Art Authentication

When a team of art experts recently sought to determine whether or not "The Holy Infants," an obscure painting purchased for $1,500 at a London auction, was in fact a lost work by Leonardo da Vinci, they didn’t just rely on their voluminous knowledge of art history or aesthetic sensibilities honed by countless hours spent studying the Renaissance master’s body of work. Instead, they took the painting to a laboratory. They examined the painting’s surface under a microscope, extracted a fragment of wood from the frame so that it could be subjected to carbon dating, and bombarded the artwork with high-energy protons to identify the elements present in the pigments. They even brought in a fingerprint expert, who lifted a 500-year-old print from the painting and compared it to one found on an early sketch of Leonardo’s "Last Supper."

That’s just one example of how art authentication — that is, evaluation of a painting or sculpture to determine whether it is the genuine work of a master and not, say, a student’s copy or a clever forgery — has become an increasingly high-tech endeavor. Long gone are the days when authentication was the province of aficionados such as Bernard Berenson, the flamboyant 1930s American art critic and collector, whose judgments sometimes were based upon the emotional and physical sensations that he felt when staring at a painting. (If he experienced vertigo or felt depressed, it suggested to him that a painting was a forgery.) Instead, many of today’s art authenticators are part scholar, part scientist and part detective, and their tools range from particle accelerators to computer programs that analyze an artist’s brushstrokes for telltale patterns. However, while technology promises to eliminate much of the educated guesswork and perhaps solve some of the long-standing mysteries of the art world, some of the new tools are themselves controversial. Even advocates of scientific art authentication caution that their methods are not yet infallible. Though recent advances in computer analysis of paintings hold great promise of someday providing the definitive last word, at present it still remains easier to spot a fake than it is to conclusively identify a genuine masterpiece.

Museums and collectors need the services of art authenticators because, as the late art authenticator and microscopic analysis expert Walter McCrone wrote, copying and fakery are among the art world’s most ancient traditions, dating back to the Phoenicians’ counterfeit Egyptian pottery and the Roman forgers who churned out copies of coveted Greek paintings and sculptures. The myriad Italian Renaissance forgers included Michelangelo himself, who as a youth sculpted a sleeping cupid, treated it with acidic earth to simulate the appearance of an antiquity, and then sold it through a dealer to a cardinal. (When the fraud was discovered, the cleric actually let Michelangelo keep his percentage of the sale, because his sculpting was so impressive.) By the 18th century, according to McCrone, art fraud had evolved into a highly sophisticated profession, with forgers baking their canvases in ovens and mixing chimney soot into their paints in order to simulate the effects of age. (Some fakers even went through the charade of restoring the intentionally damaged paintings.)

In the 1930s and 1940s, Dutch painter Henricus Antonius van Meegeren — perhaps the most infamous art forger of all time — set out to embarrass the critics who’d snubbed his own work by creating a series of paintings in the style of 17th-century master Johannes Vermeer, and then passing them off as the real thing. Sure enough, the arbiters of taste praised the "lost" Vermeers, and collectors were willing to pay hefty sums for the fakes, because van Meegeren was no ordinary crook. He spent years developing a method for applying phenol and formaldehyde to paint to simulate aging, and even used the same sort of brushes as Vermeer in order to more faithfully replicate his technique.

Such fakery continues to this day. Thomas Hoving, former director of the Metropolitan Museum of Art in New York, once estimated that 40 percent of the artworks on the market were either outright fakes or "half-forgeries," paintings by obscure artists that have been altered to make it appear that someone famous created them. Art forgery has grown so audacious that some even dare to counterfeit relatively recent artists. In the 1990s, a high-profile auction of works supposedly by artists such as Jasper Johns and Georgia O’Keefe had to be cancelled, after authenticators exposed them as forgeries. In January 2006, a painting attributed to Andy Warhol, valued at nearly $2.6 million, became worthless when a board of art authenticators challenged its genuineness.

Conversely, when a previously unknown work by a famous artist is identified and authenticated, whoever originally bought it cheaply will make out pretty well. In 2001, for example, the Duke University Museum of Art acquired a painting in France with only two marks of identification: the initials "FG" and the year "1787." Art authenticators determined that it actually was a work by Francois Gerard, and as such probably worth two to three times what the museum had paid for it. Another painting was thought to be just an anonymous copy of a famous 18th-century Diego Velazquez painting, until Walter McCrone determined that it had been painted by 19th-century French painter Edouard Manet. Not surprisingly, the newfound "lost" Manet skyrocketed in value.

McCrone’s examination of the suspected Manet painting, which he described in a 2001 article in The Sciences, showed the extent to which scientific analysis has augmented — if not quite superseded — traditional authentication methods such as documenting an artwork’s history or submitting it to visual inspection by aficionados. McCrone began by using a fine-tipped needle to remove microscopic samples of pigment from various spots on the painting. Using a microscope, McCrone then studied the size and shape of the pigment particles. McCrone discovered that although the basic pigments were common to artists of the mid-19th century, these particular samples also had idiosyncratic characteristics — the white pigment, for example, was made up of a rare form of lead carbonate, and its particles were arranged in an unusual column-like fashion. McCrone then persuaded museum curators to allow him to take microscopic pigment samples from two known Manet paintings. He subjected all the samples to electron-probe X-ray microanalysis, an even more exacting method in which a beam of focused electrons is aimed at a pigment particle to generate X-rays, which then are analyzed to measure the precise amounts of various elements present. The result? The white pigment in the suspected Manet had come from the same production lot — and possibly the same paint tube — as the two known Manets.

But paint is just one of the physical clues perused by scientific art authenticators. If a painting is on canvas, the cloth’s fibers can be analyzed under a microscope to determine their probable age and origin. (McCrone once discovered that a purported 18th-century Russian religious painting contained Dacron polyester, a synthetic that didn’t become available until the 1950s.) Paintings done on wood panels can be dated using dendrochronology, the study of how tree ring patterns were affected by historical changes in weather. Ultraviolet light can reveal portions of a painting that have been altered — and possibly uncover a forger’s attempt to make an obscure artist’s work appear to be that of a master.

Recently, a Dartmouth College team headed by computer scientist Hany Farid unveiled a new tool — computer analysis of paintings. According to a 2004 article published by Farid and his colleagues in Proceedings of the National Academy of Sciences, the method uses a process called "wavelet decomposition," in which a painting is converted into a digital image and then subjected to mathematical and statistical analysis. In a manner similar to handwriting analysis, the scientists can map the distinctively personal pen or brush techniques used by different artists — details that are invisible to the unaided eye, but which set apart the work of a master from, say, one of his pupils or a forger. Farid’s team tested the process by analyzing eight drawings by 16th-century artist Pieter Bruegel the Elder, along with five drawings that until recently were attributed to the master. The researchers found that the authentic Bruegel sketches all had a similar digital signature, while the imitations were not only different from the genuine drawings but different from each other as well — suggesting that they were done by multiple imitators. A similar analysis of "Virgin and Child with Saints Jerome and Francis," by 16th-century Italian Renaissance master Pietro Perugino, confirmed art historians’ long-standing suspicions that Perugino himself probably did only part of the painting. Similarly, at the University of Oregon, professor Richard Taylor has used fractual geometry to break down the distinctive drip patterns of Jackson Pollock paintings — and to call into question the authenticity of some newly discovered works.

While digital analysis of artwork has not yet gained wide acceptance in the art world, it could be the wave of the future. But even now, science already plays a major role in differentiating frauds and copies from genuine masterpieces. As McCrone once put it, whether a scientific authenticator judged a painting to be genuine or a fraud, the expert "can be more confident in that belief than anyone who lacks objective scientific evidence."