JAIC 1997, Volume 36, Number 1, Article 5 (pp. 59 to 81)
JAIC online
Journal of the American Institute for Conservation
JAIC 1997, Volume 36, Number 1, Article 5 (pp. 59 to 81)




Documentary evidence indicates that Puvis's murals were restored on at least three previous occasions. The first documented treatment occurred in 1930 when they were cleaned and coated by Charles Durham. Commenting on that early restoration, Edward Forbes wrote that “the protective surface that was put on … was a mixture of the white of egg combined with some other ingredients, which makes a sort of varnish and has given the paintings a slight gloss” (Forbes 1940). In 1940, the restorer R. Arcadius Lyon cleaned the murals with a weak solution of castile soap (sodium carbonate and olive oil) and water. At that time he found that the previously applied egg white coating was tough and well adhered and that it did not respond well to his cleaning efforts. Fearing possible damage to the paint layers, he chose not to try stronger cleaning solutions and decided to leave the coating on the surface (Forbes 1940). Finally, Finlayson Brothers restored the murals in 1953, but they do not appear to have left any documentation of the treatment.

As part of a major restoration program for the library, the murals were scheduled for conservation in 1994. They were generally in a very good state of preservation but did require surface cleaning, consolidation of minor flaking, and retouching of minor abrasions. During the early phase of the renovation, however, an unfortunate event necessitated the immediate conservation treatment of The Inspiring Muses. In the early morning hours of February 3, 1992, the painting was badly damaged by water condensing on its surface as a result of an accidental steam valve release. Fortunately, the other eight murals on the staircase were not damaged, as they had been covered with multilayer barriers to protect them from the renovation work. Kate Olivier, who was the first conservator from the Straus Center for Conservation to arrive at the scene, found the area still full of hot, white steam and described it as “like being in a Turkish bath.” The first RH reading, taken that morning shortly after 9:00 a.m. was 87%, but it surely must have been very near 100% earlier that morning. Fans were installed in the staircase windows to extract the damp air. By midafternoon the RH had dropped to 77%, and by the next morning it had dropped to 55%. Later, a protective polyethylene barrier was draped around the scaffolding in front of the mural, and humidifiers were placed in the enclosure to control the humidity level.

The steam caused three specific types of damage. The most serious was extensive lifting and tenting of the paint layers that affected approximately 40% of the painting (fig. 7). Most of the damage occurred on the right half of the mural; fewer and smaller sections were affected on the left half. Why this occurred is not entirely clear, but a likely explanation is that variations in air currents increased the amount of water condensation on the right. Fortunately, the canvas did not shrink very much because of its strong attachment to the wall. The areas of lifting varied in severity over the surface of the painting. The worst damage occurred along the bottom edge, where most of the water collected, and in the blue areas on the right side of the mural. There was moderate flaking in the middle section and minor flaking in the upper section. Generally, the white paint layers exhibited less damage because of their greater thickness and the slow absorption of water by lead white. In spite of their critical condition, the paint layers remained surprisingly intact with very few visible losses. A second serious consequence of the water condensation was overall blanching and the formation of dark vertical stains over much of the paint surface. As the warm water ran down the mural, it dissolved surface material that was re-deposited and dried in the form of disfiguring drip marks (fig. 9). The third problem encountered was the detachment of small areas of canvas from the wall (fig. 11). These areas were located near the bottom of the mural where the water had collected and in six of the cuts made by the mural installation crew in 1895.

Fig. 7. Detail of an area of tented cleavage caused by steam damage in The Inspiring Muses. Courtesy of the Boston Public Library. Before treatment

Fig. 9. Detail of stains and drip marks on The Inspiring Muses. Before treatment

Fig. 11. Detail of detached canvas around an original installation slit in the arm of one of the Muses, before treatment. Also visible is the extensive cleavage in the surrounding paint layers.

Given the urgency of the situation, consolidation tests were begun immediately. Access to the entire surface of the mural was provided by the fixed scaffolding that had been installed for other renovation activities. Of the many adhesives considered, the two adhesives finally selected for testing on the mural were sturgeon glue and BEVA 371. Both were effective in readhering the paint layers. The BEVA 371, however, held the flaking paint more effectively during the removal of the excess adhesive. In addition, it would not be affected by water-based cleaning solutions that would later be required to remove the surface grime and stains; it would not cause further canvas shrinkage during application; and it would be less affected by fluctuations in temperature and relative humidity.

The BEVA 371 adhesive, diluted (1:1) in benzine and heated to increase its flow and penetrating properties, was applied to the affected areas with sable brushes. Because of the vertical surface, the adhesive did not always flow easily under all of the flaking paint. To assist penetration into difficult areas, benzine was lightly sprayed onto the paint surface to wet it out before applying the BEVA 371. Two or three applications of the adhesive were required in most areas to ensure adequate consolidation. The BEVA 371 was allowed to dry at least 24 hours prior to paint reattachment using Willard heated spatulas. The severity and extent of the flaking over the mural made this a difficult and lengthy procedure, but in the end the paint layers were successfully readhered (fig. 8). Then a xylenes/water emulsion was applied by brush to remove the excess BEVA 371 adhesive (for the formula, see appendix 3). The emulsion and the gelled and partially dissolved BEVA 371 were then lifted from the paint surface with dry cotton wads. This procedure reduced the amount of solvent penetrating into the paint layers that might otherwise have dissolved the adhesive. The emulsion had the added advantage of removing some of the less tenacious grime layers. To eliminate the possibility of continued solvent action on the paint layers, the emulsion residues were rinsed several times with water-moistened cotton wads. Once dry, the paint surface was rinsed with benzine and cotton wads.

Fig. 8. Detail of The Inspiring Muses. Scale is 1:1 with fig. 7. After treatment

After the consolidation was complete, the fixed staging was replaced in favor of rolling scaffolding, which allowed more flexibility and the ability to view the work as a whole. The steam had completely disrupted the grime layers and surface coating (fig. 9). Even before the steam damage occurred, the surface grime was known to be difficult to remove, t attempts to clean the mural were further complicated by the redeposited drips. Coating samples were analyzed with Fourier transform infrared spectroscopy (FTIR), and distinct amide groups for protein were found (see appendix 2). The source of the amide groups is most likely the egg white coating that was applied in the early restoration but could possibly be a component of animal glue extracted by the steam from the canvas support. To date, no further analysis has been carried out on these samples. Numerous cleaning solutions and solvents routinely used in painting conservation were tested without success. A cleaning solution recently devised by Richard Wolbers was tried and, unlike the others, gave very encouraging results. The effectiveness of this solution was confirmed by an ultraviolet fluorescence staining analysis undertaken by Wolbers (see appendix 3). Further testing indicated that if the solution were used in a controlled manner, the surface grime and blanched remnants of egg white coating could be effectively and safely removed. The solution also caused no noticeable blanching of the paint layers, a finding that had important implications for 74 later decisions concerning the possible application of a protective coating. The solution was a citrate and detergent gel, buffered to a pH of 8.5 (for the formula, see appendix 3). The cleaning gel solution was brushed onto the mural and gently worked over the surface for approximately 10 to 15 seconds (fig. 12). Much of the grime and staining was dislodged during this first application, which was removed from the paint surface with large dry wads of cotton. Several more applications of the gel were required to remove the grime and surface coating remnants (fig. 13). Finally, the mural was rinsed several times with water followed by petroleum benzine (fig. 10).

Fig. 10. Detail of The Inspiring Muses. Scale is 1:1 with fig. 9. After treatment

Fig. 12. Catherine Rogers applying the diammonium citrate gel solution on the surface of The Inspiring Muses

Fig. 13. Detail of The Inspiring Muses during the removal of surface grime and stains

A number of conservators and conservation scientists have expressed concerns about the use of some of Wolbers's cleaning solutions. They question the solubility effects of various components on the paint layers, the difficulties of clearing surface residues, and the long-term effects of residues that might remain on the paint layers. These issues led to repeated, painstaking rinsing of the mural surface. Given its size and the quantities of cleaning solution used, it is still possible that some residues remain. However, no other cleaning solution was found to remove the accumulated grime and staining.

In the white drapery of the Muses the redeposited accretions and streaks were also removed, but it became evident after cleaning that the paint layers had been darkened slightly by the condensing steam. While the nature of this darkening effect was not determined, possible causes include: (1) the conversion of lead carbonate into lead sulfide, which is typically seen in fresco paintings, (2) the much rarer conversion of lead carbonate into lead dioxide (Gettens et al. 1967), and (3) the recrystallization of the lead carbonate, which would give it a darkened appearance. Attempts were made to convert what was thought to be darkened lead sulfide into lead sulfate (white form) using hydrogen peroxide. The very limited success of this procedure suggested that lead sulfide was not present, and this conclusion was later confirmed by analysis.

The detached canvas areas were plasticized by the application of local moisture and pressure. To apply the necessary pressure, a jig made from a veneer press was secured to the scaffolding using C-clamps (fig. 14). Dampened blotters, sections of foam (to allow the blotter to conform to the distortion), and a Masonite board were placed against the delaminated canvas, and gentle pressure was exerted. After some 40 minutes the paint and canvas were sufficiently plasticized, and the dampened blotters and foam were removed. They were replaced with dry blotters, a 1/4 in. thick Ethafoam section, and a Masonite board and left in place overnight to allow the canvas to dry flat. The detached canvas was then readhered to the wall with Jade 403 (polyvinyl acetate emulsion), and the adhesive was allowed to set under the pressure of Mylar, blotters, and Masonite board for 24 hours. Although Jade 403 is basically an irreversible adhesive, it is sufficiently weaker than the lead white paste and was chosen for its strength and flexibility.

Fig. 14. Pressure jig applied to the arm of a Muse to readhere a pocket of detached canvas (see fig. 11 for detail)

In comparison to the scale of the mural, the losses were small and relatively few. Filling of the losses was not necessary since they occurred in the thinnest paint layers and thus would not be noticeable from a normal viewing distance. The losses and irreversible remnants of the drip marks were inpainted with Bocour Magna Colors (fig. 15). These paints were chosen for their reversibility and appropriate degree of opacity to match the original colors. Although the colors have been discontinued, the Straus Center for Conservation still has a full complement. Retouchings that dried too matte were glazed and adjusted with Acryloid B-72 in xylenes.

Fig. 15. Detail of one of the nine Muses after inpainting

Once conservation of The Inspiring Muses was completed, treatment proceeded on the other eight murals in the staircase. Apart from Philosophy, which has had a history of structural problems, these murals were in good condition and required very little consolidation. Surface grime was thick throughout, but especially obscuring in the murals nearest the windows (fig. 16). The consolidation, cleaning, and retouching of these murals was carried out using the same methods and materials described for the treatment of The Inspiring Muses.

Fig. 16. Aeschylus during the removal of surface grime

Although the Boston murals were not originally varnished, consideration was given to the possibility of applying a protective coating. Issues of public accessibility, vandalism (thus far not a problem), and the redeposition of grime were the main concerns. Natural resins and most synthetics were considered but rejected because of their high saturating properties and poor aging characteristics. Acryloid B-72 was chosen for testing because of its stability and lower saturating properties. It was applied by brush in concentrations ranging from 2 to 6% in xylenes. It was clear to conservators and art historians who were consulted that varnishing would impart far too much saturation and gloss to be aesthetically acceptable. Another concern was the solubility effect that aromatic hydrocarbon solvents would have on the BEVA 371 that was used to consolidate the paint layers should it become necessary to remove the varnish in future treatments. Furthermore, Puvis's clear desire for a matte surface dictated that the murals should remain unvarnished. After the grand staircase was reopened to the public, low display cases were placed against the dado below The Inspiring Muses to create a partial barrier between the mural and visitors to the library.

Copyright 1997 American Institute for Conservation of Historic and Artistic Works