Keeffe's career, made such a possibility unconvincing.

Another inconsistency Walsh noticed: The technique in some of the "Canyon Suite" paintings was relatively crude. That didn't make sense, Walsh says: O'Keeffe had been making watercolors since 1902, "so her works from this period were not made by someone who was groping for ways to use the medium. She was exploring form and subject matter, but she'd already mastered her medium. She didn't slop over her edges."

It apparently wasn't until one of the final meetings that the managers of the catalogue raisonne project, notably Fine of the National Gallery--who had early on enthusiastically endorsed the "Canyon Suite" as a major find--decided it wasn't good enough to be included.

"It took many years before I went over to the other side," Fine said at the time. "It was not one thing. But there was a moment when I stopped going back and forth. And it was late, after a lot of torment."

But even if in retrospect it becomes clear that these are not O'Keeffes, there is no hard evidence that they are deliberate fakes, a word gallery officials (all but one of them, who uttered it inadvertently) assiduously avoid. The only suggestion that they could be fakes, that there was some attempt at deception, has come from Mark Stevenson, a private paper conservator and former National Gallery Mellon Fellow hired by Kemper after the "Canyon Suite" was shot down by the catalogue. He has seen something sinister in the fact that some of the paper in "Canyon Suite" looks salvaged.

"The first thing a forger looks for," observes Stevenson, "is old paper." Meaning someone looking for vintage paper for its age alone might not be able to be choosy about quality, and might have to include paper of wildly different types, just as in the "Canyon Suite."

Based on Walsh's findings, he says, it is clear O'Keeffe was not the sort to scavenge for paper. Still, in 1916, she was teaching in a Texas frontier town, where cabs were still horse-drawn wagons and shootouts took place within earshot of her boardinghouse. What if she had to swallow hard and use what paper she could dig up until the next shipment came in?

Not likely, say Stevenson and Walsh. There is ample correspondence between O'Keeffe and her friends in New York to show that they were keeping her well supplied with the paper she almost obsessively preferred.

Asked whether she might someday be proved wrong about "Canyon Suite," Lynes replied in writing: "Of course new documentation could materialize, but I find the chances of that happening highly unlikely."

Not Quite by the Book

So if these aren't O'Keeffes, what are they?

We can only guess. Perhaps they were somehow gathered from the same school files and storage rooms where other Caballero works were found and represent a similar mix of things, including attempts by O'Keeffe's students, or other students, to copy or draw inspiration from her work. The truth, for now at least, remains unknowable.

Many people have suffered as a result of the "Canyon Suite's" wild ride, dealer Peters above all. Owner of the world's largest commercial gallery with branches in New York and Dallas, Peters today is a chastened man.

Terry Caballero, who never claimed the paintings were O'Keeffe's, got to keep her sale price. Kemper got a full refund and $2 million interest in the form of the important early O'Keeffe. The National Gallery and the other O'Keeffe experts turned embarrassment into a triumph of scholarship, which highlighted Walsh's significant advance in identifying art paper.

Peters, however, is out millions and has no recourse. Eager to acquire the watercolors before anyone else did, he agreed to buy them "as is," with a legal disclaimer freeing Terry Caballero from any responsibility for what they finally turned out to be.

"I'll never make a deal like that again," Peters says.

In fact, Peters is still trying to satisfy his own nagging curiosity about the Canyon paintings. He has commissioned further technical studies on the paper, which he hopes may determine whether the paint and paper are of similar vintage or whether old papers were used with new paint--a classic forger's trick. He also has a detective on the case, but doesn't expect much there, either.

"I don't really expect to find anything," he says, "but I had to try. This cost me a fortune, not just in money but in reputation. I haven't gone to sleep one night since this happened without thinking about what I did wrong."

And what has he concluded? "We did all the right things," he insists. "We just came to the wrong conclusion."

Well, not all the right things. One wonders whether any of the scholars who heaped early praise on the "Canyon Suite"--or Kemper, for that matter--might have hesitated if they knew about the 29th watercolor--the one Peters bought from Terry Caballero that never made it into the "Canyon Suite."

After making the purchase, Peters sent the original lot of 29 to the late Keiko Keyes, whom O'Keeffe herself had employed to restore early works that had suffered damage over the years.

Way back then, Keyes pointed out that one of the watercolors had a late watermark and couldn't have been painted during the Canyon years. To keep the "Canyon Suite" idea intact--that these were early watercolors from O'Keeffe's years as a humble art-college teacher--Peters then did something that proved to be his undoing. Instead of pursuing further paper-dating tests, he simply eliminated the errant work from the group, put it in his private collection and convinced himself that nothing had changed.

"I just so wanted them to be right," Peters says today. "I'd become their advocate, which was a big mistake. That was the smoking gun, and I just didn't see it."

Another one who didn't see trouble coming is Terry Caballero. Despite her financial windfall, she can't wait for the whole episode to blow over.

"This hasn't been a lark for a long, long time," she said wearily last week. "It's been so traumatic."

Caballero says it has also taken a heavy toll on her father-in-law, Emilio Caballero, and his wife, now both in their eighties and in failing health. (Through her, they declined to be interviewed for this article.) Plus, the whole episode has led to estrangement in the Reid family because of suspicions that family assets in the form of O'Keeffe originals--if they ever existed--should have been shared, but weren't.

"It was just a fluke," Caballero said wistfully.

Some fluke.

Postscript: In 1976, in her only book, Georgia O'Keeffe wrote, "When I knew I was going to stay in New York, I sent for things I had left in Texas. They came in a barrel, and among them were all my old drawings and paintings. I put them in with the wastepaper trash to throw away and that night when Stieglitz and I came home after dark, the paintings and drawings were blowing all over the street. We left them there and went in."