When the panel entered the Prado in 2011 the surface was covered by various areas of repainting, one of which, at the lower left side, concealed the figure of the kneeling donor and his patro saint, Agnes, identifiable by the lamb at her feet. On completion of the restoration it was possible to appreciate the artist’s unique and innovative style, in particular with regard to the novel depiction of the pictorial space, which reveals stylistic advances comparable to the work of various miniaturists such as the Boucicaut Master (active 1390-1430).
The identification of the donor as Louis I d’Orléans has allowed for the suggestion that the artist is Colart de Laon (documented 1377-1411), Louis’ painter and valet de chambre. Following the Gospel of Saint Luke (22, 39-46), the painter has depicted the kneeling Christ praying to God the Father before a rock on which rests the chalice, a metaphor of his Passion as the cup from which he must drink. Due to the long, narrow format of the work, the inclusion of the donor and of the saint in the foreground obliged the artist to increase the distance between Christ and the sleeping Apostles and to reduce them in size.
The panel had been the subject of previous interpretative restorations of an inappropriate type and its appearance was thus very different to its original one. The textures and finish that it would originally have possessed were concealed by overpainting and it could not be affirmed that they were in an appropriate condition. The colours had become dulled, losing their subtle nuances and transparency. The present text will offer an account of how the restoration was undertaken, explaining the decisions that were taken as the panel started to reveal the full complexity of its state of preservation and the richness and subtlety of its original qualities.
An initial examination with ultra-violet light (fig. 19) revealed that the surface was covered with a thick layer of varnish and large areas of overpainting that produced a blackened, opaque image. This information was confirmed by technical studies carried out in the Museum’s Technical Documentation Department.33 As noted above, in the technical analysis of the work both the Infra-red reflectography and X-radiography (figs. 20 and 21) confirmed the existence of a large area of overpainting on the left side of the panel and showed the presence of a female saint with a halo, a martyr’s palm and a lamb at her feet, identifiable as Saint Agnes. Also revealed was a donor holding a scroll. The existence of underlying painting was visible to the naked eye in some areas where the paint surface had been damaged, as it was in an area of overpainting that had become thinner, thus allowing the letters on the scroll and a fold of the donor’s clothes to show through.
The thick overpainting covered a diagonal zone that ran from the two mountains framing the figure of Christ to the bottom of the panel but did not affect the mountain with the Apostles or the sky. The overpaint was particularly visible in the spaces between the trees in the centre of the painting, making their branches thick and heavy. The work also had some conservation problems including areas of wear and paint loss, which are referred to in the section on technique in this text.
Having exhaustively analysed the problems presented by the panel and with the help of the technical resources available at the Museum, we embarked on its restoration in order to return the work to its original appearance, eliminating the imbalance resulting from the overpainting between the two areas into which the composition had practically become divided, revealing the hidden figures and recovering the work’s pictorial quality and aesthetic unity.
The fact that the support was well constructed, the outstanding quality of the oak and its appropriate thickness for the dimensions of the painting meant that the structure of the panel had survived well over time and therefore did not require restoration.
The technical studies undertaken provided key information on the underlying paint layers such as the existence of the figures beneath the overpaint. They did not, however, yield any information on the state of preservation of these figures. X-radiography and Infra-red reflectography confirmed the quality of their design while the small number of isolated stratigraphic micro-samples that were taken showed that the painting underneath was quite thick. Nonetheless, three options had to be considered: that the painting underneath was in very poor condition and had therefore been covered over; that during that procedure the surface had been deliberately eroded in order to make the paint subsequently applied on top of it adhere better; and finally, that this was simply a change to the composition or a pentimento on the part of the artist and that the figures were never completed.
In the light of this situation, and once again with the ongoing support of technical analyses, we decided to proceed with the elimination of the overpaint, bearing in mind from the outset of cleaning that the limits of the treatment could only be established through careful visual analysis, i.e. attentive observation of the physical difference between the materials of the lower layer and those of the overpaint.
Restoration began in January 2012 with an initial cleaning of the varnish that covered the entire panel. Fortunately, and due to the fact that it was a natural resin varnish, it could be removed using a light dissolvent. From the outset it was evident that the original paint layer was extremely thin and was so worn that even rubbing it was not a possible option. We then proceeded to remove the overpaint, which consisted of two layers; a lower one with a high proportion of lead white in its composition that made it difficult to dissolve, and an upper one that was relatively recent in date and made up of pigments that were only used from the nineteenth century onwards (fig. 25). It was clear that neither of the two were part of the original paint surface due to the presence of a translucent organic layer (natural pine resin varnish), which separated them from the painting underneath (fig. 26), proving that the painting had been completely finished before the overpaint was applied. Analysis of the painting underneath corroborated this: it consisted of various compact layers of egg tempera made of costly pigments such as lapis lazuli with gold details added on top to the drapery and haloes. Also applied on top after the painting was completely finished was the phrase written on the scroll, as I will discuss later. In addition, the agglutinant of both layers of the overpainting was not egg tempera but oil.
The two layers of overpainting were thus easily distinguishable from the underlying painting but the appropriate method to remove them had to be established. Given that the original paint was extremely fragile it was not advisable to use dissolvents, both due to the lengthy amount of time that they take to act and because they have to be rubbed on. Also inappropriate was the use of a dissolvent in a water-based gel as the fact that the painting was worn in numerous areas had left the priming unprotected, which could thus also dissolve.
The key to appropriate elimination lay in the above-mentioned layer of varnish, which functioned as an isolating barrier between the original and subsequently added paint. This layer of varnish allowed us to opt for the most highly controlled and safe procedure which was that of removing the added paint without moisture and by purely mechanical means. We thus decided to use a scalpel that could be passed over the intermediary layer of varnish, knowing that this varnish would protect the original paint from any possible erosion. The use of this method was also possible due to the small dimensions of the work, which meant that a stereoscopic microscope could be used. The highest magnification (x40) is essential for precisely identifying all materials that are not part of the original work, distinguishable by their texture and the size of the grains of the pigments, which are always larger than the original tempera. Some details, such as the cross borne by the lamb and the letters on the scroll were more difficult to distinguish as they were subsequently added in relief on top of the varnish over the underlying paint layer. It was therefore necessary to continually compare what was seen through the microscope with normal observation in order not to lose spatial references. Cleaning guided by a microscope is only possible with paintings like the present one that have precisely defined and directly applied colours, without gradations or glazes and thus easy to identify.
I then proceeded to work on the area of the sky, which had not been concealed by the overpainting, undoubtedly because it contained a large amount of costly lapis lazuli as well as gold. This intervention was thus limited to the elimination of a dense layer of dirt and of thick oxidised varnishes. I achieved this through the application of a light dissolvent that softened this layer and allowed it to be removed dry, avoiding any wear to the gold. This layer concealed the original technique of the gilding to the stars, which was painted on in imitation of the pastillagetechnique.34 Cleaning returned both the glitter and relief to these areas of gilding.
After the overpaint was removed the figures of the female saint and the donor once again became visible to the naked eye and were found to be perfectly preserved. As well as leading on to the art-historical research on the painting, the study of these figures suggested that what initially seemed to be chemical deterioration or simply dirt in the areas not affected by the overpainting could also have been due to later restoration. When cleaning the sky it was the figure of the God the Father that first revealed this grey-brown covering layer, which was slightly thicker than the one simply consisting of dirt. This provided us with the clue to the fact that all the figures in the painting might have been re-touched. Christ and the Apostles were also curiously painted with a rougher texture and opaque colouring, giving them a different finish to the precise original tempera of the two figures that had had their overpainting removed.
The green of Saint Agnes’s mantle provided a good reference for cleaning Saint John’s and once this had been carried out it was possible to confirm that what had initially seemed to be chemical deterioration (due to the oxidisation of the copper resinate) was another very thick area of repaint (fig. 27 a-d). Staff at the Laboratory confirmed this observation through the study of various stratigraphic samples and detected the presence of two paint layers separated by a layer of varnish: an original one, which was that of the newly revealed figures, covered by varnish, and this new repaint. Once again this intermediary layer allowed us to mechanically remove the repaint with the assistance of the microscope. The appearance of this other area of overpainting indicated that it had been applied by a competent painter who had undoubtedly been requested to update the appearance of the painting to conform to the style of the day. When doing so he carefully reworked all the objects and details as well as the shadows and facial expressions, simply giving them more intense colouring but without aiming to paint his own work on top, a decision that resulted in the better preservation of the original paint (fig. 28). This artist, who undoubtedly appreciated the highly quality of the painting on which he was working, actually applied the principle of modern restoration of respect for the original work. This fact was essential when arguing for the elimination of his intervention, as would be the case with any modern restoration that affects a correct reading of a work of art.
This procedure began with samples taken from each of the figures in order to analyse the state of the underlying paint and the presence of gradations and glazes or of drawing in the folds etc, as it was only with full and complete knowledge that we could embark on the removal of overpaint in a work as unique as the present panel.
A different case was that of the letters on the scroll that reproduce the opening words of Psalm 51: “Miserere mei deus secundum magnam misericordiam tuam”, which are not the original letters as they are painted over the isolating layer of the first layer of aged varnish, as noted earlier. In this case we opted not to remove them as the original letters had barely survived (only the first M, which was greyer in colour, and a few other minimal strokes). It is logical to think that whoever painted them was looking at the original letters and copied them.
By studying the gilding on the panel through an analysis of the X-radiographs, it was noted that three different techniques had been used: one involving gold leaf, to be found on the chalice and in the haloes, and which did not produce radiographic contrast; another, mentioned above, carried out with a brush, which showed up as white contrast in the X-radiographs on the rays of God the Father and in the stars; and another technique, which, although seemingly painted, did not produce radiographic contrast and was only detectable in the decoration of the Apostles’ mantles. Restoration confirmed that the latter passages of gilding, which are much more delicate than those on Saint Agnes’s mantle, are replacements of the originals, of which only a few traces survive. All the gilding was applied with gold and an oily adhesive.35 Similarly not original were the red reinforcements that decorated the Apostles’ haloes. These were removed during cleaning in order to return them to their original, simple design. Also revealing traces of gold leaf were the leaves decorating the donor’s clothing, which are not intended to imitate embroidery but are possibly intended to suggest gilt-metal applications, as indicated by the logical shadow that they project on the cloth of the sleeves.
With regard to the vegetation in the landscape, both the background trees and the middle-ground shrubs had been considerably reinforced by overpainting as they had further deteriorated due to the fact that they were executed on top of a dark priming that had contracted over time. During the cleaning process the stronger and more noticeable non-original leaves were removed, thus returning the appropriate position to each element within the pictorial space.
Having removed all the areas of overpaint, the work was found to be exceptionally well preserved. The figures were almost perfect and only the background areas, particularly the central zone around the shrubs, showed wear. It was evident that the figures only required very specific, minimal retouching and the application of a few subtle glazes in the landscape in order to replace the visual unity that had been lost. Study of the original pictorial treatment in the well preserved parts provided us with the necessary references for treating the worn zones.
Areas of lost pigment were replaced with watercolour with the assistance of a stereoscopic microscope in order to avoid covering even the tiniest part of the original paint and to ensure the greatest possible degree of precision.
When choosing a protective layer we looked to manuscript illumination, which employs matte pigments of extremely intense colour that last well due to the use of a non-oil-based technique. It was clear that a similar appearance was the appropriate one for this Agony in the Garden in order to recover its intense illumination and vivid chromatism. It was also important that the varnish should revive the original vibration of the gold against the blue sky and emphasise the effect of relief created by the gold on the clothes. For all these reasons we selected a natural dammar varnish that was applied in a single, very thin layer.
The restoration of the frame followed the same criteria and methodology as those used for the painting, given that they are one and the same object, both stylistically and in material terms. After the frame was cleaned and later metallic paint removed, a minimal aesthetic readjustment was made by glazing losses with watercolour, albeit without replacing the few small missing fragments in order to respect the object’s material history to the greatest possible degree.
On completion of the restoration process it was clear that the painting had recovered its remarkable original quality, its slightly greenish flesh tones, transparency, and the sense of volume in the clothing, as well as the composition and relationship between the pictorial planes. The mountains now no longer seem separated in terms of distance and the shrubs create a linking element between them. The original concept of the pictorial space between Christ and the Apostles, which was flattened out by the overpainting, has also been re-established. In addition, it is now possible to appreciate the representation of the ground with its stepped terracing so characteristic of the world of manuscript illumination. The present intervention has revived the original lightness of the paintwork, particularly in elements such as the trees, recovering the effect of the wind passing through the branches. Finally, ensuring that the figures are all in a similar state of preservation means that they are once again harmoniously related.