True colours

Close Examination at the National Gallery looks at 40 problematic works — including outright forgeries, misattributions, pastiches, copies, altered or over-restored paintings, and works whose authenticity has wrongly been doubted. Curators Ashok Roy and Marjorie E. Wieseman have taken on a huge subject — the range of possibilities museum professionals take into consideration when they investigate a picture's status and the variety of technical procedures conservation scientists use to establish authorship and date.

The case histories they discuss have a single common denominator. In whatever direction and to whatever conclusion the combined disciplines of connoisseurship, science and art history may lead, the study of any work of art begins with a question: Is the work by the artist to whom it is attributed?

A good example is an Italian painting on panel that the National Gallery acquired in 1923 as the work of an artist in the circle of the Italian 15th-century painter Melozzo da Forli. Today we find it incredible that anyone was ever fooled by a picture that looks like it was painted by a Surrealist follower of Salvador Dali. But that is to forget how little was known about Melozzo 90 years ago, and how little could be done in the conservation lab to determine the date of pigments or wood panel.

Even so, from the moment the picture was acquired, sceptics called its status into question. Nothing could be proved until 1960 when a costume historian pointed out the many anachronisms in the clothing. When technological advances enabled the gallery to test the pigments they were found to be 19th century.

Scientific evidence can be invaluable but it has to be used with caution and in tandem with historical research. Corot's ravishing plein-air sketch The Roman Campagna, with the Claudian Aqueduct has always been dated to about 1826, soon after the artist's arrival in Rome. But the green pigment called viridian that Corot used in the picture only became available to artists in the 1830s. The landscape wasn't a fake and for stylistic reasons couldn't have been painted later than the mid-1820s. All became clear when art historians discovered that the firm that sold artists' supplies to Corot in Paris started making the newly developed colour available to selected customers in the 1820s, long before it came into widespread use.

The flipside of a fake — but capable of doing equal violence to an artist's reputation — occurs when an authentic work is mistakenly labelled a forgery. Back in 1996, the former director of the Metropolitan Museum of Art, Thomas Hoving, declared that Uccello's lovely little canvas of St George and the Dragon was forged. The gallery therefore X-rayed the picture and tested paint samples before concluding that it was a rare survival of a work by Uccello dating from the early 1470s.

Anyone can label a picture a fake or copy but their opinions are worthless unless they can support them with tangible proof.

Conservation science is the glamorous part of making attributions and the aspect of a museum's work that receives the most publicity. But it will never entirely supplant the human eye — or touch. The National Gallery acquired a small portrait on panel attributed to an unknown artist working in Germany in the mid-15th century. The sitter is Alexander Mornauer, town clerk in the Bavarian city of Landshut. At the time of purchase the painting showed the burgher against a plain, anti-naturalistic blue background, much as Hans Holbein was to do in the next century.

But the moment Ashok Roy ran his hand over the background's smooth surface, alarm bells rang. In the Renaissance, deep blue pigment was made by pulverising hard stone, which made the paint surface gritty to the touch. As Roy suspected, ensuing tests showed that the background had been painted over in the 18th century. Once it had been removed, the original picture was revealed to be typical for a work of about 1464-68 but not a proto-Holbein.