THE STUDY of art history is nowadays perceived as a delicate pursuit, but the 19th-century scholar, saddlesore and dusty, did not think it such. He rode from town to town searching out old pictures. He studied them by candlelight. Struggling to memorize the works of art before him, he often spent long afternoons drawing in the gloom of some old Italian church. To Succeed at connoisseurship a century ago four things were essential: sharp eyesight, a good memory, a sketchbook and a horse. The game, of course, is different now. The late Erwin Panofsky, that paragon of scholars, described the main change tersely: "The man who has the most photographs wins."

New photographs, though flawed, have forever altered the study of old paintings.

The photograph in black-and-white frequently misleads, it says little about color and nothing about scale. But as a kind of note, as a crucial aide to memory, it has become the art historian's second most important tool. The book remains the first, but except for the printing press, no other innovation -- not the light bulb, or the telephone, or the horse-replacing car -- has done more to expand our knowledge of old art.

The National Gallery of Art was founded with a gift of about 100 paintings in 1941. The public knows its drawings, its many prints and paintings, but its largest hoard of pictures -- of Western art and architecture -- will never be displayed.

The Gallery today owns about a million photographs. Until 1970 there were only a few thousand photos in its archive. By the end of this decade there may well be three million.

The Gallery's new Center for Advanced Study in the Visual Arts will next week appoint its first half-dozen fellows, but the two main tools they'll use there while conducting their researches are already being forged. One is an art library that already holds more than 75,000 volumes. (It, too, is due to triple.) Beneath it, in the basement, expanding almost daily, is a photographic archive -- to be used by scholars only, not open to the public -- that one day will compare with the finest in the world.

There are other photo archives here -- at the Library of Congress, at Dumbarton Oaks, at the National Collection of Fine Arts -- but the Gallery's is growing faster than the others.

By the hundreds, by the thousands, in envelopes and cartons, photographs are pouring into the East Building. Today they cost, on average, say, $5 each, though they used to cost much less. Most now at the Gallery were purchased with the aid of the Kress Foundation grant, which has, so far, provided for that purpose nearly $1 million. Ruth Philbrick and her staff of seven -- five art historians and two typists -- are working there full time now.

Once refined by scholars, and enhanced by high technologies, such an archive, theoretically at least, could give art historians an almost universal museum without walls. Any college student in Idaho or Kansas someday may be able to summon to his TV screen, say, all the prints of Rembrandt, or 10,000 crucifixions, or all of the surviving Medieval pictures showing devils that have horns. Should advanced computers, and miles of tape or color film, be plugged into his system, he may be able also, while still sitting at his console, to take a tour of Venice, pausing at each church that catches his attention, inspecting its interior, even calling up its plans. Such a television journey is more than just a pipe dream. Henry A. Millon, professor-in-charge of the Gallery's new Center, says that such a TV tour -- of the present and the past of Aspen, Colo. -- may even now be taken on the screens at MIT.

Machinery far less complex already is expanding the photographic archive. In addition to the million photographs (on paper) now stored in its stacks, it owns 400,000 more images that have been miniaturized on microfiche by other institutions. Photographs collected by the Witt Library in London, by Germany's Marburger archive, and by the Dunlap Society (which has begun to publish drawings, prints and photographs of this city's major monuments), may already be seen in the gracious reading room of the photographic archive. It is even possible that all important images in all important archives someday may be reproduced, computerized, and then combined in some super archive accessible to all.

But that day has not yet come. A visit to the archive brings one back to earth. No computers hum there. Of the million pictures there, no more than 10 percent have been catalogued and filed to the scholars' satisfaction. And that painstaking, mind-numbing work is being done by hand.

Many photos there have not yet been identified. There are perhaps as many more that have been identified incorrectly. The authorship of paintings often is in question. Is that a Leonardo, as was once believed, or is it a studio work, or a picture by a follower? Before it can be catalogued by country, period, location, artist, subject, title, it must be scrutinized by scholars. Are the works of Copley American or English? Are Picasso's French or Spanish?

The photographer who photographed the objects sold at auction at Sotheby Parke Bernet between the 1930s and 1972 has sold his prints to the Gallery. So Have Seligmann's and Durand-Ruel's, the well-known dealers. Heavy cardboard boxes filled with photographs submitted to Art International, the modern magazine, also have been purchased. These images, and countless more, must be scrutinized and catalogued before they're filed away.

When the archive of art photographs becomes encyclopedic, achieves what one might call a kind of critical mass, it can become a near-essential resource. Jerry Mallick of the Gallery the other day arranged a little demonstration of his archive's usefulness. Spread out on a table there were more than a dozen pictures that, at least at first glimpse, all looked much the same.

All showed two infants hugging: Christ and John the Baptist. The first photo showed a drawing made by Leonardo; the second showed a painting from Milan, done by someone else, but in the master's style; the fifth one reproduced a later Flemish painting; and the last one in the series portrayed an old woman leering at the youngsters who had, by now, become a necking boy and girl.

The legends on the photographs were not to be trusted -- one picture, now in Naples, had been variously assigned to "School of Leonardo," to Cesare da Sesto, and to Joos van Cleve. But though the words misled, the photos on the table convincingly delivered a complex visual message. They spoke of Leonardo's influence on others of the transmission of images from Italy to the Low Countries, and of the fading line between the sacred and lascivious. It is difficult to know how without sets of photographs such points could be made.

The photographic archive, a development of our country, has yet to be perfected. Because color printing has not yet been perfected to the Gallery's satisfaction, because they go green or pinkish and their colors fade, most of the photos stored there are still black-and-white. Computerization of the archive has yet to be attempted, and even preliminary cataloguing is far from complete.

Historians need no longer ride from church to church searching for old frescoes, books can tell them much, but given a great archive, an archive far superior to any now existing, they may someday discover undetected patterns and new sorts of order in that sea of images that is the history of art.