The art sleuths gathered early Wednesday in the National Gallery's West Building, hoping to pick up George Inness' tracks.

Inness, the 19th-century American landscape painter whose work will be the focus of a National Gallery exhibition opening tomorrow, had a reputation for painting over his works several times, experimenting until he came up with the precise combination of elements that satisfied him. "There are many, many stories of how Inness would impulsively paint and repaint his works, changing landscapes to seascapes, adding and taking away figures," said Nicolai Cikovsky Jr., the gallery's curator of American art, who headed the investigating team, "and we'd like to see if we can verify those stories."

Using a video camera equipped with an infrared tube, Cikovsky and his crew sought to look at the paintings -- already in place for the show -- as Inness originally conceived them, by revealing the sketches underneath the finished work. At Cikovsky's order, gallery officials bathed selected works with high-intensity light and turned the camera on them, one by one. As the initial images came into focus, the assembled group -- Cikovsky, Painting Conservator Sarah Fisher, exhibition organizer Michael Quick of the Los Angeles County Museum, and various gallery staffers and interns -- craned their necks toward the monitor.

"Nice cows," quipped Fisher as the camera zeroed in on one painting, "The Rainbow," revealing only a cluster of cows already visible to the naked eye.

"But they look exactly like they do in the pictures," sighed Cikovsky. "No hidden cows."

Cikovsky said he had picked out certain paintings because they showed visible evidence of alteration. In "Sundown," for example, which features a solitary figure carrying hay against a pastoral background, there are faint vertical lines around the figure, suggesting that it was painted on top of a tree.

The infrared technology, first used by the National Gallery in 1982, is especially effective in revealing underlying charcoal sketches, since black is highly visible under an infrared camera. Cikovsky and the other investigators didn't know, however, whether Inness sketched with charcoal.

*The same procedure was used to find an additional boy and more sails in the background of Winslow Homer's "Breezing Up," one of the gallery's best-known masterpieces, and similar inspections revealed significant changes in charcoal drawings underlying other Homer paintings.

The alterations in Homer's work detected by the infrared camera, Cikovsky believes -- particularly the ones in "Breezing Up" -- have provided valuable clues from which art historians can evaluate the painter's ideas about his work.

In "Breezing Up," Homer "was obviously trying to make a more complicated image -- more boats, sails, evidence of the harbor -- and thought it was too crowded," Cikovsky said. "That he removed a boy and added on an anchor -- a 19th-century symbol of hope -- suggests he intended the painting to convey a sense of optimism.

"If you can see an artist doing something, you've caught him in his tracks," he continued. "You can then ask why. Why did he make this change? If there was a pure landscape underneath and you found he added a figure, it tells you something about his ideas about landscape. If, say, you found a third figure instead of two in 'Gossip,' it would change the narrative flow of the painting."

In fact, the infrared reflectogram did not show an extra figure in Inness' "Gossip," only some evidence in an underlying paint layer that a different style of hat had been placed on the head of one of the existing figures. Turning the camera on "Winter at Montclair," a bleak, snow-covered scene, the investigators found a painted-over sapling protruding through the icy ground.

Neither discovery involved an underlying drawing, just layers of paint, and neither proved to be of great interest to the assembled group.

Infrared reflectography, unlike some other scientific methods of examining paintings, is mobile and relatively cheap, costing only between $5,000 and $10,000 for equipment comparable to that used by the National Gallery. The technology first became available in the late '70s with the invention of the infrared vidicon tube, which can turn an ordinary television camera into an infrared scanner.

X-rays, which museums have used to examine paintings since the turn of the century, do not reveal drawings as well as infrared does, although an X-ray generally gives a better image of underlayers of paint. The major disadvantage of X-ray technology, however, is that paintings must be carefully transported to the machine, making the X-raying of a large exhibition impractical. The infrared scanner can be transported to wherever the paintings are hanging.

Laser technology, yet another way of probing beneath artists' finished work, was used by the Louvre to determine that Leonardo da Vinci's "Mona Lisa" once wore a string of beads. But it too is more expensive and less mobile than an infrared scanner and monitor.

*For several hours, the National Gallery investigators continued searching for evidence of drawings beneath the Inness paintings. They wheeled the machines from painting to painting, always staring expectantly at the screen and finding . . .

Nothing.

Inness apparently did not sketch with charcoal; the minor alterations between layers of paint were all that turned up. The artist's original notions about his paintings will remain hidden, at least until some stronger technology comes along to reveal them.

"There was an absence of underdrawings in those particular paintings," said Gary Carriveau, head of the National Gallery science department. "When they're there, they're very, very clear."

Or, as Cikovsky said while disappointed gallery staffers put away the video equipment: "Art triumphs over science once again."