Materials science

The art of restoration

Article metrics

There are various techniques for the restoration of artwork — how effective and safe these are also varies. ‘Reversible’ gels could, however, provide a less risky way to reverse the ravages of time.

The cleaning of paintings is one of the most controversial activities that can be conducted in a museum. Unlike many artefacts, much of the value of a painting (both aesthetic and monetary) depends on its appearance, which in turn depends on the condition of its surface, especially the first few micrometres. The removal of dirt, yellowed varnishes and later overpaint is a critical step in the restoration of a painting, to return it to as close to its original state and appearance as possible. In Langmuir, Carretti et al.1 report the development of ‘rheoreversible’ gels for the removal of non-original layers. These solvent-containing gels can be converted back to a free-flowing fluid state in situ for easier and more complete removal of the cleaning mixture. This work illustrates both the value of applying science to the preservation of artwork and the necessity of understanding the practical problems involved in determining the safety of any restoration technique.

The problem lies not in removing unwanted material from the surface of a painting, but in doing so without removing, disturbing or otherwise altering the original design layer. This can be extremely difficult, especially if the solubility of the overlying layer is similar to that of the paint binder (medium) — or, even worse, if the layer to be removed is less soluble than the original layer in any reagent. Traditionally, reagents such as solvents, spirits, alkalis, acids and soaps have been used, as well as simple mechanical action (scraping with a scalpel). More recently, scientific approaches have been brought to bear: solvent theory, explaining the action of solvents, has taken much of the guesswork out of preparing solvent mixtures of the desired strength; enzymes can remove very specific types of materials (proteases for proteins, lipases for oils, and so on). Lasers can simply vaporize the layer to be removed if it is otherwise intractable or if the original layer is too sensitive to other reagents; concurrent spectroscopic monitoring of the plasma plume of material removed by the laser indicates when the interface between different layers has been reached2. Still, the cleaning of paintings relies as much on the skill, experience and judgement of the conservator as it does on science, as any technique improperly applied can do damage.

In recent years, a number of gelled reagents have been introduced3. These polymer-based materials make it easier to localize the action of cleaning, to combine the action of otherwise immiscible components, and to avoid the problem of quick evaporation of volatile solvents. Unfortunately, much of the literature on these materials consists of anecdotal case studies that have not been peer-reviewed and which often contain unsubstantiated, even incorrect, statements regarding their safety, specificity, mechanism of action and the roles of the individual components. The safety of gelled reagents has not yet been adequately demonstrated, and even more serious problems than those with volatile solvents have been documented4.

One major problem is the difficulty of completely removing the viscous gelled reagent and any non-volatile materials such as soaps or high-boiling-point liquids contained in the mixture. This is especially true when the material has been forced into cracks or a porous layer during the cleaning process. But Carretti et al.1 have an ingenious approach: their gelled solvent mixture reverts to a free-flowing liquid on addition of a small amount (a few microlitres) of a weak acid solution (0.05 M acetic acid). The resulting ‘degelled’ mixture is more easily and completely removed than are viscous gels, in this case by absorption into a cotton swab. Further clearance of any remaining residue is facilitated by the fact that the gelling agent is now readily soluble.

Working on a small area of a fourteenth-century icon (Fig. 1) from the National Gallery in Siena, Italy, Carretti et al.1 have demonstrated that their rheoreversible gels can remove material from the surface of a painting. The spectrum of X-rays absorbed by the removed material indicated that no mercury was present, which implies that none of the vermilion pigment (mercuric sulphide) of the original paint layer had been removed. However, removal of inorganic pigments is not often the major problem, because most are insoluble in the standard cleaning agents. More serious is the extraction of organic material such as organic dyes and pigments, or the low-molecular-weight compounds in the medium that function as plasticizers in maintaining the flexibility of the paint film.

Figure 1: Brighter future? This fourteenth-century icon, held at the National Gallery of Siena, Italy, has become dulled over time.


Tests on a small area (not shown) by Carretti et al.1 suggest that ‘rheoreversible’ gels could be effective cleaning agents that would not damage the artwork.

Carretti and colleagues' infrared analysis of the removed material showed that the varnish was a natural resin. But infrared analysis does not find the minor amounts (relative to varnish) of paint media that are typically extracted. Infrared spectra of the common resin varnishes are quite variable, and also change with age, making the interpretation of resin spectra difficult to begin with, and the detection of minor components even less likely. Methods such as gas chromatography are much more useful in this context for evaluating the amount of paint media removed (with chromatography it is quite easy to look for compounds, such as fatty acids in oil paint, that are specific to the paint rather than to the resin). Similarly, infrared analysis of the cleaned surface cannot adequately demonstrate that no traces of cleaning mixture remain on the surface. Even scanning electron micrographs of the surface that show a surface similar to an uncoated paint film cannot unequivocally demonstrate that components of the paint film have not leached out.

The development and testing of new cleaning techniques and reagents for the restoration of paintings is a daunting task. Both techniques and reagents must be shown to be safe, causing no significant alteration of, or damage to, the original layers of the artwork. And this must be proved before advocating or promoting their use, which has not always been done by other researchers. Carretti et al.1 have made significant progress in developing their new method, but much still remains to be done in evaluating the results and refining the technique to minimize any undesired effects. The rheoreversible gels developed by this team are a positive step in the development of gelled cleaning mixtures that are safer than those presently in use. The field should look forward to their further contributions on the subject.


  1. 1

    Carretti, E., Dei, L., Macherelli, A. & Weiss, R. G. Langmuir doi:10.1021/la0495175 (2004).

  2. 2

    Scholten, J. H., Teule, J. M., Zafiropulos, V. & Heeren, R. M. A. J. Cultural Heritage 1, 215–220 (2000).

  3. 3

    Wolbers, R. Cleaning Painted Surfaces, Aqueous Methods (Archetype, London, 2000).

  4. 4

    Erhardt, D. & Bischoff, J. Studies Conserv. 39, 3–27 (1994).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Erhardt, D. The art of restoration. Nature 431, 410–411 (2004) doi:10.1038/431410b

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.