‘The tears wept by our windows’: severe paint loss from stained glass windows of the mid-nineteenth century
This feature article is based on the dissertation submitted by Alison Gilchrist to the University of York in September 2010 as part of her MA in Stained Glass Conservation and Heritage Management. The full text of the dissertation is available on the CVMA website and also in the University of York library. Alison is now working as a stained glass conservator at Barley Studio, Dunnington, near York.
Introduction
The aim of the study was to investigate the well-known and widespread problem of severe paint loss from stained-glass windows made by many firms in the mid- to late nineteenth century. It has become known as the ‘borax problem’ (borax being a commercial product that was used to lower the melting temperature of glass), and its effects were identified in a letter sent by William Morris to George Howard, around 1880, in which he writes: ‘We (and I believe all other glass painters) were beguiled by an untrustworthy colour, having borax in it, some years ago, and the windows painted with this are going all over the country. Of course we have taken warning and our work will now be all right. We have given instructions to our man to take out the faulty glass, which we will – restore! – at once, and pay for that same ourselves – worst luck! Borax is the name of the culprit: the colour makers, finding that the glass-painters wanted a colour that would burn well at a lowish temperature, mixed borax with it to that end; but unluckily glass of borax is soluble in water, and hence the tears wept by our windows – and our purses. We use harder colour now, so that if any window of ours goes now it must be from other causes; bad burning or the like; I don’t think as things go that this is like to happen to us.’
As Morris suggests, many (although not all) stained-glass firms of the period experienced the problem of paint loss. Harrison suggests that James Powell & Sons suffered particularly badly – as did Burlison & Grylls, Lavers, Barraud & Westlake, and C. E. Kempe & Co. – while Heaton, Butler & Bayne and Clayton & Bell apparently had fewer problems with their paint. The problem was not confined to the English studios: many windows made by the important nineteenth-century Belgian studios of Capronnier and Bethune are now suffering from severe paint loss, as are many nineteenth-century windows in the United States of America.
Later writers have also blamed the addition of borax for the failure of glass paints, with Newton commenting that ‘in the 19th C. some purveyors of paints added much too much borax with the result that the painted linework would dissolve in any condensation which occurred on the inside of the window!’ It is interesting to note, however, that borax is still used as an additive for glass paints today, suggesting that the problem cannot be quite so simple. More generally, the use of ‘incorrectly formulated glass paint’ that ‘did not fuse correctly with the glass’ has been blamed for the failure of the Saunders & Co. windows at Cork, Skelton and Studley Royal. Underfiring of paint (the ‘bad burning’ of Morris’s letter) and inconsistency of firing have also been suggested as causes for failure. Few technical studies of the problem however have been reported.
In this work a multi-disciplinary approach was taken, looking at the historical background and undertaking a technical study before discussing conservation issues. As many windows created by different studios over the period in question are now suffering from paint loss, it was decided to narrow the focus to selected case study windows made by a single firm. The chosen firm was John Hardman & Co. of Birmingham, whose substantial surviving archive is held in Birmingham Archives and Heritage. During the nineteenth century, Hardmans were prolific producers of stained-glass windows, including many that are now suffering from severe paint loss. Three windows, showing differing degrees of deterioration, were selected for detailed study: the former west window of Sherborne Abbey (1851, removed 1997, severe paint loss); the west window of Beverley Minster (1859 and 1865, moderate paint loss); and the former north transept window of All Saints’ Church, Emscote, Warwick (1889, removed when the church was demolished in 1967, paint in good condition). This article will focus on the study of the Sherborne Abbey window, to which access was kindly granted by the Worshipful Company of Glaziers, since the window is currently held in the London Stained Glass Repository. (On the Repository, see this Vidimus feature article.)
Historical Background
The medieval abbey church of St Mary, Sherborne (Dorset), underwent a major restoration under the architect Richard Cromwell Carpenter (1812 –1855) between 1849 and 1858, during which time a number of stained-glass windows were commissioned from Hardman & Co. The first Hardman Glass Sales Day Book (November 1845 – January 1854) includes the windows ordered by Carpenter for Sherborne:
A large West Stained Glass Window of 27 lights and 85 pieces of tracery £301 [12 June 1851]
3 windows of Stained Glass for North Aisle of 4 lights each £210 [23 July 1851]
Subjects in above 3 windows The Twelve Apostles
1 window of Stained Glass subject groups for South Transept of 4 lights £70 [23 July 1851; this actually refers to a fourth north aisle window]
A Stained Glass Transom Window of 8 lights and tracery, for South Transept Window £320 [19 April 1852] Subject The ‘Te Deum’
Correspondence between Pugin and Hardman shows the direct involvement of Pugin in the design of these windows, with the cartoons for the west window being drawn by Frederick Hill and John Hardman Powell. The difficulty of fulfilling the large order in the short time demanded by Carpenter, as well as the relative inexperience of Powell in preparing the cartoons, is also clear from these letters. Both Carpenter and Pugin however appear to have been pleased with the final results, with Carpenter writing to Hardman: ‘Mr Pugin has written to me to say the work makes a splendid job … let me tell you how much I like the Sherborne glass … altogether it is very fine’. The west window for Sherborne Abbey depicted figures of Old Testament prophets in its twenty-seven main lights, with decorative roundels in the tracery lights [Fig. 1]. The lack of canopies, which might be expected over such figures, has been explained by the relative shortness of the lights in the window, which means that the figures almost fill their openings. The rather repetitive design of the tracery lights was commented upon by Carpenter, who wrote to Pugin that ‘in the West window the effect is much injured by a row of red stars which glare unfortunately & the people say look like railway signals. I have spoken about this to Hardman who says an alteration shall be made.’ It is not clear whether any such alteration was in fact made.
Problems of water penetration and paint loss from the west window were noted by the abbey architect Kenneth Wiltshire in the 1970s, although no action was taken at that point. A full survey undertaken by Wiltshire in 1989 noted that the leadwork was in poor condition, and that there were ‘a considerable number’ of broken and missing pieces of glass, leading to ‘appreciable water penetration’; in addition, ‘many of the faces, inscriptions and swags can no longer be read’. The supporting masonry was also reported to be in need of substantive repair. After various options had been considered, and after a long legal case, the window was removed and donated to the London Stained Glass Repository; John Hayward’s ‘Incarnation’ window was installed in the Abbey in 1997.
An indication of problems with the painting of these early Hardman windows in Sherborne Abbey appears in correspondence regarding the north aisle Apostle windows. On 16 September 1865, only fourteen years after the windows were installed, the Revd E. Harston wrote to Hardman.
A curious and mortifying circumstance has happened here with which I feel it right to acquaint you. A workman employed this week in dusting this Abbey Church disobeyed his orders and dusted the glass of your windows in the North Aisle containing the 12 Apostles – one of whose faces he has entirely obliterated so that nothing now remains but a piece of clear transparent glass where the face should be, showing the wire guard through it from the outside. I believe you erected these windows in 1850 or 51 during the life of the late Vicar Mr Parsons. And they have stood quite well to this time – and how a workman with a feather brush (clumsy and stupid though he was) could so entirely obliterate the features of S. James, and more or less damage some of the other faces – is a mystery which we cannot understand, assuming that the colouring was burnt in as usual. Can you at all suggest an explanation of the injury? The man excused himself by saying that the glass was damp, & being a north wall, this is true of the masonry – but how came the pigment to be so easily got off? I shall be glad to hear from you on the subject – and whether you can remedy the damage.
Hardman’s reply, dated 18 September 1865, placed the blame firmly on the glass paint used.
We are exceedingly sorry to hear that the colour of the faces of the Apostles at Sherborne has become soft & we will of course make them good. The reason is that about 1850 the seller who had made our colour for 12 years previously, became very ill, he did not tell us so but went on supplying us for some months with inferior colour, without giving us the least reason to suspect its being changed. It was only after sixty works had been executed that we detected the fault, as at first it appeared all right after the burning. It has been a very serious loss & annoyance to us (as of course we are bound to make all good) & more anxiety to those who had works from us at that period. Our foreman shall come down & make a report when we will arrange to make all necessary repairs.
The ‘necessary repairs’ were made in October 1866, and entered into the Glass Sales Day Book as ‘Faces, hands and feet of the 12 Apostles restored’ [no charge]. No indication is made of how the restoration was carried out, but sadly it has proved as impermanent as the original [Fig. 2].
Technical Examination
Three main-light panels from the former Sherborne west window (2d, Moses; 2e, Joshua; and 2f, Aaron) and six tracery lights (D1–6) were made available by the London Stained Glass Repository for examination and analysis during the study. For the purposes of this article, however, only panel 2d (Moses) will be discussed in detail, as representing the condition of the whole window [Fig. 3].
Condition of Panel 2d
Moses is depicted with his traditional ‘horns’ of light, holding a rod and tablets with the numbers I to X, representing the Ten Commandments. Overall, the panel is in a rather poor condition, with several cracked or missing pieces of glass, many lead and solder fractures, and much surface dirt. The most severe problem, however, as can be seen from Fig. 3, is the loss of painted detail. Very little paint survives on the face, hands and tablet of the commandments, leaving only ‘ghosted’ images with some flesh-tone shading. Other areas, however, seem to have survived slightly better, such as the pattern on the halo, areas of the diapered background, and some of the drapery [Figs 4 and 5]. This suggests that there is some difference between the paints used, or the manner in which they were applied. It is noticeable that the best-preserved areas throughout the panel are those painted on a deep yellow potmetal glass, used for Moses’ undergarment. For some reason, these pieces are all reversed, such that the main painted detail is on the outside face rather than the inside; this reversal, and the better-preserved paint on these yellow pieces, are also noticeable in other panels.
When viewed in reflected light, the surviving paint can be seen to be rather reddish brown in colour. Even those areas on the inside face of the glass that appear to have survived relatively well show some deterioration, possibly salt efflorescence, of the painted surface [Fig. 6]. The outside face shows extensive use of back-painted shading, as well as the detail on the reversed yellow pieces, both of which appear to be in relatively good condition [Fig. 7]. Interestingly, there is no back-painted shading on the face or hands (presumably on account of the use of flesh-tone shading in these areas), which could partly explain why these areas stand out for their paint loss; even though the drapery areas have lost their surface paint, they retain some of their modelling in the back-paint.
Viewing the paint surfaces under a digital microscope shows the paint on the inside face of the panel to be loose and powdery, rather than a coherent layer; where the layer does survive, the microscope shows the surface to be full of pinholes [Fig. 8]. This broken surface will allow ingress of water and other contaminants that can attack the paint layer, causing further deterioration.
The paint surfaces on the reverse face are generally in much better condition than those on the inside face, though viewed under the digital microscope they can also be seen to contain pinholes in the painted surface [Fig. 9]. In fact, the paint surfaces on the outside face appear very similar to the best-preserved paint surfaces on the inside face; this might suggest that the same paint and firing conditions were used for both, but that the paint on the outside face has remained nearer to its original condition, while that on the inside face has deteriorated over time.
Sampling and Analysis
In order to examine more closely the physical structure of the paint layers, and to determine their chemical composition, small samples were taken from various pieces. The panel was partly dismantled, taking advantage of glass pieces that were already broken and fractured solder joints across the centre of the panel, to allow access to suitable pieces (where the piece had surviving paint at the edge, and where the removed sample corner would be hidden under the surrounding lead leaves). Unfortunately, it was only possible to sample a limited selection of types of glass in this way; in particular, the face, hands, tablet and inscription, where paint loss was particularly severe, were not sampled.
The sample preparation and analysis was carried out in the University of York Nanocentre. The samples were mounted as cross-sections and examined using a scanning electron microscope (SEM) linked to an energy-dispersive X-ray spectrometer (EDS). The combination of SEM and EDS allows both imaging and elemental analysis of samples, and allows different areas of a sample to be analysed individually, in this case permitting the compositions of the paint layer and the underlying glass to be determined individually. Unfortunately, it is not possible to detect all elements using this system, and in particular it proved impossible to detect the element signal for boron (the element characteristic of borax). The presence, or absence, of borax in the samples analysed could not therefore be confirmed.
Examination of the Sherborne samples under the SEM showed the paint layers to be rather deteriorated (as expected), with large variations in the thickness of the layer across the sample [Fig. 10]; a granular appearance with large particles within the layers [Fig. 11]; and numerous vertical micro-cracks through the paint layer, often running from the surface of the paint right down to the interface with the glass [Fig. 12].
Analysis of the chemical composition of the samples showed that the glasses used to be of two different types. The clear glass and the clear base of the red flashed glass are soda-lime silicate glasses, whose compositions are comparable to recipes of the time and other published data. The yellow glass, in contrast, is a lead silicate coloured with iron. The red layer of the red flashed glass is also a lead glass, but coloured with copper.
The paint used for all the samples analyzed is essentially the same, being composed of lead silicate glass (generally known as the flux) mixed with iron-oxide pigment, again as would be expected from recipes of the time. Further analysis of the SEM images shows that the large particles present are the rather coarsely ground iron-oxide pigment particles embedded in the surrounding flux. The level of pigment is very high, approaching 50% of the paint by weight; this corresponds to the maximum pigment loading for a good-quality paint layer, requiring the use of a flux soft enough to fill the small gaps between pigment particles. Experimentation by Schalm has shown that it is very difficult to obtain a durable paint layer with such a high pigment loading. The ratio of lead to silica in the flux is around 3:2, suggesting that the flux was quite hard, with a relatively low lead content. Although the analysis undertaken was not capable of detecting boron, it is unlikely that borax (sodium borate) was added to soften the flux, as the amount of sodium present is very low. It is interesting to note that the paint used on both the clear and yellow glasses has the same chemical composition, but has deteriorated to markedly different extents.

Fig. 10. Paint layer from Sherborne, sample 5 (clear glass with paint on inside face), showing variation in thickness of the paint layer.
Comparing these results with those obtained for the other Hardman windows examined in the study shows that very similar glasses (both soda-lime silicate and lead silicate) were used in each case. Similar glass paints were also used, although the level of pigment present varies; the Sherborne window, which shows the most deterioration, has the highest amount of pigment. The flux used (the ground-glass component of the paint) is consistent between the various windows, being a lead-silicate glass with the ratio of lead to silica varying between around 1:1 and 3:2. This suggests the use of a flux recipe similar to that given in Porter’s Treatise on the Origin, Progressive Improvement, and Present State of the Manufacture of Porcelain and Glass of 1832, composed mainly of flint glass (3 parts silica to 2 parts lead oxide) with the addition of lead oxide to soften the flux. Porter also suggested the use of equal weights of pigment and flux, as used at Sherborne. There is no evidence for the presence of borax in these formulations; its use cannot be ruled out however by this analysis. It is interesting that there is relatively little difference in the paints used by Hardmans for these three windows, executed over a period of nearly forty years, although correspondence in the Hardman archive suggests that they were buying paints from at least three different suppliers over this period.

Fig. 11. Paint layer from Sherborne, sample 4 (yellow glass with paint on outside face), showing large particles within the layer.
The use of lead-silicate glasses in both the Sherborne and Beverley windows suggests that the paint applied to these glasses must have been fired at a relatively low temperature; the softening point of lead glasses is significantly less than that of soda-lime glasses, and the base glass should not deform during the firing process. If all the pieces of each window were fired together, at a similar temperature, then the paint on the soda-lime glass pieces would have been left under-fired compared with that on the softer lead glass pieces. This, combined with the relatively coarse pigment particles and high pigment loading seen in some of the case-study examples, would have resulted in a fired paint layer that appeared to be well fixed, but which was in fact vulnerable to deterioration. The paint would have been attached more firmly to the softer lead-silicate glasses, and it is notable that the colourless glasses used for the flesh areas (which show the worst deterioration at both Sherborne and Beverley) have the harder soda-lime-silicate composition.
Analysis of samples from the west window of Beverley Minster suggested that corrosion of the glassy phase of the paint has occurred. Atmospheric moisture will leach alkali metals (and also lead, especially if the moisture is slightly acidic) from the paint, creating an alkaline solution on the surface of the paint, which is then able to attack the silicate matrix. Cracks and pinholes in the paint layer, as seen in the Sherborne samples, would also encourage water to enter, and be held in, the paint layer, causing further deterioration. The gradual breakdown of the paint structure would lead to loss of surface material, causing the paint layer to become thinner over time and resulting in the visual fading effect.

Fig. 12. Paint layer from Sherborne, sample 5 (clear glass with paint on inside face), showing large particles and vertical cracks in the layer.
The involvement of water in the deterioration process suggests that the environment of the window may also be important. Both leakage of rainwater and the formation of condensation can lead to the inside face of the glass’s being damp for long periods of time (as noted by the Revd Harston in his complaint to Hardmans regarding the north aisle windows at Sherborne), thus encouraging the corrosion process. Although the outside face of the window is exposed to the weather, this is a different environmental cycle, as the glass is washed clean by rainwater and then dries until the next rainfall. The inside face, by contrast, becomes damp by condensation (allowing the corrosion process to begin) and then dries by evaporation, leaving corrosion products on the surface which will continue the process as soon as condensation recurs (most likely on a daily cycle). As there is no significant difference between the chemical composition of the paints used on the inside and outside faces of the Sherborne glass, it seems likely that these different inside and outside environments, in combination with the different base-glass compositions, are responsible for their different levels of deterioration.
Concluding Remarks
The work described in this article, and in the full dissertation, represents only a brief foray into this complex and wide-ranging subject. Much more research is needed to understand fully the various factors underlying severe deterioration and loss of glass-paint from windows of this period, but it does seem even from this brief study that the phenomenon cannot be attributed simply to the use of borax in the paint. The full dissertation text also outlines and evaluates techniques currently available for the conservation and restoration of windows suffering from paint loss; further discussion is required however to prevent more windows from being lost in the manner of the west window at Sherborne. In particular, the installation of protective glazing systems (now generally accepted for the preservation of medieval glass), as a method of protecting vulnerable paint surfaces from further deterioration, would be an important step forward in the treatment of these windows.
Thanks
• London Stained Glass Repository, for access to the Sherborne panels
• Church Buildings Council, for access to the files on Sherborne
• Ian Wright and Prof. Ed Boyes, University of York Nanocentre, for facilitating SEM-EDS analysis
• David Dungworth, English Heritage, for access to analytical data on Beverley Minster west window.
Further Reading
A catalogue of the Hardman archive is available here.
The Roy Newton bibliography is available on the CVMA (GB) website here.
• D. Dungworth, H. Bower, A. Gilchrist and R. Wilkes, The West Window, Beverley Minster, Beverley, East Yorkshire: Chemical Analysis of the Window Glass and Paint, English Heritage Research Department Report Series, 25, 2010
• M. Fisher, Hardman of Birmingham: Goldsmith and Glasspainter, Ashbourne, 2008
• M. Harrison, Victorian Stained Glass, London, 1980
• J. Hayward, ‘Sherborne Abbey: The Great West Window’, Journal of Stained Glass, xxi (1997), pp. 92–97
• R. G. Newton, The Deterioration and Conservation of Painted Glass: A Critical Bibliography, CVMA (GB), Occasional Paper II, Oxford, 1982
• J. W. MacKail, The Life of William Morris, London, 1922
• O. Schalm, ‘Characterization of paint layers in stained-glass windows: main causes of the degradation of nineteenth century grisaille paint layers’, Ph.D. thesis, University of Antwerp, 2000
• S. A. Shepherd, ‘The West Window of Sherborne Abbey’, Journal of the British Society of Master Glass Painters, xix/3 (1994–95), pp. 315–22
• S. A. Shepherd, The Stained Glass of A.W.N. Pugin, Reading, 2009