2 CATALOG OF TRADITIONAL FILLING MATERIALS

2.1 INTRODUCTION

Filling compounds can be divided into two main categories: mixtures of adhesive substances and bulking agents (sometimes referred to as fillers in the historic literature) that do not otherwise interact chemically, and mixtures that undergo a chemical reaction or crystallization. This distinction is an attempt to make sense of a complex topic; it was not made or understood in the past because of limited knowledge of chemical reactions. Moreover, in practice, there is often overlap between these categories. Many recipes are so complex that a variety of interactions are likely. Nomenclature is inconsistent, but the term “cement” usually refer to a heavy-bodied compound that can be usefully employed for filling voids, “mastic” refers to thermoplastic filling compounds, and “lutes” refers to sealing and filling compounds that resist heat and/or chemicals. Use of the terms “filling compounds” and “fills” as the modern conservator uses them appears to be recent.

Restorers have apparently long understood that most adhesives can be made into effective filling agents with the appropriate bulking agent and that the degree of bulking is the chief or only distinction between an adhesive and a filling compound. This understanding is explicitly stated by Leland (1896, 1), who explains that “stickers” (adhesives) and “cements” (by which he means filling compounds) are effectively the same in most instances. He further states that “this principal of mixing a powdered substance with glue or gum or an adhesive runs through all the arts of mending. The powder of cocoanut shells, slate, of paper, plaster of Paris, of leather, clay, lime, fine sand, and many other substances can all be combined with adhesives.”

Binding agents were chosen for their strength and longevity under particular service conditions, and bulking agents were apparently chosen for their color and perceived compatibility with the substrate. Sawdust was often chosen for fills on wooden objects, chopped leather for leather objects, minerals for ceramics, glass powder for glass, metal filings for metals, and so forth. These materials were, of course, readily available to specialist craftspeople and may have often been the best choices from a technical and aesthetic point of view. But that is not to suggest that all the recipes given in old books are simple and rational in their design. In many cases the developers of filling mixtures seem to have been guided by the principle that if a few ingredients are good, then several more would be even better.

The increasing pace of technological innovation, beginning in the 18th century and accelerating throughout the 19th and 20th centuries introduced an increasing variety and complexity to the options for adhesives and filling materials. Those materials that appeared to be the result of scientific inquiry may have been particularly attractive in an age enamored of science and innovation. A profusion of specialized treatises on the fundamental chemistry and properties of various materials were published in the 19th century, and these were studied and cited by various writers on restoration. One writer on repairing (Leland 1896, 4–5), for example, cites in the course of two pages, four German-language treatises on such subjects as natural resins and spirit lacquers, glue and gelatin, egg and blood albumen, and lime and mortars, and stresses his reliance on “the three hundred volumes of the Chemical Technical Library of A. Hartleben.” The 19th-century American publisher Henry Carey Baird made an effort to translate and publish a long list of such European treatises for the benefit of American manufacturers (Green 1991).

The following description of materials includes the filling compounds, binders, and bulking agents most usually reported in the historic literature, but in an article of this length it is not possible to be comprehensive. Aside from the most common ones such as plaster (numerous references to plaster fills are found in Daniels 1988), confirmation of such materials on actual objects is rarely noted in the recent conservation literature. In the practical world of the conservation studio and lab, old and degraded or unsightly fills are most often simply removed as expediently as possible. Even if sophisticated analytical techniques were brought to bear, the results might well be misinterpreted without a knowledge of the historic possibilities. Sorel's cement, for example, would most likely be interpreted as corroded zinc metal. Organic binders present many difficulties in analysis, particularly after aging, and can often be sorted into broad categories only.

Filling agents that deserve greater discussion due to their frequency of use and those that fall outside the general classification scheme of binder/bulking agent because of more complex chemical interactions are listed first.

2.2 COMMON AND COMPLEX FILLING COMPOUNDS

Plaster—There are two classes of calcium compounds known as “plaster”: “lime plaster” and calcium sulfate or “gypsum plaster.” Both have been in use since antiquity. Lime plaster is made by heating calcium carbonate (CaCO3) to produce calcium oxide (CaO or “quicklime”) and mixing it with water (called “slaking”) to produce calcium hydroxide (Ca(OH)2 or “lime putty”). The lime putty is mixed with various aggregates and binders to create mortars and plasters and hardens by the evaporation of water and gradual subsequent reconversion to calcium carbonate over a period of years. Gypsum plaster (also called “plaster of paris”) is made by heating calcium sulfate (CaSO4 • 2H2O, “gypsum”) to remove part of the chemically bound water, creating a hemihydrate (CaSO4 • � H2O). When mixed with water it reconverts to hydrated calcium sulfate, setting in the process.

Unlike lime plaster, gypsum plaster sets without shrinking. In fact, gypsum plaster expands upon setting in most instances, making it virtually unique among materials available to restorers of the past. In some cases, the insertion of high-expansion plasters into closed and rigid voids has resulted in damage to the object.

Lime plasters have been preferred for use in building restoration for many centuries, but their alkalinity and shrinkage upon drying would generally have made them undesirable for use on smaller objects meant for indoor use and display. Exceptions might include fills to stone sculptures, as the use of lime mortars and putties was traditional in the stone trades (Cassar 1988; Hanna and Lee 1988), and the use of lime as a bulking and denaturing agent for casein and other protein-bound adhesives and fillers.

Hard-finish plasters (patent plasters)—During the 19th century, various proprietary gypsum-based cements were introduced. These were generally made from gypsum that had been heated past the usual point for the production of plaster. Such “dead burned” gypsum (CaSO4 “anhydrite”) resists setting unless a metal salt is added as a catalyst. These proprietary cements took longer to set than ordinary plaster of paris, allowing for manipulation over a longer period of time, and they were also harder and less porous (Eckel 1928). These qualities would have recommended them as filling materials in some instances. Examples are Keene's cement made with alum, Martin's cement made with potassium carbonate (“pearlash”) and alum, Mack's cement made with sodium sulfate (“Glauber's salt”), Parian made with borax (Eckel 1928), and a similar product, Keating's, made with borax and alum (Spon's Mechanics own book 1893). While these may well be found as fills on objects, it would be difficult to distinguish them from ordinary plaster unless they were specifically analyzed for their additives.

Portland cement—Although the term “Portland cement” dates to the patent of Aspdin in 1824, it was the natural successor of the socalled hydraulic mortars or Pozzolanic mortars that had been in use since the Roman period (Torraca 1981). All of these materials set by rapid hydration and crystallization and do not need access to air to harden. Modern Portland cement is made by heating a mixture of calcium carbonate and alumino-silicates (usually clay) until it partially vitrifies. The resulting “clinker” is ground to a powder. When mixed with water, it sets firmly, hardens further by drying, and the lime component reconverts to calcium carbonate over a period of years due to the agency of carbon dioxide. These materials can be used in large fills or castings, particularly when mixed with sand and gravel aggregates (a mixture called “concrete”), and they are resistant to water. For these reasons they have been used for filling large voids in all sorts of outdoor objects. Although Portland cement-based mortars and concretes are most likely to be encountered on stone (Cassar 1988; Hanna and Lee 1988), conservators have found Portland cement fills in metal and wood sculptures and structures, particularly in the outdoor environment.

Sorel's cement and magnesite cement—In 1853, a chemist named Sorel discovered that if zinc chloride was mixed with zinc oxide and water, it set rapidly into a very hard cement (Eckel 1928). This substance was described in subsequent technical treatises as “Sorel's cement.” A subsequent compilation of recipes (Hopkins 1906) gives Sorel's cement as zinc white and fine sand combined with zinc chloride solution and applied immediately. Other variations from the same source include borax. It seems likely that hardening of such mixtures is the result of the formation of zinc oxychlorides, silicates, and borates. It was later discovered that magnesium chloride and magnesium oxide had the same properties, and this material was called “Sorel's magnesium cement.” “A newly invented cement” (Hopkins 1906) was described as being made from calcined magnesite (magnesium carbonate burned to produce magnesium oxide) mixed with sand and water. These materials were too expensive to have been used on a large scale, but became important as refractory cements (“lutes”) and for small-scale applications including dental fillings. Eckel (1928) devotes a chapter (“Magnesia Bricks and Oxychloride Cements”) to their production and use.

Gesso—Traditional “gesso,” consisting of animal glue as the binder and calcium sulfate or carbonate as the bulking agent, has been used as a preparation layer for gilding and as a filling compound by the ancient Egyptians and in every era since. In addition to being cheap and available, animal glues have the property of gelling at room temperature and suspending the filler particles in a uniform manner without settling. Gesso putties are easy to make, use, and level and continue in the repertoire of many conservators of various specialties (Thornton 1991). Animal-glue-bound fillers shrink considerably upon drying however, and are notoriously unstable if exposed to changes in relative humidity. Because they are soluble only in water, the use of traditional gesso as a fill for gesso or other water-soluble substrates may cause damage during application and will create problems of reversibility.

Composition—The material or category of materials called “composition” (often abbreviated to “compo”) was used most extensively from the fourth quarter of the 18th century to the present to create sculptural relief in architectural interiors and on picture frames. The basic recipe, consisting of animal glue, linseed oil, and a natural resin (usually rosin) bulked with a calcium filler has been fairly standard since then, but was clearly based on earlier press-molded compositions that had been in use since the medieval period (Thornton and Adair 1994). Composition has been most extensively used as a fill material in picture frame restoration, and it still has some notable advantages for this purpose, but it has probably found use on other objects as well.

Composition ornament was created by pressing the warm and pliable material into rigid negative molds of wood, metal, sulfur, or resin compositions. After the material cooled it was tough and rubbery. The artisan had considerable latitude in gluing it onto a substrate, stretching or compressing the material to make it fit or come out right, and freely composing with bits of detail taken from the same mold. Composition continues to be used as a filling material by modern conservators, particularly on wood and composition picture frames (Thornton 1991).

Sulfur—Elemental sulfur can be melted and poured into voids, where it hardens to a rigid, plasticlike mass due to cooling and crystallization alone. In this form it is hard but brittle and has only a slight smell. Sulfur was used by sculptors for making molds as early as the Renaissance, and is also noticed as a useful fill material in various works on restoration and mechanical arts. Sulfur occurs in two crystalline forms, one black and one canary yellow. The latter was used as a decorative inlay material in furniture (in what are commonly called “Pennsylvania Dutch dower chests,” for example). Black sulfur was mixed with iron filings for use in filling iron objects (Pilkington 1881), the color being sympathetic with black finished iron.

Metal—As already noted, metals of all sorts have a tradition of use for fills and repairs on metal objects, but low-melting temperature alloys, particularly those usually characterized as “soft solder” (lead-tin alloys) are occasionally found on other objects such as those of ceramic and stone (Williams 1988 and informal discussion with colleagues). A eutectic alloy of lead and tin melts at 361�F. Low-melting alloys of lead, tin, bismuth, cadmium, and mercury were extensively explored in the 19th century (Ure 1842; Buchanan 1910; Brannt [1890] 1919), and became known by the names of their inventors. Wood's metal and Lipowitz's metal (lead, tin, cadmium, and bismuth), to give just two examples, had melting points of only 140–160�F and could be melted in hot water. The addition of mercury to Wood's metal (called “Mackenzie's amalgam”) yielded a novelty alloy that could be melted by the friction of rubbing it (Buchanan 1910). Such low-melting-temperature alloys could be safely used to fill gaps in many types of material (including teeth, as dentists discovered), and it would be surprising indeed if no restorers had thought using them.

White lead and linseed oil putty—Barthelet (1884, 32) says “We shall not despise the old and venerable putty, that has rendered our ancestors so many signal services because it is an affair of the past.” He goes on to say that putty composed of linseed oil, white lead (lead carbonate), and sometimes a little litharge (lead monoxide) to “speed the hardening” is the most durable for objects in everyday use and the most resistant to water of all temperatures. The 15th-century artist Cennino Cennini (Thompson [1933] 1960) recommends a cement for mending ceramics and glass that consists of varnish and white lead (colored green with verdigris if the glass is green). “Putty” appears to have been commonly understood to consist of linseed oil and chalk or white lead (often both) for a few one hundred years at the least (Leighou 1925). Lead compounds catalyze the “drying” of linseed and other “drying oils,” resulting in final products that are hard but also flexible and tough, with excellent gap-filling capabilities (as a primary use in window glazing attested). The disadvantages were the unsightly mechanical fasteners such as staples and rivets that were necessary to hold ceramic objects together while the putty was drying and the seepage of the linseed oil into adjacent areas of porous objects, causing staining. Most conservators who work on ceramics, stone, and other hard materials have encountered repairs made with white lead putty and staples, a method so old that Leland (1896, 18) comments that those who claimed to have invented varieties of this type of repair “might as well apply for a patent for having discovered the art of mixing brandy with water.” Another writer (Seaman 1899, 205) states: “There is nothing better for mending china than white-lead,” meaning putty. “It resists water and heat.”

Glycerin-litharge cement—This cement, a mixture of litharge and glycerin, was used for cementing stone and metal and as a “luting” (sealing cement) that would resist a wide range of chemicals (Brannt 1886). The mixture hardened within minutes, probably due to polymerization of the glycerin and the formation of metallic soaps. Glycerin (also called “sweet oil”) was a by-product of the manufacture of soap and was isolated by Scheele in 1779 (Fieser and Fieser 1956). It would have been easily available from druggists or chemists throughout the 19th century.

Protein-lime mixtures—Various proteins can be made into useful adhesives and binders for fills by the addition of lime (calcium oxide) or some other strongly alkaline substance (ashes and water-glass, for example). Alkalinity destroys hydrogen bonding within the proteins, causing them to denature, uncoil, and then harden by evaporation of water.

• Casein: The making of “cheese glue” with milk protein (casein) and lime is described by Theophilus (Hawthorne and Smith 1963) and many subsequent writers of technical treatises.
• Albumin: Egg white (albumen) was often chosen as the protein where light color was required. So-called Chinese cement, an item of commerce in the 19th century, consisted of egg white, lime, and glass prepared by powdering it and “sieving it through silk” (Hasluck 1904, 6:306). An influential 19th-century writer on technical subjects, Andrew Ure, states that cement made of lime and egg white was “much employed for joining spar and marble ornaments” (Ure 1842, 391).
• Blood: Blood could be used where the color was acceptable or desirable. Ure (1842, 391) also reports a cement “used by coppersmiths to secure the edges and rivets of boilers” consisting of quicklime and bullock's blood.
• Gluten: The protein component of wheat flour called gluten was denatured by a strong alkali and used as the basis of such recipes as the “lime putty for wood,” listed in a 19th-century recipe book (Brannt 1886, 67), which contained slaked lime and rye flour. Acid would also denature the protein, and Leland (1896, 242) describes a mixture of breadcrumbs and nitric acid that became very hard and was used to imitate meerschaum and bone articles.

2.3 BULKING AGENTS AND FILLERS

2.3.1 MINERALS

Minerals are obvious choices as bulking agents due to availability and variable properties such as color, grit size, translucency, hardness, etc. Stone dusts might consist of a variety of mixed minerals, and powdered ceramics and glass can be thought of as man-made minerals in terms of their function in filling compounds.

Calcium carbonate—Marble dust, limestone dust, chalk (called “whiting” and “gilder's whiting”), and ground marine shells and eggshells all consist of calcium carbonate, and all are found in various fill recipes. Leland (1896, 144) recommends ground eggshells for ivory fills because they “are even less likely to turn grey” than fills made of ivory dust. In Northern Europe, gesso and gesso-filling putties were usually bulked with calcium carbonate (whiting) because it was cheap and locally available.

Calcium sulfate—In Southern Europe, “gesso” was normally made with calcium sulfate because of local availability. Three forms of this mineral would not react with water and set, and thus could be used as inert bulking agents in combination with dilute animal glue size: gypsum, consisting of calcium sulfate dihydrate (CaSO4 • 2H2O); selenite, a massive and transparent crystalline form of calcium sulfate dihydrate; and anhydrite (CaSO4), which might consist of natural anhydrite or be made by completely desiccating calcium sulfate dihydrate, in which case it might be called “dead burned plaster”. Cennini (ca. 1437) describes how finely divided calcium sulfate dihydrate for gesso was sometimes made by mixing plaster of paris (see discussion of plaster above) with excess water and stirring it while rehydration occurred so as to prevent it from setting into a hard mass (Thompson [1933] 1960).

Barium sulfate—This mineral, also called baryta and barytes, was only occasionally mentioned as a bulking agent (Leland 1896; Brannt [1890] 1919), possibly due to relative expense. Unlike other highly toxic compounds of barium, barium sulfate is too chemically inert to be toxic and could be safely used as a bulking agent. It has the consistency and appearance of chalk, but much greater weight.

Clays—Clay minerals have extremely small particle sizes and are naturally variable in color. Pure primary clays are white to gray (called “pipe clay” or “china clay”). Darker grays and blacks may be due to organic staining or carbon in the form of graphite. Reds and yellows are due to the presence of iron oxides (called “red and yellow boles,” “yellow ochre,” and “red ochre”). A disadvantage of clays when used in aqueous binders was the large degree of drying shrinkage due to the loss of the numerous water films present between the flat and hydrophilic clay particles.

Ground ceramic—Already-fired ceramics were often ground to make effective bulking agents. These were used by the ceramics industry in the form of “grog” and so would have been commercially produced. A mixture of ground ceramic and lime was used as a traditional stone adhesive and coating in Malta (Cassar 1988) and probably other places as well. “Brick dust” is often specified in fillers for wood because the color matched timbers with reddish coloration such as mahogany (Sheraton [1803] 1970).

Ground glass—Glass powder was used to make light and translucent fills for china and glass. It was the bulking agent in “Chinese cement” and was also used with potassium and sodium silicate water-glasses.

Soapstone—Hydrated magnesium silicate stones were called “soapstone” and “steatite” because of their softness and slippery feel. In powdered form they are known as “talc” and “French chalk.” In the United States, large quantities of soapstone dust was produced by the manufacture of soapstone sinks, stoves, and bed warmers, and Leland (1896, 34) states that in America putty “is made from pulverized soapstone and oil.”

Graphite—This mineral, one of the crystalline forms of carbon, was also called “plumbago” and “black lead.” It was used primarily in compounding filling agents for iron objects due to its gray-black color and sympathetic metallic sheen.

2.3.2 METAL POWDERS

Iron filings—The chief value of iron filings as a bulking material was that the rusting and concomitant expansion would tightly fill voids and lock parts together (as any auto mechanic knows). Such filling compounds usually included a chloride-containing rusting agent such as acid, urine, or ammonium chloride (called “sal ammoniac”). A recipe from Barthelet (1884, 36) for “plugging iron into stone” consists of iron filings, acetic acid and garlic. Ure (1842) describes “the iron rust cement” as consisting of iron filings and sal ammoniac moistened with water. A later treatise (Spon's Mechanics own book 1893) refers to such rust cements or “cast iron cements” as being commonplace and describes them as consisting of iron powder, sal ammoniac, and flowers of sulfur, which when mixed together and dampened begin to heat and set into a hard mass. Rusting and expansion were not always the intent. Scott (1926) recommends a proprietary filling compound for iron called “Medeesi” (made easy?) consisting of iron filings, calcium sulfate, and calcium phosphate and claims that it will not rust—probably due to the formation of stable iron-phosphates.

Metal flake powders—“Bronze powders,” usually consisting of brass alloys (copper and zinc) became available in large enough quantities to be used in adhesives and fills after the 1840s, when Henry Bessemer began manufacturing them on an industrial scale. Formerly, they had been made by hand-grinding hand-beaten metal leaf to fine powder, and the expense would have precluded their use as bulking agents (Bessemer 1905). A wide range of colors was available providing a good match for many metals. By the end of the 19th century, aluminum was also made into flake powders, followed by stainless steel. The proprietary adhesive/filler called “Liquid Solder” and the material called “Liquid Stainless Steel” were bulked with these flake metals (Brady and Clauser 1977, 748).

2.3.3 CELLULOSE

Sawdust—Sawdust was always, of course, easily available, and it is commonly found in recipes for wood fillers and for making various moldable composition materials such as the cases for early photographs. Most of these employ a nonaqueous binder such as shellac due to the extreme swelling and shrinkage that sawdust (or any cellulose filler) undergoes in an aqueous binder.

Paper pulp—Three-dimensional paper objects (usually called “papier-m�ch�”) were made of paper pulp as early as sheet paper, and paper pulp or sheet paper adhered with glue or gum, generally referred to as papier-m�ch�, has been extensively used as a sculptural material (Thornton 1993). It is likely that the material has been used as a fill material in some instances for an equally long time, and Leland (1896) describes many filling and composite molding materials based on paper pulp. Numerous other practical manuals give instructions in the use of papier-m�ch� and related materials.

2.3.4 PROTEINACEOUS NATURAL ORGANICS

Bone and ivory dust—Recipes for artificial ivory found in many recipe treatises (Spon's mechanics own book 1893 contains several) often called for bone and ivory dust. These bulking agents were also a natural choice for compounding fills for bone and ivory objects due to good color match and perceived compatibility. Ivory and bone were the basis of extensive industries prior to their replacement with synthetic polymers, and the waste from these operations would have been available in many cities. Leland (1896, 144) recommends a fill for ivory made of ivory dust, “such as can be bought of every ivory turner,” combined with gum arabic and alum or “silicate of potash” (see discussion of water-glass, below).

Ground leather—In keeping with the natural impulse to fill like with like, ground or macerated leather has been used to fill leather objects when mixed with a variety of binders (Waterer 1973). Leland (1896, 185) extols his “leatherpaste,” which he used for filling losses in leatherwork, made of macerated leather and “Caoutchouc or indiarubber in solution.” Ground leather was also used in the same way as papier-m�ch� to create sculptural relief. An ornate Tudor-era “leather-m�ch�” ceiling (previously thought to be papier-m�ch�) exists in the apartments of Henry VIII at Hampton Court (private communication with curator).

2.4 ADHESIVES AND BINDERS

2.4.1 PROTEINS

Aside from albumen, gluten, blood, and casein which have been described above, the most important category of proteinaceous adhesives are the animal glues. Over many centuries, the term “glue” meant only animal glue, and this was so well understood that it is used without explanation in most old texts. Barthelet (1884) says, “Of all the cements that are compounded, glue is still the most serviceable and the most universally employed.”

Animal glues are hydrocolloids derived from collagen (colla, meaning “glue” in Greek and gen, meaning “creator”), which is the protein present in skins, bones, and connective tissue. Other starting stock such as bones (bone glue), fish skins (fish glue), and rabbit skins (rabbit skin glue) have also been used to produce distinctive adhesives. Glue is made by hydrolyzing the initial collagen, a process of breaking the polymer chains into smaller extractable units by cooking the animal parts (glue stock) in water, then concentrating the resulting broth, and finally drying it.

It has long been understood that glycerin added to animal glue will give it greater toughness and elasticity when dry. If glycerin is added in sufficient quantities (weight percent equal to the dry weight of the glue), it will prevent the glue from hardening altogether, the result being a rubberlike substance. A hard and nonshrinking putty for wood repairs is given by Leland (1896) as consisting of glue, glycerin, and cocoa-nut powder.

Hide glue—Most animal glue is made from cattle hides and called hide glue. Animal glues of the 19th century and earlier were of uneven quality, and many craftspeople chose to make their own by boiling parchment clippings and concentrating the broth. The resulting parchment size was often specified in recipes for gesso and gesso putty. Present-day gilders often use a high molecular weight and relatively high fat content glue called rabbit skin glue, but no specific notice of this material can be found in historic treatises and it is likely to be of fairly recent origin. Gelatin consists of high-quality glue that has been purified and freed of color (Fernbach 1921; Mills and White 1994).

Isinglass—Isinglass is an animal glue with a rich lore. It consists of the swim bladders of sturgeons (but other fish are sometimes substituted). Even though isinglass is not generally classed as a fish glue, it behaves as one. Very high molecular weights are reported in comparison to other types of animal glue, but solutions of isinglass resist gelling at room temperature. Gelation is a highly complex chemical phenomenon in which a number of factors aside from molecular weight operate. Compared to gelatins and rabbit skin glues, isinglass forms brittle dried films and is relatively moisture sensitive. Light color and resistance to gelling have been the chief characteristics that make it desirable for restoration purposes, and it shows up in many recipes for adhesives and fills, including the “diamond cement” that was an apparently celebrated restoration material in the 19th century. One source describes diamond cement as being composed of isinglass and various natural resins in an alcohol solution (Ure 1842).

2.4.2 CARBOHYDRATES

These are a class of organic compounds generally derived from natural sources that usually have the general formula Cn (H2O)n. For more information on the chemistry of starches and gums, the reader is advised to consult an organic chemistry textbook (Masschelein-Kleiner 1985; McMurray 1988, for example; see also Mills and White 1994).

Starch—Starches have always been important foods and have been readily available in virtually all places and cultures. It is not surprising then, that they were often used as adhesives (Williams 1988) and are found in many recipes for filling and modeling materials. One 19th-century formula book contains a recipe for “French putty for wood” consisting of potato starch, gum arabic, and water (Brannt 1886, 66). Leland (1896, 145) claims that the 16th-century Italian “bas reliefs for small caskets” (pastiglia) were made of rice and lime.

Dextrin—This material has the same general formula as starches, but a lower molecular weight. Dextrin was made by treating starches with acids and heat. It dissolves readily in water and was used in various filling putties in the same way as gum arabic would have been used.

Gum arabic—This is the gum exuded by various Acacia species. It is freely soluble in water and has been an item of commerce since ancient times. It has been frequently used as a binder for filling putties, particularly those used on ceramics (Parsons and Curl 1963), and because it does not gel, it has advantages over animal glue when used in this way. (One of the author's first instructors in object conservation, Rostislav Hlopoff, preferred a filling putty for ceramics made of gum arabic and whiting over any other fill material, having used it as a restorer in Paris in the 1920s and 1930s, where it was apparently in common use.)

Gum tragacanth—Gum tragacanth, variously rendered in historic treatises as “gum adragant,” “gum dragon,” etc., is the exudate of various species of the genus Astralagus. The gum swells in water rather than dissolving and would have been more difficult to use as a binder for filling compounds than gum arabic. It was used, however, to make various complex modeling and molding compositions in the 19th century, often incorporating whiting and white lead as bulking agents (Betzler 1996).

2.4.3 NATURAL RESINS

Natural resins from angiosperms and gymnosperms, both living and extinct (amber and some varieties of copal), have been traded and used in various arts for many centuries. They include damar, mastic, sandarac, shellac, rosin, copal resins (the so-called fossil resins), and the various oleo-resins with volatile components called “turpentines” and “balsams” (known as “Strasbourg turpentine,” “Venice turpentine,” “Bordeaux turpentine,” and “Canada balsam”). Natural resins consist of repeat units of isoprene and are classified as monoterpenes (2 isoprene units: includes oil of turpentine, lavender oil), sesquiterpenes (3 units: includes shellac), diterpenes (4 units: includes rosin, sandarac, copals), triterpenes (6 units: includes mastic, dammar), and long chain polyterpenes (n units: includes natural rubber). It is outside the scope of this article to discuss them all in detail, and the reader is referred to several excellent references on the history, chemistry, and analysis of these materials (Masschelein-Kleiner 1985; Horie 1994; Mills and White 1994). As with most natural materials, detailed basic books on occurrence, trade, nomenclature, and use date from the end of the last century and the beginning of this one (Parry n.d.).

Natural resins dissolved in oils were generally called “varnish” and show up in fill recipes fairly frequently. Resins dissolved in solvent (called “lacquers” or “spirit varnishes”) usually show up in fill and adhesive recipes as minor additives, possibly because of the expense, inconvenience, and other unpleasant aspects of using solvents. When added to animal glue recipes, for example, as in composition and “diamond cement” (see discussion of isinglass, above), natural resins and varnishes increase water resistance and help to mitigate shrinkage.

In the thermoplastic “mastics,” natural resins are the major constituents. A typical example is the filling mastic for flaws in marble consisting of “yellow wax, rosin, burgundy pitch and a little sulfur and plaster” colored with suitable pigments to match (Spon's mechanics own book 1893, 450).

Only a few of the natural resins seem to warrant greater description here because they show up so commonly in fill recipes.

Rosin (colophony)—Rosin is the solid component of raw pine resin after the volatile “spirits of turpentine” have been distilled off. An alternate name for rosin is “colophony” (from the Greek words for “glue” and “sound”) which refers to its use on bows for stringed instruments. It was a cheap natural resin for many centuries and was used extensively for low-quality varnishes, for composition, and for other filling materials. Culter's cement for filling the hollow handles of silver knives and holding the blades in place was commonly made from colophony, sulfur, and any convenient bulking agent (Seaman 1899).

Pitch and tar—These are inexact terms that may refer to a variety of substances including various naturally occurring minerals (also called “bitumen” and “asphalt”), to the products of the pyrolysis or destructive distillation of both hardwoods and softwoods, or to the less volatile compounds present in fossil coal and crude oil (Langton 1925; Mills and White 1994). All of these materials are dark brown to black in color, sticky, and cheap in comparison to other natural resins, including rosin. They have all been used since antiquity (with the exception of refined petroleum tar) for caulking seams, hafting weapons, waterproofing textiles and cordage, and generally filling gaps and sticking things together. Bitumen fills have been found as ancient repairs to ceramics (Williams 1988). All of these materials can be and have been mixed with a variety of bulking agents. One historic recipe for “repairing defective places in castings” (presumably of iron) consisted of rosin, “black pitch,” and iron filings (Brannt 1886).

Shellac—Shellac is the resin exuded by the scale insect Laciffer lacca (Kerr). It has been used as a coating, adhesive, and binder in various bulked compositions, including those from which photograph cases and phonograph records were made (Katz 1984). Dissolved in alcohol, it has been a popular adhesive for adhering ceramics, stone (Hanna and Lee 1988), and other materials, and sticks of shellac, sometimes softened by the addition of wax or some other softer resin, were used as a thermoplastic fill material (Barthelet 1884; Robson 1988). The first use of shellac as a coating has been the subject of considerable debate, but it was commonly used as a metal lacquer in the 18th century (Dossie 1758). One of the major exporting firms (Angelo Bros 1956) reports in its treatise on shellac that as late as mid-19th century, its chief commercial interest in shellac resin was the water-soluble lac-dye extracted from the crude encrustation called “stick-lac.” Large-scale export of the resin increased rapidly after 1870. This record indicates that shellac as a common and cheap fill material (as opposed to an adhesive) may well postdate the 1870s.

2.4.4 OILS AND WAXES

Oils—There is an extensive literature on oils, including sources that are specific to the needs of the conservator (Mills and White 1994), and a detailed description of all the oils of commerce over the last few centuries is outside the scope of this article. In general it can be said that when oil is called for in fill recipes, linseed oil is so commonly given that it is essentially the only drying oil that conservators need to be concerned with in this context. Linseed oil was ubiquitous and cheap, and it oxidized reasonably rapidly, especially with appropriate metallic salt dryers.

Waxes—Beeswax is generally what is meant when the word is used alone in craftspeople's recipes because, like linseed oil, this was the cheapest and most commonly available wax—particularly before the advent of the petroleum waxes. Wax has been used in various arts since antiquity. It was the primary ingredient of a type of relief ornamentation called Pressbrokaat (press-molded brocade) used to enrich medieval sculpture (Richardson 1991) and panel paintings (Kesner et al. 1993). Most waxes, including beeswax, are suitable for fills that will not be subject to much heat, but they have been used in a variety of sealing and filling applications for many centuries. Barthelet (1884, 38–39) recommends a wax and rosin mixture for filling translucent objects made of ivory, alabaster, and white glass but warns that the mixture is difficult to manipulate and impossible to paint over.

During the 19th century, a large number of other waxes besides beeswax came into commerce. One comprehensive book on waxes (Knaggs 1947) describes these in detail, including waxes from plants (carnauba, candelilla, esparto, ouricuri), insects (shellac wax, Chinese insect wax), whales (spermaceti), and sheep wool (lanolin). Waxes also have geological origins. There are natural deposits of mineral wax called ozokerite in a few parts of the world (Utah and Galicia), and “montan wax” began to be extracted on an industrial scale from lignite in Germany around 1910 (Knaggs 1947). Waxes extracted from crude oil became available after the beginning of commercial oil drilling in 1859. The low molecular weight waxes of larger crystalline size are referred to as paraffin in the United States, and the higher molecular weight waxes with smaller crystalline size are referred to as amorphous or microcrystalline waxes. The chemistry of many of these materials as well as additional references can be found in Mills and White (1994).

2.4.5 EARLY SYNTHETIC AND SEMISYNTHETIC POLYMERS

Even though some of these materials are reasonably old they are not commonly seen as materials for filling or restoration compounds but seem to have been used as moldable materials for creating new objects. Vulcanized rubber was introduced in 1839, the cellulose nitrate called Parkesine in 1855, and Celluloid in 1869 (Katz 1984). They are similar to most filling compounds in consisting of binders (the polymer) and bulking agents.

Cellulose nitrate—Cellulose nitrate dissolved in solvent was termed “collodion” and as such was sometimes used as a binder. By itself, collodion was excessively brittle and was often plasticized with other materials. One “elastic collodion” cement was made with gun cotton (cellulose nitrate) dissolved in ether and alcohol with the addition of Venice turpentine and castor oil as plasticizers (Youman 1876). Celluloid was plasticized with camphor. Cellulose nitrate mixed with wood dust was the original composition of the popular proprietary filling compound called “Plastic Wood.” A series of references to cellulose nitrate used in conservation are given in Horie (1994) and begin in 1899.

Rubber and gutta-percha—Natural rubber and gutta-percha are both derived from hydrocolloid plant latexes. They have identical chemical structures (poly-isoprene) but are stereoisomers of each other, with rubber being cis, and gutta-percha and a related natural latex called balata. Natural rubber, also called “India rubber” or “caoutchouc,” is elastic and rubbery, while gutta-percha is soft only while warm, becoming rigid and flexible at room temperature. These materials were the subject of lucrative trade and fervent technological experimentation during the 19th century, and whole books were written on their properties and the various compositions that could be made from them (Brannt 1900). Gutta-percha is firm enough to have been used as a molding and gap-filling agent on its own but was also mixed with various other resins. A compound called Davy's Universal Cement was composed of equal parts of pitch and gutta-percha (Brannt 1886). Natural rubber tends to be too sticky in heat and too hard in cold, but it can be made harder, more stable, and less sensitive to temperature changes by mixing it with sulfur and heating it, a process that was named “Vulcanization” by Charles Goodyear, who patented it in 1839. The necessity of heating this mixture would certainly have limited its use as an in situ fill. Gutta-percha and rubber were sometimes dissolved in benzene and chlorinated solvents such as ether for use as binders, but these must have been quite objectionable to work with. Both natural rubber and gutta-percha are so prone to degradation through oxidation and depolymerization that very old fills composed of these materials are likely to have reversed themselves over the years. Natural rubber first becomes sticky as it deteriorates, then eventually becomes hard and crumbly, while gutta-percha degrades by falling into chunks and eventually powder.

2.4.6 WATER-GLASS (SOLUBLE GLASS, SILICATE OF SODA)

If glass is made with an intentional excess of alkaline fluxes (usually sodium and potassium), it can be dissolved in water (generally by cooking under pressure). One source states that the substance was first observed by Van Helmont in 1640 (Crookes 1892), but Leland (1896) claims to have found the earliest mention of it in the works of Paracelsus (1490–1541) as a substance described as “destillatio crystalli.” Water-glass had a great many uses during the 19th century, when it appears to have had its greatest vogue, including fireproofing wood and paper, weighting textiles, rot-proofing eggs, making artificial stone and as a medium in a type of mural painting called “stereochromy,” said by one source to have been invented by a Professor Schlotthauer of Munich in 1848 (Spon's Mechanics own book 1893). Water-glass was extensively used as an adhesive (called “mineral glue”) and serving as a binder for fills on a wide variety of substrates. It was alkaline enough to denature casein in the same way that lime would. A cement for meerschaum, which “also forms a mass closely resembling genuine meerschaum” consisted of water-glass, casein, and magnesia or powdered meerschaum (Brannt 1886, 65). Leland (1896, 29) associates sodium-silicate filling materials with B�ttger, a key figure in the European development of porcelain at Meissen and gives “cement of B�ttger” as containing “purified chalk and thick solution of silicate of soda.” He also states that it “occupies the first place as an adhesive for glass, nor is it surpassed as a cement in solid form,” and that a French-language book on repairing he had consulted consisted entirely of water-glass recipes. Despite the popularity of water-glass, it had detractors. Barthelet (1884) warns against its use, stating that it absorbed water and fell apart, was prone to corrode glass (a warning that we can confidently predict from a modern understanding of glass corrosion), was difficult to work with owing to the speed of hardening, and was difficult to remove from surfaces when the joining had not been successful.