STEVE INSKEEP, host:
Art curators may have a new way to address an old problem: identifying forgeries. A mathematician at Dartmouth College has come up with a technique that could help determine whether you're buying an excellent copy or the real thing. NPRs Joe Palca has more.
JOE PALCA: Sometimes you choose a career. Sometimes it chooses you. Take Daniel Rockmore. Hes currently chairman of the mathematics department at Dartmouth College.
Professor DANIEL ROCKMORE (Dartmouth College): I joke a lot that I am a mathematician by mistake. It was something that I had an aptitude at, but I've always had lots of other interests.
PALCA: One thing Rockmore is particularly interested in is art. And hes found a way to integrate his interest in art with his career in mathematics. Specifically, he's come up with ways of analyzing digital images.
Prof. ROCKMORE: A digital image is a collection of pixels. So the list of those pixels actually describes an image - for example, the digital representation of a painting as a coordinate in a very high - I mean thousands of dimensional space.
PALCA: And who better than a mathematician to make sense of something that has thousands of dimensions? One day, back in 2001, Rockmore was talking to a friend of his who was an art historian in New York City.
Prof. ROCKMORE: And a friend of hers, Nadine Orenstein, is in the drawings department at the Metropolitan Museum.
Ms. NADINE ORENSTEIN (Metropolitan Museum of Art): And I was working on a Bruegel exhibition at the time.
PALCA: Orenstein is talking about Peter Bruegel, the Elder, a Flemish painter of the 16th century.
Ms. ORENSTEIN: The exhibition was of Peter Bruegels drawings and prints.
PALCA: Orenstein says she invited Rockmore to come to the museum to see the exhibition. It featured works by Bruegel, but Orenstein says there were also several drawings on display that were once thought to be Bruegels but werent.
Ms. ORENSTEIN: Our exhibition was really the first time that we were getting together in one place all of the drawings by Bruegel and the ones that were no longer considered to be by him.
Prof. ROCKMORE: And so Nadine was explaining to me the various particular pen strokes that seem to be characteristic of the way Bruegel works, the way he creates a scene.
PALCA: And Rockmore realized that if he had digital images of these drawings, he could use his math skills to design a computer program that would analyze those pen strokes and characterize which were Bruegel-like and which weren't. So he and some colleagues got hold of the digital images, and they wrote the program.
Prof. ROCKMORE: And we did a pretty good job.
PALCA: That was in 2004. Now, as he reports in a recent edition of the journal�PNAS, hes done a better job using a different kind of statistical sampling technique.
For the moment, Rockmore has only tested his program on Bruegel drawings, but theres no reason it couldnt be used for other artists. But Rockmore says even if the program did indicate that a particular work seemed fishy, that wouldnt prove it was a fake. He says his program only says whether a particular drawing is consistent with an artists style, not whether its a fraud.
The Metropolitan Museums Nadine Orenstein agrees.
Ms. ORENSTEIN: That might put a little light bulb in the head of, you know, the scholar and say, oh, you know, here's something we have to further investigate.
PALCA: For his part, Rockmore isnt particularly interested in designing a mathematical tool for tracking down art forgeries. He hopes his work will offer art historians a new way of deconstructing art a way of describing what it means to be Picasso-like or Bruegel-like.
Prof. ROCKMORE: You get at deeper questions about the creation of art, and our experience of art. But, you know, it's probably more fun to report on whether or not that's a fake Bruegel.
PALCA: Although reporting on the science of how we experience art is fun too.
Joe Palca, NPR News, Washington.