Until last month, a quiet surprise was tucked into the dimly lit rotunda of the Morehead Planetarium on the UNC-Chapel Hill campus. Among a circle of portraits in the small and little-known gallery was the painting of a young 17th-century woman identified as Rembrandt's sister, Liesbeth van Rijn, poised as if in the very shadows of the room. Despite its subdued colors and plain subject, the portrait seemed like an exciting part of the collection: the only Rembrandt painting known to exist in North Carolina. If it were truly a Rembrandt. But it wasn't.

In June a university employee removed the placard that read "Rembrandt van Rijn" and replaced it with one saying the portrait had more likely been painted by Isack Jouderville, an apprentice from the master's workshop in Amsterdam.

Reattributions have happened quietly, without fanfare, hundreds of times over the past few decades in museums across the country. The Morehead's might have been just one more unnoticed downgrade had the plate been changed 20 years ago when a team of experts first concluded that the painting wasn't a Rembrandt after all.

But it's hard to let the masters slip through your fingers.

N.C. Museum of Art curator Dennis Weller knows that sinking feeling. He was an assistant curator at the National Gallery of Art in Washington in 1993 when two supposed Rembrandts in the collection were reattributed. NCMA has seen 10 paintings that it had purchased as Rembrandts or Rubenses reattributed to lesser hands.

"Some of those stars have dimmed," said Weller, whose specialty is northern European art.

The upheaval has been the result of an ongoing reappraisal of works once thought to be by those artists. Now, in the 400th anniversary year of Rembrandt's birth, about half of what he was once thought to have painted has been credited to his students and other artists -- about 200 paintings in all.

Recent research casts an increasingly skeptical eye on the few experts who attributed so many paintings to Rembrandt and Rubens around the turn of the last century. Some of their scholarship is considered outdated or even suspect because they were paid for their opinions.

One of those experts was the North Carolina art museum's first director, the late Wilhelm R. Valentiner, who helped stock its initial collection, drawn from the artists of that era. Valentiner, working under very different standards and without today's technology, was at the center of a series of mistaken identifications that have disrupted collections around the world, including in the Triangle. Two recent publications have questioned his motives.

"It's just the nature of the beast," Weller said. "Thankfully, otherwise art historians would have nothing to do 300 or 400 years later sorting it all out."

Masters had their factories

It's baffling to think that history could so carelessly lose track of what Rembrandt and Rubens, or van Dyck and Vermeer, painted themselves or left to their pupils. But the golden age of Dutch and Flemish masters is closer than you might think to Andy Warhol's Factory and Thomas Kinkade, the mall outlet mass-marketer.

When Peter Paul Rubens, the star of the early to mid-17th century, was commissioned to paint, contracts spelled out whether he would be the primary painter or would merely touch up the piece after his best pupil or other students did most of the work. At the least, Rubens typically did an oil sketch and spelled out the composition and sometimes the colors, and someone in his studio would work it up to scale, Weller said. Or someone in the studio designed it and Rubens did the painting. He employed specialists in still-life, landscapes and other forms.

Similarly, Rembrandt van Rijn ran a mid-17th-century workshop of at least 100 pupils. His best pupil at one time was Anthony van Dyck, who became nearly as famous as his teacher.

It was often difficult to distinguish the masters' works from the best of their pupils'. The paintings were bought, sold and dispersed throughout the Western world, often without documentation. As time went on, the paintings' origins became harder to trace.

By the early 20th century, German-born Valentiner, his colleague Ludwig Burchard and a few others were widely acknowledged as the experts on whether paintings were done by the masters or their contemporaries. They were the go-to guys for art dealers, galleries and collectors.

America's new wealthy class had a big appetite for European paintings but little experience discerning their worth, so the Fords and the Fricks relied on the European experts. The more cautious among them only bought paintings that came with certificates of authentication, according to Kasia Pisarek, an independent researcher in England.

"The money was in America, the expertise in Europe," she said.

Valentiner once marveled in his diary, now stored at the state archives in Raleigh, that Americans were willing to pay any price for art that anyone who hadn't grown up in Europe would have difficulty understanding. And so what should have been independent scholarship was sometimes driven by the marketplace.

Pisarek found this comment by American art historian John C. van Dyke, writing in 1907, about the prevalence of forgeries:

"Our American millionaires, whose brains have stood them in such good stead in the accumulation of money, seem to part with common sense when it comes to the buying of pictures. They are a shining mark for European sharpers, who have found that an enormous price asked is considered a guarantee of genuiness in America."

No one has ever called major scholars such as Valentiner and Burchard sharpers. But they couldn't help being influenced by their relationships with the art dealers and commercial galleries that paid them to render opinions, says Timothy Riggs, curator of collections at the Ackland Art Museum at UNC-CH.

There was an "expansionist" mood among the experts, driven in part by a resurgence of Rembrandt's reputation during the 19th century, Riggs said.

"People were discovering paintings and were very pleased to discover them," Riggs said. "Of course, the owners of the paintings were very happy to have them."

And so were the museums that eventually received them.

Tracking down the real thing

A calamitous mix of anguish and triumph, "Gideon Overcoming the Midianites" hangs in the 17th-century Flemish room at the N.C. Museum of Art in Raleigh.

Burchard didn't think much of the painting at one time. In 1950, Pisarek found in her research, he warned a colleague not to buy it, saying it looked like it had been painted by a contemporary of Rubens.

Four years later, Burchard wrote a certificate of authenticity to a commercial dealer in London affirming the painting as a Rubens. That's when the Raleigh museum bought it as part of a $1 million spree authorized by state legislators to get the museum started, with Valentiner giving the final OK to each piece.

With estimates of as many as 400 Rembrandts floating around, the art world came to recognize the confusion between work from a master's hand and work from someone else's. The Dutch in 1968 launched the Rembrandt Research Project, which has been re-evaluating and cataloguing the artist's oeuvre ever since. A similar project, Corpus Rubenianum, which focused on Rubens, has been going on almost as long.

Project teams tracked down paintings in museums with significant holdings, such as the N.C. Museum of Art, and in such little-known corners as the Morehead Planetarium gallery.

"Gideon" attracted the attention of visiting scholars who cast increasing doubt on its authenticity. In the 1980s, the museum changed the attribution to a "follower" of Rubens. During that decade, NCMA had to downgrade nine other works from Rubens and Rembrandt to unknown or lesser-known artists.

Reattribution doesn't necessarily reduce a work's value or popularity, Weller said. But the fact remains that NCMA at one time thought it had four Rembrandts and seven Rubenses. Now it has a single Rubens.

Museums and private collectors don't always relinquish their claim of an old master without a fight. Some museums have resisted the research project's findings. The Raleigh museum saw a "Rembrandt" vanish when its authenticity was challenged.

"Portrait of a Woman With a Black Veil" had been given to NCMA on an extended loan in Valentiner's memory in 1959. When the Rembrandt Research Project later decided it wasn't the master's, the owners took it back rather than have it relabeled as the work of an anonymous painter in Rembrandt's circle.

How is a museum-goer to know?

The never-ending game of second-guessing poses a problem for the average museum visitor: how to tell if what you're looking at is what it is advertised to be.

"It's not an exact science," Riggs said. "It's a matter of knowing. Connoisseurship is a little like espionage in the world of diplomacy. It has this reputation as secret knowledge: 'Who are these guys who walk into a gallery, and how do they know?' "

Valentiner and Burchard perfected an expertise in evaluating paintings based on their ability to discern the work itself rather than relying only on documentation. Today researchers have laboratories and online photo archives. Experts are no longer allowed to accept money for their opinions, and the field has become self-policing.

The turn-of-the-century experts are still widely respected, although it is acknowledged that Valentiner was overly optimistic in identifying Rembrandts. Pisarek is even more skeptical. The Sorbonne graduate has been studying Rubens' paintings for a decade and is completing a doctoral degree on the relationship between connoisseurship and trade in the 20th century.

Her most recent research was published in April in the journal ArtWatch UK and covered in the London Times. She says she has found more than 80 such misattributions by Burchard.

"Burchard was the biggest authority on Rubens, yet he attributed extremely weak paintings to Rubens," Pisarek said in a phone interview from her home near London. "By today's standards they would be considered imitations, not even studio works, but imitations. If he couldn't tell the difference, how could we trust him?"

She also finds Valentiner's attribution of more than 180 paintings to Rembrandt troubling. Many of the 10 demoted paintings Valentiner brought to the Raleigh museum were purchased from a single gallery, she notes.

Another skeptic is Stephen C. Jordan, author of a biography published last year, "Bohemian Rogue," about Hollywood art forger John Decker. Jordan and other researchers believe Decker painted "Bust of Christ," which hangs in the Fogg Art Museum at Harvard as a Rembrandt.

Jordon says Valentiner authenticated the painting in exchange for $600, although it has no records documenting its history before 1939. Harvard concedes that it might not be a Rembrandt but insists that it is a 17th-century Dutch painting and not a fake.

Experienced historians such as Weller and Riggs are not as skeptical of Valentiner's motives.

"It's difficult for us to put ourselves in the position of these scholars who were often paid minimal salaries, who depended on a comfortable income from the privilege of giving opinions," Riggs said. "The tendency now is to discount those opinions, even though those scholars were not fools, not crooks."

Clinging to its false lineage

For half a century, the small portrait of Liesbeth van Rijn with her sly expression hung in a forgotten corner of the art world, stubbornly holding on to its false lineage. It was a secret gem for art lovers.

"The first time I walked in I was surprised -- this is a Rembrandt?" said Lee Sorensen, an art librarian at Duke University.

The Morehead Foundation's records track the portrait back to private French owners in the later part of the 18th century and follow it through a marquis' collection sale in Paris in 1912. Newspaper stories reported that it fetched the highest price of the day, $73,000, before it traveled to private collections in the United States.

Records don't tell us whether Valentiner had anything to do with authenticating the portrait.

"Between 1910 and 1920 he discovered 120 Rembrants," Pisarek said. "It could have been one those."

And he certainly knew about it. As director of the Detroit Institute of Arts in 1930, he chose it for the first major exhibition of Rembrandt works in the United States, and the next year included it in a book.

The painting ended up with the wealthy North Carolina industrialist John Motley Morehead and his wife, Genevieve. After his wife's death, Morehead donated the portrait and six other works to establish the planetarium gallery in 1949. Valentiner tried to borrow it again in 1956 for his first big splash at NCMA, a show of 25 Rembrandts, but UNC refused, correspondence at the state archives shows.

Scholars with the Rembrandt Research Project visited the planetarium gallery in 1970 to inspect the unsigned and undated picture using ultraviolet radiation, according to their report. Thirteen years later, when the painting was moved to the Ackland Museum for an exhibition there, the Ackland's director provided project scholars additional information about its dimensions.

When the portrait was exhibited at the Ackland, it was no longer called a Rembrandt, but was attributed to someone in his circle, Riggs said. But when it returned to the rotunda, it apparently took its place beside its old label identifying it as the Dutch master's own work.

By then researchers had begun to see how similar the portrait was to other paintings by Jouderville, which fell short of the master's skill. They also began to think it might not even be Rembrandt's sister, suspecting that his assistants freely duplicated the portrait and might have used different models.

The Rembrandt project published its report in 1986 and said the painting was well-preserved and was probably done by Jouderville, whose work never ventured far from his master's influence. In 1991, a Dutch documentary about the Rembrandt project mentioned the Morehead painting as a case of mistaken identity, and the next year the Chapel Hill Herald newspaper ran a story about it. The article quoted the planetarium's director at the time saying the painting would be relabeled to reflect the change in identity "very soon."

Fourteen years later, UNC was still calling it a Rembrandt. The painting had simply fallen between the bureaucratic cracks.

The planetarium does not have an art curator, since it is not a museum. The paperwork belongs to the university's historic collections office. Staffers there knew about the controversy as early as 1992 but never changed the label because they are not art curators, either.

And it's hard to let an old master go.

"If you own the painting, you're naturally inclined to trust the guy who says the nicest things about it," Riggs said. "As long as there is doubt about an opinion, there is an inclination to give the painting the benefit of the doubt, to leave the Rembrandt ticket up until somebody comes along and says, 'Quite frankly, there's no one that takes this seriously any more. You really should take that down and put up a label that more accurately reflects it.' "

After The News & Observer inquired about the matter last month, the university relabeled it as Jouderville's apparent work.