RESTORATION TECHNIQUES OF STAINED GLASS    
 
 
BRUCE S. HUTTON BACHELOR OF ARTS FINE ART (CRAFT) 1987. CHISHOLM INSTITUTE OF TECHNOLOGY SUBMISSION FOR GRADUATE DIPLOMA IN CERAMIC DESIGN PENINSULA SCHOOL OF ART, MONASH UNIVERSITY. 1996  

For the most part the well being of glass lies in the hands of the person delegated the task of finding someone to ‘fix this’. Stained glass restoration is a specialist field, the person undertaking the works needs a sound knowledge of the history of glass and design as well as access to a stock of the many new and old glass varieties. The wrong person often leads to important glass being removed, copied poorly, then the original discarded.

INTRODUCTION
The following document is a narrative focusing on two projects undertaken in the past two years. They have been used as vehicles to describe, in some detail, restoration processes and combine a mixture of experience and research. These techniques however are specific to the applications described, and should be viewed as such.

If nothing else this paper illustrates the difficulties in dealing with old material and may perhaps help a practitioner faced with similar difficulties.
 

Chapter One: RESTORATION OF AN EXTENSIVELY DAMAGED PAINTED STAINED GLASS WINDOW.

 
The restored window.

The following narrative is a log of the restoration of the Constance Emily Fanning window. It illustrates not only the difficulties involved in such a task, but also the need for research and improvisation using modern materials to maintain the integrity of an historic window.

Christchurch in Acland St, St.Kilda, Melbourne was burgled in November 1994. According to the Reverend Philip Hutchinson, the successful thief stole one antique Gothic chair. In accessing the piece, the offender caused massive amounts of damage to the Constance Emily Fanning Window, dated 1874. To comprehend the damage, one only needs to imagine a person smashing a hole in a fully painted stained glass window large enough to drag through both themselves and the chair.
The window was originally fabricated in three segments; it was the lower two that ere damaged, the lowest of these two suffering the most damage and the first dealt with.
 

  Panel ‘A’, As Found.
 
Restoration of panel ‘A’.

The pieces of the window were retrieved and the many hundreds of fragments gathered. The lead matrix and all the necessary photographs were taken. The middle segment of the window was left in situ, with only the pieces that were in danger of falling athered and their proximity noted.

The first step was to lay the sections of the window on a light box, manipulating the lead to its approximate position. Once bent back into shape, this provided the most significant clue as to the panels’ original form. Almost without exception there was a glass fragment still wedged in the cement. These pieces later became the missing shard in a whole piece that was being assembled, and in turn gave the proximity of that piece in the lead work.
An unfortunate aspect about the ‘gathering of clues’ part of this restoration, was that all the windows of the Church had been substantially photographed, except this one. Monash University and the churches likely sources were contacted without success. This raises a critical point. There are many windows and other valuable artefacts in Melbourne that have not been accurately documented. With stained glass, photographs should be taken using both colour and print film in reflective and transmitted light, copies of these should be stored along with the negatives in a safe place, one set in the church and another in a repository well away from the site. It is also recommended that if the windows are relatively easily accessible, that rubbings be taken of the lead matrixes. The arduous task of sorting out the hundreds of  fragments was the next process to be tackled. Firstly, working on the light box, the shards were assembled into colour groups, then those into design groups. This created piles of fragments that were then moved systematically, registering edge shape and design in the same fashion as a jigsaw puzzle. Steadily the pieces emerged and although very successful for the most part, highlighted a mistake. A few pieces still had one or two missing shards. It is regrettable that during the gathering of the fragments from the church, minute pieces were overlooked, them appearing to be unusable. This project would have been more successful if even the tiniest fragments were retrieved.
Photographs were then taken of the lead matrix before any further adjustment. The lead, further manipulated to a state reasonably similar to its original form, was then transferred onto paper as both a rubbing and as hand drawn lines. The rubbing was then placed on a board and the process of slow dismantling began. This involves a lead knife and lead snips to release the pieces then placing them along with their matching assembled fragments on the appropriate space on the rubbing. Care was taken to keep the lead in a sound condition, as this provides a guide during the leading up process.

 
The window laid out after dismantling.

 The lead matrix.
 

In this section of the original window design there were one hundred and sixty pieces of glass; only eighteen remained unbroken.

At this point the ethical questions arose, as they do with any restoration / conservation project. Virtually every article of the Burra Charter had some relevance to the dilemma at hand, however the most relevant are articles 13, 14, 15, 17, 19, 23, 27 and 28. This was fundamentally a restoration and reconstruction project, with the bias upon the reuse of as much fabric as technically possible. Only the loss of aesthetics and the readability of the window, created the need for accurate reconstruction. Each piece was deliberated upon individually, considering the complexity and human expression of the brush strokes, then the feasibility of enabling this fragment to prevent wind and rain from entering the church. The decisions were made after discussion with  Penny Edmonds from the conservation laboratory of the State Museum. Her input on both the technical and ethical questions was invaluable, giving great focus to the principles of conservation, and sound backing to the use of  HXTAL NYL-1, a high grade epoxy resin developed specifically for the restoration of glass.

The next process was the reconstruction of the lead matrix on paper. This involved taking measurements of the glazing space, templates of  the curves that still remained intact, as well as noting the position of borders in the middle panel, that still remained in situ. This information, and the fact that the central design was essentially symmetrical (a quarter of this was in relatively good condition and accurately noted before dismantling), provided a rigid guide in which to move pieces into their original configuration. The removed pieces of glass retained most of their putty around the edges, a much clearer indication of their glazing space than if the putty was removed. Experience has indicated, that more often than not, glass pieces such as these are inaccurately cut and can give a false indication of a lead line position.
These pieces were then traced from the middle of the plan moving outwards, using a 1.1mm marking pen. This steadily built a plan of the lead structure, enabling missing pieces to be re-cut, and ultimately the releading of the panel. In accordance with the ethics of the Corpus Vitrearum,  it is considered wrong to groze or cut original pieces of glass to make them fit; the creation of a good plan enables leading without any further damage to the glass.

The plan after leading.

Cleaning was the next process undertaken.
All fragments were put through a bath of de-ionised water mixed with 2% non-ionic pH neutral detergent. The pieces were placed in the bath in the same formation and allowed to soak for forty eight hours. At the end of this time the cement was easily removed with a sharp tool, running along the edges. Care was taken not to touch any part of either surface or paint. The flat surfaces were then cleaned with a chamois.
 
Paint loss is an important factor in the selection of a cleaning process, as various Victorian artists and studios chose to use quantities of borax in the paint mixture. The long term disadvantage of this, and legacy to glass restorers, is that borax is water soluble and has been the cause of terrific paint loss; even within the life time of people such as William Morris. In this particular case however, only two scroll pieces suffered very slight paint loss, most likely attributable to under firing or too much gum in the trace. These pieces were only very lightly wiped and left to dry with the others.
In the following section individual case studies will be used to illustrate restoration techniques and reasons for selecting these specific techniques. The aim, as mentioned earlier, is to retain the original material, without compromising the readability and intentions of the original artist.

THE SCROLL.
 
.
    The scroll during restoration. The sections left and right of the ‘H and E’ in the
    centre of the frame, are the original fragments already edge bonded.

The scroll was considered a piece that displayed much of the artist's character; the objective being to preserve all the material available. The first step in this process was to ascertain what the inscription says. Of all the individual pieces in the panel, these were the most extensively smashed due to their length.
It was difficult to decipher and needed the help of the Vicar to find out which verse was likely to contain the words ‘BLESSED’ and ‘WHICH’. With subsequent moving about of fragments the words ‘ARE’ and ‘DEAD’ became apparent. That was enough clue as to the remainder and hastened the development of further words. At this point it is important to describe in some detail, the use of HXTAL NYL-1.
There is a sufficient amount of documentation condoning the use of this epoxy, but unfortunately very little to indicate how to use it. The suppliers themselves provide only a very basic description, and upon further research by questioning conservators, it appeared to be something that requires great practice. Penny Edmonds of the Museum of Victoria, provided information on its dangers, its potential applications, and the use of other materials in conjunction with it, to ultimately achieve the desired results. As with any learning process, the first attempts are inevitably clumsy; the importance of this glass dictated the need for testing on non-valuable glass first. This was simply done by breaking and then mending new float glass. The difficulty in using the adhesive made itself apparent immediately, the predominant question being, how can a material that retains such low viscosity for such a long period of time, be effectively controlled?  The technique used for these pieces was soft dental wax. The nature of  HXTAL, because it needs seven days to set, requires an area for it to sit without interruption and for it to be provided with a suitable reservoir, as it will flow to all crevices until stopped.

STEP 1.   The edges of the pieces to be glued should be carefully cleaned of residues and the edges wiped with acetone in preparation for the pieces to be assembled. The edges should not be handled.
STEP 2.   Soft dental wax sheets are laid on a float glass sheet large enough to contain the assembled piece, then evenly heated with a hair dryer until softened.
STEP 3.   The glass fragment is carefully embedded into the surface of the wax only ensuring that the neighbouring piece is locked into perfect registration.  If the wax is too molten it can squeeze between the glass and effect the join.
STEP 4.   Assuming that the bottom side of the crack is now sealed, reservoirs can be built up the sides. If the piece has also suffered shelling, this can provide a technique for infill if the fragments are not available.
STEP 5.   The HXTAL is then run along the crack with a pin, then as it works with capillary action, moves itself into all the space available to it. If the break is clean, very little HXTAL is needed. The process is aided by a light box, which allows the movement of the epoxy to be viewed; the crack effectively almost disappearing.
STEP 6.   These pieces are then placed in a dust free environment to cure.
STEP 7.   After three days, provided there is no potential for paint dislodgement, any raised HXTAL can be carefully removed with a scalpel. There should be no unwanted resin underneath if the seal is good. The HXTAL can also be dissolved at this stage with cotton wool soaked in acetone. This is a gentler method than the use of a scalpel and a better approach for painted surfaces.

In the case of “Are the Dead Which” (above) and “Die in the” pieces, not all the material was available. This required infilling with a section of new glass painted, stained and fired to match the original, in both reflective and transmitted light. This was then prepared for HXTAL infilling and joined to the other two already substantially repaired pieces, by the same process earlier described. There was less than half the material left in the final word “Lord” that to infill using HXTAL would be unsympathetic to the final result. This piece was repainted to match the original as closely as possible.
                                                                                

 
  The piece ‘Lord’ repainted and date marked.
 
TITLE - ST. MARK C.5.V.47.

This piece was important and required the use of all available material. Like the scroll it needed extensive epoxy repair work, and again there were large pieces missing. These were filled with accurately fitted glass pieces of a similar colour which had been painted and fired with missing details. Very small pieces were infilled with epoxy then a film of transparent colour laid on the surface of the cured epoxy. It too was secured by epoxy. A problem with pigments is their potential of U.V. colour adjustment over a period of time. By using the pigment on the surface only, not throughout the material, any future reversal can be made without the need to dismantle the piece. The restoration process of this piece is well documented in the accompanying photographs.
 

Title piece in the process of being repaired.
 
The Title Piece set into wax with hand painted and fired infills in place.

Detail of HXTAL work, showing wax reservoir and infilling.

CURVED BORDER
These pieces were smashed very badly with many fragments missing. The effect of these pieces in the design was significant, the artist having used the colour gradation in a sheet of antique glass to aesthetic advantage. The original pieces however were beyond repair and, coupled with the fact that they had no painting, the decision was made to remake them. The likelihood of finding a piece the same colour was minimal, to find one with colour and gradation to this extent, was deemed impossible. Studios which stock old glasses were approached and the expected answer received - the only option was to make the glass - and without going into kiln processes to any length the problem was solved as follows.

 
 A comparison between original and fused replica.

 
1.      Locating the colour by looking at all the options available in modern glasses; colour and thickness being important.  Two layers of turquoise Desag machined antique created a colour match acceptable to the dark or thick end of the old piece. The light, or thin end, was very close to one layer of turquoise Desag.

2.        The two layers were fused at 780 degrees Celsius enough to stick-fuse them only. There was no need to full fuse.

3.        The sheet approximately 300mm x 150mm was then sandblasted professionally to achieve perfect gradation of colour.

4.        The piece was fire polished in the kiln to retrieve the finish of the old sample.

5.         Cut to shape in the normal method (glass cutter).

BORDER.
The green border was matched in colour with one layer of green water glass - Spectrum product - and one layer of turquoise machined antique. As these glasses are not fusible, a technique involving the use of copper foil was necessary. The two pieces are cut from the plan, then effectively sandwiched together and held in position by the foil which is wrapped around all edges. It is then leaded into the matrix with the use of high heart lead. There are many applications of this plating technique. It can be used to add missing features to pieces suffering paint loss as well as increase the potential of matching various old glass types. One glass can be used for colour, and the other for texture. When used with historic glass however, attention must be given to the created inter space. The presence of water in this closed or nearly closed environment can lead to the chemical degradation of the glass surfaces. Ultimately after engaging these techniques numerous times the board that had been used to store the fifteen hundred glass fragments contained 162 pieces of much handled glass ready to be leaded. This process was aided by the already developed plan based on the pieces themselves and the remains of the lead matrix. The old lead is valuable because it indicates the way in which one came intersects another; i.e. the way that it is tapered. It is also crucial that the craftsman leading the work is highly experienced. Bridging bad lead joins with solder is completely unacceptable, resulting in solder being hard against the glass surface. If the panel is flexed by whatever force, its presence can lead to the development of a new crack.

 
 Lead work prior to soldering.

After leading the panel was hand puttied in such a manner as to touch as little of the painted surfaces as possible. Even though the paint was secure, it is still porous and if cement or putty were used haphazardly, would create a need to again clean much of the glass, (hydraulic syringes are available to inject the correct amount of putty under the lead leaf, this means that there is no need to touch the surface at all during this process- an obligatory technique if paint loss is evident). The excess was removed with the use of a sharp wooden stick, not a nail, as is the practice for general lead lighting. The oil was removed with the use of whiting and a soft cloth. The final detailing was done in cold colour to the internal surface of the window. By taking a sample of the old painted glass to a paint shop and having the surface computer analysed, an indistinguishable match is achievable. An acrylic based paint is used as it may be removed at a later stage with the use of methylated spirits.

 
  Completed lower panel.

 
 The Middle Panel in situ.

THE MIDDLE PANEL.
This panel although quite broken was still secure in its glazing position and left in situ until the lower panel was completed. It was removed and a plywood temporary glaze was installed in its place. One section was hanging from the bottom by one lead. This was disconnected. The remainder of the panel was still in relatively good condition with solder joins still strong. The panel was easily carried and transported with the use of a board and suffered no further damage. The following report on the restoration of this panel will be divided into treatments and their descriptions.

TREATMENT ONE.
This panel was 1.5 metres high and was bowing, an ailment common to lead light. This would have been present to some degree prior to the damage but grossly exaggerated by the break-in, with many subsequent cracks. It was impossible to work on the window in this state and it had to be flattened. The principle of curing this problem is simply to push the window flat, however this operation must be undertaken with delicacy. The window is covered by a firm rubber foam mat about 20mm in thickness, the purpose of which is to even out the pressure so no one point receives all the weight. A board is then laid onto that and telephone books placed on top. The bow decreases steadily and generally within three days or so, the window becomes flat without any further damage to the window. A rubbing is then taken off the lead matrix. More complicated methods involving the partial dismantling of a window, may need to be applied to other cases. The window in question, prior to the break in, had a maximum deflection of approximately 30mm from flat, and would most likely not have required any intervention

TREATMENT TWO.
There were two areas in the panel that sustained significant damage. The bottom left hand side and the lower section of the painted scene. The remainder of the window needed only very passive intervention, such as cleaning. ‘Treatment two’ deals with the work to the painted scene. All this material was deemed very important and as it was all present, HXTAL edge gluing was the preferred technique. As is virtually always the case, the broken pieces had moved out of registration within the lead matrix, which had to be dismantled before repair was possible. This is done by carefully cutting through the leaf of selected lead came joins on both top and bottom. The panel, if possible, should be turned as infrequently as possible, as this always increases the potential of  breakage. The cement needs to be scraped from the came that the glass is to be released from, with care not to damage paint or cause movement within the pieces. One edge pushed up and against another can create shelling.

 
  The panel on the bench accessing pieces for epoxy treatment.
 

Prior to edge gluing.

Glued and cold colour matched.

Once removed, the pieces are assembled on silicone release paper. Then held in position by pressure sensitive tape. The HXTAL is then applied carefully and sparingly along the cracks. Any surplus resin was removed after three days cure, as mentioned previously.

TREATMENT THREE.
Treatment three involved similar techniques employed in the lower panel, accumulating matching pieces, HXTAL repairs and the remaking of several damaged and lost pieces.  New lead was added where necessary and the panel made back into its original form, puttied and then finished with new ties ready to install.

 

Leading the restored areas.

 
 The restored panel.

The window was refixed using the original bars, and repointed using a mortar mix of, three parts sharp sand to one part cement, as suggested by E. Liddall Armitage.

There is certainly scope for further research in the topic of mortar mixes and relationship between the glass and the stone. the A slaked lime and sand mortar mix is often the original used, however the alkaline content of lime is much more readily leached from this mortar which may be detrimental to the glass.

At present a lime based mortar mix is preferable.
 
CONSERVATION DOCUMENT.
 
After the completion of the restoration of the two panels, all the notes and photographs made during the process were collated and a plan taken from the rubbings of the completed pieces.  Symbols, in the form of letters were annotated in the space of that piece, describing the techniques applied. These symbols are generally recognized and described in Newton - Conservation of Glass.  The codes applied are the ones recommended by the Council for the Care of Churches.

 
   The restoration report.

These documents were accompanied by before and after photographs, a background of materials and substances used and the reversal technique or agent. The original material that was unused in the restoration was wrapped in acid free tissue paper in glass and design varieties, then archived.  
FINAL NOTE:
The window has been monitored over the past four years and even though at this point it is only the beginning of its restored life, the window is extremely stable.
 

 
Chapter Two: CONSERVATION / RESTORATION ISSUES ARISING DURING THE FLINDERS
STREET STATION STAINED GLASS ARCHES PROJECT.

 
  Accessing the windows during restoration.
 

The four arched windows of Flinders St. Station facade were restored by Almond Glassworks, from July through to October 1995. The following describes some of the issues that arose during this large project.
 
The works consisted of four very large windows glazed with old Chance Brothers pale green Flemish. The pieces were generally around 800mm x 400mm, with stencilled, Art Nouveau design, fired glass paint and stain treatments. It was believed to have been installed with the original building, which is dated 1908. The glass itself confirms this assumption. The general scope of the work was to clean the windows in situ and replace any missing original pieces with glass that matched the existing design as close as possible.

 The building was made at a time in engineering history when the use of steel window frames in construction was popular. Railway stations provided an ideal opportunity for the provision of such techniques. Stained and painted glass being produced in 1908 was generally limited to churches, with lead light structures more common for other public buildings and residences. These factors combined indicate a unique decorative glass installation, and although perhaps not the most important windows in Melbourne from an aesthetic view point, they are an integral part of a very important building.  The unfortunate reality of this uniqueness is that few, if any, comparable restorations can be accessed from which to research potential difficulties. Even relatively common stained glass restoration projects, such as the lead light structures of church windows, still require project specific analysis although the fabrication technique maybe essentially the same as many other church windows. These windows demanded special considerations throughout the entire project.

Each of the four windows required different quantities of restoration work and three of the four required access equipment and had public safety issues. ‘Window 4’ which had internal floor level and verandah access, was the first window worked upon. This window had less than half the original panels remaining with some of these panels required edge gluing. All existing original glass required cleaning. Even though this window was ultimately the last completed, it still allowed the discovery and the refinement of technique without the use of access machinery.

CLEANING
As mentioned in the previous chapter, cleaning is a fundamental and often controversial treatment in a glass conservation project. There is constant deliberating between professionals, some believing that windows are less ‘antique looking’ when cleaned and those who believe that you can not see the artists’ intentions for the grime. The situation is complicated by the fact that so much painted character has been lost because of unsound cleaning techniques, however R.G. Newton states, in his book, The Deterioration and Conservation of Glass: A Critical Bibliography, “it is the regular washing of modern windows which prevents their deterioration through weathering.” he further adds, “Regular washing took place on some buildings in the Middle Ages and it seems a great pity that this practice did not continue!” The second statement refers to this process being discarded in later years and the glass subsequently deteriorated. Weight was added to Mr. Newton’s statements by the glass in question in this project.  
Window four was in such a state of disrepair that window cleaners for many years had considered this entire window (and broken panels in any of the other three windows), unworthy of cleaning. Because these windows were painted on the smooth side the texture was glazed to the exterior, the uneven surfaces provide a plane for dust to settle upon. It was presumed that the present stain was caused by iron deposits and further assumed that these were from the train and tram traffic. It was anticipated that this was going to be a difficult substance to remove.
Fortunately the most recent panel to be smashed in this window had not been totally cleaned away and the veranda roof was littered with its fragments. These were gathered and subsequently used to ascertain the deposits on the glass surface as well as suitable cleaning solutions, with no threat to the original material. All these fragments combined were less than 10% of the unbroken sheet.
Some general testing was undertaken, starting with the gentlest techniques such as water and a clean chamois. This progressed to the use of detergent and some other general solvents such as acetone. Even rust converter was suggested but not considered as it simply alters the state of the oxide but does not remove it. A wooden scraper and nylon brush were also tested with no favourable result. It was noted that the upper surfaces, where the most significant amount of residue was found, were suffering very bad degradation and pitting, unusual for such a comparatively young glass. The under surfaces of the texture of these pieces showed significantly less. The panels that had been constantly cleaned throughout their life had no such scaring. The dust deposits had chemically stained the glass and provided a porous condition for which water to remain, this creates an alkaline environment, which in turn encourages the break down of the silica network.
Almond Glassworks sought to have fragments tested; report he stated “ The SEM trace indicated a very weather and deposit laden outer surface.” The results proved to be iron as expected. The chemists successfully removed these deposits by boiling the sample in hydrochloric acid. This process was required as the glass had to be cleaned for further testing. It could not be considered for historic glass as there is a strong likelihood of this chemical adjusting the surface and in practical terms, the implementation of in situ cleaning by this method is impossible. However it was important to find a suitable method to remove this substance as it was further damaging the glass by its presence.
Chemists, again were consulted to ascertain whether there is a substance that will dissolve rust but not further damage this glass. A variety of acid solutions were suggested, including an acid that was used to restore the finish on slightly rusted new cars. The principle was similar to the problem at hand, in that the rust needed to be removed without disturbing the other surfaces. All suggestions were noted, with a view to test for suitability. Our attempts to find a suitable answer to our problem highlighted a deficiency in dissemination of knowledge within the glass conservation community world wide. A data base system documenting the findings and experiences of firms and projects would be of fantastic benefit. It is possible that the solutions applicable to this project, were found before hand.
 Two industrial chemical companies were also contacted. It can be difficult to interest people in industry in conservation projects, as the quantities required are so often well below the production quantities of most factories. Small to medium sized organizations can be the most co-operative, with the larger companies requiring perhaps a sponsorship arrangement. In this case it was two medium sized chemical manufactures, both were interested and forthcoming with ideas.
 A group of five acid solutions were prepared for testing. These were to be low in toxicity, for health and safety requirements, and attack the iron, but not the glass. The solutions were applied by a pipette to one glass sample that had a significant and consistent rust deposit, the piece had been cleaned previously of general dirt with water and non-ionic detergent. The acids were allowed to stand for 20 minutes then removed with distilled water. The proximity of the acids were marked with a pen to define treated and non-treated areas, then the sample studied with a micro-scope. A Solution of  7.5% glycolic acid and 2.5% sodium chloride was the most effective. It by no means left a pristine surface, but the light quality was significantly improved.

The next step was to ascertain whether the solution was damaging the surface of the glass. For this, samples were viewed with the use of an electron microscope. Surfaces with and without acid treatment and silver stained surfaces were tested. Many levels of magnification were used with the pitted state of the glass making it difficult to determine one surface from another. Ten random images were taken of the treated area and ten of the untreated area at a magnification of X 22,000. It is a purely visual technique and it appeared that there was no difference between the sets. The photographs were examined by a glass technologist for verification. The results indicated that there was no visible degradation of the glass which had been exposed to the treatment. For the solution to work it had to remain on the glass for about twenty minutes. This in turn posed another problem because of the vertical plane of the in situ panels. The acid was blended with vegetable gum to create a sticky gel which satisfactorily adhered to the glass.
The chemicals use was endorsed stressing that all surfaces should be thoroughly rinsed with water after the removal of the gel. An application to the health and safety officer in charge of the railway station also had to be approved, with a plan detailing its use, product safety data sheets and the proximity's in which it was to be used. Fortunately the most important and public window had been cleaned fairly often in its history and did not suffer from significant iron deposits.
The substance was applied with a brush, then at the completion of the twenty minutes removed with chamois and water rinse. The contractors were cautious not to allow the liquid to contact any metallic surfaces. The actual application of this solution was effective to a degree, removing approximately 70% of the stain without abrasive measures.

   
 Illustration of a cleaned surfaces.

The other instance of the unpainted glass surfaces requiring an uncommon cleaning procedure was the interior of Window one. At some stage these panels had been cleaned by a substance that left smears upon all the glass. The maintenance people employed by the station were contacted and relayed that a firm that had been sub-contracted some years ago had cleaned the windows with a de-greaser.
Again, many substances were tested, including water, water and detergent, acetone and methylene chloride, none of which made any impression upon the stain. The only success was an ammonia solution. As it was unlikely that the smear was damaging the glass, it would only be an aesthetic advantage to remove it. Factors such as the constitution of the glass, and the essence of the design relying substantially upon the contrast between refractive glass and paint were considered. In the publication, Conservation and Restoration of Stained Glass: an owners guide, it is stated that ammonia-based window cleaners should be avoided. W. Lowe a former conservator at the Victoria and Albert Museum used an ammonia solution followed by a thorough water rinse, the solution was not recommended for glass paint.
Ammonia is not the ideal solution for the cleaning of historic glass, however the fact that it had been used in a careful manner by such an institution, for us was enough to justify its use in this case. It is a fine line, but the use of common sense, research and a weighing up of the influencing factors (the predominant question being, ‘will this process damage the material?’), a correct solution can be achieved. It is fundamental that even gentle techniques be researched and/or tested.
The application was with the use of spray bottles and chamois. The surfaces were then cleaned again with a water rinse to remove any ammonia salt deposits.

The areas that were affected only by common dirt, were cleaned in a similar fashion to the glass in the Christchurch window, with the use of detergent. However in this case the resistant nature of some of the residues, required some selected cleaning with 000 grade steel wool. The gentle use of this very fine steel wool left no evidence of damage.

It is worth noting that synthetic scouring pads, i.e.- the green scourer backed with a sponge, did show evidence of scratching when tested. Surface damage to glass can be readily detected because of the very obvious change in the light refraction. A scratch will show clearly and scratching at a minute level will appear similar to etching. This glass had a very hard surface, which is not the case for many other glass types, and as a general rule, cleaning should be restricted to the use of mild non-ionic pH neutral detergents. Note: If all the windows had been cleaned periodically and with the use of a correct method, there would have been little need for extensive cleaning.
 
The painted areas of all the windows were declared to be unstable. However the retention of the paint is essential to the readability of the window. It was therefore decided that these surfaces be only dusted with a light brush to remove any dust and loose dirt. There was significant potential of hastening the further deterioration of the surfaces by cleaning with water or any other products.

 
Detail of typical painted sections.

EDGE GLUING
Hxtal epoxy resin was the edge gluing medium employed in this project, with copper foiling being the other technique considered. The latter involves running black backed copper foil along the edges to be joined, then trimming the copper to reduce the thickness of the line. This lessens the interruption of the join. Solder is then ran along the copper with both pieces in perfect registration.
This process has the advantage of being easily reversed if required, however is unsightly in many applications, leaving a black line that can interrupt the flow of design. This technique also has strength limitations. When used in a network of pieces, such as a Tiffany lamp it gathers some strength as the solder holds the pieces in place. Even when used in a lead matrix, it is unlikely that the join in a repaired piece will be separated if the lead is sound. However when used on pieces as large as the panels in question and if the join is straight, essentially the pieces are held together by the weak backing adhesive of the copper foil, designed to hold the foil in position while the solder is applied.

Hxtal epoxy was used only once in the three windows where the general public walk below. A panel in Window 2 had only one crack in a top corner of the panel. Because of the cracks proximity, the fragment was supported by the beading and even if the join did fail, the glass would be still held in position. The piece was repaired without the removal of the body of the panel, a process that because of the hardness of the putty, renders the panel liable to further breakage. The glazing beads were removed from the two sides affecting the area. The surrounding putty was carefully scraped away from the corner ensuring no movement of the glass, which could lead to shelling, and the fragment dislodged. The surfaces were then cleaned with acetone and the epoxy applied to one edge. The pieces were aligned and pressure sensitive tape used to secure them until the epoxy was semi-cured. The adhesive of this tape is water soluble gum. The beading was reattached to the frame to prevent any likelihood of the glass falling into the public space below.
After four days cure the tape was removed and any residual resin dissolved with acetone. There was no damage to existing paint by the use of this tape: the finished join was very close to invisible.
As mentioned previously, broken panels from Windows 2 and 3 were removed and remade in accordance with the specification. The originals were used in Window 4 (the only one of these four windows that does not have a public safety issue) after epoxy treatment. We thought this to be a very successful piece of conservation as it enabled original material, that would have otherwise been archived, to be used and seen by the public.
These panels were worked upon in the studio and because of their size and weight demanded the instigation of another setting technique.
Paint firings during the original manufacturing process, had significantly warped the panels, creating difficulties in holding these pieces in position long enough for the epoxy to set. Methods such as the use of pressure tape were initially used. These involve the alignment of the pieces by the application of gum base adhesive tape to the front and back of the glass. The tape is placed at intervals that are opposed from front to back i.e. if one surface had two pieces of tape on one side, a piece of tape on the other side would be visible through the glass in the middle of the two upper pieces. This process had been very consistent on work previously undertaken; the fragments being attached were light enough for the tape to defeat gravity. Two moderately damaged panels had been successfully repaired using this method, but unfortunately a very warped and damaged piece, demanded a different technique.
The fragments of this damaged piece had been set up and the resin applied on a Thursday and the following afternoon (Friday), everything was still in position. The panel was not checked over the weekend and unfortunately by the Monday two edges had moved upwards about two millimetres, quite significantly out of alignment. The resin cured beyond the stage of physically being able to pull the panels apart. The reversibility issue of this material was to be explored first hand.

 The restored panel, glazed.
 

The suppliers of the epoxy were contacted and proceeded to recommend the use of Methylene dichloride, which causes the resin to expand and in turn breaks the join. An enclosed atmosphere needs to be created to enable the epoxy to be subjected to significant fumes. This panel was one of the larger sheets approximately 890mm x 460mm x 4-6mm. Plastic is attacked by the substance, so a stainless steel tray was fabricated to hold the panel then a sheet of glass sealed by plasticene used as a lid. Cotton wool soaked by the liquid was placed on the joins: it was sealed and left over night. All joins had separated by the next day, the resin easily scraped from the exposed edges.
The glass and painted surfaces were completely unaffected by this procedure. It is however worth noting that if the shape of the pieces to be dismantled create an arch, there is some potential that the force of the expanding epoxy will create a new break in the glass.

Explanation of crack caused by expanding epoxy during reversal.

Another technique considered involves standing the largest fragment in a sand box vertically, then again with the use of tape, placing the pieces in registration upon the mother piece. This is a very useful method, however the bulk, the comprehensive damage and deformity of the panels made it impossible in this case. The subsequent new technique involved the use of fifty or so small plasticene pyramids (for a piece of glass this size) placed at even intervals on a board or bench. The pieces were put into alignment by laying the largest piece down then gently moving the other pieces into registration in accordance with the pattern of the break. The plasticene absorbs the various undulations and irregularities of the glass. Care must be taken if pieces require pressing down to create the correct level. Glass edges moving against each other often create shelling. The epoxy was applied without the use of tape with very good results. A subtle problem of this technique, which is usually only visible on polished, or float glass when the glass is in a reflective state, is a slight change in the glass plain caused by inaccurate setting. This can be overcome in the studio by using light, during the positioning stage, in such a manor as to create a reflection. The subject being reflected will be adjusted if the pieces are not correctly aligned.

A more general problem with the use of epoxy is a star fracture caused by projectile impact. The glass may have many fractures, with perhaps six continuing across the sheet to effectively break the glass. The remaining faults can be partially filled with epoxy in some cases, however the faults remain active and often complete their course through the glass at a later date. This is negative in that it requires further action, however the glass panel generally does not fall out of registration which enables in situ application to the now receptive crack. This process should be done before residues work into the crack from exposure to the elements.

THE DESIGN WORK.
The missing panels were replicated by copying the designs from the coinciding panel within either itself or another window. The interior of Window 3 faces a floor level passage and retained all panels except two. This window was the source for all the designs used for Window 4. The broken panels were replicated directly from the original fragments. All designs were transferred onto tracing paper and then replicated in the studio in the traditional technique.
Because of the repetitive nature of the windows design, the original craftsmen had used a stencilling method to achieve with ease the required gentle curves. Upon analysis, the outline of the trace appeared to be first laid, then later filled. This may have been fired in, then the matting applied, or potentially all processes were achieved in a single firing. It is difficult to guess whether the residues of the silver stain used in the original panels would fire into the glass at 650 degrees Celsius (a recommended paint firing temperature for modern paints) as some stain products do presently. If it was the case, it may have been incentive enough to lower the firing temperature to around 600 degrees Celsius, to achieve this single firing. The side effect being slightly porous paint surface, a potential cause for the windows intrinsic fragile paint.

 
Tracing the design from existing panel.

The paint colour was a crucial factor in the success of the mimic panels, particularly Windows one, two & three. These panels are often active from transmitted light and visible from all sides at any time of the day. At night the reflective colour of the paint was a consideration from the interior and at mid day the windows are often lit by direct sunlight and obviously often in shadow. In all these moods, the new paint had to match the depth and consistency of the originals, the stippled paint matte being the most important factor.
Glass paint is very different in appearance before firing and requires the mixing and firing of many samples using fragments of the glass type to be used. These must be viewed next to the original in all the various light conditions until a suitable mixture is obtained. The testing has to be documented to reproduce accurately the quantities of the suitable sample.

The original silver stain was applied to the external face of the glass, as is generally the case and as the windows are often viewed from outside during the day, the reflective quality of the stain had also to be matched.
 

 
           Reflective paint and stain colours, window four.

There are many varieties of silver stain, all reacting differently to different glass, even the side in which the stain is applied to can make a dramatic difference, due to that glass's manufacturing technique. Essentially silver stain is silver nitrate mixed with a clay medium to allow application to the glass surface. It gives no indication of its inevitable colour when being applied, the powders being a variety of shades of black or brown. Generally, the more medium is applied the deeper the shade of amber. It is a common practice to deepen the colour by applying more stain and refiring a glass piece. It is not so forgiving to over-application as hydrofluoric acid is required to remove excess colour: leaving an unwanted etched surface.
The silver stain used for the reproductions was a Blythe product that had all the characteristics required, including the rare yellow colour in reflective light. As this stain is no longer available it was fortunate that sufficient old stock could be accessed for the project. The stencilled trace painting was done using a masking technique not unlike acid or sand blast resist. The glass to be painted was covered by adhesive clear vinyl sheet, then laid upon the prepared design. The heavy trace lines were cut and the film removed, the exposed glass was then painted with the matching paint. The stain was also applied at this stage on the opposite side. A badger brush was used to achieve the gradations of colour present in the originals.
The panels were fired at 600 degrees Celsius. This temperature enables the securing of the paint without burning the oxides of the stain irreversibly into the under surface. The matting was then applied on a light box beside an original for intensity and stipple match. This was then fired at 650 degrees C giving a non-porous lustre to the paint surface. Silver stain, once the oxides are removed, is stable and generally does not adjust with subsequent higher temperature firings.

In the restoration of glass and probably in all fabrics found in old buildings, one should assume nothing. The initial paint colour matching was done using a sample from Window 2. Believing this original paint type to have been used for all the windows, trace lines were done on one small panel for Window 1. Before any further treatments, this panel was held in position to check for size, registration and surprisingly, paint colour. The panel was still used as the subsequent matte firing correctly adjusted the reflective colour.

 Glazing a finished panel.
 

The apex panel in Window 1 was one of the eleven inappropriate replacement panels found in that window, with no painted design. As Window 1 was of a different structure and slightly different design to the other three, it was impossible to copy all the missing design off another original. The Public Transport Corporation keep a photographic archive detailing the changes that the building has undergone throughout its history which includes an excellent photograph taken in the 1940s or 50s of the window in question. The missing panel was clearly visible enabling scaling for the recreation of this panel. The actual detail was a mixture of a variety of stencils used else where in the windows, these again being copied onto a piece of tracing paper. The connecting lines to the lower panels were finalized by placing the drawn design into it's glazing position, then with the use of a marking pen drawing in the line positions. It was necessary to stand in the viewing space below to confirm the flow of those lines in relation to surrounding panels.

 
      Window 1 completed.

THE WORKING CONDITIONS
The logistics of this project were fundamental in terms of safety and the economics, influenced by labour and plant use. It was a condition of the works that the public are not in any way endangered or unduly inconvenienced by the proceedings. Window 1 was the most complicated in regard to these requirements. The work area was only permitted to consume half the space of the steps area at any one time. It had to be barricaded and at the completion of a days work all was to be packed away, and the area left clean. Access was by 45 foot snorkel lift. The machine was stored on a less used railway platform at night, then towed and pushed into position at the beginning of each day. It was rolled down the footpath and back into storage at the completion of the day. The plan was to set up the work at seven a.m. before the peak hour. Then dismantle at three thirty before the next peak period. The works were programmed to prevent the unnecessary need for the pulling down of the machinery, with as many processes as possible done to the area of window being accessed during that day. Forethought was given to the requirements of the other side, i.e. while the exterior was cleaned the glass was scored for safe removal by breaking the inappropriate panels, and temporary ‘glazing’ was installed in the form of clear plastic taped to the frames exterior. Even a process as mundane as this had to be undertaken with safety considerations. The use of board (a more usual temporary glazing technique) could fall to the crowd below during the night.
The function of the plastic was two fold, preventing weather from entering the interior as well as, during the removal of glass from the inside, preventing any glass fragment from falling away from the access equipment. Public safety was a constant issue throughout the project, that in other situations would not necessarily have needed to be considered. It was appropriate to protect the public from the potential outcome of a piece of glass being dropped into the space below, and there was significant potential of airborne glass splinters, if such an event took place, travelling well beyond the barricades and into public space. This was done by adhering contact film to the glass. Generally it was the new panels and the inappropriate glass being removed that required this action. The process of the relocation of the broken old panels was the only instance of the film being applied to original glass. This was adhered to the exterior surface to prevent endangering the fragile paint; the silver stain was stable and unaffected by its a use.

THE REPLICATING OF CHANCE BROTHERS PALE GREEN FLEMISH GLASS, MIDDLE SIZED PATTERN.
The panels that over the years had been inappropriately replaced posed a familiar and the most difficult problem of this project, finding a convincing match of the existing pale green Flemish. It is approximately sixty years since this glass was produced, which makes the likelihood of accessing the required twenty square meters, unlikely, and there are no truly satisfactory modern glass alternatives. This was always going to be the most difficult and inevitably disappointing, segment of the works.

 
  Typical bay in window four prior to intervention.

In the station complex at the Elizabeth St. entrance, were two areas glazed by the required glass type. One bay had ten panels of varying sizes with three panels either broken or cracked. The other bay had seven panels with two broken pieces. These panels had no painted design, and two had to be replaced for public safety reasons. These two panels further adding to the Flemish dilemma. The decision was made to allow the use of all the glass in these bays to prevent compromise in the restoration of as much of Windows one, two and three as possible. It was considered that the decoration and position of the major windows out weighed the glasses present secondary glazing position.
 
  From well before the signing of the contract, research was being undertaken to attempt to effectively manufacture a comparable glass. Window four required (at an initial count) twenty six panels, this was reduced to twenty two by the movement of repaired original panels, but still well short of the requirements. Initially the attempt was to be made by a glass technologist that had been contemplating the concept of small batch rolled glass for some time, this project provided the incentive. The chemical components and their quantities of the sample were analysed by ACI. A batch was mixed and a non patterned piece of glass was rolled. It was an interesting piece of glass, not without potential for other applications, however the colour was quite unlike the original. The difficulties of producing a new glass by this method are worth pursuing but the time frame here would not allow for the requisite research and development. The next attempt was by kiln forming. Good results had previously been achieved with this technique, with the mimicking of glass types such as old cathedrals, antiques and drawn glass.

The first step was to remove in one piece one of the unpainted sheets glazed at the Elizabeth St. entrance. This sheet became the model for castlable refractory material. Kiln forming is steadily becoming more of an exacting science, however even with the involvement of leading practitioners in this field, the road to the development of a mould capable of repetitive firing and to this degree of accuracy was difficult. Pilkington pale green cathedral was selected as the only the glass which produced a similar colour. This imposed restrictions as the glass type would not withstand temperatures far beyond 800 degrees Celsius without devitrification on both surfaces.
Eventually the problems with the moulds were solved and undulating surface of Flemish achieved, unfortunately though, the similarity was restricted to reflective light conditions only. Transmitted light revealed almost non of the refractive qualities so important to the effect of the glass.
Another option for the imitation of Flemish, was to coat a modern Pilkington clear Waverly glass with an almost transparent enamel mixed to the appropriate colour. A further option was another modern glass called pale amber Flemish, which although not the same colour is of a very similar tone and related texture. A further solution was one sheet of glass for colour (the cathedral version of the amber Flemish), vented double glazed behind a piece that mimicked the texture. The textured piece was modern Pilkington Waverly, and was the glass painted upon, the texture being glazed to the outside.
Samples of these alternatives were made and presented to the architect and the P.T.C. project manager. The appearance was studied from the interior, and the exterior from street level.

The comparison between the direct and indirect views.

The double glazing method was decided upon. This was successful from inside and from outside during the night. However the new glass surfaces when viewed strait on from outside during the day, are more reflective than the original glass and quite conspicuous.

   
     Before and after photographs of a typical bay in Window four.
     (symbols- H = epoxy repair from another window;  N = new panel using the
     double glazing technique).

AUTHORS NOTE
I believe that, perhaps with the exception of window four, this was a successful project. All concerned, from architect and project manager, to the people who undertook the work, did so with the importance of the building and integrity of the glass in mind. The results are there for all to see.
 

CONCLUSION

Philosophy varies so much, in a conservation / restoration project. Not only in technique, but also in the actual goal sort after at the beginning of the works. The re-emergence of a window from a state of disrepair, has the ulterior effect of losing some of it's evidence of age. The decision is a difficult one with considerations such as inappropriate repairs standing as both history and, if preserved, adding value to what was most likely a case of bad workmanship. This is further complicated by the question of the actual value, in artistic terms, of an installation. Are the blemishes left by previous restorations affecting the original intentions of the artist, and, are those intentions significant?

I  looked at a small project at a Victorian mansion that involved a logistically difficult cleaning situation. I was asked to propose my concept of how the problem could be solved, and at the same time a conservation architect presented his proposals. The solution was a mixture of both methods, both individual approaches improved because of the exchange of ideas and knowledge.
Discussions of this nature address the many issues present in restoration and conservation, alleviating the need for decision making by an individual and adding the involvement of the many trains of thought that are present in the heritage community.

Research is of fundamental importance to restoration at all levels, as well as the need to be up to date with the state of  practices in restoration in both Australia and abroad. The involvement of glass technologists is important in any situation that involves chemistry. Their understanding of the composition of the fabric and the likely influences of other chemical factors upon the material, is of great value. It is inevitable and desirable that science becomes an integral part of glass conservation.

In the general research undertaken for this study I found that although there is plenty of material published on the subject of stained glass, there is a lack of detailed information about specific conservation techniques, in particular the use of Hxtal epoxy adhesive. This study has addressed this to some degree and anyone intending to use the material will benefit from reading the methods described in this document.
Finally I would like to invite any feedback upon the techniques and methods described, including negative comments. Conservation techniques are continually evolving, and better methods and materials are a likely outcome from projects, as the two described here. If better methods, or other relevant techniques are known to improve the preservation of stained glass, they are better shared.

Main Page| Thesis| C.V.| E-mail| Links|