Volume 18, Number 1 .... January 1996

Letter to the Editor

To the Editor:

This letter is in response to your invitation for a constructive discussion of the articles on "Climate Control in Museums" (WAAC Newsletter, January 1995 issue). This discussion is prompted by the environmental guidelines announced by the researchers at the Conservation Analytical Laboratory (CAL) of the Smithsonian Institution.

As two American researchers who have been investigating the effects of environmental changes on paintings since 1981, we are concerned that the temperature and humidity changes suggested by the CAL researchers might be considered established by published research. This, of course, is not the case and, in fact, numerous cases can be described which contradict the CAL guidelines.

It is imperative that environmental guidelines be based on a consensus of published research and practical experience. At a minimum, these guidelines must include the permissible magnitudes and rates of variation of temperature and relative humidity. In addition, exceptions to the guidelines should be noted for particularly sensitive materials or conditions. The importance of time and the rate of variation of environmental conditions should be emphasized. For example, many materials which behave elastically in the short term will behave plastically, or creep, in the long term. Also small environmental variations which occur quickly can have greater effects than larger variations which occur slowly. These principles can be illustrated by the following experiments and cases.

A biaxial stress tester for stretched canvas was built by Gustav A. Berger and William H. Russell in 1982, and equipped with automatic data logging in 1984. The testing arrangement, whereby the samples are subjected to controlled environmental changes, closely resembles the conditions to which canvas paintings are actually exposed. The resulting stress in the warp and weft directions is measured automatically, and without interruption, by separate load cells. Thus, it is possible to observe, measure and record changes in tension (stress) in response to fluctuations in the environment as they occur in the center of actual paintings, while eliminating the possibilities of human error or bias (1). These investigations and their results were published in many professional magazines, following peer review.

As an example of an environmental test, Figure 1 shows stress changes in response to a change in RH from 20% to 60% and a temperature change from 10 to 20°C (50 to 72°F)." The sample was drawn from a 50 years old oil painting with a heavy paint layer which has cracked already. The temperature change from 10 to 21°C causes a drop in tension of 70N/m or about 50% of its initial tension. This temperature change, however, amounts to only slightly more than one half of the "permissible range" of 52-88°F recommended by the CAL scientists.

Other tests have shown that a rise in temperature within the CAL guidelines would have caused the tension in the sample to drop to nearly zero. This means that in the sample tested in Fig. 1, at 88°F (31°C), the pressure exerted by the expanding paint film would overcome the tension of the stretched canvas. Consequently, the pressure of the expanding paint film at 88°F would be equal to the tension (stress) previously borne by the stretcher and the canvas stretched on it.

Figure 1

Figure 1 The top graph shows 300 hours of controlled fluctuations of temperature and relative hunmidity. The bottom graph shows the resulting changes in tension (stress) in a sample drawn from a 50-year old painting with a heavy paint layer which is supported by a light-to-medium canvas. These graphs show that small fluctuations in temperature cause much more violent stress reactions than do larger changes in humidity. The small temperature fluctuations of only 3°C (between the 60th and 120th hours of the test) cause considerable stress changes. Yet, a large rise in humidity (38% at the 30th hour) causes a smaller stress reaction than another 3°C fluctuation in temperature between the 180th and 200th hours. These results directly contradict the commonly held belief that canvas paintings are relatively insensitive to moderate temperature changes. Ours was a revolutionary finding in 1982 when first published. In spite of being annually repeated in national and international magazines, it still requires further dissemination.

The elevated temperature causes the paint to soften and expand. Since expansion of the paint is restrained, it becomes compressed with every cycle of rising temperature. When the laminate cools, the paint film contracts but, because it was compressed, it is now shorter and comes under tension which is now higher than it was before. As a result, the paint film either cracks, or previous cracks enlarge. Since paint films are never uniform, a heavy plate of paint would exert an even stronger compression along its edges than would a thin and brittle one. On the other hand, a crack in the paint film would provide no resistance to tension and the full force generated by its contraction would have to be carried by the adjacent film. This creates a stress concentration which elevates the stress considerably above the average. Therefore, cracked or uneven materials need a much higher degree of protection from fluctuations in the environment. Cracks can be found on most old objects. The resulting leverage at the stress concentrations causes even low stress to generate forces capable of damaging the materials due to the plastic deformation described. The effect of environmental changes on the surface and support of any object can be compared to a wire being bent back and forth. The more often the wire is being bent the faster it will break at its weakest point.

To prove that even small environmental changes, such as turning on and off the lights in a gallery, can have a noticeable effect on canvas paintings, the test in Figure 2 was performed. A sample was acclimatized for several days in the environmental chamber under nearly constant temperature and RH. It was then irradiated by a 100W reflector lamp from a distance of about 50cm (20") for a period of 5 to 7 minutes, keeping the ambient temperature and RH constant (2).

Figure 2

Figure 2 shows that this radiation has little effect on the temperature within the environmental chamber. However, it leads to an instant drop of tension in the sample caused by the thermal expansion of the paint. The paint film becomes compressed with every expansion and, upon cooling, it takes up less space than it did before heating. Accordingly, the canvas pulled together by the paint becomes smaller, and the tension in the sample rises. After six such cycles, Figure 2 the tension rises from 160N/m to 187N/m, i.e. about 15%. If this process is repeated too often with samples of old paint, cracks will form because the rise in tension causes the paint film to rupture in a new place or an old crack becomes larger, with a concurrent loss in tension in the sample.(2)

It can be readily observed that even the small protection provided by stretcher bars reduces the cracking of a paint film along the edges of a painting. A stretcher cannot protect the face of the painting from changes in the environment, and can delay only slightly the effect of fluctuations on the reverse. Therefore, the reduced cracking in the areas covered by the stretcher bars is very impressive evidence that fluctuations in temperature cause cracks to develop. In a similar way, panel paintings warp because their face is protected by paint while their reverse is exposed to environmental changes. It has been demonstrated by Buck (3) and by Berger & Russell (4) that when one side of a panel (or the surface of most materials) is covered, changes in temperature and RH lead to compression and eventual curvature of the unprotected side. It is a similar mechanism to the one which leads to cracking and cupping of the paint film discussed earlier in this paper.

Most organic materials, such as wood, textiles, skins, plastic, and paint have their own version of RH, which is called "water absorption". Absorbed water is retained by these materials in proportion to the RH in the surrounding air. Since the moisture content influences both the dimensions and mechanical properties of these materials, their reactions to stress are too complex to generalize (Nathan Stolow) (5).

Let us take an extreme example within the range considered "safe" by the CAL researchers and assume that a panel painted on one side is kept in storage at 11°C (50°F) and 65% RH. It would absorb an amount of moisture from the air which would be in equilibrium with the surrounding air at 65% RH. Figure 3, adapted from an engineering Table

Figure 3

Figure 3 in Reference 6, shows that the water quantity absorbed by cypress wood kept under such conditions would be 17.5% of its dry weight (Point A). If this panel were moved to a room at 32°C (88°F) and 65% RH (Point B1), the percentage of moisture in the panel (17.5%) would correspond to a moisture content in the air of over 80%. The panel would begin to give off water to conform to its equilibrium moisture content which at this temperature is about 14% (Point B). The unpainted and unprotected side would start drying and it would shrink. The unpainted side would come under tension but, since wood resists tension better than compression, the inside of the panel would be compressed. Ehrhardt et al, (7) show that wood yields in compression when the RH is above 70%, and at 80% RH -"(compression set), failure and breakage follow" ( Reference 7, Fig.5, p.22).

The above example describes a rapid temperature change. However, a similar reaction would result at a more gradual temperature and RH change over a period of weeks because a wooden panel painted on one side gives off moisture very slowly. This is shown in Point N (Figure 3) which represents "normal museum conditions" of 21°C and 50% RH. If the panel had been brought from storage where it was kept at 11°C and 65% RH (Point A), into museum conditions it would still be exposed to a moisture content of about 70% RH, a point at which plastic deformation occurs in a restrained sample (7). This is without any other stress concentrations caused by the uneven material of the wood or previous damages.

It is significant that in all of the above cases, changes in temperature had a larger and more decisive effect than changes in RH. Moisture is absorbed by osmosis only slowly but a change in temperature has an immediate effect even on materials which do not absorb water.

It is important to note that all homogenous materials, such as blocks of synthetic materials, glass, etc., consist of an outer layer and an inner core. The outer layer reacts first to changes in the environment while the inner core lags behind and resists the reactions of the outer layer (4).

The thin layer of canvas, wood, and paint which was used in tests can well be compared to the thin outer layer of all materials exposed to environmental changes. Thus, while measuring the interactions of thin layers of support and paint, an insight has been gained into the mechanics of the outer layers of all materials exposed to fluctuations in temperature and RH. If relatively flexible materials, such as oil paint, cause the violent stress changes shown in our tests, then one can expect much greater forces to be created by the expansions and contractions of more rigid materials such as hard lacquer, enamel, glass, etc. which are all known to deteriorate rapidly once their surface is cracked. In some of these cases, water absorption and condensation are likely to occur when a cool object is brought into a warm and humid room. There is a danger of condensation particularly in the presence of hygroscopic salts which are often found not only in old enamels but also in papers, paint, and adhesives (5).

Materials have been well preserved not only in the salt mines used by the National Gallery in London but also in such diverse environments as the caves of Provence, the swamps of Northern Germany, the glaciers of the Alps and Siberia, the Pyramids in Egypt, the caves of Chile, and the Dead Sea. What all of these materials had in common is a stable and unchanging, although often severe, environment. As soon as the caves, for example, were open to the public and to fluctuations in the environment, the materials began to decay.

Reduction of fluctuations in temperature and RH to the lowest level that can be achieved under local conditions works best for most collections. However, such a nearly constant environment must be adjusted to the particular needs of certain objects in the collection, as has been suggested by well known investigators in Europe, Japan, and the USA for many years.

At any given time of the year, the lowest temperature and RH which can be comfortably maintained with minimal fluctuations and which is optimal for most items in the collection seems to be the best and most economical choice. If necessary, these values could follow the seasons by very gradually cropping in winter and rising in summer.

It seems clear that rapid changes, such as produced by cycling radiators and air conditioners cause considerable stress on the surface of many brittle materials, and that a RH variation of 35 to 65% allows no safety margin even according to calculations shown by Erhardt himself. Gradual and infrequent changes within the CAL limits might be acceptable for many objects, but their limits must also specify how gradual this transition should be, how such a gradual change could be accomplished, and whether such changes could be realized and maintained with inexpensive equipment.

Gustav A. Berger and William H. Russell

References
1. G.A. Berger and W.H. Russell, "Deterioration of Surfaces Exposed to Environmental Changes", JAIC, v. 29, no.1, 1990, 45-76.

2. G.A. and W.H. Russell, "Interaction Between Canvas and Paint in Response to Environmental Changes", Studies in Conservation, v. 39, no. 2, 1994, 73-86.

3. R.D. Buck, "Some Applications of Rheology to the Treatment of Panel Paintings", Id., v. 17, no.1, 1972, 1-11.

4. G.A. Berger and W.H. Russell, "Investigations Into the Reactions of Plastic Materials to Environmental Changes. Part I: The Mechanics of Decay of Paint Films", Id., v. 31, no. 2, 1986, 49-64.

5. N. Stolow, "Conservation and Exhibitions", Butterworths (1987), 132-142, 11-14 (Parchment).

6. K. Toishi, "Relative Humidity in a Closed Package", Recent Advances in Conservation, Butterworths, (1963), 13-15.

7. D. Erhardt et al; "The Determination of Allowable RH Fluctuations", WAAC Newsletter, v. 17, no. 1, 1995, 19-23.