JAIC 1998, Volume 37, Number 1, Article 4 (pp. 35 to 47)
JAIC online
Journal of the American Institute for Conservation
JAIC 1998, Volume 37, Number 1, Article 4 (pp. 35 to 47)



ABSTRACT—This article describes materials and methods for gap-filling transparent and translucent artifacts. By far the greater number of objects in this category are those made of glass. Consequently, the methods and materials described are those that have been developed and used to gap-fill glass vessels. These, of course, can be adapted for use on other transparent and translucent materials, such as some enamels, shells, and semiprecious stones.

TITRE—Rebouchage reversible de f�lures sur des mat�riaux transparents ou translucides. R�SUM�—Cet article pr�sente des m�thodes et des mat�riaux destin�s � reboucher les f�lures sur des objets transparents ou translucides. Ces derniers �tant, pour une grande majorit� d'entre eux, en verre, les m�thodes et mat�riaux d�crits ont �t� mis au point et utilis�s sur des r�cipients en verre. Ils peuvent toutefois s'employer sur d'autres mat�riaux transparents ou translucides: certains �maux, nacre ou pierres semi-pr�cieuses, entre autres.

TITULO—Rellenos reversibles para materiales transparentes y transl�cidos. RESUMEN—Este articulo describe materiales y m�todos para rellenar faltantes en artefactos transparentes y transl�cidos. Por mucho, el mayor n�mero de objetos en esta categor�a est�n fabricados de vidrio. Consecuentemente, los m�todos y materiales descritos son aquellos que han sido desarrollados para rellenar faltantes en vasijas de vidrio. �stos, por supuesto, pueden ser adaptados para usarse en otros materiales transparentes y transl�cidos, tales como: algunos esmaltes, caracoles, y piedras seimpreciosas.


As with all aspects of restoration, there are a number of important factors to be addressed when considering gap-filling of incomplete glass objects. First, the condition of the glass itself has to be assessed, since restoration is essentially a molding and casting operation that will entail the application of wax or silicone rubber directly onto the glass. It therefore stands to reason that the glass and any applied decoration must be in a robust condition; loose flakes and low-fired or unfired painting and decoration cannot be subjected to the molding and casting processes. Second, an assessment needs to be made as to whether restoration is possible, given the shape, amount, and thickness of glass remaining. Third, it is important to decide whether restoration is desirable on ethical grounds, particularly if no full profile exists. Restoration may be undertaken to improve the stability of an object such as a glass vessel or simply to improve its appearance for display purposes. If a great deal of the vessel is missing, it may not be worth restoring in terms of the time taken to carry out the work, unless the glass is historically important or is of particular value to a collection (bearing in mind, however, that glass that is of less importance to one collection may be valuable to another). Last, consideration has to be given to the availability, properties, cost, and appearance of synthetic materials with which to carry out the restoration (Davison 1981, 1984; Jackson 1984; Fisher 1988, 1992; Eckmann 1995). (See the discussion in section 3, Materials, below.)

Before the adoption of synthetic materials by the conservation profession some 30 years ago, gap-fills in glass were carried out using natural materials such as beeswax and with plaster of paris, which was often painted silver in an attempt to improve its appearance. It was also not unknown for pieces of glass from similar vessels to be cut and married to produce an apparently complete object. This type of restoration was especially suited to archaeological glass, where joins between the fragments could be disguised with flakes of iridescence and with mud.

Currently, gap-fills in glass objects are effected by making molds formed from dental wax sheets or in silicone rubber, whether on existing areas of the object or from a modeled clay or Plasticine form representing the missing area. The molds are secured over the areas to be replaced, and a clear resin cast into them. Once the resin has cured, the molds are removed and the casts trimmed with a sharp scalpel and, if necessary, smoothed with fine abrasives and polished or given a coat of clear lacquer.

If a simple cast turns out to be distorted or there are significant differences in the levels of the glass and resin (i.e., steps), it is normally less time-consuming to remove the cast and start again. On more complicated casts excess resin can be removed by the careful use of a dental drill, and the resin cast can, if necessary, be filed and abraded to some extent, though care must be taken not to crack the cast by applying too great a pressure. After using progressively finer grades of abrasive paper, the surface may be polished using, for example, Solvol Autosol (kieselguhr or “infusorial earth,” a methylated soap, white spirit, and ammonia) on a tiny buffing wheel held in the drill head. However, care must be taken in the use of abrasive pastes, since they may become irremovably trapped in small deficiencies in the cast, such as air bubbles, or at the junction between the cast and original glass. During any operation involving the use of a dental drill, it is wise to step down the current with a rheostat to prevent friction from melting the surface of the resin and the restored object from being accidentally spun out of the conservator's hands. Final finishing can be undertaken using successively finer grades of abrasive papers such as Micro-Mesh, and polish applied with cotton wool swabs. It is rarely possible to obtain the original clarity of the cast after working on it. If all else fails, the cast can be given a coat of clear lacquer—although, of course, there are then two materials that may discolor (resin and lacquer). Some restorers do, in fact, always cast colorless resin in colored glasses and apply color in a sprayed lacquer.

It is part of a conservator's task to adapt restoration techniques as required since each restoration is an individual undertaking; indeed, the processes described in this article are the result of such adaptation. However, restoration processes may be conveniently classified as follows:

  • gap-filling with casts from molds taken from the glass itself
  • replacement with casts from a mold taken from a modeled section or a previous restoration
  • gap-filling where the interior of the vessel is inaccessible for working
  • gap-filling with preformed casts
  • partial gap-filling.
Outlines of these processes are given below. Full details are published in Newton and Davison (1989), except for the section on gap-filling with detachable casts, published by Hogan (1993).



Where tiny fragments of glass are missing from a vessel—for example, part of a rim or edge of a foot—their shape may be copied by taking a mold from a similar area of glass and then repositioning the mold behind the area to be filled. The molding material may be dental wax in the form of sheets or silicone rubber, which takes fine detail. Resin may then be applied to the mold in thin layers, each one being allowed to set in turn. When cured, the resin can be abraded and polished as necessary.


In the case of glass vessels having extensive, or more complex, missing areas, such as a raised design to be copied, a form is constructed of modeling clay or Plasticine onto which the existing fragments are placed. The design may be modeled, and the whole is then molded in silicone rubber and the modeling material removed so that resin may be cast into the areas of missing glass. If a previous restoration needs replacement for aesthetic reasons, but is dimensionally sound, a silicone rubber mold may be taken from it before removal (as in the case of the Auldjo jug—see section 2.3 below). This procedure will dispense with the need to model the missing area.


If the vessel to be restored has a narrow neck, so that the interior cannot be reached, it will be necessary to adopt special methods of restoration. If the vessel is damaged only in the body— that is, the neck has not become detached—the fragments can be taped in position and epoxy resin introduced to all the cracks except those around the perimeter of the damaged area. Once the resin has set, the damaged area can be removed as one piece, missing areas within it backed with tape, wax, or silicone rubber, depending upon their size, and filled with resin. When the resin has hardened and any necessary cleaning and polishing carried out, the repaired section may be taped in position on the vessel and the perimeter cracks sealed with adhesive. If the fragments were simply bonded and the gaps filled out of the body of the glass, any slight misalignment would prevent their eventual reinsertion.

If, however, the vessel is very damaged and has a large area of glass missing from the body, it may still be possible to reconstruct it with epoxy resin as described above, leaving a large area unbonded around the edges so that it can be removed. This area allows access to the interior of the vessel for ease of working. When gap-filling of the missing glass is complete, the glass can be replaced and the perimeter cracks sealed with adhesive.

A more complicated version of the silicone rubber molding technique had to be devised in order to restore the Auldjo jug, a Roman cameo glass on display in the British Museum. The jug has a narrow neck, and therefore access to the interior for molding purposes was severely restricted. The jug had been broken at some time in the past, and a considerable amount of the body was missing. It had previously been repaired, probably with animal glue, and gap-filled with plaster of paris colored dark blue to match the remaining glass. The plaster restoration had become damaged and unsightly, so it was decided to remove the plaster in order to effect a more accurate, lightweight, and aesthetically pleasing restoration. The previous plaster restoration was dimensionally correct, and it was possible to take a silicone rubber mold from it after some repair had been done. Two silicone rubber molds were produced, one conforming to the inner and one to the outer profiles of the jug. The outer mold was supported by constructing a two-part resin and fiberglass mold, which could be bolted together during the casting process. The inner mold was supported by filling the interior with vermiculite particles. The details are published in Newton and Davison (1989).


An alternate method of replacing missing areas of glass that has been suggested (Gedye 1968) is to cut the shapes from preformed acrylic sheets such as Perspex (U.S. Plexiglas), bend them to the required curve after softening with a hot air blower, and attach them to the glass with an adhesive. However, this lengthy process requires accuracy in cutting and filing, and the finished result is in no way as aesthetically pleasing as casting in the missing fragments with a clear resin; it is therefore not recommended for general use.

A method of gap-filling glass with preformed cast of epoxy resin was devised at the British Museum (Hogan 1993). A glass bottle required considerable conservation work to make it stable enough for exhibition. The green glass vessel, which is approximately 25 cm high, had an unusually thin base (1 mm approximately) in comparison to a thicker, and hence heavier, neck and rim (3 mm approximately). About a third of the original vessel is missing.

The vessel was reconstructed in two parts. The main body of the vessel and the neck and shoulders were reconstructed using a UV-curing acrylic (Loctite 350 engineering adhesive). The joins between the top and bottom of the vessel were not substantial, and some filling was necessary to give support. A method of support with minimum use of strategically placed resin infills was devised. Detachable resin fills were constructed away from the object and stuck into position once cured. This procedure had the advantage over casting the resin in situ of not creating undue stress on the small join areas during molding and casting of the resin. It also minimized contact with the surface of the glass during the finishing processes such as sanding and polishing.

Sheets of epoxy resin AY103/HY956 were cast to correspond to the varying thickness of the glass vessel in simple hexagonal molds made from sheets of dental wax with added support from wax sides (any suitable molding material could be used). When the cured resin was removed from the mold, a thin film of wax remained on the surface of the resin. This film was removed with a spatula, and the resin surface was then cleaned with white spirit.

A sheet of dental wax was placed across the gaps in the glass to be filled, and the outline of the gap was scribed onto the wax using a pin vise. The wax shapes were then cut out and placed in situ in the vessel to ensure a good fit, and they were secured with Sellotape. Once this process was completed, each wax shape was removed in turn, using acetone to release the Sellotape, and placed onto a resin sheet of appropriate thickness. Its outline was scribed onto the resin with a pin vise. The resin was then gently heated with a hot air blower to make it pliable enough for the shape to be cut out with scissors. The edges of the resin shards were filed where necessary with needle files to ensure close contact with the glass. Where appropriate the resin sherds were reheated and the curvature was modified to comply with the contour of the glass vessel. The finished pieces were placed in position and secured with HMG (cellulose nitrate) adhesive. The resin sherds were then painted with one coat of Rustin's clear plastic coating (urea formaldehyde) colored with Maimeri Restoration Colors to match the color of the glass.

As this method of gap-filling proved to be very successful, variations upon this technique using alternative resins were used to restore other glass vessels. A piece of Anglo-Saxon glass was gap-filled using the same methods but substituting clear HXTAL NYL-1 epoxy resin for the AY103 epoxy resin. The resin was heated more gently by immersing the pieces in warm water. The pieces were then bonded into position with HMG.

Another small delicate flask needed support for its neck in order to connect it to the body and base. An internal mold of the neck was made using a coil of dental wax. The wax was removed and sealed at one end, and superfine casting plaster was poured into it to produce a solid core of the internal shape of the neck. A sheet of Ciba-Geigy Araldite 2020 (epoxy resin) was made into a wax mold as previously described. Before the resin had completely cured, it was removed from the wax mold. While still flexible, it was formed around the plaster core using cling film as a release agent and left to cure fully. Once cured, the plaster core was removed by splitting the resin with a scalpel blade. The resin backing, now in two sections, was inserted into the neck of the flask, and adjustments were made to secure a good fit. The backing was then bonded with HMG, giving full support to the neck. To prevent discoloration of the epoxy, the HXTAL NYL-1 and the Araldite 2020 must be fully cured before being bonded in position with HMG.

The method proved to be extremely effective, not just in the final appearance of the glass vessels but also in the support it provided. Being able to work on the resin fills away from the glass surface, avoiding the need for individual molding and casting of sherds, makes this a quick and safe method of filling fragile glass. The fact that the resin pieces are bonded into position with cellulose nitrate makes future removal easy and safe for archaeological glass. A disadvantage may be in the heating of the resin, which could accelerate yellowing. Yellowing was not a problem in the case of AY103 epoxy as the glass was green and a perfect color match was not essential. In the cases of HXTAL NYL-1 and Araldite 2020, although no yellowing occurred in the instances described, excess heating of the resin should be avoided. A previous case of heating HXTAL NYL-1 epoxy with a hot air blower caused yellowing of the resin. In view of this effect, the use of warm water to soften the resin is preferred. This method could, in theory, be used as a complete gap-fill on a vessel, depending on the intricacy of the missing shapes.


Where a glass vessel has large areas missing but does not warrant total reconstruction because, for example, it will remain in storage, it may be partially restored for safe handling, e.g., during study, photography, or drawing for publication. Strips of fine glass fiber tissue cut to size and impregnated with cellulose nitrate adhesive or epoxy or polyester resins are used to bridge gaps in the glass and to hold floating fragments in their correct positions. Total reconstruction of small vessels is also possible by this method. The result is not aesthetically pleasing but may be useful as a temporary measure.


In general, the materials used for making molds and casts for filling gaps in transparent and translucent objects are the same in Europe and North America, i.e., Plasticine, modeling clay, dental wax, silicone rubber, epoxy, and polyester resins. However, the exact chemical composition of the materials will vary and it cannot be assumed, for instance, that epoxy resins or silicone rubbers available in one country will necessarily produce the same results as those available in another. Also, materials used in temperate climates may not be available for use in tropical climates. Thus it is appropriate to quote materials in terms of chemical rather than trade names in order that appropriate materials can be found in or near the country in which they are used. Discussion between the conservator and the manufacturer or supplier should be possible. One consideration is the age of the material, which may require that there be no prolonged storage or shipping times.


Missing areas of glass may be modeled up in situ prior to molding using one of several commercially produced modeling materials. Those most commonly used are potters' clay and Plasticine. Plasticine is a putty composed of petroleum jelly, fatty acids, and whiting. These oily substances enable the Plasticine to be worked to a smooth surface. However, it does not adhere very well to glass and will contaminate the surface. Plasticine residues should be removed with cotton wool swabs moistened with a degreasing agent such as acetone before gap-filling with resin begins.

Newplast is a modeling material similar to Plasticine in texture but formulated especially for use against polyester resins. It is available in a buff or dark blue form. The dark blue variety has a tendency to discolor the resin cast and is therefore not recommended. Damp clay adheres very well to glass and is easily worked to a smooth surface with a spatula dipped in water. In addition, any fragments of glass that can be positioned accurately but do not actually join to the body of a vessel—i.e., floating fragments—may be held in position by placing them in situ on a clay former. The disadvantage of using clay is that moisture contained within it can separate cellulose nitrate adhesive (if used in the repair) from the glass after several hours. Backing joins with tape, rubber latex, or thin sheets of wax does not prevent this separation from occurring. In fact, adhesive on Sellotape and masking tape breaks down with moisture to form a messy substance, removal of the wax sometimes causes the joins to fail, and latex flows into tiny cracks and chips and is difficult to remove without dismantling the glass.


Materials used for making molds from glass artifacts should have the following properties. They should not harm the object physically by adhering too strongly to the surface, by pulling off glass projections, or by generating heat. They should not harm the glass chemically by contaminating or reacting with it. Molding materials should reproduce all the fine details or the original without distortion. The viscosity and thixotropic properties should be sufficiently variable by the manufacturer or conservator to allow the materials to be adapted to meet different requirements. They should preferably be available at reasonable cost and have an adequate shelf life. Molds must be able to withstand the heat of polymerization of the proposed casting material and must not react with that material.

For molding glass, the most suitable molding material is generally silicone rubber, which is available in several grades of thixotropicity. However, the cheaper materials—plaster of paris and dental wax—are frequently used with good results. The majority of silicone rubber products used for molding cure by catalytic elimination of alcohol to form cross-links between chains. When the alcohol evaporates, the rubber shrinks, but the amount of shrinkage is small (less than 1%) and occurs over a period of a few days. Shrinkage of 2.2% has been observed to occur over a number of years; however, this shrinkage is rarely a problem in the case of small objects restoration, where molds are unlikely to be reused after a period of time has elapsed. Silicones are insoluble in solvents but can be swelled considerably by the use of aliphatic, aromatic, and chlorinated hydrocarbons (Noll 1968). Wihr (1977) has suggested swelling silicone rubber back to size by exposing it to organic solvents, but this method seems to be unreliable.

In the majority of cases silicone rubber requires no release agent between it and the glass surface. However, instances of silicone rubber adhering to glass and porcelain have been known (Morg�s et al. 1984), and therefore preliminary tests must be undertaken. A release agent such as petroleum jelly or an organic lacquer must be used if silicone rubber is to be cast against a cured section of silicone rubber or the two will adhere. Thin layers of silicone rubber may tear when peeled off an object, but thick layers are hard-wearing and the molds are reusable. If necessary, the rubber may be overcatalyzed to shorten its setting time, for instance when silicone rubber is being used to reattach a silicone rubber mold to glass. Overcatalyzing silicone rubber does result in its being extremely brittle, but brittleness is not a problem in this instance.

If silicone rubber is improperly mixed, it will not cure or will cure in patches. Uncured rubber is best removed with small paper tissues a little at a time to prevent it from being smeared over the object. The final residue may be removed with small cotton wool swabs soaked in acetone. If, however, the uncured rubber has become trapped within cracks or joins in the glass, the object may have to be dismantled to be cleaned thoroughly.

Inert fillers such as kaolin, talc, or aerosol silica may be added to thicken mobile grades of silicone rubber, which can then be applied over the first layer of rubber to produce a thicker, self-supporting mold that does not tear easily while being removed from the object. This procedure is particularly useful in cases where small molds are being made, since it obviates the necessity of buying two grades of silicone rubber (i.e., mobile and viscous grades). In the case of large molds the rubber mold may require the support of a mother mold made of plaster of paris or even resin and fiberglass matting.

Rubber latex shrinks too much at the time of cure and in the following weeks to be of use in molding such a precise material as glass. Even a small shrinkage will mean the details such as trailed threads on the glass will not match up with those on the cast. A mold must remain stable for several days or weeks while restoration is in progress.

However, rubber latex has been used for making molds on glass, and, as it tears easily, it is possible to find brittle remains trapped in cracks and chips. A repaired glass may have to be partially or wholly dismantled to release the traces of rubber latex.

Hot-melt preparations such as gelatine, Formalose, and Vinamold (PVC) should be avoided for direct use on glass since the heat may cause damage. However, Formalose, a gelatine material containing glycerine to keep it flexible (Wihr 1977), is useful for reproducing the interior shape of an object with a narrow neck where plaster or rubber cannot be introduced because of their viscosity and, on setting, their rigidity. The Formalose can be poured hot into a plaster mold and, on cooling, it sets and begins to shrink uniformly. The process must be watched, and when there is a gap between the Formalose and the plaster, representing the thickness of the glass vessel to be reproduced, resin can be cast into the gap. The Formalose core must be supported away from the plaster walls. The method is described by Petermann (1969).

Dental waxes are composed of a number of different waxes, contain an unidentified pink or red dye, and are often supplied as sheets measuring 180 � 82 � 1.5 mm. Dental waxes are available in various grades of hardness, in sheet sizes up to 305 � 203 mm and in thicknesses of 0.3 to 3 mm. For making molds the sheets or parts of them can be softened by gentle heating in warm air or water before shaping them over the glass object. Before casting, the wax mold should be coated with a proprietary solution of poly(vinyl alcohol) (PVAL) to act as a release agent between the wax and the resin to be cast against it. Without this separating layer, dye from the wax tends to be absorbed by the resin. The PVAL has a tendency to gather in pools, but this development can be prevented by adding a few drops of Synperonic N to the solution, which breaks the surface tension on the wax. The PVAL must be brushed continuously until it has almost set, then left to dry before the wax is attached to the glass. If the cast being made is large, it may be advisable to apply a second coat when the first has dried. The second coat must be brushed on with a light touch so as not to disturb the first.

Plaster of paris (calcium sulphate) is prepared by heating gypsum (CaSO4 � 2H2O) to drive off some of the combined water, forming 2CaSO4 � H2O. On adding more water, the calcium sulphate rehydrates, forming interlocking crystals, which set to a rocklike mass with very slight expansion, typically 0.5%. Various grades of plaster are available with different setting times, expansions, and particle fineness. Some dental plasters set to form very hard solids with very slight expansion. The material is available, cheap, and useful for the construction of case molds over silicone rubber. However, it requires release agents between plaster-to-plaster surfaces and plaster-to-resin ones. It is rigid and hence any undercuts on the glass must be molded first, preferably in silicone rubber. If incorrectly made, a plaster mold may prove very difficult to remove without causing damage to the glass inside.


Release agents must prevent adhesion among objects, molds, and casts; the agents chosen will depend upon the materials being used. As previously mentioned, silicone rubber requires the use of a release agent such as petroleum jelly or organic lacquer only when it is being cast against a section of cured silicone rubber. The surface of plaster of paris mold pieces, however, must be sealed, as each is made to prevent it from adhering to the adjacent pieces.

Shellac in a solution of industrial methylated spirit (IMS) (U.S.: grain alcohol) or a solution of PVAL can be used to seal the surface of dry plaster of paris mold pieces, after which the surface is coated with soft soap, Vaseline, detergent, petroleum jelly, or a wax emulsion (see also Koob 1979).

It is worth noting that silicone release agents, which are available as liquids or in spray cans, can be used although they are not recommended. Traces of silicone oils will remain on the cast, and, if they are not removed completely, they can prevent paint or adhesives from bonding properly to the surface.

Polytetrafluoroethane (PTFE) dispersions in aerosol cans have also been used as release agents in the past. However, they are not recommended as the dispersions become embedded in the resin cast. PTFE dispersions are also thought to contribute to the destruction of the earth's ozone layer.


Over the past few decades many epoxy and polyester resins, designed for different purposes, have been manufactured and distributed by hundreds of outlets. The majority of these cannot be used for glass restorations since they are opaque and/or colored, because they contain fillers, or because they cure only at high temperatures. The ideal resin for use on glass objects should be a water-clear, room-temperature curing product. It should be possible to pour the resin into molds; it should set with minimal shrinkage to form a hard solid; it should be crystal clear, not yellow with age, and remain colorless indefinitely. No materials meet this specification completely (Down 1984), although some come close to it. Before restoration work begins, the resin to be used should be tested by mixing and casting it into a mold of the same material to be used in the final reconstruction. This precaution will ensure that the resin's shelf life has not expired and its behavior is as expected, i.e., the formula of the product has not been altered by the manufacturer since it was last ordered. A further consideration is the possibility of future removal: epoxy and polyester resins cannot be dissolved but can be softened with dichloromethane and removed mechanically. Care must be taken that the swelling of the resin, which accompanies the softening, does not cause damage to the glass.

In chronological order the following epoxy resins have proved useful adhesives and fillers for small gap fills in glass: Araldite AY103/HY956; Ablebond 342-1 (marketed from 1979); HXTAL NYL-1 (made and marketed for conservation purposes from 1984); Epo-Tek 301 and 301-2 (U.S.); Fynebond (U.K.) and Ciba-Geigy XW396/XW397. The latter was so named during its experimental stage and is now marketed as Araldite 2020. The properties of this resin are well documented. Currently the author uses Araldite 2020, as it is readily available and has good working properties, whereas Fynebond is expensive and the hardener crystallizes at room temperature (as does Ablebond 342-1) and the use of HXTAL NYL-1 is impractical due to the high cost of importing the resin from the United States. Although developed for use as adhesives, epoxy resins have been used for small gap fills in glass. Large gap fills made in epoxy resin would be expensive compared with polyester resin. Epoxy resin casts do not polish as well as those made of polyester resins.

A number of articles have been written concerning the matching of colorless adhesives for glass to the glass itself (Tennent and Townsend 1984; Messinger and Lansbury 1989). Augerson and Messinger (1993) have experimented with mixtures of HXTAL NYL-1 and Ablebond 342-1 epoxy resins to alter and control refractive index.

There are also many clear polyester resins on the market, and whereas the epoxy resins mentioned above seem to be universally available to conservators, different polyester resins are used in different countries. The difficulty of transporting highly flammable materials may have a bearing on this fact. The author has had experience of Tiranti's clear embedding resin (U.K.), Trylon Shallowcast embedding resin EM400PA (U.K.), and Poly-pol PS-230 (Netherlands). All are clear or become clear with the addition of the organic peroxide hardener. These resins have been chosen as they are initially water-clear and cure in thin layers. Provided that they have been accurately mixed, the rate at which the cured resins yellow depends to some extent upon the conditions in which restored items are displayed or stored.

Disadvantages in the use of polyester resins are the shrinkage of 8% during curing (though this shrinkage can partly be compensated for by topping up the mold as the resin polymerizes); the emission of styrene for some considerable time after curing; and the fact that the resin surface often remains tacky for some time. Reasons for this final phenomenon are interference from atmospheric moisture, aging of the hardener, or, if the resins are stored under refrigeration, the use of the resin and hardener before they have reached room temperature, thus preventing the complete chemical reaction.

Hardening of the surface is aided by polymerizing the cast in a dry atmosphere such as a sealed cabinet containing trays of silica gel. Warming the cast in an oven is not recommended since it may cause premature discoloration of the resin. Polyester embedding resins abrade and polish easily.

Although cured epoxy and polyester resins can be reversed, they cannot be truly dissolved. Because solvents cannot break down and dissolve the cured polymer network, the resin merely swells to the point that it can be mechanically removed. This procedure has potential dangers. The swelling process may cause so much tension that damage is caused to the glass. Thus several applications of solvent (normally methylene dichloride), each of which is removed mechanically after a few minutes, are preferred.

The methacrylate Plastogen G with Lumopal hardener used by Wihr (1963) and Errett (1972) is transparent and mobile and is therefore normally cast into closed molds (see above). The liquid resin is mixed with 0.25 and 0.5 hardener (powder), which is difficult to assess in small quantities, and the addition of too much hardener may cause premature discoloration of the resin. Plastogen G has a 15-minute pot life but cannot be worked during that entire time because a skin quickly forms over the surface. It also has an extremely powerful, unpleasant smell. After mixing, and if necessary coloring, the resin should be covered and left to stand while air bubbles escape. Notman (1973) used a color acrylic resin to build up grozed edges in stained glass so that the pieces could be edge-joined.

Technovit 4004A (poly methacrylate) is occasionally used as a gap filler, but, being translucent, it can be used only with opaque glass. The polymer (powder) is mixed with the liquid monomer in the ratio of 5 parts to 3, but the proportions are not critical and the setting time may be varied by changing the amount of the powder. When mixed in the recommended proportions, Technovit 4004A sets at room temperature in 10–15 minutes but can be worked with a spatula during this time. This product is guaranteed by the manufacturer not to shrink or expand on curing. It emits heat if cast in large amounts. Technovit adheres well to glass, is relatively hard, and can be abraded and polished.

The thermosoftening property of Technovit 4004A has been exploited by Errett (1972) and Jackson (1983) to reshape small casts and thus dispense with the need to make complicated molds.


Materials used for coloring fall into two categories: those used for mixing in with resin and those applied to the resin after it has cured. Colors for mixing in with the resin may be transparent or opaque depending upon the desired effect. When incorporating color with resin, enough colored resin must be produced to complete the restoration, and allowance must be made for areas that may need to have more resin added to them or others that may have to be cast more than once. Preparing sufficient color is important since a color can rarely be exactly matched a second or third time. Hardener is then added to small amounts of the colored resin as required for use, and the resin is allowed to stand before use to enable air bubbles to escape.

Most glass on display in museum cases is exposed to high levels of illumination, and hence light-resistant pigments are needed in any restored portions. Improvements in pigment technology have provided the conservator with a fairly wide palette of colors, and a list of light-resistant colors has been given by Thomson (1972). Other pigments are used by Wihr (1963), Davison (1981), Errett (1972), and Staude (1972). Occasionally pigments can produce adverse effects with some reactive polymers, and it is then more satisfactory to purchase ready-mixed colors. There are colors for polyesters, silicones, and epoxies; their light stability must be checked before use. Manufacturers or suppliers may issue data, but conservation scientists in major museums have tested many materials and published the results either in-house or in conservation literature.

Unfortunately, the transparent epoxy and polyester adhesive and casting materials often require the use of dyes rather than pigments to color them in order to retain their transparency, and there is a much smaller range of light-stable dyes available for use in polymers.


The lacquers usually employed for retouching the restored portions of glass objects are those used in the restoration of ceramics, porcelain in particular, and their availability varies from country to country. Transparency and retention of color are important, as is any short- or long-term interaction with the resin substrate. Enameled decoration may be copied on resin casts using any pigments and media used for ceramic restoration, provided that they adhere to the cast and the solvent does not damage it. Gold decoration may be copied in leaf gold applied on a size or as liquid metallic paints, though the latter will discolor on aging.


As previously mentioned, properties of materials vary from country to country. Moreover, since restoration is not normally the prime purpose for which they were intended, materials may be altered by manufacturers without notification. Thus it is important that before work commences experiments be undertaken on materials in use for the first time to confirm that their performance will be as expected. A full discussion of materials used in conservation can be found in Horie (1987).


Augerson, C. C., and J. M.Messinger. 1993. Controlling the refractive index of epoxy adhesives with acceptable yellowing after aging. Journal of the American Institute of Conservation32: 311–14.

Davison, S.1981. New materials in the service of glass restoration. Annales du 8e Congress Internationale d'Etude Histoire du Verre. Liege, Belgium: 369–75.

Davison, S.1984. A review of adhesives and consolidants used on glass antiquities. IIC preprints, 10th International Congress, IIC, Paris. London: IIC. 191–99.

Down, J. L.1984. The yellowing of epoxy resin adhesives: Report on natural dark ageing. Studies in Conservation29: 63–76.

Eckmann, C.1995. Ein schnellh�rtender silikonkautschuk auf vinylpolysiloxan-Basis als Manschettenmaterial bei Erganzungen von Glas�rn. Arbeitsbl�tter f�r Restaruratoren: 72–76.

Errett, R. F.1972. The repair and restoration of glass objects. The Bulletin of the International Institute for Conservation—American Group, no. 12. Washington. D.C.: IIC. 48–49.

Fisher, P.1988. Advances in the restoration of glass vessels. UKIC preprints, United Kingdom Institute for Conservation 30th Anniversary Conference, London. London: UKIC. 81–83.

Fisher, P.1992. HXTAL NYL-1, an epoxy resin for the conservation of glass. In Glass and enamel conservation. UKIC Occasional Paper 11. London: UKIC. 6–9.

Gedye, I.1968. Pottery and glass: The conservation of cultural property. Museums and Monuments11:109–13.

Hogan, L.1993. An improved method of making supportive resin fills for glass. Conservation News50:29–30.

Horie, C. V.1987. Materials for conservation. London: Butterworths.

Jackson, P. R.1983. Restoration of an Italic glass oinchoe with Technovit 4004A. Conservator7:44–47.

Jackson, P. R.1984. Restoration of glass antiquities. ICOM Committee for Conservation preprints, 7th Triennial Meeting, Copenhagen. Paris: ICOM. 20:13–17.

Koob, S. P.1979. The removal of aged shellac adhesive from ceramics. Studies in Conservation24:134–35.

Messinger, J. M., and P. T.Lansbury. 1989. Controlling the refractive index of epoxy adhesives. Journal of the American Institute for Conservation28:127–36.

Morg�s, A, J.Nagy, and L.P�lassy. 1984. New silicone rubber mold-making materials: The addition type silicone rubber. ICOM Committee for Conservation preprints, 7th Triennial Meeting, Copenhagen. Paris: ICOM. 20:19–20.

Newton, R. G., and S.Davison. 1989. Conservation of glass. London: Butterworths.

Noll, W.1968. Chemistry and technology of silicones. London: Academic Press.

Notman, J. H.1973. Restoration of a stained glass roundel, St. Anne with Virgin and Child. Scottish Art Review15(2):10–13.

Petermann, R.1969. Nachbildung antiker Gl�er. Arbeitsbl�tter f�r Restaruratoren1(18): 9–14.

Staude, H.1972. Die Technik des Zusammensetzens und Erg�nzens antiker Gl�ser. Arbeitsbl�tter f�r Restaruratoren1(5): 20–27.

Tennent, N. H., and J. H.Townsend. 1984. Factors affecting the refractive index of epoxy resins. ICOM Committee for Conservation preprints, 7th Triennial Meeting, Copenhagen. Paris: ICOM. 2:26–28.

Thomson, G.1986. The museum environment. London: Butterworths.

Wihr, R.1963. Repair and reproduction of ancient glass. In Recent advances in conservation, ed.G.Thomson. London: Butterworths. 152–55.

Wihr, R.1977. Restuarieren von Keranik und Glass. Munich: Callwey.


Tennent, N. H., and J. H.Townsend. 1984. The significance of the refractive index of adhesives for glass repair. In Adhesives and consolidants, ed.N. S.Bromelle, E. M.Pye, P.Smith, and G.Thomson. London: IIC. 205–12.



Araldite AY103/HY956, B & K Resins Ltd., Unit 2, Ashgrove Estate, Ashgrove Rd., Bromley, Kent, BR1 4TH U.K.

Araldite 2020 (XW396/XW397)

Ciba Polymers, Duxford, Cambridge, CB2 4QA U.K.

Distributed by, B & K Resins Ltd., Unit 2, Ashgrove Estate, Ashgrove Rd., Bromley, Kent, BR1 4TH U.K.

Epo-Tek 301 & 301-2

Conservation Materials Ltd., 1165 Marielton Way, Sparks, Nev. 89431


Fyne Conservation Services, Airds Cottage, St. Catherine', Loch Fyne, Argyll, Scotland PA25 8BA U.K.

Ablebond 342-1 and HXTAL NYL-1

Conservation Materials Ltd., 1165 Marielton Way, Sparks, Nev. 89431


Trylon Shallowcast embedding resin EM400PA, Trylon Ltd., Wollaston, Northamptonshire, NN29 7QJ U.K.

Tiranti's clear embedding resin

A. Tiranti Ltd, 27 Warren St., London W1 U.K.

Poly-pol PS 230 embedding resin

Poly-Service B. V., Hoogeveenenweg 83, Postbus 160, 2910 AD Nieuwerkerk A/D Ijsse, Netherlands


Technovit 4004A, Rubert & Co., Ltd., Acru Works, Demmings Rd., Cheadle, Cheshire, SK8 2PG U.K.

Plastogen G and hardener

Alfons Schmidt, 6-12, Speyer, St. German Strasse 14, Bavaria, Germany

Silastics (silicone rubber)

I.C.I. Silastics, Imperial Chemical Industries Plc., Cleeve Rd., Leather head, Surrey, KT22 7SW U.K.

Dow Corning Silastics

Dow Corning Corporation, Midland, Mich. 48640


Deka transparent colors, Deka Textilfarben AG, Munich, Germany

Maimeri Restoration Colors

F. Ili. Maimeri & Co., Mediglia (M.I.) Italy

Available at good artists' suppliers


Astynax toughened pink dental modelling wax, Associated Dental Products Ltd., Purton, Swindon, Wiltshire, SN5 9HT U.K.

Cab-O-Sil (fumed silica)

Cabot Corp., 125 High St., Boston, Mass. 02110

Glass fiber surfacing tissue

Resin manufacturers

HMG (cellulose nitrate)

Henry Marcel Guest Ltd., Riverside Works, Manchester, M40 7RU U.K.

Loctite 305 (UV-curing)

Loctite U.K. (Division of Loctite Holdings Ltd., Consumer Products), Watchmead, Welwyn Garden City, Herts, AL7 1JB U.K.

Micro-Mesh cushioned abrasives

Micro-surface Finishing Products, Inc., 1217 W. Third St. (or Box 818), Wilton, Iowa 52778

Newplast synthetic modeling material

Trylon Ltd., Wollaston, Northamptonshire, NN29 7QJ U.K.

Polyvinyl alcohol (separating agent)

Resin suppliers, e.g., A. Tiranti Ltd.

Synperonic N (Non-ionic detergent)

Conservation Resources Ltd. (U.K.), Units 1, 2, and 4, Pony Rd., Horspath Industrial Estate, Cowley, Oxon, OX4 2RD U.K.

Inert fillers plasticine, modeling clay, plaster of paris, shellac

Sculptors' suppliers

Acetone, IMS, etc.

Chemical suppliers

Sellotape (pressure-sensitive tape)



SANDRA DAVISON gained a diploma in archaeological conservation and worked in the British Museum for 14 years, specializing in ceramic and glass. She trained in glass restoration in Mainz, Germany. She has worked for museums in the United Kingdom, France, Malaysia, and Saudia Arabia; and lectured and published widely, including a definitive work on glass, Conservation of Glass, with Professor Roy Newton. She has taught glass restoration at the conservation schools in Copenhagen and Amsterdam, West Dean College, Chichester, the Institute of Archaeology, London University, the Getty Conservation Institute, Los Angeles, and the Smithsonian Institution of Washington, D.C. The Conservation Studio was established in London in 1984 and relocated to Thame, Oxfordshire in 1991. In 1979 she was elected a Fellow of the International Institute for the Conservation of Historic and Artistic Works (FIIC). Address: Conservation Studio, 68, East St., Thame, Oxfordshire, OX9 3JS, U.K.

Section Index

Copyright � 1998 American Institute for Conservation of Historic and Artistic Works