ART / The appliance of science: Dalya Alberge dons her lab coat and explores a technological approach to the dating game

THERE IS an X-ray of someone's rib-cage on the laboratory wall, lit by a huge light-box. In adjacent rooms are scanning machines, microscopes and other bits of medical equipment that might be used for treating the case. But this patient is not human. Take a closer look and you can make out the skeletal details of a badly flaking Old Master. It is in poor condition and awaiting the attention of staff at the Courtauld Institute's laboratory for the technical examination of paintings.

Here art meets science as pictures from both public and private collections arrive for rigorous tests and treatment. Some need to be restored and conserved; others to be analysed and authenticated. Some undergo the full works.

Although, in some academic quarters, it is felt that science is obscuring the old-fashioned idea of simply looking at pictures - and that some restoration programmes are making pictures too perfect - hi-tech equipment does allow conservators to see things beyond the reach of the naked eye. With X-ray vision, so to speak, conservators can spot where an artist changed his mind and reworked a composition; and by magnifying a minuscule speck of pigment 1,000 times and subjecting it to chemical analysis by X-ray, they can sometimes even date a work.

The Courtauld Institute has a long research tradition, but it is seeking to increase the emphasis on science - using it to complement and reinforce historical and stylistic evidence. Three months ago it appointed Aviva Burnstock, who previously worked at the National Gallery's scientific department, as head of its in-house scientific section for conserving easel paintings. She lectures on the three-year postgraduate course and liaises with the outside clients who submit works of art for treatment.

There is no shortage of paintings: she can afford to pick and choose. She tends to focus on museum-quality works - both in the interests of furthering art historical research and because they make interesting case-studies for her students. Last week, she was particularly excited about a 17th-century painting - a sensitively- portrayed peasant scene by Le Nain - sent in by the National Trust from Petworth House in Sussex. Its authenticity is not questioned, the Trust just wants to know more about it.

Burnstock's first brief is to find out who actually painted it. There were three Le Nain brothers - Antoine, Louis and Mathieu - and they all collaborated on the same pictures. The task is to work out which brother painted which bit. The original hangs in the lab, next to its X-ray. Within the criss-cross patterns of the canvas stretcher is an image so clear you can see the canvas weave and cusping between nails. Seeing whether a canvas was prepared in the workshop, as was normal before the 19th century, or commercially, as thereafter, can help with dating.

But the most exciting revelation is of two ghostly figures, a child and a bearded man, who peer out from the X-ray but are not visible in the finished painting - unlike the figure of a second bearded man, who is visible in the painting but not in the X-ray. Burnstock explains that some pigments - chalks, for example - do not show on X-rays. This suggests to her that the peasants were painted by different hands - artists who preferred different palettes.

Next, she must determine whether the indoor scene was originally an outdoor one. She has taken pigment samples from the upper edge of the painting and from one of the figures. Under the microscope, these samples - barely visible to the eye - are magnified 200 times. They reveal colours that are not apparent on the surface: blue and white beneath the architectural details of the composition may indicate that sky was originally there; a speck of green may have come from foliage that was painted over. She needs to do more tests to be sure.

Any study of an Old Master must, she says, take into account that it is likely to have undergone several sessions at a restorer's hands, sometimes within the artist's own lifetime. Previous restoration can hamper analysis. 'You must be sure about what's original and what's not . . .' she says, citing the example of the Victorians, who used to cover up nudes. Sometimes, earlier restoration has actually damaged a work. On an easel in Burnstock's lab is a portrait attributed to Van Liewens, a pupil of Rembrandt, from a large gallery in London. It has a huge crack down the sitter's face caused by a 19th-century restorer who tried to strengthen the thin panel with a criss-cross wooden structure. It proved too rigid and, as the wood couldn't move, it split.

Advances in technology have helped restorers avoid such mistakes. Whereas they used literally to iron crumpled canvases, crushing the impasto, they now straighten them with an air-table that draws air through a perforated table-top. In one room, a 17th-century portrait that came in, she says, as 'a crumpled wreck' is now in tip-top condition and ready to go home.

Although the air-table is one of several pieces of sophisticated equipment in use, the Courtauld hasn't got the money to buy any more machinery. Burnstock says that her years at the National Gallery have spoilt her (the best equipped gallery in the land, she says); now much of her time is spent fund-raising and negotiating loans and donations just to cover running costs (about pounds 5,000 a year).

The Courtauld does not charge museums for restoration work (apart from expenses). But it can take up to three years before pictures are returned - analysed, studied and restored by students under Burnstock's guidance. Private clients, who are charged, make up only a small proportion of the Courtauld's work. They can, however, go to another research department, based at University College, London.

Libby Sheldon is co-director of UCL Painting Analysis, an independent company based at the college. She deals mostly with auction houses and dealers, who, she says, are getting increasingly worried about 'selling something that isn't what they think it is . . . In the past few months, Sotheby's has routinely been getting us to test every single work . . .' Because the company is generally offered things about which a client is doubtful in the first place, as much as 15 per cent of the 130 or so objects they test each year turn out to be fake. Some are 'found out' by pigments tests: in one case, cobalt blue, a 19th-century pigment, was discovered on a '16th-century' picture; in another, a 'Goya' nude was found to have a painting of a 19th-century steam- ship underneath. The company also tests works for the police.

Although some clients might want to use a lab report as a guarantee of authenticity, both the Courtauld and UCL are careful to issue disclaimers. Burnstock says she does not make firm attributions. She might suggest instead that 'the analysis is consistent with the suggestion that . . . I wouldn't attribute something to Rembrandt, for example, but say that the material is consistent with it being a Rembrandt. Science cannot prove that it is not by one of his pupils.'

Another academic research centre that doubles up as a commercial venture is the Oxford University Research Laboratory, best known for carrying out tests on the Turin Shroud. Oxford concentrates on archaeology and art objects rather than paintings - making firm attributions is even harder in this area. Indeed, up to 30 per cent of the 700 objects it tests each year prove to be fake. When the technicians take a sample from something like an Ancient Greek vase, they must be careful that it is not an assemblage of different periods and that they are not sampling just the oldest bit. A technician demonstrated to me how she drills a hole into the base of a vase, using what looks like an ordinary electric drill. The sandy grains produced by the drilling are scanned to date them. Results cannot be exact - the margin of error is plus or minus 20 per cent. The lab says it is easier to say what something is not than what it is. When they tested a bronze figure of a dancing faun, spotted by a Sotheby's expert in a garden sculpture sale, tests suggested a date of between 1450 and 1630; that reinforced Sotheby's dating of around 1610. The piece sold for pounds 6.2m.

The Oxford department, a rabbit-warren of small rooms spread across three Victorian buildings, is just like a chemistry lab - something out of a Sixties sci-fi film. It is full of test-tubes, jars with weird solutions and machinery sprouting thousands of wires and covered in buttons marked with ominous warnings such as 'Danger High Voltage'. There is no sign of a single work of art: for security reasons, once samples are taken, objects are returned to their owners.

As Professor Michael Tite, the lab's head, showed me round technicians were at work in most of the rooms. One was scanning sherds. Another was looking at a computer graph of thermo-luminescence dating, a process that can measure natural light accumulated within a piece over thousands of years. As we entered another room, there was the sound of liquid being stirred. Another chemical solution, I assumed. No, just someone making coffee.

(Photographs omitted)