IN the back rooms of the Louvre in Paris, an argument has been raging over what to do with the Mona Lisa. Like most Old Masters, Leonardo鈥檚 painting is protected by varnish that has yellowed and darkened with time. Restorers say that if they are allowed to remove and replace the discoloured layer, the colours Leonardo painted nearly five hundred years ago will once again shine through. But some art historians won鈥檛 hear of it. They say the varnish is partly responsible for the softening 鈥渟fumato鈥� effect that gives the subject of the Mona Lisa her beguiling smile. If restorers are allowed to wipe off the varnish, they might easily destroy the mystique of the painting too.
Restorers regularly face awkward choices like this. Many of the paintings they look after have been restored several times before, sometimes with disastrous results. Take, for example, Tommaso Masaccio鈥檚 The Virgin and Child (right) in the National Gallery in London. Mary has black blotches on her blue robe, and Jesus has a nasty red rash on his foot. These blemishes are the result of attempts over the centuries to make good damage done to the 15th-century original. But ugly as these marks may look, not all experts agree that they should be removed. 鈥淪ome people would argue the retouchings are important as a part of the painting鈥檚 history, and should be kept,鈥� says Martin Wyld, chief restorer at the National Gallery.
More questions will surface if the gallery does decide to remove the blotches. Should restorers repaint the missing parts in the original 鈥� a process called 鈥渋nvisible infilling鈥� 鈥� to make the painting appear complete? Or should they do what Italian restorers do, and simply show with hatched areas the parts where damage has been removed? If the decision is to infill, the restorers will have to use their historical and technical knowledge as well as their aesthetic judgment, to guess what the artist painted. They must then assess whether their restoration techniques are good enough to achieve the result they want.
Advertisement
Solvent abuse
One of the paintings that the National Gallery has already restored is Veronese鈥檚 The family of Darius at the feet of Alexander. The gallery used solvents to clean away discoloured layers, then tried to restore it to look the way the artist intended it. This approach has drawn fierce criticism from Michael Daley, the director of Art Watch International, an organisation that monitors restorations. 鈥淭hey take off all the varnish until they鈥檙e down to the paint, so they produce a whistle-clean painting鈥ut wreck it,鈥� he says. 鈥淒aley says that the gallery has removed or changed some of the elements that could be seen in the painting before it was restored. He points out that one figure鈥檚 robe has changed from brown to green, and the blueness of the sky has faded. Daley attributes the fading of the sky to leaching of paint by solvents during cleaning.
The gallery defends the changes it has made to the look of the painting. It used X-ray analysis to reveal drawings by Veronese underlying the painting, and says that restoration has simply recaptured the artist鈥檚 original intention. Neither does the gallery accept that solvents damage the original paint. 鈥淭he arguments are based on aesthetics and emotion,鈥� says Mark Richmond, the former head of the defunct Science and Engineering Research Council who now sits on the gallery鈥檚 board of trustees. He admits that it is difficult to prove that solvents are not damaging. 鈥淎ll you can say is that to the best of our ability, we don鈥檛 think there is any damage.鈥�
Elsewhere in Europe, however, this uncertainty has made restorers much more cautious in their use of solvent. The upper layers of old paintings are composed of glazes, and these may be attacked by solvent without producing an immediate effect, says Jacques Franck, the art historian who last year persuaded the Louvre not to restore Leonardo鈥檚 Sainte Anne. The danger comes when deeper layers of varnish are removed 鈥淭here鈥檚 some pigment migration between the upper part of the paint layer and the lower part of the varnish, and this intermediate zone is very fragile,鈥� says Franck. 鈥淪o we just reduce the varnish layer rather than remove it. This is the distinction between the Franco-Italian and Anglo-Saxon methods.鈥�
Dusky legend
Restoration of Sainte Anne was to have been a prelude to restoring the Mona Lisa itself. But Franck says that the darkening of the Mona Lisa has become 鈥渁n essential element in the perception of an image whose dusky quality has become legendary鈥�. His biggest worry, and one the Louvre accepted when it decided not to go ahead, was that solvents used to clean the varnish might penetrate and damage the fine layers of glaze that give the flesh its mystical colour.
Much of Daley鈥檚 evidence for solvent damage comes from a 1990 conference organised in Brussels by the International Institute for Conservation. One paper from the conference, by Stefan Michalski of the Canadian Conservation Institute in Ottawa, warns that 鈥渃leaning science barely exists鈥�. It catalogues experiments with organic solvents which show that they can leach binder from paint, or make the paint medium swell upwards, sometimes leaving a void underneath. 鈥淲ater and alcohols attack inorganic pigments especially, while aromatics and ketones are more likely to attack organic pigments,鈥� Michalski says.
Another of the papers cited by Daley as evidence of the damage solvents can cause was cowritten by Aviva Burnstock, an art historian at the Courtauld Institute of Art in London. But Burnstock says that many critics of restoration have read too much into the results. 鈥淣one of the research shows damage to Old Masters,鈥� she says. 鈥淎ll the tests are on pieces of relatively unaged material that are dissimilar to real paintings in many respects.鈥� She says that the studies Daley quotes are on paint just 12 years old. 鈥淚t doesn鈥檛 compare with Old Master paintings.鈥� To extrapolate these findings to Old Masters, 鈥渨ould be making inferences that might be unreasonable鈥�.
Conservators at the National Gallery agree. Research published so far has either been on new samples of paint, or on paints that have been artificially aged by exposure to strong light and high temperatures. They add that some of the colours subjected to these tests were never used in Old Masters. 鈥淲e know that young paint films behave in the way observed, even if they鈥檝e been artificially aged,鈥� says Ashok Roy, the gallery鈥檚 chief scientific adviser. 鈥淏ut it鈥檚 not the way paints in Old Masters behave鈥e are talking about real paintings, and what鈥檚 proposed is not a model for what happens in real life.鈥�
The first studies that demonstrate damage to paintings can be found in the book, On picture varnishes and their solvents, published in 1959 by Robert Feller, Nathan Stolow and Elizabeth Jones. The authors report that samples of paint immersed in solvents for up to 72 hours could lose as much as 70 per cent of the oil medium. But Raymond White, head of organic analysis at the National Gallery, warns these results 鈥渞eport the worst possible case鈥�. White says that this figure should be seen as 鈥渢he theoretical maximum, or infinity point, of what damage could be done in terms of leaching of material鈥�. Burnstock is equally wary of these results. The samples of paint tested are just two or three years old, she says. 鈥淥ld Masters are at least 100 years old, so you may be looking at a completely different set of problems.鈥�
Roy and White are in the process of collecting and analysing data that could move the argument about solvent damage onto new ground. For the first time, they have analysed the effects of solvent on samples from genuine Old Masters, comparing the damage caused by solvent to that from abrasion. 鈥淲e found that solvents are not leaching materials from films on real paintings,鈥� says White. So far they have experimented with samples from several paintings.
Test samples
鈥淭he oldest, dating from the 1400s, is by the Florentine painter Lorenzo Monaco, while the youngest is a Degas,鈥� he says. Others included a Rubens and samples from Holbein鈥檚 The Ambassadors, which is being restored at the gallery. 鈥淲e tested whether there was any difference if varnish was removed mechanically or with solvents.鈥�
Using gas chromatography and mass spectrometry to analyse the chemical makeup of the paint that was left behind, says White, they found 鈥渘egligible differences鈥� between the results of the two techniques for removing varnish. Roy meanwhile, used a scanning electron microscope to inspect the surfaces of the samples. 鈥淎gain, it showed no discernible change between the area changed by solvent or mechanical means,鈥� he says. 鈥淚t鈥檚 very encouraging, because you can鈥檛 see any material being used up, or cracks forming.鈥� Roy and White plan to publish their results.
Daley remains to be convinced, but is pleased that new evidence will be made available. 鈥淚t鈥檚 certainly the first bit of research they鈥檝e agreed to publish on this,鈥� he says. 鈥溞影稍磗 elsewhere will be able to respond to it, and attack it or be persuaded by it,鈥� he says.
But Roy does not see an end to the dispute. 鈥淢y feeling is that publication won鈥檛 lay this ghost to rest,鈥� he says. The passions that restoration work arouses and the jumble of technical and aesthetic judgments that are called into play provide fertile ground for argument. 鈥淚 don鈥檛 think anything will dispel the doubts,鈥� Burnstock says.