JAIC 1987, Volume 26, Number 2, Article 1 (pp. 65 to 73)
JAIC online
Journal of the American Institute for Conservation
JAIC 1987, Volume 26, Number 2, Article 1 (pp. 65 to 73)


Barbara Appelbaum

ABSTRACT—The author reconsiders the “Principle of Reversibility” by clarifying its definition and by examining the variables that make a treatment reversible. Different criteria for different parts of a treatment are recommended. The relationship between cleaning and reversibility is examined. There are degrees of reversibility; even irreversible treatments should be considered in terms of whether they allow for future treatment (“re-treatability”).

The “Principle of Reversibility”1 is one of the basic tenets of the modern field of conservation. The Code of Ethics of the American Institute for Conservation, however, was written mainly with the treatment of paintings in mind. Clearly, varnishing and inpainting are expected to be reversible processes. And, when the Code was written in the 1960s, wax linings were such an improvement over glue linings that we can assume that the reversibility of the lining process was not considered in any way problematic.

Reversibility is still a major criterion of good conservation treatment, one that sets conservators apart from skilled restorers or repairers. The “Principle of Reversibility” is one of the factors which establish our unique intent to project our work into the distant future. Conservators have an obligation to assure to the best of their ability that the condition of an object remain unchanged long after treatment is completed. Knowledge of how conservation materials age, how they interact with the object, and how the object responds to its environment is therefore necessary to fulfill this obligation.

Since the Code was promulgated, there has been little discussion about the meaning of reversibility in relation to treatments other than those of paintings, although the principles in the Code have since been applied to the treatment of a much wider range of materials, including machinery and whole buildings. The range of treatment materials and techniques has also expanded tremendously. It is important, therefore, to examine the idea of reversibility in relation to all parts of a treatment, to many different kinds of objects, and to a wide variety of treatment techniques.

“Reversibility” is often used inaccurately as a catch-all term for a variety of treatment criteria. These include such varied issues as the appropriateness of a treatment material to the aesthetic requirements of the object and the compatibility of a treatment material to the physical requirements of the object. More precise and sophisticated terminology is necessary to produce more realistic evaluations of conservation treatments.


In order to be as clear as possible about the meaning of the term “reversible,” it is important that conservators confine its use to the description of a process rather than of a material. The idea that a material can be reversible is not logical. The incorrect use of “reversibility” to mean “solubility” as applied to resins has however become common. It is certainly tempting to refer to a material like Acryloid�B-72, for example, as a “reversible material” in order to indicate a group of properties which we find desirable. These include chemical inertness, unchanging solubility over long periods of time, durability, and lack of color changes. However, the use of the term “reversibility” to cover all of these properties may produce serious error if it leads us to assume that any treatment using a soluble material is a reversible one, or that any use of an “approved” conservation material constitutes a proper treatment.

In order to avoid this confusion, I propose that Robert Feller's terminology for describing the photochemical stability of thermoplastic resins2 be put into common use and extended to other conservation materials. Acryloid�B-72 and the polyvinyl acetate resins can then be referred to as Class A materials, with a useful lifetime of over one hundred years, polybutylmethacrylate (e.g., Elvacite� #2044) as a Class B material, with a useful lifetime of twenty to one hundred years, etc. This terminology is particularly meaningful because, unlike many terms we take from other fields, it describes our particular needs for long-term stability, requirements much more stringent than those of the industrial fields from which we appropriate materials.

In this paper, I shall reserve the term “reversibility” to denote the property of a treatment that allows a knowledgeable conservator to “turn back the clock” on a treatment. In functional terms, this does not require that the object be identical to what it was,3 only that we can return it to a state where our treatment choices are as broad as they were before the treatment in question was performed.


In order to clarify how reversibility relates to different parts of a treatment, it might be useful to divide treatment procedures into separate parts (these refer to the original treatment under question). These are given in the usual order of procedure: 1) cleaning, the removal of non-original materials either those accidentally deposited (i.e. dirt) or those applied purposely (varnish, inpaint, fillings); 2) disassembly, the separation of fragments, the removal of adhesive repairs, sewing, and mechanical constraints such as linings, patches, mounts, etc.; 3) internal consolidation of structural weakness, including impregnation, consolidation of interlayer cleavage, and the sizing of paper; 4) changes in original materials, as in the reduction of metals, bleaching, or sanding the reverse of canvas paintings; 5) repairs and reassembly, including adhesive repairs, sewing, mechanical constraints such as linings, patches, and mountings; 6) additions, largely cosmetic, as in varnishing, fillings, compensation.


Cleaning is obviously not reversible; the exact material removed cannot be replaced. It is therefore vital that the conservator be sure that the material being removed is not original to the maker of the piece or important to any historic use of it, or that any information that such material can provide is not lost during the process. Cleaning of archaeological bronzes or ethnographic materials must be considered carefully before significant material is removed. However, once it is ascertained that the materials being removed are not a purposeful addition by the artist or user, the cleaning does not detract from the integrity of the piece. If there is a possibility that the removed material might at some time be analyzed to provide information about the history of the piece, it can be saved. Cleaning does not necessarily destroy information.

Even though cleaning is not technically reversible, the capability of reversing the visual effect can be important. An argument of the partial and selective cleaners4 in the recurring controversy on the cleaning of paintings is that varnish layers must not be completely removed so that a thin wash of discolored resin is left on the surface to harmonize color schemes which have become unbalanced because of the different rates of deterioration of the pigments. This is unnecessary, since cleaning is visually reversible, and the technical identity of the material being removed is unimportant as long as we are sure that it is not the artist's design layer. All discolored varnish which can be safely removed should be, since it will usually continue to cross-link or oxidize, causing additional color change and a decrease in solubility. Varnishing and inpainting of losses should be carried out before any decision can be made about the color balance of the painting. If, after an appropriate period of consideration, the color still seems inappropriate, a toned varnish can be applied to the whole or to parts of the painting. This process is actually a step in compensation. The use of a modern toned resin rather than the existing varnish for this purpose will provide less need for re-treatment, since the natural resin will continue to darken, while a Class A conservation material will retain exactly the color that is applied.


Even if an internal consolidant is easily soluble, it is unlikely that much can ever be removed, particularly since objects that need consolidation are by definition so weak that repeated applications of solvent may cause damage. Even if solvent vapors are used, the deeper the impregnant has penetrated, the less likely removal is.5 Even a minor treatment like injecting warm gelatin under loose flakes of paint is not reversible, as there is no physical access to the gelatin lodged between the layers of paint after the paint is laid down. If internal consolidation with easily soluble materials will not be reversible in the future, then what criteria should be applied for the choice of material?

Impregnation is one example of a sometimes necessary but irreversible treatment. As with any treatment, impregnation must satisfy the requirements of aesthetic appropriateness and physical and chemical compatibility, but if it is irreversible upon completion of the treatment, reversibility during the course of the treatment must also be considered. Can drips or pools of impregnant be removed from the surface of the piece before the material sets? What can be done in the course of treatment to adjust the gloss of the surface? How much control is there over the appearance of the piece? What happens if the treatment does not proceed exactly as the conservator had expected?

An equally important question is: what will happen when the piece needs treatment again, particularly if the problem that necessitated the treatment recurs? Can the same treatment be repeated? Can a different material be used with the first one still in place? What can be done with written condition and treatment records to make it more likely that a future conservator can find out what was done? The undertaking of an admittedly irreversible treatment does not absolve the conservator of responsibility for the future of the object, but increases the importance of a factor we might call, for want of a more elegant term, “re-treatability.” The notion of re-treatability is one that is often more helpful in evaluating treatments than the idea of reversibility itself. This is particularly true in the impregnation of badly deteriorated materials, since the treatment strengthens what is left of the object but may not prevent further deterioration of original material, and re-treatment may not be far in the future.


Obviously irreversible treatments like bleaching and structural changes in metals are, properly, subject to a great deal of controversy. An important issue in their use is the degree of predictability the conservator can bring to the choice of treatment methods and materials, and the amount of control that can be exercised in the course of treatment. Issues related to treatment techniques like these perhaps belong to specialists in the relevant conservation field; it is noteworthy, however, that many common conservation techniques (e.g. changes on the pH of paper, sanding the reverse of a canvas support for a painting, attempts to flatten wood panels, etc.) are beyond question irreversible, yet have not often been discussed in these terms.


Although mechanical additions like patches, mounts, and linings should be easily removable, for many objects durability may be a more important consideration. Joint failure in glass or ceramic objects could result in substantial additional damage to the piece, and reversibility may therefore be a lesser consideration. Ease in undoing existing repairs may, however, depend on a lack of durability. We may be able to take apart old repairs, glue linings, etc., simply because they have become very weak. This is not an acceptable standard for modern conservators. Modern techniques should be both durable and reversible.


In this area, the highest requirements for reversibility should be applied. Compensation and protective coatings should be easily removable without removing or weakening structural parts of a treatment.


Once we have standards of reversibility for different parts of a treatment, by what means can we judge in advance the future removability of the materials we add? In order to establish a method for evaluating the reversibility of treatments, it would be interesting to try to define what makes the removal of added conservation materials possible. To a first approximation, these are the main points of consideration: 1) the solubility of the components of the piece under treatment; 2) the solubility of the conservation material; 3) the physical nature of their interface; 4) the amount of material to be removed. The theoretical relation between numbers (1), (2) and (3) is sometimes a simple one: in order to remove a material by dissolving it, it must be soluble in a solvent that does not soften the substrate, and it must be physically accessible to the solvent. Acryloid� B-72, for example, cannot be removed safely from an acrylic emulsion painting if the paint is soluble in the same solvents that dissolve the B-72.

The nature of the interface can be important apart from any question of solubility: if the bond between a coating layer and a substrate is weak, mechanical removal is possible, and sometimes preferable, regardless of the solubility of either material. This is the case with the removal of a deteriorated glue lining from the reverse of a canvas painting. In situations like this, putting the glue into solution is less desirable than mechanical removal, since the solution would penetrate the porous textile, from which it probably could not be removed. Another example where mechanical removal might be preferable is in the case of removing excess adhesive from glass or porcelain. As long as the object surface is strong, mechanical removal may be harmless. Dissolving the adhesive would spread the solution over the surface, making the remnants difficult to remove, and the solvent could penetrate into breaks to soften the adhesive already in position. Easy solubility of an added conservation material may therefore be irrelevant to the reversibility of a treatment.

Solubility is also a negligible factor with a material like gesso. Many gessoes can be softened considerably with cold water. Those made with clear sheet gelatin soften much more easily on the palette than those made with rabbit-skin glue, but when used as a filling material, gesso usually becomes impregnated with varnish, paint medium, or consolidants, making the original differences in solubility irrelevant. Removal is aided to varying degrees by water, but the removal is largely mechanical. Removal is also aided by the fact that dried gesso is more friable than most ceramics and most paint films. Since it has very weak adhesive qualities, its removal tends not to pull off any original surface.

Consideration of the amount of material to be removed (4) is an important but often neglected point, thus illustrating the gap in the field of conservation between practice and theory. A small amount of a material can often be removed with a skilled hand and a sharp scalpel. Large amounts of material create quite different problems. If a material is to be removed by dissolution, the dissolved material must be removed without its dripping or flowing into places it is not wanted, like pores or cracks. If a great deal of material is to be removed mechanically, problems of contaminating the object with crumbs and dust may be encountered.


In practice putting a material into solution involves more than bringing a solid and a liquid into physical contact. The statement that one material is soluble in another may be technically correct, but it does not in itself indicate the conditions required for dissolution. It might be useful to look at the ways conservators make adhesive solutions in vitro, and to compare that process with procedures used to dissolve those same adhesives in situ, that is, on the object, in preparation for removal. Researchers who test conservation materials have recognized that there is a difference between chemical or technical solubility and solubility in practice, and have coined the term “removability” for use in a particular testing situation.6

One factor not covered in charts of solubility is the time required for dissolution. Making a resin solution in the laboratory, for example, often requires several days. Some objects can be soaked, or at least exposed to fumes for long periods; most cannot. The range of times available for safe exposure of certain objects to solvents is so limited that the actual chemical solubilities of materials may be irrelevant to conservation treatments. The time ranges that are involved in many treatments are very limited compared to procedures common in chemistry laboratories. On the other hand, these time limitations allow us the safe use of solvents that, technically speaking, could dissolve the material of which the object is made.

Another factor in producing a solution is agitation, probably because it helps promote removal of the dissolved surface layer, which in turn provides better access by the solvent to the as-yet undissolved material beneath. Yet the amount of abrasion to the substrate caused by agitation can make such a procedure harmful. Normal cleaning procedures involving the removal of resins from the surface of an object may require significant amounts of friction to shorten the time the resin takes to dissolve. Anything sensitive to abrasion, like the surface of soft ceramics, rigging lines on ship paintings, or very lean contemporary paint films, makes us aware of the potential danger of even a small amount of abrasion in our cleaning procedures.7 A common mistake of conservation students when finding small amounts of color on their swabs during cleaning tests on lean paint films is to assume that they are dissolving the film rather than abrading loosely bound particles. A dry swab may remove the same amount of color. Dissolving resin films off the surface of extremely abrasion-sensitive objects by simply dripping a solvent over the surface and wicking up the liquid will remove some resin, but not as much as the use of a cotton swab. On the other hand, the removal of a resinous coating does not necessarily entail the complete chemical dissolution of the resin. In most cases putting only a small percentage of the resin into solution is enough to break it up so that it can be wiped away; resins which swell rather than dissolve are also removable. Many removals are actually combinations of softening or breaking up a material with a solvent, and mechanical removal.

Another factor which promotes the solution of materials is heat. Preparation of starch paste or a gelatin solution requires elevated temperatures. Few works of art can withstand the range of temperatures necessary to dissolve gelatin or make starch paste. Fortunately, these materials often soften enough with moisture to be mechanically removed; removal may be aided by temperatures significantly lower than those used in adhesive preparation. However, conservators who do not specialize in works on paper may be surprised that the removal of starch paste linings may require prolonged immersion in water, and that quite hot water may be needed. The wide reputation of starch paste as a “safe” adhesive does not imply ease of removal. However, paper is so sensitive to materials in its surroundings that the chemical compatibility of starch paste with paper and its long-term stability are overriding criteria for the choice of adhesive.

Heat may be useful in the removal of some materials where heat was not used in their formulation or application. Because of the relationship between the ease of solubility and the second-order transition temperature,8 it may be that slight heating of a resinous coating would increase the rate of penetration of the solvent through the film, and therefore, the speed of dissolution. In my experience, slight heating is not a common tactic in removing difficult films, but it may be one that should be tried more often.

The relationship between reversibility and solubility in polyvinyl acetate emulsions is a controversial topic. There seems to be no agreement in the conservation literature on the actual degree of solubility of these materials.9 The large number of ingredients used to formulate proprietary emulsions, individually untested by conservators, results in a startling variety of softening temperatures, pH, and other properties10 and makes the understanding of these properties extremely difficult. For practical purposes, dried films can be softened in a wide variety of solvents, but never dissolved to form a liquid of low enough viscosity to make removal easy. They tend, even when softened, to remain sticky, so their safe removal from fragile surfaces is virtually impossible. The removal of softened emulsion from the edges of a soft or grainy ceramic almost inevitably involves some loss of original material.

There seems to be a great deal of confusion about why polyvinyl acetate emulsions are often difficult to remove. Some of the difficulty of removal is due to the nature of the resin, not usually to cross-linking11 the resins used in formulating PVA emulsions are of a much higher molecular weight than the polyvinyl acetate resins conservators ordinarily use, so that their properties are quite different. Difficulty in removal is also caused by changes in behavior with time, due to the loss of water and other volatiles rather than to changes in the resin. The temperature required for heat-seal bonding when an emulsion film is touch-dry is therefore far lower than that required when most of the volatile materials have evaporated from the film. Color changes seem to be due to materials other than the polyvinyl acetate resin, and are not associated with cross-linking. For all these reasons, equating cross-linking, discoloration, and loss of solubility in emulsions prevents a realistic understanding of their properties and appropriate uses. Because the properties of dried emulsion films continue to change for several years after application, the later removal of heat-seal emulsion linings, unlike the removal of heat-seal resin linings, can be much more difficult than expected.

Acryloid� B-67 is another material where the assumption that difficulty in dissolution is due to cross-linking leads to serious errors. The glass transition temperature of B-67 is above room temperature. At room temperature it has a glassy dense surface. Removal with xylene may be difficult, not because the resin is insoluble—it is not—but because the surface is relatively impermeable to solvents. Slight warming should remedy this problem. Acryloid� B-67 is a fine conservation material, one which, I believe, is under-used due to misunderstandings of its solubility characteristics.

In short, it can be extremely difficult in a given case to project our knowledge of the behavior of newly-applied materials into the future and predict the reversibility of any particular conservation treatment. Only by providing a wide margin of safety and using materials which fulfill the most stringent aging tests can the reversibility of a treatment well into the future be assured.


Reversibility is not a simple “yes” or “no” proposition. Within the wide range of treatments of which the results can be undone, there are degrees of reversibility, depending on how much time and trouble are involved, and on what risk it poses for the object, since a troublesome or time-consuming job for the conservator almost inevitably involves an ordeal for the object under treatment. How much flexibility for the future is built into the treatment? Is it possible to undo some of the treatment without undoing the rest? Specifically, can external supports, like patches or linings be removed without weakening internal consolidation materials?

The choice of auxiliary support in the restretching of a painting is an example of a treatment procedure which offers alternatives similar in other respects but differing in their degrees of reversibility. If a painting stretched on a stretcher were scratched or dented, ease of access to the reverse would usually permit local treatment. However, if the painting had been stretched around, or adhered to, a solid support, complete removal would probably be necessary to allow access to the reverse. This could entail a major treatment instead of a minor one. The sewing of textiles to a stretched fabric is analogous. With time, particularly if a textile is on exhibition or stored vertically, the textile may stretch slightly and “belly out” from the support. Small fragments may come loose. If the textile has been kept on its mounting strainer, more sewing can be done easily. If, however, the mounting fabric has been cut loose from its strainer and wrapped around a solid support, sewing can only be done with a curved needle, if at all, a procedure extremely tedious and quite stressful on the object. The ease of re-treatment could make reversing the original treatment unnecessary, thus avoiding major stress on both the object and the conservator.

Ease of reversibility is also an issue with textiles because the mounting is both an aesthetic setting (equivalent to the mat of a work on paper or the frame of a painting) and a structural support (equivalent to a lining). As exhibition conditions, ownership, or styles change, the color or texture of the mount could become objectionable, but changing the appearance could necessitate complete removal and redoing of the treatment. Although there are seldom technical problems involved in un-sewing a textile, it can be extremely time-consuming and therefore extremely costly. The removal of sewing threads can cause significant powdering of the original, and the handling required can cause additional loss. The need for re-treatment in this case can cause exactly the kind of damage that the first treatment was designed to avoid. In general the reversibility of sewing as a treatment needs some critical re-examination.12

Proper consideration of the principle of reversibility in a particular treatment will answer the following questions: what is the relationship of the solubility of the added material to its removal? If it cannot be removed by dissolution, are its physical properties different enough from those of the original to make mechanical removal possible? If a treatment is reversible, how difficult, expensive, time-consuming, or risky to the piece would the process be? If removal of an added material is impossible, is the piece re-treatable with the same or different materials? To what extent can aesthetic changes be made without undoing the structural aspects of the treatment? What future events might necessitate re-treatment, and how could they be handled most efficiently? Conservators' attempts to project the behavior of complex systems far into the future is, as seen above, extremely difficult. As well-informed as we try to be, we can only guess at the pitfalls of our treatments. It is for this reason that reversibility is such an important concept. Our obligation is to make treatments as easily reversible as possible.

As important as the concept of reversibility is in the modern fields of conservation, it does not necessarily have a direct connection with the propriety or advisability of a treatment. An easily reversible treatment may damage an object, and an irreversible treatment may be the best under a particular set of circumstances. Many desirable attributes of a conservation material in a particular treatment do not relate directly to reversibility, but to other issues entirely. Some are concerned with the compatibility of added materials with those of the original. Such properties include response to changes in temperature and relative humidity, development of physical stresses from shrinkage, and the production of potentially harmful byproducts of deterioration. Possibly the most important criterion in judging a treatment is whether it provides the help the piece needs; in medical terms, whether it cures the disease. This must be judged on a case-by-case basis.

Even if a treatment is, unavoidably, not reversible, the conservator is not absolved from responsibility for the future of the piece, or to those who must treat it in the future. The fundamental reason we do our work is to insure that the pieces we treat will last forever. Therefore, unless it is destroyed first, every piece we treat will be treated again, and some provision must be made for future treatment. Particularly when novel or complex treatments are proposed, we have an obligation to future custodians to consider in detail the choices they will have to make when they deal with the products of our labors.


1. AIC Code of Ethics, section II. E.: “PRINCIPLE OF REVERSIBILITY. The conservator is guided by and endeavors to apply the ‘principle of reversibility’ in his treatments. He should avoid the use of materials which may become so intractable that their future removal could endanger the physical safety of the object. He also should avoid the use of techniques the results of which cannot be undone if that should become desirable.

2. R. L. Feller, “Standards in the Evaluation of Thermoplastic Resins,” Paper presented April 16, 1978, Fourth Triennial Meeting, ICOM Committee for Conservation, Zagreb. The following is adapted from this article:

3. It has been noted (C. V. Horie, “Reversibility of Polymer Treatments,” p. 3–2 in Resins in Conservations, Proceedings of the Symposium, Edinburgh, 1982) that it is likely that no treatment is reversible on the molecular level. When traces of modern materials might interfere with sophisticated analytical methods, this is significant. In terms of future treatments, however, this caveat can be disregarded.

4. Gerry Hedley, unpublished article, “On Humanism, Aesthetics and the Cleaning of Paintings,” January, 1985.

5. C. V. Horie, “Reversibility of Polymer Treatments,” pp. 3–1 to 3–6, in Resins in Conservation, Proceedings of the Symposium, Edinburgh, 1982, The Scottish Society for Conservation and Restoration. As a test of the reversibility of consolidation treatments, modern earthenware was impregnated with polymethyl methacrylate. It was then washed in acetone in a Soxhlet extractor for eight hours. About 50% of the resin remained.

6. R.L. Feller, M. Curran, “Changes in Solubility and Removability of Varnish Resins with Age,” AIC Bulletin 15#2 (Summer, 1975): pp. 17–26.

7. One kind of object which illustrates a difference in reactivity of a material with and without abrasion is plaster sculpture. Cleaning the surface of plaster with damp cotton can produce easily observable loss of surface detail. However, long-term soaking of plaster sculpture in water is a common approach to the treatment of these pieces.

8. Personal communication with Robert Feller.

9. Jane L. Down, “Adhesive testing at the Canadian Conservation Institute, past and future,” IIC Paris Conference, 1984, p. 20: “the polyvinyl acetate emulsions…are insoluble.” Rachel Howells et al., “Polymer dispersions artificially aged,” IIC Paris Conference, 1984, p. 39: Changes in solubility are discussed, although the authors note, “Strictly speaking, the method [of testing] assesses removability rather than solubility.” Many other authors (e.g., E. De Witte et al, “Influence of the modification of dispersions on film properties,” IIC Paris Conference, 1984, pp. 32–35) do not make the distinction. Removability as these material scientists have defined it in a laboratory setting is, of course, very different from removability in a practical treatment context.

10. Rachel Howells, ibid.

11. R. L. Feller, “Polymer Emulsions,” Bulletin, IIC-AG 6 #2, (May, 1966), p. 27.

12. See Paul Himmelstein, “A Re-examination of Sewing Used in the Treatment of Textiles,” pp. 33–34, and Pat Reeves, “Re-examining Textile Conservation Techniques,” pp. 35–38 in Textile Treatments Revisited, The Harpers Ferry Regional Textile Group, Symposium, November 6 & 7, 1986.

Section Index

Copyright � 1987 American Institute of Historic and Artistic Works