Until recently, the common factor in all the science used to figure out if a piece of art was forged was that it was concerned with the medium of the artwork, rather than the art itself. Matters of style. and form. were left to art historians, who could make erudite, but qualitative, judgments about whether a painting was really good enough to be, say, a Leonardo. But this is changing. A paper in this week's Proceedings of the National Academy of Sciences by Hany Farid and his colleagues at Dartmouth College in New Hampshire uses statistical techniques to examine art itself—the message, not the medium. Dr. Farid employed a technique called wavelet analysis to examine 13 drawings that had at one time or another been attributed to Pieter Bruegel the Elder, a 16th-century Flemish painter. He also looked at Perugino's 'Madonna with Child', a 15th-century Italian masterpiece lodged in the college's Hood Museum of Art. He concluded, in agreement with art historians, that eight of the putative Bruegels are authentic, while the other five are imitations. In the case of 'Madonna with Child', he analysed the six faces in the painting (Mary, the infant Jesus and several saints) and found that three of them were probably done by the same painter, while the other three were each done by a different hand. The view that four different painters worked on the canvas is, he says, consistent with the view of some art historians that Perugino's apprentices did much of the work, although there is no clear consensus among art historians. As sceptics will doubtless point out, this is a small number of images. Furthermore, Dr. Farid knew before performing the analysis what results he expected. But he is the first to acknowledge that it is early days for his methodology. He hopes to study many more paintings. By looking at large numbers of paintings that are universally believed to be authentic, Dr. Farid hopes to be able to examine doubtful cases with confidence in the future, Even with the Bruegels—real and imitation—though, Dr. Farid's results are persuasive. It is tricky to describe exactly what it is that distinguishes the real ones from the imitations, but Dr. Fetid says that it can be thought of as the nature of the artist's brushstroke. Unlike some analyses of Jackson Pollock's work that have been done over the past few years by Richard Taylor of the University of Oregon, Dr. Farid says his technique could, in principle, be used for any artist. What Dr. Farid did was to convert each work of art into a set of mathematical functions. These so-called wavelets describe particular parts of the image as a series of peaks and troughs of variable height and wavelength. By expressing an image this way, it is possible to compress that image while losing very little information. The sums of the wavelets from different images can then be compared. Once he did this, Dr. Farid found that the types of wavelets used to express authentic Bruegels were noticeably different from those used to express the imitations. (The Perugino was analysed by treating the six faces as distinct paintings.) It seems that curators may soon be able to add another weapon to their anti-forgery arsenal. The message Dr. Farid's work focuses on is close to ______.