Probing prehistoric art to the bone.

In the dry heat of southwest Texas, where the Pecos River melds with the mighty Rio Grande, there stands an archaeological site known as White Shaman.

More than 10,500 years ago, hunter-gatherers dwelled along those riverbanks, seeking shelter in shallow limestone caves abutting the water's edge. Performing rituals there, tribesmen decorated their shelter walls with symbolic paintings.

Archaeologists know little about the lives of these migrant people, so they can only speculate about what the paintings mean. Scientists also wonder about the kinds of tools and materials they used, including the ones adapted to artistic purposes.

Marvin W. Rowe, a chemist at Texas A&M University in College Station, and his colleagues have brought DNA analysis to bear on paint samples from the rock art. Using a technique known as polymerase chain reaction, the team made many copies of ancient DNA fragments taken from two pictographs dated between 3,000 and 4,300 years ago. This yielded large amounts of the so-called histone  4 gene.

Genetic sleuthing led Rowe's team to conclude that the paint's binder, or the base holding the pigment, most likely came from the bone marrow of local deer or bison.

"We are certain that the biological material came from an animal in the order Artiodactyla," Rowe reported last month in Anaheim, Calif., at a meeting of the American Chemical Society. "That order contains the family of even-toed ungulates."

Ungulates include such mammals as bison, deer, elk, rabbits, cattle, sheep, goats, and antelopes, which are native to the Southwest. Camels, giraffes, and llamas--also ungulates-- seem less likely to have populated prehistoric Texas.

Further DNA study and some common sense have led the scientists to focus on white-tail and mule deer, elk, American bison, pronghorn antelopes, javalinas, and rabbits as candidates for the bone marrow that apparently served as the paint binder. Ancient bison teeth from the nearby Bonfire slaughter site provided benchmark DNA from 12,500 years ago.

"At first, we had no idea what to look for," Rowe says. "So we chose the histone gene because it changes little from one species to another. We then compared the DNA sequences from the rock art with known sequences, looking for a match. This led us to the ungulates."

To ensure that they had truly ancient DNA fragments, not recent contaminations, the researchers amplified some samples tainted with known DNA and others with no added fragments. They also homed in on DNA from mitochondria, which evolves nearly 10 times as fast as DNA from cell nuclei, says Rowe. "Mitochondrial DNA shows wider variation within an order, so it should provide species identification."

This approach also rules out the problem of contamination by bacteria, which have no mitochondria, Rowe adds.

In the absence of other ethnographic information, rock art, Rowe believes, constitutes an important "window into prehistoric thought."