Technical Research Bulletin

VOLUME 5

2011

The Noli me Tangere: study and conservation of a Cretan icon Lynne Harrison, Janet Ambers, Rebecca Stacey, Caroline Cartwright and Angeliki Lymberopoulou

Summary A seventeenth-century icon of the Noli me Tangere (1994,0501.3), in the collections of the Department of Prehistory and Europe at the British Museum, has been the subject of detailed technical examination from scientific, conservation and historical perspectives. The aims were to investigate its original materials and manufacture, its treatment history and, if possible, to look for indications of function and use in Orthodox practice, with the ultimate purpose of informing the conservation required to stabilize the icon’s structure and address the damaged condition of the image. Research into the icon’s original context was also undertaken to strengthen its provenance. The wooden panel was identified as cypress and the original palette was found to comprise lead white, gypsum, carbon-based black, red and yellow earths, a basic copper carbonate green, verdigris, indigo, azurite, vermilion, red lead and a red lake. The painting was found to have been originally coated with an oil and resin varnish and two campaigns of restoration were also identified. The results of the study enabled a suitable conservation protocol to be devised and applied. This involved removal of the degraded restoration varnish and the unstable restoration from the early twentieth century to reveal the fine original painted surface. Those areas of restoration thought to date from the painting’s early history, including repairs to the edges and the complete regilding of the background, were left in place. The results of the study also support a Cretan origin for the icon.

INTRODUCTION A seventeenth-century icon of the Noli me Tangere (1994,0501.3), currently held by the Department of Prehistory and Europe at the British Museum, was purchased in Chania, Crete (probably in 1894) and donated to the National Gallery, London in 1924, before being transferred to the British Museum in 1994 [1; Note 3]. Painted on a wooden panel measuring 606 × 472 mm and coated with a glossy and darkened varnish (Figure 1), the image shows the announcement of Christ’s resurrection organized in six successive scenes accompanied by Greek inscriptions. A detailed technical examination of the icon was carried out, the aims of which were to investigate the original materials and manufacture, the treatment history and, if possible, to look for indications of function and use in Orthodox practice, while informing the conservation needed to stabilize the icon’s structure and address the damaged condition of the image. Research was also undertaken into the icon’s original context in an attempt to strengthen its provenance and to investigate a possible link to the post-Byzantine

Cretan artist Michael Damaskinos (1530/35–1592/93) [2; p. 458 No. 100].

TECHNICAL EXAMINATION Technical examination was carried out using X-radiography, emission radiography, ultraviolet (UV) and infrared (IR) examination and photography, optical microscopy, Raman spectroscopy, scanning electron microscopy with energy dispersive X-ray analysis (SEM-EDX) and gas chromatography-mass spectrometry (GC-MS), see the experimental appendix for details. Layer structures were investigated and the materials used in the original production and later interventions were identified. Unfortunately it was not possible to sample all areas of the painting so the results given below, while as comprehensive as possible, cannot be viewed as exhaustive. Large quantities of data were generated in the process of this study and it is not possible to include them all here. 25

LYNNE HARRISON, JANET AMBERS, REBECCA STACEY ET AL.

figure 1. Icon of the Noli me Tangere (1994,0501.3) before conservation. Note the highly glossy varnish seen clearly at the bottom edge

Instead, the most significant results have been selected for inclusion in Tables 1 and 2, and for discussion and interpretation below. Full details from the study are contained in a British Museum internal report that can be accessed through the online catalogue of British Museum icons [3].

Original materials A single wooden panel of Cupressus sempervirens L., cypress (sometimes called Mediterranean or Italian cypress) was cut in the radial longitudinal axis and prepared so that it was smooth on the front surface, with a rougher finish (with traces of adze work visible) on the reverse [4; p. 33]. Two well-fitting wooden battens, also of cypress, were then attached horizontally to the reverse 26

of the panel, at equal distances from the top and bottom edges. No nails are visible on the reverse, but X-radiography revealed that short, flat-headed nails had been inserted from the front of the panel, equally spaced along the length of the battens. A layer of plain (tabby) weave cloth was glued onto the smoothed front surface of the panel. While it was not possible to sample or identify either the cloth or glue used, linen soaked in animal glue was usually employed for this purpose [5; p. 28]. A white ground of gypsum bound in a proteinaceous medium was then applied to the whole front surface of the panel and burnished smooth. No ground was applied on the reverse of the icon.1 Evidence from the IR reflectograms, which reveal elements of the underdrawing, suggests that the layout of the image, including the positions of the figures, drapery

Raman + optical microscopy GC-MS

Physical tests

Paint (pigments) Paint (binder)

Ground (binder)

Raman + optical microscopy GC-MS GC-MS

GC-MS

Pigments Paint (binder) Indigo overpaint (binder)

Ground (binder)

SEM Raman

Optical microscopy

Raman + optical microscopy

GC-MS

Raman + optical microscopy Staining tests

Visual assessment

Optical microscopy

Gold Bole

Mordant

Paint (pigments)

Paint (binder)

Ground Ground (binder)

Cloth

Wood

Cupressus sempervirens L., cypress

Plain/tabby weave

Gypsum Protein

Vermilion; lead white; carbon-based black; azurite; indigo; copper carbonate green (verditer?);c red lead; goethite, hematite; and gypsum Verdigris; earth pigments; and red laked Conifer resin; drying oil

Thick, pigmented and organic-rich (pigments and medium not identified)

Gold alloy (c.93% Au, 4% Ag, 3% Cu; c.22.4 carat equivalent) Goethite- and hematite-containing red earth

Conifer resin and drying oil

Animal glue (gelatine)

Red lead; hematite; goethite; carbon-based black; vermilion; indigo; gypsum; and red lake Fat (egg?)/oil; conifer resin; and sugars (gum?)b Conifer resin and oil

Conifer resin; drying oil; and Pistacia resin (tr)

Water soluble

Cadmium yellow; vermilion;a Prussian blue; and carbon-based black Conifer resin; drying oil; and Pistacia resin

Pistacia resin; drying oil; and conifer resin (tr)

Results

‘tr’ indicates a trace amount was found. a The small particle sizes suggested that the vermilion was produced using the wet process. b The presence of sugars is not understood at this time. c This may be natural malachite, but the regular, small and spherical particles make the use of artificial green verditer more likely. d Given the techniques available it was not possible to determine the organic colorant in the red lake.

Notes

GC-MS

Varnish

Original layers

GC-MS

Varnish

Early restoration materials

GC-MS

Analytical method

Varnish

Later restoration materials

Material

table 1. Summary of the main analytical results

THE NOLI ME TANGERE: STUDY AND CONSERVATION OF A CRETAN ICON

27

28

CS10 Angel’s grey robe, with old varnish residues. Ground layer missing

CS9 Green of angel’s wing with red underpaint. Upper varnishes removed

CS7 Overgilding and overpaint over original paint from edge of cliff to left

Restoration 10. Varnish, which shows white luminescence under UV illumination; to the left it is over layer 8 and penetrates through a crack to the level of the original paint layers; 9. Paint layer comprising finely ground pigments (over original layers at right) – earths, red lake, carbon-based black (Raman); 8. Paint layer between two thin UV-luminescent coatings (at the left of the sample); and 7. Thick varnish showing cream luminescence under UV illumination

CS2 Christ’s left ankle, area of gilded decoration. Sample includes all layers of varnish and possibly recent overpaint

Original 2. Dark grey thin paint layer – carbon-based black, lead white (both by Raman); and 1. Light grey thicker paint layer – carbon-based black, lead white and azurite (all by Raman)

Restoration 5. Some fragmentary particles on surface, possibly overpaint; 4. Non-luminescent surface layer – possibly calcium oxalate (visual examination, unable to confirm by Raman); 3. Remains of varnish layer above and below the oxalate layer showing cream luminescence under UV illumination;

Original 4. Varnish, possibly original, showing faint white luminescence under UV illumination; 3. White highlight paint – lead white (Raman); 2. Green paint – manufactured green copper carbonate and azurite (both by Raman); and 1. Red paint – vermilion, lead white (both by Raman) and red lake

Restoration 6. Remains of two surface coatings to the right. Some particles show luminescence under UV illumination; and 5. Dark layer penetrating into crack

Original 5. Dark layer – dirt(?); 4. Thin varnish layer showing luminescence under UV illumination; 3. Original dark green paint – azurite, lead white and carbon-based black (all by Raman) and yellow (unidentified); 2. Original light yellow-green paint (unidentified); and 1. White paint – lead white (Raman) plus fragment of gold leaf

Restoration 10. Two upper varnish layers showing white luminescence under UV illumination; 9. Gold leaf; 8. Red ‘bole’ type layer; 7. Dark overpaint, medium showing luminescence under UV illumination; and 6. Lower thick varnish that shows cream luminescence under UV illumination and penetrates into crack in original paint

Original 6. Metal leaf – gold (SEM-EDX); 5. Thick, pigmented organic-rich mordant; 4. Greenish-yellow layer of flesh paint – earth pigments and verdigris (optical examination only) topped with thin pinkish-red layer of flesh paint – vermilion and lead white (Raman); 3. Warm white paint – lead white, carbon black and yellow ochre (Raman); 2. Dark layer; and 1. Gesso – gypsum (Raman) [protein]

Results (layer structure from surface down)

Sample

table 2. Selected paint cross-sections showing stratigraphy

LYNNE HARRISON, JANET AMBERS, REBECCA STACEY ET AL.

Pigment identifications given are based on the method given in brackets and nature of the organic medium / varnish is based on microchemical staining tests.

Note

Restoration 6. Upper varnish showing white luminescence under UV illumination; 5. Pigment and gold overpaint – black and red particles (unidentified); and 4. Thick brittle varnish showing cream luminescence under UV illumination

CS13 Gold overpaint layers from area of damage, top centre of gilded background and including original layers beneath

Original 3. Gold leaf; 2. Red ‘bole’ layer – hematite (Raman); and 1. Gesso

Original 4. Crust – degradation visible on surface; probably a calcium oxalate crust (unable to confirm by Raman); 3. Red glaze – red lake; 2. Red paint – vermilion (Raman) and red lake; and 1. Brownish paint – earth pigments (goethite and hematite) and carbon-based black (all by Raman)

CS12 Damaged red paint from kneeling figure to left. Restoration varnish removed

THE NOLI ME TANGERE: STUDY AND CONSERVATION OF A CRETAN ICON

29

LYNNE HARRISON, JANET AMBERS, REBECCA STACEY ET AL.

figure 2. Details from the IR reflectogram in the area of Christ’s drapery showing carbon-based spots from pouncing of an original anthibolon. The image to the right is an enlargement of the region bounded by the rectangle in the image to the left

and architecture, was transferred to the ground layer from an anthibolon (a cartoon or pattern used for tracing the basic designs of works) by pricking and pouncing through the pinholes with a carbon-containing material. The resulting dots were then joined by drawn lines, Figure 2 [6; pp. 56–60 and 76–79, 7; pp. 169–170, 8]. From the X-radiographs it is clear that some of the lines were then further incised into the ground with a sharp implement so that they could be seen throughout the painting process, Figure 3. Interestingly, the trees and rocks of the background, the angels’ wings and the faces of the figures were not incised, Figure 4. Further evidence of underdrawing, without accompanying incised lines, can be seen in the IR reflectogram, for example in the branches of the trees and in the complete drapery of the left figure and folds of the sleeve of the middle figure in the scene to the upper right showing the Myrrophoroi (unguent bearers), Figure 5 [1; p. 191]. Only slight changes in design (pentimenti) between the underdrawing and the finished image are evident, as in the hem of the drapery of the middle figure in the Myrrophoroi scene, Figure 5. A dark, patchy wash of colour on the faces and hands is also apparent in the IR reflectograms and is interpreted as an underpainting or proplasmos [9; p. 234]. Following the production of the underdrawing, the areas to be gilded (the background and halos) were first coated with a bole based on red earth, burnished and then covered

figure 3. Details from the X-radiograph in the area of Christ’s drapery showing incised lines. The image to the right is an enlargement of the region bounded by the rectangle in the image to the left

30

figure 4. Diagram indicating the positions of incised lines as deduced from the X-radiographs

with thin metal leaf of soft, high-purity gold (see Table 1), which was further burnished. Examination of the layer structure revealed details of the composition of the original paint layer and two additional restoration layers; in some places the ground preparations were also present, see below. The original painting technique was distinct from that used in the restoration layers in pigment choice, preparation and colour mix, making it easy to distinguish original workmanship from later additions. As is common in icons, the use of restoration to maintain a complete and functional image for worship has produced a complex and confused layer structure with penetration of solvents and media from layer to layer making it impossible to identify the original paint medium with certainty [3; Section 4.1]. Conifer resin and drying oil were found in the original paint layers but no proteins or fats were identified, although their presence, perhaps only in small quantities, may have been masked by other materials. The original palette consisted of lead white, gypsum, carbon-based black, red and yellow earths (coloured by goethite and hematite), a basic copper carbonate green, verdigris, indigo, azurite, vermilion, red lead and a red lake, see Table 1. The image was built up in layers, with simple paint mixtures of two or three pigments. Selected passages of the painting have been studied in greater detail; see Table 2 for details of the cross-sections taken from these areas. The flesh tone of the large figure of Christ was painted with a yellow-green paint containing a mixture of yellow earth, verdigris and an unidentified white pigment, CS2: Table 2. Highlights of lead white were applied over this layer with the addition of small amounts of vermilion for areas of warm flesh tones. A similar yellowishgreen paint mixture was used for some of the background

THE NOLI ME TANGERE: STUDY AND CONSERVATION OF A CRETAN ICON

cliffs (CS7: Table 2), where a lighter layer was laid down first and then darker passages applied on top, with the addition of azurite and carbon-based black to the paint mixture. A fragment of gold leaf was found within the lowest paint layers. The grey robe of the angel to the left of the empty tomb was painted using a mixture of a carbon-based black, lead white and azurite with the darker folds added on top using a mixture comprising only lead white and carbon-based black, CS10: Table 2. The angel’s wings were first painted with red (a mixture of vermilion, lead white and a red lake) and then edged with greenish wing tips of a mixture of malachite (or perhaps its artificial analogue green verditer) and azurite, with white highlights on top, CS9: Table 2. To the lower left, in the scene of the Chairete (‘All Hail’), the red drapery of the kneeling figure of the Virgin Mary is quite distinct from the other reds used in the image. This effect was produced using three paint layers, CS12: Table 2. The first brownish-red layer was coloured with earths and carbon-based black and this was coated with a second thin layer of semi-opaque red (a mixture of vermilion and red lake) to which a comparatively thick third layer of a translucent red lake was finally applied. After painting, mordant gilding was used to decorate Christ’s cloak and sandals, and the angels’ wings. The mordant from an area of Christ’s sandal, visually identified as an organic binder bulked out with inorganic pigment, was applied to the painted surface and then coated with gold leaf that was left unburnished, CS2: Table 2.2 The surface of the painting was then coated with a varnish containing a mixture of drying oil and conifer resin, the remains of which were located in the paint cross-sections during analysis and confirmed during conservation.

Later changes (damage and restoration) At some point, possibly relatively soon after completion, the icon suffered extensive damage to the wood panel and painted surface by wood-boring insects, probably attracted to the glue-impregnated cloth layer as a potential food source. In general the damage was restricted to the uppermost surface of the panel, with only few areas of activity visible on the reverse, which are discussed below. As a result, original paint and ground were damaged or lost and restoration was clearly undertaken. The painted surface was cleaned to remove the original varnish, resulting in some damage to the paint (particularly the Virgin’s red robe in the Chairete scene) and etching of the surface of the gilded halos and mordant gilding, Figure 6. The lost areas – particularly at the edges – were replaced with a white ground layer bound in animal glue and an attempt was made to complete the detail of the lost image using pigments that were probably bound in egg or an egg/oil mixture, Table 1. An exception to this was the indigo paint used to cover areas of loss in the trees, which was oil-based. Curiously, the restoration layer was not built up to the same level as the original surface, but sat just below, forming a step

figure 5. Detail from the IR reflectogram showing underdrawing in: (a) the tree branches; and (b) the scene of the Myrrophoroi

around the edges of the damages. Some of the insect flight holes were also plugged with white fill and overpainted. The early restoration palette included red lead, vermilion, indigo, ochres, a red lake and gypsum, and differs from the original palette both in pigment mixtures and the size of the pigment particles. For example, restoration of the red coffin in the Myrrophoroi scene, originally painted in vermilion, was carried out in a mixture of red lead and a red lake. Larger areas of loss of original ground in the gold background were also replaced and the whole of the background was regilded. Of the original gilding, only the halos and traces in the background remain (see below). The profiles of the mountains and architecture were repainted and a black inscription added on top of the gold background between the mountains to the right. The original crosses were completely overpainted. A comparatively thick coating of an oil and conifer resin varnish was then applied across the whole surface including the restorations and the original paint.

figure 6. Photomicrograph of the mordant gilding on Christ’s sandal strap showing loss of original mordant gilding beneath the cracked and brittle restoration varnish. Image size 12 × 9 mm

31

LYNNE HARRISON, JANET AMBERS, REBECCA STACEY ET AL.

figure 7. Icon during conservation showing those areas (shaded in pink) that comprise earlier restorations or regilding, plus the black inscription; all of these were left in place

A later, probably early twentieth century, restoration campaign has also been identified, which concentrated on the lower left edge and left side of the icon, but included other scattered areas of retouching across most of the surface. The losses were replaced with white, water-soluble filler and inpainted using pigments that were probably bound in oil. The palette included traditional pigments such as vermilion for the overpainted red border and some post-eighteenth-century materials such as cadmium yellow and Prussian blue; the latter was used to retouch the Magdalene’s blue robe that had originally been painted with a mixture containing indigo and gypsum, Table 1. This paint was applied directly on top of the older degraded oil and conifer resin varnish from the previous treatment and concealed areas of original paint and earlier restoration. Unlike the earlier repairs, no attempt was made to recreate 32

the lost parts, using instead simple blocks of colour to fill the losses. A further, thinner, layer of oil and mastic resin varnish was then brushed over the whole surface.

CONSERVATION TREATMENT

Condition before conservation As a result of its history of construction and change, the painting was in urgent need of conservation, both to stabilize it and to clarify the heavily restored and damaged image. The icon was structurally unstable and visually compromised by the previous restoration treatments. The

THE NOLI ME TANGERE: STUDY AND CONSERVATION OF A CRETAN ICON

two oil and resin varnish layers from previous restorations were brittle and flaking and original paint was often attached to these flakes. There were localized areas of recent paint loss, some showing cleavage between the paint layers while others included the ground and exposed the wood support. Two rigid paint blisters in the centre were raised out of plane and hollow to the touch. The later restoration was particularly unstable along the left edge and was separating from the wood. Insect damage had caused areas of the original surface to become soft and hollow. These had collapsed in places, causing surface undulations and loss. The painted layers were riddled with insect flight holes, many of which had been plugged with restoration material. X-radiography revealed the extent, depth and severity of a crack running vertically through the centre of the panel; this was particularly evident in the electron emission radiograph, which also provided a very clear view of other panel damage and the restoration of the painted surface. The panel had remained relatively flat with only a minor twist from top to bottom, although shrinkage and expansion could readily be observed in the behaviour of one of the paint blisters in the upper centre right, which altered in height in response to changes in relative humidity (RH).

Treatment Conservation treatment was undertaken to render the icon structurally stable and improve the clarity of the image. The outermost oil and mastic resin varnish was removed with a solvent mixture comprising propan-2-ol and white spirits in a ratio of 2:3 (v/v), applied on small cotton wool swabs rolled across the surface; this revealed the lower, older varnish and the most recent restoration. Where these restorations extended over the original surface they were removed mechanically with a scalpel. The lower varnish was then removed with propanone (acetone) applied in the manner described above, uncovering the remains of a relatively insoluble coating that had been applied directly onto the painted surface and which was, therefore, interpreted as an original material. The remainder of the most recent restoration and associated fills, along the left edge and where present on the other edges, were removed mechanically using a scalpel, revealing the wood beneath and exposing the insect damage. The remains of a cloth layer between the wood and ground were discovered at this point; its presence was not previously known, as it had not been revealed by any of the imaging methods applied to the painting. Traces of earlier restoration were also uncovered along the left edge, matching those present on the right, including the step in level. It appears that prior to the later restoration the panel had lost most of its previously restored left edge, together with more of the original paint at the bottom and top left corners, and that this may be the reason for the later treatment to repair the left side.

The earlier restorations were left in place except where they covered original material and obscured the original surface, where they were removed mechanically. The regilding and the added black inscription between the mountains in the background were left, as neither the extent of any surviving original gilding nor the presence of an earlier inscription could be determined, Figure 7. Localized areas of flaking original paint were consolidated with an acrylic dispersion (Lascaux® 4176) applied beneath the lifting flake with a small brush. The treated area was then warmed with a heated spatula to approximately 40°C through release layers of lens tissue (closest to the paint layer) and lightweight Melinex® polyester film to relax and reattach the paint. The area was weighted with sandbags until the adhesive dried. Localized areas of cleavage between the ground and the cloth or wood were reattached by injecting a warm solution of gelatine (approximately 10% w/v in de-ionized water) through losses in the original surface with a small syringe and then weighting with sandbags placed over release layers as described above. The large rigid blister discussed earlier was not treated, as the long-term stability of this reattachment could not be guaranteed unless the icon is henceforth stored permanently at a raised RH to prevent any panel shrinkage. Were the panel to be subjected to a period of low or fluctuating RH there would be increased risk of the blister lifting again or, more worryingly, of the paint layers becoming compressed, resulting in active flaking and loss. For this reason the blister, which was considered stiff enough to support itself, was not further treated. The reverse of the panel was cleaned with ‘smoke sponge’, a vulcanized rubber molecular trap that contains a minuscule percentage of a mild soap (