The Conservation of Heat Damaged Micro-Cracked Stained Glass: A York Minster Case Study 

 

Susanna Wyse Jackson MA, The York Glaziers Trust

 

This article outlines an emerging approach to the conservation of heat damaged, micro-cracked stained glass. This has developed out of the conservation of York Minster’s window S5 (Fig. 1). The York Glaziers Trust (YGT) began work on this window in November 2023, under the direction of Principal Conservator Nancy Georgi MA ACR. The project is part of the Minster’s twenty-year strategic plan for the conservation and protection of unprotected medieval windows, launched in 2017.

Fig. 1: Window S5, eastern choir clerestory, York Minster (c.1390). Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 1: Window S5, eastern choir clerestory, York Minster (c.1390). Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Past Restorations and Current Condition

Dating to c.1390, window S5 is located in the eastern choir clerestory of the Minster (Fig. 2). It is one of three surviving windows depicting the Prophets’ and Apostles’ Creed. The panels of window S5 bear evidence of multiple phases of intervention and replacement during the eighteenth and nineteenth centuries. One of these phases, evidenced by graffiti found throughout the choir windows (Fig. 3), can probably be attributed to the restoration of extensive damage sustained during the near-catastrophic 1829 fire in the choir of the Minster (Fig. 4).

Fig. 2: Ground plan of York Minster, with window S5 highlighted in red. © The York Glaziers Trust

Fig. 2: Ground plan of York Minster, with window S5 highlighted in red. © The York Glaziers Trust

Fig. 3: John Browne, A Representation of the Choir of York Minster, as it appeared from the Organ Screen, on the 3rd of February 1829. The day after the Fire. Lithograph on paper (1829). © York Museums Trust, object number YORAG:2010.392.2. Image in public domain.

Fig. 3: John Browne, A Representation of the Choir of York Minster, as it appeared from the Organ Screen, on the 3rd of February 1829. The day after the Fire. Lithograph on paper (1829). © York Museums Trust, object number YORAG:2010.392.2. Image in public domain.

During past restoration campaigns at York Minster, corroded and fragmented glass was often replaced with new glass, resulting in the loss of valuable historic material. By contrast, today’s ethical codes of practice, encapsulated in the Corpus Vitrearum’s Guidelines for the Conservation and Restoration of Stained Glass advocate for the retention of original material wherever possible.((International Corpus Vitrearum, Guidelines for the Conservation and Restoration of Stained Glass, Nuremberg, 2004 (2nd edn). Online: https://www.cvma.ac.uk/CVConservationGuidelines2004.pdf)) This includes material introduced during previous campaigns, provided its retention does not pose additional risk to the original glass and lead. Window S5 has been subject to heavy restoration which involved, among other interventions, the replacement of a large proportion of the medieval glass (Fig. 5).

Fig. 4: Inscription on the exterior of panel 2c, reading ‘Thomas Dalton 1829’. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 4: Inscription on the exterior of panel 2c, reading ‘Thomas Dalton 1829’. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 5: Detail of panel 3b, with corroded blue glass replaced by newer blue glass during a previous restoration campaign. Photo taken following cleaning. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 5: Detail of panel 3b, with corroded blue glass replaced by newer blue glass during a previous restoration campaign. Photo taken following cleaning. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

On close examination of the window in the studio, it was discovered that numerous panels bore evidence of micro-cracking. The combination of intense pitting and heat damage – the latter likely sustained during the 1829 fire – has resulted in the development of significant numbers of irregular fractures in some of the medieval glass (Figs. 6 and 7). The micro-cracking is primarily concentrated in the blue glasses, though it has also been found in white glass, and in several instances is accompanied by a loss of material. Despite the damage, the paint layer is stable and robust, with only minor loss where the glass has cracked.

Fig. 6: Detail of the exterior of panel 5a, photographed in reflected light, showing advanced pitting. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 6: Detail of the exterior of panel 5a, photographed in reflected light, showing advanced pitting. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 7: Detail of the interior of panel 5a, photographed in transmitted light, with microcracks highlighted in red to aid visibility. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 7: Detail of the interior of panel 5a, photographed in transmitted light, with microcracks highlighted in red to aid visibility. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

 

In a number of cases the glass has degraded to a thickness of less than 1mm, which has led to the development of holes (Fig. 8). Despite the severe damage sustained by these fragments, retention of the material in situ is always a primary objective. However, without the proactive stabilisation of these cracks, it would be ethically irresponsible to return the panels to the clerestory, even with environmental protective glazing (EPG), due to the risk of further deterioration and consequent loss of material. The selection of an approach that best achieved the stabilisation of these cracks required the examination of a number of considerations, including but not limited to: The long-term stability of the conservation materials introduced; the visual impact of the intervention on the window; and the re-treatability of the window in the future.

Fig. 8: Detail of the interior of panel 3a, photographed in transmitted light, showing a significant loss of thickness and formation of holes. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 8: Detail of the interior of panel 3a, photographed in transmitted light, showing a significant loss of thickness and formation of holes. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Reviewing Conservation Approaches

For the repair of broken stained glass, epoxy resins such as Araldite 20/20 or Hxtal NYL-1 are well-attested to provide excellent adhesion, longevity and optical properties, negating the need for disfiguring mending or strap leads, provided they are used in environmentally controlled settings.((S. Centenaro, G. Franceschin, E. Cattaruzza and A. Traviglia, ‘Consolidation and Coating Treatments for Glass in the Cultural Heritage Field: A Review’, Journal of Cultural Heritage, 64, 2023, 135.)). However, the micro-cracked glass in S5 is far too friable and thin to justify the use of epoxy resins. The resin bond would be much stronger than the glass itself, presenting risks to the surviving glass, were the window to sustain more impact or stress.((E. Jägers, H. Römich and C. Müller-Weinitsche, ‘Konservierungsmaterialien und Methoden’, in A. Wolff (ed.), Restaurierung und Konservierung historischer Glasmalereien, Mainz, 2000, 129–66. English translation available online at: https://www.cvma.ac.uk/conserv/conservation.html))

Another option would be to provide the cracked glass with a physical support structure. However, the number of cracks in the glass (upwards of 5-6 per fragment of glass) precluded the use of strap leads or copper foils, as these options would have greatly reduced the translucency of the glass, and thus posed significant visual interference throughout the window. Conservation trials undertaken on York Minster’s nave aisle window n27 in 2018, under the direction of Nancy Georgi, where several different options for the consolidation of micro-cracked glass were trialled, found that a copper wire matrix, while offering less visual disturbance than strap leads, nonetheless provided insufficient strength to the compromised glass (Fig. 9). Double plating, widely used at York Minster in the period 1920–1970, was also discounted, as the plated unit can trap dirt and moisture, leading to the development of a microclimate in the interspace between the two plates (Fig. 10). Water ingress into the interspace can trigger the growth of micro-organisms on the historic glass, which can cause significant damage to paintwork and the glass surface. Following deliberation, it became evident that an alternative solution was required, and that further investigation was necessary to determine the best course of action.

Fig. 9: Photo from a 2018 trial on window n27, assessing the efficacy of a copper matrix for the stabilisation of microcracked glass. Cyclododecane has been used temporarily to keep the wire in place. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 9: Photo from a 2018 trial on window n27, assessing the efficacy of a copper matrix for the stabilisation of microcracked glass. Cyclododecane has been used temporarily to keep the wire in place. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 10: Glass plate of c.1930, removed from panel 2a, showing the deposition of dirt on the interior of the plate, where it has accumulated along the lines of paintwork. Photo: York Glaziers Trust.

Fig. 10: Glass plate of c.1930, removed from panel 2a, showing the deposition of dirt on the interior of the plate, where it has accumulated along the lines of paintwork. Photo: York Glaziers Trust.

Drawing upon the work of Dr Stephen Koob, an expert in the use of Paraloid B-72 in glass and ceramics conservation, it was decided that Paraloid B-72 would successfully meet the conservation requirements of S5’s micro-cracked glass.((S. P. Koob, Conservation and Care of Glass Objects, London, 2006; S. P. Koob, ‘Manipulating Materials: Preparing and Using Paraloid B-72 Adhesive Mixtures’, Objects Specialty Group Postprints, 25, 2018, 1-8.)) Widely used as an adhesive in cultural heritage conservation, Paraloid B-72 has many advantages including stability, ease of use, reversibility, fast setting time, and miscibility with several solvents.((S. Chapman and D. Mason, ‘Literature Review: The Use of Paraloid B-72 as a Surface Consolidant for Stained Glass,’ Journal of the American Institute for Conservation, 42/2, Summer 2003, 381; Koob, ‘Manipulating Materials’, 1.)) Compared with that of epoxy resins, the bond strength of Paraloid B-72 is considerably weaker, which allows for easier removal and renders it more suitable for stabilising heavily cracked, friable glass with low tensile strength.((J. Podany, K. M. Garland, W. R. Freeman and J. Rogers, ‘Paraloid B-72 as a Structural Adhesive and as a Barrier within Structural Adhesive Bonds: Evaluations of Strength and Reversibility,’ Journal of the American Institute for Conservation, 40/1, Spring 2001, 15.)) It can be used both as a bonding adhesive for cracks and a surface consolidant for loose paintwork, suggesting its potential suitability for safely securing a consolidating material to the surface of the glass. Furthermore, it retains the optical properties of historic glass due to its refractive index of 1.49.((Centenaro et al., ‘Consolidation and Coating Treatments’, 135.)) Given that Paraloid B-72 is known to fail at high humidity levels, the control of the window’s surrounding environment is instrumental to the success of this conservation approach.((J. Brandt et al., ‘Evaluation of the Composition, Thermal and Mechanical Behavior, and Color Changes of Artificially and Naturally Aged Polymers for the Conservation of Stained Glass Windows’, Polymers, 15, 2023, 16.)) As part of its conservation programme, window S5 will be fitted with internally ventilated environmental protective glazing, thereby maintaining the medieval glass in an environment with stable temperature and humidity, protecting it from condensation, rain and wind pressure. As well as protecting the stained glass itself, this external layer of glazing will also prolong the life of any contemporary conservation materials used in the window’s conservation.((For further information on YGT’s approach to environmental protective glazing, see: http://www.yorkglazierstrust.org/our-work/environment-and-protection/ . A broader overview of the current state of protective glazing technology is provided in S. Oidtmann, J. Leissner and H. Römich, ‘Schutzverglasungen’ in A. Wolff (ed.), Restaurierung und Konservierung historischer Glasmalereien, Mainz, 2000, 167-209. English translation available online at: https://www.cvma.ac.uk/conserv/glazing.html))

 

Current conservation scholarship strongly informed our decision-making process for this project. A recent Cologne Cathedral study which explores the conservation of ‘craquelé’ or microcracked glass using ‘Glasfasergewebe’ (fibreglass webbing or mesh), found this to be a useful solution for securing irregular glass cracks.((U. Brinkmann et al., ‘Anwendungen Innovativer Restaurierungsmaterialien und Methoden zur Sicherung Craquelierter Glasmalereien Modellhafte Anwendung an Glasfenstern des Kölner Domes (Weltkulturerbe)’, German Federal Environmental Foundation funded project, June 2013. Available online at: hornemann-institut.de/de/epubl_txt/2013_DBUProjekt_Brinkmann.pdf.)) Similarly, the conservation of craquelé glass at Naumburg Cathedral in Germany, utilised Paraloid-B72 and fibreglass webbing to great success.((S. Jarron, J. Hildebrandt, M. Maquiné and E. Yates, ‘The Inherited Condition: The Complex Methodology Required to Conserve the West Choir Windows of Naumburg Cathedral, Germany’, Proceedings of the 11th Forum for the Conservation and Technology of Historic Stained Glass, Barcelona 2022, 20-28. E-Book Online: https://publicacions.iec.cat/PopulaFitxa.do?moduleName=cataleg&subModuleName=cerca_avanzada&idCatalogacio=38528)) During the conservation trials undertaken on Minster nave aisle window n27, the use of a thin mesh cut in strips, or cut to the shape of the fragment and affixed to the glass surface with Paraloid B-72, was found to be a very promising solution, with Polyester Stabiltex and fine steel mesh being the preferred stabilising materials (Fig. 11).

Fig. 11: Polyester Stabiltex cut to the shape of the fragment in order to test its visual appearance during the conservation trial undertaken on York Minster’s window n27. Photo: York Glaziers Trust.

Fig. 11: Polyester Stabiltex cut to the shape of the fragment in order to test its visual appearance during the conservation trial undertaken on York Minster’s window n27. Photo: York Glaziers Trust.

On the basis of these studies, it was determined that the use of a paper, fabric or fibreglass mesh, adhered with a solution of Paraloid B-72, would be the conservation approach likely to  provide the greatest levels of structural stability, translucency and reversibility in the conservation of S5. This method has a significant number of advantages. It facilitates the stabilisation of the broken fragments and prevents loss of material, while allowing for future conservation interventions to take place when required, as Paraloid B-72 can be easily dissolved with the use of solvents. The process is relatively non-invasive, as the introduced materials remain on the surface of the cracked glass, rather than penetrating the cracks themselves. It also negates the need for glass support plates, strap leads or copper foil repairs, all of which require either heat or interference with the lead matrix, which would be risky given the fragility of the glass. Finally, using strips of mesh rather than pieces cut to the shape of the whole fragment would allow for selective placement, thus avoiding the application of adhesive to the entire surface or areas of paintwork. It was also decided that the stabilising material will be attached to the interior of the glass, given the stability of the paint layer and the advanced pitting that renders the exterior too highly textured for the application of adhesive. Furthermore, environmental monitoring has shown that the relative humidity in the interspace between stained glass and protective glazing is significantly higher than in the interior of the Minster, making it safer to introduce new conservation materials to the interior surface.((For more information see YGT’s approach to environmental monitoring at York Minster, online at: yorkglazierstrust.org/our-work/environment-and-protection/.))

With this decision made, the challenge was then to choose between a number of variables, namely the mesh, the solvent, and the strength of the Paraloid B-72 solution. The rest of this article outlines the experiments undertaken to assess and select the appropriate materials for the conservation of window S5.

Selection of Solvents

Four solvents were considered for this experiment: ethanol, toluene, acetone and methyl ethyl ketone, all deemed suitable solvents for dissolving Paraloid B-72.((For solvent data sheets see: fishersci.co.uk/gb/en/catalog/search/sdshome.html.)) In addition, they are readily available to conservators and are relatively inexpensive. All four are hazardous, to different degrees. The risks posed by ethanol, acetone and methyl ethyl ketone, when used with appropriate personal protective equipment, are lower than those presented by toluene, which poses a far more significant danger to health. Ease of use, accessibility and the safety of the conservator were identified as priorities in the selection of materials, meaning that the use of toluene was ruled out in this instance.

Selection and Preparation of Substrates

Four materials were chosen for this experiment. Firstly, Japanese paper, a tissue commonly used in the conservation industry, though primarily within the paper and books conservation sector.((M. Mizumura, T. Kubo and T. Moriki, ‘Japanese paper: History, development and use in Western paper conservation’ in Adapt & Evolve 2015: East Asian Materials and Techniques in Western Conservation. Proceedings from the International Conference of the Icon Book & Paper Group, London 8–10 April 2015, London 2017, 43-59.)) Secondly, Polyester Stabiltex, a very finely woven polyester fabric with a sheer, lightweight texture. Finally, two types of fibreglass fabric were chosen, one pre-treated by the manufacturer (Schlösser and Cramer KG) with a coating of the inorganic-organic hybrid polymer consolidant ORMOCER, and the other left untreated. These fibreglass fabrics are similar to Polyester Stabiltex in their texture, weave and translucency. Although fine steel mesh was identified as a useful stabilising material in the conservation trials for window n27, it was not suitable for our purposes as it requires soldering to the perimeter of the fragment, whereas the intention with S5 was to use isolated strips of mesh to stabilise individual cracks and avoid unnecessary interference with the surrounding lead.

The Japanese paper was very easy to handle, cut and manipulate. It demonstrated no tendency to warp or fray, and it retained its shape. The Polyester Stabiltex was rather difficult to cut and manipulate due to its floppy texture and its tendency to stretch along the bias of the weave. However, it was not overly prone to fraying, and generally held its shape. The treated fibreglass fabric was relatively easy to handle, cut and manipulate, primarily due to its stiff texture. It demonstrated a slight tendency to warp and fray, but generally held its shape. The untreated fibreglass fabric was very difficult to handle, cut and manipulate. Its floppy texture, strong propensity to warp and stretch along the bias of the weave, and tendency to fray at the lightest touch, rendered it a challenging material to use.

The three chosen solvents were applied in unadulterated form to samples of the four selected substrates to test the impact of the solvents on the stability, condition and visual appearance of the substrates. Assessments were made on initial application, after two hours and again the following day. In general, none of the samples showed signs of significant damage, deterioration or discolouration. All the substrates were subject to a degree of lifting and warping when treated with the chosen solvents, however this appeared to be least pronounced in the case of ethanol. Of the four substrates selected for the trial, Polyester Stabiltex showed the least propensity to warp, although the sample treated with methyl ethyl ketone did display some lifting once dry. Ultimately, none of the samples displayed significant enough results to preclude their use in further trials.

Preparation and Application of Solutions

Two solution strengths were selected: 25% and 6%, as this would allow for clear comparison between the effects of a strong, viscous solution and a weaker, runny solution. Stirred intermittently, the rate at which the Paraloid B-72 dissolved varied between the solvents, with the acetone and methyl ethyl ketone quickly rendering the adhesive viscous and sticky, while in the ethanol, it remained in solid beads (Fig. 12). Ultimately, ethanol failed to dissolve the Paraloid B-72 after 4 hours, in either 25% or 6% concentrations, and was therefore discounted from further experimentation. In the case of the acetone, the 25% solution fully dissolved the Paraloid B-72 after 70 minutes, and the 6% solution after 15 minutes. The methyl ethyl ketone, at 25% solution took 80 minutes to dissolve the Paraloid B-72, while the 6% solution took 20 minutes. These results are roughly comparable with one another, and both solvents were considered suitable for further investigation. 

The four solutions of Paraloid B-72 (acetone 25%, acetone 6%, methyl ethyl ketone 25%, methyl ethyl ketone 6%) were applied to samples of light-coloured glass and dark-coloured glass. On each sample of glass, half of the surface had been abraded using a Dremel tool to replicate, in so far as is possible, the texture of glass corrosion (Fig. 13). It was anticipated that adhesion of the substrate to the surface of the glass samples may differ between uncorroded and corroded glass, so this was accounted for in the preparation of the test materials. The solutions were applied using a small paintbrush. A layer of adhesive was applied to the glass sample and a square piece of substrate was applied on top. A second layer of adhesive was then applied, but only to the upper half of the substrate. This was done to facilitate comparison of the results of a single adhesive underlayer versus a double ‘sandwich’ of adhesive layers in order to evaluate several criteria:

  1. Ease of application (both the substrate and the solution)
  2. Level of adhesion
  3. Textural impact (i.e. smooth or rough finish, development of bubbles etc.)
  4. Visual impact (i.e. development of white cast, darkening, impact on translucency etc.)
Fig. 12: Solutions of Paraloid B-72 in acetone, ethanol and methyl ethyl ketone. Photo: York Glaziers Trust.

Fig. 12: Solutions of Paraloid B-72 in acetone, ethanol and methyl ethyl ketone. Photo: York Glaziers Trust.

Fig. 13: Abrading the surface of the test glass using a Dremel tool. Photo: York Glaziers Trust.

Fig. 13: Abrading the surface of the test glass using a Dremel tool. Photo: York Glaziers Trust.

Results: Ease of Application

In the case of both solvents, the runny 6% solutions were much easier to apply. The 25% solutions were very viscous and challenging to apply in a single, smooth layer. The thick nature of the 25% solutions meant that the paintbrush tugged at the substrates, in particular the Polyester Stabiltex and untreated fibreglass fabric, resulting in warping. This was most pronounced with the acetone solution, but also occurred with the methyl ethyl ketone solution. The difference between the 25% and 6% solutions was most notable with the acetone solutions, given the solvent’s propensity to evaporate rapidly. Methyl ethyl ketone offered more working time.

Results: Level of Adhesion

The 25% solutions offered excellent adhesion irrespective of substrate or glass surface. The level of adhesion provided by the 6% solutions was consistently lower, except in the case of Japanese Paper which adhered well, possibly due to the paper’s absorbency.

Results: Textural Impact

There was a stark contrast between the textural impact of the 25% and 6% solutions. The 6% solutions offered results that were uniformly smooth and clear, whereas the 25% solutions tended to appear rough, clogged or lumpy. This was very conspicuous on the grid-like textures of the Polyester Stabiltex, treated fibreglass fabric and untreated fibreglass fabric, as the solution tended to clog the weave and trap air bubbles. However, between the 25% acetone and 25% methyl ethyl ketone solutions, the latter performed notably better, with acetone 25% being the overall worst-performing solution with regards texture.

Results: Visual Impact

The solutions fared moderately well across all categories regarding visual impact. One notable exception is the combination of Japanese paper with methyl ethyl ketone 6% on dark glass, which was highly conspicuous due to its white cast in reflected light. At the other end of the spectrum, Polyester Stabiltex with acetone 6% and methyl ethel ketone 6% were the most successful combinations in visual terms, offering a very discreet covering to the glass. Aside from these three exceptions, mid-range results were achieved by all other solvent-substrate pairings.

Results: Light versus Dark Glass

The colour of the glass had moderate impact on the outcome of this experiment. Results were generally consistent across light and dark glasses, with the exception of the assessment of visual impact. Here, results for a number of samples showed the light glass scoring more highly than its dark glass counterpart (acetone 25% with Polyester Stabiltex; acetone 6% with treated fibreglass fabric; methyl ethyl ketone 6% with Japanese paper and treated fibreglass fabric). This is probably because the white cast presented by the substrates in certain solutions is more prominent on darker glass than on lighter glass. The only exception to this is the 25% acetone solution on untreated fibreglass fabric; the negative visual impact was more significant on the light glass than on the dark, due to the clogging of the material with the thick adhesive.

Results: Corroded versus Uncorroded Glass

As mentioned above, the texture of the glass surface appears to have had no identifiable impact on the outcome of this experiment.   

Discussion

While the experiment was ultimately successful, there is no obvious winning solvent-substrate combination from among those tested. The majority of pairings have significant advantages and notable disadvantages. Each may be appropriate to use in a given circumstance, and very few options can be decisively ruled out. This section, therefore, will outline the decision-making process for the use of the tested materials in the context of York Minster’s window S5 specifically. Neither the 25% nor the 6% solutions were without their drawbacks. The former provided excellent adhesion but was viscous and visually unsightly. The latter applied well and looked inconspicuous but provided relatively poor adhesion for our purposes. If the conservation methodology called for light adhesion, there would be few reasons not to advocate for 6% solutions of either acetone or methyl ethyl ketone. The textural and visual impacts of these solutions were negligible, rendering the materials suitably inconspicuous for successful conservation work. Both may be appropriate in a museum setting, for instance, where ongoing visual monitoring is possible. In the case of York Minster S5, however, the location of the window at clerestory level, and the need for a secure solution that will last many decades, makes a firmer and more reliable adhesion necessary. Despite the installation of EPG, it is anticipated that the 6% solutions would be too weak to provide much long-term stability. The 25% solutions provided excellent adhesion. However, the visual impact and viscous texture precluded their use in this context. All of the substrates, with the exception of Japanese Paper, snagged in the viscous solution, resulting in the weaves being pulled out of alignment. It was therefore decided that a middle ground was necessary, and that a 12% solution would allow the conservator to benefit from the adhesive properties of a stronger solution, without sacrificing textural and visual benefits of a thinner, runnier solution.

The time needed to dissolve the Paraloid B-72 was shorter in the case of the acetone solution. However, the time difference was marginal and does not present a compelling reason to choose acetone over methyl ethyl ketone. In fact, while the differences between the two solvents were generally slim, there were a couple of key advantages to using methyl ethyl ketone. Both methyl ethyl ketone solutions (25% and 6%) were smoother than their acetone counterparts. Additionally, methyl ethyl ketone offered a longer working time than acetone, given the latter’s rapid drying time. This allowed for smoother, less rushed application, and ultimately led to the selection of methyl ethyl ketone for this conservation treatment programme.

With a 12% solution in methyl ethyl ketone chosen, the next task was to select a mesh material. While the Japanese paper offered good results in terms of application and adhesion, it presented an obtrusive white cast, especially on the dark glass. As a paper, its resistance to moisture damage is lower than the polyester or fibreglass materials, and so its use in an architectural setting cannot be readily justified. Furthermore, as an organic material, it is possible that the paper could be a medium for the growth of mould if exposed to moisture. It could therefore be recommended for conservation treatments in museum contexts, where the environmental conditions can be more easily controlled. The untreated fibreglass fabric was very difficult to handle, cut and manipulate. It is prone to warp along the bias of the weave, making it nearly impossible for the conservator to maintain its shape during application. The untreated fibreglass fabric has a strong tendency to fray along the edges and tangle in the brush during application, so it was easily eliminated from the running. Both the Polyester Stabiltex and the treated fibreglass fabric performed reasonably well in all areas. Although the treated fibreglass fabric outperformed the Polyester Stabiltex in terms of ease of application due to its slightly stiffer structure, the two materials offered identical results for adhesion and texture. In terms of visual impact, however, Polyester Stabiltex had notable advantages and was selected for this conservation programme due to its very inconspicuous appearance, whereas the treated fibreglass fabric had a noticeable sheen which created the appearance of a glossy film on the surface of the glass.

Implementation

Ultimately, it was decided that Polyester Stabiltex adhered with a 12% solution of Paraloid B-72 in methyl ethyl ketone was the best combination of materials for our purposes. The Polyester Stabiltex was cut into small strips, approximately 0.5 x 2cm. These strips were carefully placed to brace each crack, while avoiding overlap with any painted detail. The fabric strips were coated with the Paraloid B-72 solution using a small paintbrush. After five minutes, a second coat of adhesive was applied to ensure sufficient adhesion. Finally, a cotton swab lightly dampened with acetone was used to clean up the edges of each fabric strip, as areas of exposed adhesive beyond the fabric had a slight surface sheen.

The results of this technique are very inconspicuous from close range, and invisible from a metre’s distance, both in transmitted and reflected light (Fig. 15). As part of this project, two test samples were prepared. One will be retained in the studio for regular, ongoing examination. The other, which will be situated at clerestory level to replicate the environmental conditions experienced by the historic glass, will be examined at wider intervals. It will be useful to have these samples available for analysis, so that YGT can assess the condition of the conservation materials over time and respond proactively if the materials show any indication of failure.

Fig. 15: Detail of the interior of panel 5a, photographed: (a) in reflected light showing microcracks before application of Polyester Stabiltex; (b) in reflected light showing microcracks after application of Polyester Stabiltex; and (c) in transmitted light showing microcracks after application of Polyester Stabiltex. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Fig. 15: Detail of the interior of panel 5a, photographed: (a) in reflected light showing microcracks before application of Polyester Stabiltex; (b) in reflected light showing microcracks after application of Polyester Stabiltex; and (c) in transmitted light showing microcracks after application of Polyester Stabiltex. Photo: York Glaziers Trust, reproduced courtesy of the Chapter of York.

Ultimately, this conservation strategy has successfully met the needs of window S5 by providing stabilisation to its vulnerable micro-cracked glass. As shown, the visual impact of the intervention is negligible; it allows for the re-treatability of the window when necessary; and the process has succeeded in retaining all material inherited in the window. On a final note, the provision of environmental protective glazing is critical in ensuring the ongoing safety of the glass, the lead and the newly introduced conservation materials. The methodology explored in this article is not recommended for unprotected stained glass but would be appropriate where there are suitable environmental controls, either in situ or in a museum context.

Acknowledgements

The author would like to thank Nancy Georgi MA ACR for her guidance in developing this experiment and for offering instrumental feedback and support throughout. Many thanks also to YGT colleague Sarah Shepherd MA for providing photographs and very useful discussions.

Comments are closed.