Development of humidity recommendations in museums and moisture control in buildings

JP Brown & William B Rose
1997

The following paper is the full-length text of an article which was published, in a somewhat adumbrated form, in the APT Bulletin as: 'Humidity and moisture in historic buildings: the origins of building and object conservation', APT Bulletin, 27 / 3 (1996), 12-24.

1. Introduction

The majority of artifacts in historic houses are made of natural organic materials: wood, leather, traditional adhesives (rendered down animal or vegetable matter), natural textiles, paper products, fur, bone, ivory. Empirical observation of the reaction of such artifacts to variations in relative humidity over the last eighty or so years has become progressively codified into recommendations for 'optimum' humidity levels for the preservation of these artifacts. The central concern of the recommendations is that humidity should not fall to a level which causes brittle failure of organic artifacts, nor rise to a point where mold growth can flourish. In particular, the recommended levels for mixed collections of organic artifacts center on controlling humidity in the range 30 to 70%RH; rather tighter control — typically 50 to 60%RH — has been recommended for valuable objects such as paintings and antique furniture. However, as we shall show, these figures have not arisen from detailed research leading to a clear understanding of the effects of humidity on organic material, but are in fact codifications of pre-WWII custom and practice, modified by the practicality of indoor environmental control in specific climates.

Control of room humidity to a lower limit of 40%RH, while it may be good for artifacts, is a potential source of damage to buildings exposed to cold winters. In most of the northern hemisphere above 35° N latitude, heating to the lower range of modern human comfort levels (18°C/65°F) will reduce ambient internal humidity to below 20%RH for at least part of the winter. The humidity can be restored by using machines to add moisture to the indoor air (humidification), but this procedure may raise the dew point of the internal air above the external air temperature. In this case, when the warm humid indoor air meets building elements cooled by contact with the outdoor climate (walls, roofs, windows), water will condense out of the air leading to rotting of wooden elements, mold growth on interior finishes, corrosion of metal elements, and spalling of masonry — damage which can quite rapidly reduce building elements to the point where renovation must be performed. As we move further north and the winters become generally colder, so the likelihood of condensation becomes greater if the interior is heated to 18°C/65°F and humidified to 40%RH.

The conflict between this museum specification for objects and the need to prevent condensation in buildings in which they are housed is clear, as was recently recognized in the New Orleans Charter.[[1]] The options for resolving this conflict are also obvious:

  1. Lower temperatures and/or lower humidities during winter. If the building has plumbing or wet central heating the practical lower limit for temperature is usually around 5°C/40°F to prevent the pipes freezing and bursting. Heating to this modest level will reduce indoor humidity below 40%RH if outdoor temperatures remain below a level of about. -10°C/15°F. If external winter temperatures generally fall below this level, or if the building must be heated to a higher temperature, it becomes a question of some importance whether the empirically determined lower limit of 40%RH for humidity is correct, and what the effect of excursions below the lower limit will be in terms of the length of time the excursion is maintained and the degree to which humidity drops below the limit. Although the focus of this article is on humidity, we should note that if reduced heating is used as a control mechanism, we also need to know whether the annual temperature variation between indoor conditions in summer and winter will cause damage to historic collections, a question which is moot if temperature is kept relatively constant.
  2. Intervention on the building fabric to mitigate the effects of condensation (moisture engineering). In the US this has usually followed current building codes: insulation, double glazing and vapor barriers. We will return to this option later in the paper.
  3. Climate-controlled display cases to maintain a suitable environment around a humidity sensitive object, regardless of the ambient room conditions. In social history dioramas this option is usually limited to framed two-dimensional objects such as paintings, photographs and samplers. As with option a., microclimate control also begs the question of what degree of control (in terms of level and variation of humidity) is required to prevent damage.
  4. Not displaying humidity sensitive materials in historic houses. We should be clear that not displaying material is a real option — it makes no sense to destroy a building, just so that you can display a particular artifact in it. Again, how are we to predict in advance whether a safe level for the building is also safe for a particular object?
  5. In order to be able to evaluate these options, we must have a clear understanding of the development of the various museum standards for 'optimum' humidity levels for preservation of organic artifacts, and the bases upon which we can opt for various methods of environmental and moisture control. Michalski has traced the origin of humidity standards for paintings.[[3]] His account depicts a standard originally determined by local climate and the feasibility of humidity control at the National Gallery, London in the 1930's and subsequently transferred to all organic materials. The analysis presented by Michalski is accurate but incomplete — word limits were restrictive in the volume where Michalski's paper appears and his remit much broader than an historical survey. How, over the years, did humidity standards for organic artifacts in museums arise? And how have building professionals coped with elevated indoor humidity? Are the methods and standards which developed into current practice based on an intelligent program of research leading to successful results in practice?

2. Environmental Performance of Early Museum Buildings

The development of museum specifications for humidity must necessarily take account of the buildings that they developed in since, prior to WWII, building form was the primary determinant of environmental performance. For centuries, moisture control in buildings meant ensuring the safe discharge of rainwater away from the building; heat was provided by burning wood or coal in fireplaces, the air rising up the chimney sucked air into the rooms through whatever path was available (under doors, through chinks in window frames) creating draughts; artificial lighting was confined to candles, flambeaux and torches, and later oil lamps so large windows and skylights were the norm to provide adequate daylight. Public building design centered on massive masonry construction, mostly for reasons of prestige, but also reflecting a concern to conserve heat during the winter and to reduce thermal gain during the summer. In the nineteenth century, two fundamental changes occurred: the supply of hot water or steam to radiators through pipes from a central boiler provided draught-free heating remote from the combustion site (furnace or boiler), and the introduction of piped coal gas for lighting increased the level of indoor illumination in urban buildings. Centrally heated buildings could be more airtight, but this advantage was largely offset by the soot and sulfur compounds (and the additional heat load) generated by the burning of coal gas which made the internal environment dirty (soot collected above gas jets and on walls), toxic to humans (complaints of headaches were common) and dangerous to organic materials (the sulfur compounds rotted textiles, leather, and paper and corroded metals). In addition, overcrowding of well-sealed rooms led to 'vitiation' of the air. Thus, design for ventilation of public buildings became a priority and, in the absence of efficient mechanical fans, air was ducted into the basement and heated by passing it over batteries of hot pipes, the resulting buoyancy driving air up through a planned route of interior ducts and grilles to heat the rooms and exhaust stale air.[[4], [5]] The level of winter heating of the internal air was not primarily determined by comfort, but by the need to provide enough buoyancy to drive the ventilation system (typically 55 to 60°F). In the 1920's, heating museums to a set point around 13°C/55°F during the US winter was considered 'warm enough for moving visitors with the additional advantage that '...[i]t also largely relieves the museum of the necessity of checking wraps at the door.'[[6]: p.114] In the summer, windows or vents in the upper galleries were opened and, on sunny days, the rising of air heated by solar radiation sucked air through the building. Solar gain was controlled, if at all, by slatted wooden shutters, or by whitewashing the pitched surface of skylights during early Summer, the action of rain gradually removing the coating by Fall. [[7]]

From the mid 19th to early 20th centuries, a composite model of such a museum building would include:

Such buildings were able to withstand the modest wintertime humidification caused by respiration and perspiration of the occupants quite well, as can be seen by an evaluation using contemporary moisture analysis. If site drainage was good then basements did not become excessively moist, with the associated contribution to the building's moisture load. The layers of paint reduced penetration of indoor moisture into the wall. Radiators were placed under windows, both to prevent condensation (by raising the temperature of the indoor surface of the window), and also to prevent the staining of paint surfaces (by dirt caught in the radiator's convection currents) which resulted when radiators were placed against walls. The principal air leakage site was around the window frames, and these leakage sites were heated, indeed overheated, by the radiators; escaping air encountered warm surfaces rather than cold surfaces along the leakage path, and this ensured moisture safety. The de facto standard for the moisture performance of building envelopes was window condensation. In 1912 Lyle stated:

'Frosting [i.e., frozen condensation] on the windows obstructs very little light, while it does interfere with vision, but this is usually not of much importance. When condensation occurs in sufficiently large quantities to run down on the window sill, it is objectionable and the best cure is to provide double windows, as a humidity of 35 per cent. to 40 per cent. will condense and flow on a window which the wind strikes on a cold day, viz., 10 deg. above zero F. or below.' [[8]]

The criteria expressed by Lyle can be found scattered throughout the subsequent heating and ventilating literature:

By 1930, charts were available which showed allowable indoor relative humidities for a given window construction and for any temperature drop across the window.[[9]]

The overheating of the masonry walls reduced the likelihood of water damage to the masonry in several ways: by helping to dry out any rainwater accumulation, by reducing the likelihood of window condensation penetrating the materials below, and by reducing capillary movement. Aside from the attic area, there were no cavities with cold surfaces to invite interstitial condensation or mold growth. The poorly insulated attics were overheated more than they were cooled, and they were open to the outside by the use of roofing materials such as tile on strip supports or provided with grilles to improve ventilation. These conditions tended to prevent attic condensation. If there were skylights (as there often were in museums), they might have needed additional radiant sources between the glass roof and the laylight: 'this will not only aid in tempering the upper part of the room [i.e. gallery] but in melting snow which forms on the outer skylights and will prevent the condensation of moisture on the diffusing skylights below.' [5: p.114]

From a humidity point of view the situation was less satisfactory. The massive masonry construction was unheated from late Spring to early Autumn. Where the building was cool, indoor humidity would rise as the outdoor warm air came in contact with the building mass, e.g. against walls on which pictures were hung or in basements. This would not be true for rooms with skylights; in this case solar gain raised indoor temperatures above outdoor levels on sunny days. However, during the night the air would cool (the skylight being a poor insulator) and indoor humidity would rise. Conditions in most museums in Europe and north America would have been between 60 to 90%RH during the summer. In winter the heating was switched on and the internal RH would fall to 30, 20, or even 10%RH, depending on the winter climate and the degree of heating. For naturally ventilated historic buildings the situation in summer would be the same; in winter, assuming heating by fireplaces, ambient winter humidity would remain fairly constant in occupied zones, typically in the 60 to 70%RH range, but window condensation would be a problem.

2.1 Mechanical control of the indoor environment

In the 19th century the internal and external urban environment was highly polluted by the burning of coal and coal gas. As electric lighting was introduced (1882 - 1910's), so the level of internally generated pollution from gas lighting fell and the major source of pollution became the infiltration from the outdoor, still highly contaminated by soot and sulfur from burning coal. In ducted ventilation systems there was the potential to improve the internal environment by filtering the pollutants from the outdoor air before it reached the public areas of the building. Early attempts to purify supply air used large settling chambers, but by the turn of the 20th century water sprays were being used to wash the soot, sulfurous tarry particles, and sulfurous gases out of supply air. A full list of pre-WWII museum literature mentioning dust and dirt as a problem in museums would be extremely lengthy; suffice it to say that air washing was exceedingly attractive to museums.[[10],[11],[12]] The potential of these 'air washers' for humidification was obvious. What was less obvious was that by controlling the spray temperature to a level below the wet bulb temperature of the intake air, air washers could also be used to reduce high summer humidities. In 1906 Willis Carrier patented this principle and set the modern air conditioning business in motion; by 1910 the theory of humidity control for air conditioning was well established in the US literature.[[13]] The early developments in central control of humidity were applied in Europe and the US, but were generally confined to large unitary industrial buildings (textile mills, printing works) where stable humidity was important for efficient production and considerations such as the location of plant, and size of ductwork, were irrelevant.[[14]]

In the 1910's the practical difficulties which had to be overcome in air conditioning non-industrial buildings were formidable: the inefficiency of the available refrigerants and compressors were crude made cooling the water spray expensive; supply pressures were low and no one had come up with a way of making sharp turns in ductwork without considerable pressure losses so ductwork was bulky, making retrofitting unattractive; control of supply on a room-by-room basis for multi-room installations was not yet possible. All these problems would be solved in time, but not fully until the 1950's, and then only at a significant energy cost. On the other hand, the use of air washers to humidify air in existing ductwork to prevent the humidity dropping during winter heating, and the design of new buildings to incorporate this feature, was immediately possible. According to McCabe, the Huntingdon Building of the Boston Museum of Fine Arts installed a central humidification and air washing system in 1908, and, after a two year experimental period, 'it was found that the humidity best adapted for paintings and other works of art ranged from 55 to 60 percent regardless of the temperature or the time of year.' [[15]: p.7] This is the earliest numerical specification for museum humidity that we have been able to locate. The Cleveland Museum of Fine Art followed suit in 1915 or so, with double-glazing to control condensation, and humidification set at 50 to 55%RH; according to McCabe, this slightly lower level had been experimentally determined as optimum for the art collections in Ohio. Summer dehumidification was not in place in Cleveland by 1931, 'partly because the initial installation is so expensive and because the apparatus would be needed for perhaps only two months of the year' [14: p.7]

McCabe notes the expense of running the system, the need to run it continuously to stabilize conditions, the desirability of automatic controls, the importance of close cooperation between engineer, curator and architect in integrating the plant into the building scheme, the advantage of zone control for different collection types (midpoint humidity was causing rust on armor displayed at Cleveland), and the importance of continuous maintenance for the proper functioning of the system. McCabe achieved control by using fans and ducts to re-circulate as much air as possible — windows were kept closed and infiltration was reduced to a minimum — presumably using the building structure to help buffer the humidity changes; the penalty for this high level of recirculation and tight control of infiltration is highlighted by McCabe's desire for 'a machine to introduce ozone or radi-ion [sic] to the air to prevent its becoming rancid.' [14: p.7]

The experimental basis of the 'best humidity' figures given by McCabe is obscure — he does not describe the experiments and there is no mention of them in the Annual Report of the Museum of Fine Arts, Boston or in the Annual Report of the Cleveland Museum of Fine Art, although the latter publication reports the work and budget of the facilities department. Further, there is no clear difference between the climates in Boston and Cleveland to suggest that the levels were determined by the capabilities of water-spray systems. Whatever the basis of McCabe's experiments, his views were influential in forming the early basis of museum humidity control during the 1930's.

3. Developments in Museum Practice During the 1930's and 1940's

3.1 'Optimum' humidity for organic collections

In the UK in the late 1920's there had been preliminary work to produce a design for installing full air conditioning in the National Gallery, London (NG-L), but the cost of £5,000 — mostly for installing ductwork — proved prohibitive.[[16]: p.10] During the exceptionally dry year of 1929 there was an increase in blistering and flaking of paint from panel paintings (oil paint on a gesso ground over a wooden panel) in the NG-L's collection. The problem was thought to be due to the warping of the wooden panels by reduced humidity and was referred to the UK's Forest Products Research Laboratory (FPRL) to find the best way of stopping panels warping. The FPRL ran a series of experiments in the early 1930's on sections taken from original panel paintings to determine the cause of the flaking. The panel sections were subjected to long-term humidity fluctuations (40 to 90%RH in 5%RH steps and back again at 10%RH steps; 25°C; one cycle lasting four months), rapid humidity fluctuations (40 to 80%RH and then back in one step; 40°C; ca. 26 cycles, steady states at each extreme lasting between two days and 6 hours); and forced bending without a humidity change to the same level of warping caused by extreme humidity change (about five bends per minute; each deflection equivalent to the maximum warp occurring during the long-term humidity test). The results indicated that long-term humidity change caused cracking and flaking, many short-term humidity oscillations caused less damage (at least partly because the high temperature at which the test was conducted increased the flexibility of the paint film, also because the changes were too quick for the wood surface to react fully), and simple bending without humidity fluctuation caused very little damage. The results were confused by the limited number of samples, and the only definite conclusion drawn was that a moisture barrier applied to the unpainted side reduced warping — the mechanism of paint flaking was still unclear, but was clearly related to long-term humidity changes and not a simple function of the bending of the wood. Sadly, these intriguing results were not followed up — the museum scientists fixated on warping in panel paintings, presumably since a warped panel was unsightly.

The FPRL report was republished by the Courtauld Institute of Art in 1934 or 5, together with an appreciation of humidity control by MacIntyre (HM Office of Works engineer).[15] MacIntyre's section explained the temperature and moisture dependence of RH, gave an unambiguous description of the limitations of the recording thermohygrograph for the accurate measurement of humidity, noted that air washers would provide control between 60 and 90%RH, and showed some understanding of the effect of the building construction in absorbing and desorbing water. He echoed McCabe in explaining that refrigerant cooling for control of summer humidity control was desirable, but that the capital cost was too high. MacIntyre was aware that low humidity retarded the fading of watercolor pigments [[17]], speculated that low humidities might also be beneficial for pigments in oil media, and noted that, at the current state of knowledge, 'optimum conditions for any particular class of paintings must be more or less arbitrary.' [15: p.16] Nonetheless, he 'arbitrarily' proposed figures of 55 or 60%RH for paintings on wood and canvas, and a temperature of 60°F. He gave no clear reason for these particular humidity figures; perhaps they were based, as Michalski has suggested [2: p.624], on the levels of indoor humidity control that could be achieved in London (the text is open to interpretation), but the similarity to the figures arrived at by McCabe in America should be noted. It seems likely that, as with McCabe, the feasibility argument derives not from any peculiarity of London's climate, but from the level to which indoor climate could be controlled using unrefrigerated air washers. The temperature is more readily explained: it was simply the lower limit for winter heating at the National Gallery.

In 1934 MacIntyre had already installed a primitive HVAC system in the Orangery of Hampton Court Palace (where Mantegna's cartoons had suffered severely from conditions varying 55 to 80°F and 25 to 90%RH) to attempt to limit humidity to 55 to 75%RH with a temperature not lower than 60°F during the winter.[[18]] Since refrigerant cooling was too expensive, and heating could not be used to reduce the high summer humidities without discomfort to visitors (MacIntyre had engaged in a somewhat acid correspondence with Rosenberg, the Conservator of the National Museum, Copenhagen, on the inadvisability of heating to control summer humidity to below 65%RH [[19],[20]] ), the final design involved heating, humidifying, and re-circulating air through hygroscopic material (1,740 pounds of canvas hose) to stabilize the indoor humidity. We cannot find any record of the long-term success of the experiment (one suspects that the canvas would have become putrescent eventually), but it is probably one of the larger passive dehumidification schemes attempted in a museum context.

Some attention was also being paid to antique furniture. In 1931 Brewster, a 'lumber utilization engineer', published a very thoughtful account of the need for humidification for historic European furniture imported to the US. [[21]] The account was buttressed by the latest research from the US Forest Products Laboratory (USFPL) which had measured the changes in samples of wood exposed to different indoor environments throughout the US:

'Atmospheric humidities, both out-of-door and indoor, are much higher in Western Europe than in much of the United States, and the less efficient methods of heating generally employed [in Europe] result in considerably lower indoor temperatures during cold weather than are customary in our homes, offices and public buildings. Consequently, old furniture and woodwork comes to a balance with atmospheric humidities in European buildings at moisture contents that are rarely under 9 per cent and may be as high as 14 or 15 per cent... Woodwork in this condition cannot be subjected to the low atmospheric humidities characteristic of the winter time [in the US] without an appreciable amount of shrinkage, which is often sufficient to cause damage.' [20: p.66]

Brewster produced figures to show that the problem was an order of magnitude greater in Madison (Wisconsin, cold winters) than in San Francisco (California, warm winters), noted the time lag for dimensional response of wood to changes in atmospheric humidity, summarized recent research which showed that the time lag was increased by surface coatings such as varnishes, and put the lower safe limit for imported antique wooden objects at 40%RH. This level was empirically determined: the humidity was raised by humidification to 40%RH during winter heating and the damage stopped. Brewster also noted that humidity might have to fall to 35%RH to prevent condensation during very cold spells (note the similarity with Lyle, above). 'Fortunately, extremely cold weather is usually of short duration and the use of a higher humidity between cold spells will usually be sufficient to prevent any harmful shrinkage in furniture which is protected against rapid moisture changes by paint or varnish.' [20: p.68] It is important to emphasize that Brewster was discussing a class of composite organic artifacts, acclimated to high humidity and brought into contact with sustained low humidity. In this context, his lower limit of 40%RH, with brief excursions to 35%RH, makes sense.

Further research directed towards establishing the optimum humidity for panel paintings was conducted at the NG-L, in the 1930's. Samples of four different woods were left to equilibrate in the galleries and their moisture content was monitored by weighing — a technique which paralleled the USFPL work cited by Brewster. The work confirmed Brewster's conviction of higher moisture contents in Europe: the annual average moisture content for the samples was found to be 11% and it was therefore determined, from the EMC-RH curve for the woods, that this moisture content indicated that the average humidity had been between 55 and 60%RH.[[22]] It is worth pointing out that the research had not in fact determined the 'optimum' conditions for the panel paintings, but rather the average annual moisture content to which the wooden substrate was acclimated in an uncontrolled indoor environment (high summer humidity, low winter humidity during heating).

By 1936, a review article by Coremans, head of the Central Laboratory of Belgian National Museums, stated unequivocally that 'Experiments which have been carried out in various places seem to show that the optimum temperature and humidity is very close to 15°C (59°F) and 60 per cent relative humidity.'[[23]: p.341] The 'experiments' are not described, and the only paper in Coremans's bibliography which mentions experiments on artworks is a French translation of the McCabe's article which, as we have pointed out, gave no details. The 59°F figure is close enough to MacIntyre's lower winter limit of 60°F for Mantegna's Cartoons (also cited by Coremans) to be explained as a result of transposing this lower limit from MacIntyre's Fahrenheit to a round number in the continental Celsius, but why Coremans felt this temperature was optimum for year round control is a mystery: as he freely admitted, control to 60°F and 60%RH was not feasible in European museums during the summer — the humidity was too high and the cooling load too great. Nonetheless, this conviction of 60°F and 60%RH as the 'optimum' environment was dominant in European literature during the late 1930's and early 1940's.

With the outbreak of the Second World War, mixed collections from the British Museum and the Victoria and Albert museum were removed to quarry storage at 60%RH, 60°F and survived the war in very good condition.[[24]: p.2] The National Gallery's collections were evacuated to storage in the Manod slate quarry, Wales.[[25]] Conditions in the slate caves were a constant 47°F and 95 to 100%RH; brick shelters were built in the caves to house the paintings [[26]] and 'simple heating' was used to control the climate in the shelters to 58%RH, 63°F/17°C with 'exceedingly minor variations.'[21: p.194] Prior to the removal to Manod, a technician had been employed eight months out of every year to deal with cracking and blistering in the panel paintings; during the first year of storage in Wales (1941- 1942), his work reduced to one month and by 1945 he had nothing to do; after the War ended, the paintings were returned to the essentially uncontrolled environment of the National Gallery, and an epidemic of blistering, warping and cracking broke out.[21] This very positive experience of climate-controlled quarry storage contrasted powerfully with the experience of evacuating collections to bomb-proof storage in the tunnels of the London Underground (subway) system during the First World War, when uncontrolled damp had caused serious damage to collections.[[27]]

After the war, the Weaver Report on the cleaning of pictures in the National Gallery was commissioned by the British Parliament, and reported in 1947 that not only would the use of full air conditioning be desirable for filtration of soot from the gallery air (to reduce the need for cleaning), but also that it was pointless to restore pictures if they were subsequently damaged by uncontrolled relative humidity fluctuations. The story of the Manod Technician was cited as striking evidence of the importance of humidity control to 60%RH.[23] The Weaver report, in fact, contains two problematic conclusions. First, the report generalized from the improvement in the condition of panel paintings stored at Manod, to specifying the humidity conditions necessary for paintings as a whole — lumping the oil paintings in with the panel paintings — even though the 55 to 60%RH set-point had been determined by analogy to the substrate of the 'delicate' panel paintings, rather than the more 'robust' oils on canvas. Second, the report had implicitly concluded that the improvement in condition was owing to maintaining the level 55 to 60%RH, ignoring the other possible conclusion that it was the reduction of variation in RH to a nominal ±2.5%RH which provided the benefit, or that stable temperatures were beneficial. This level of control was relatively easy in the cave shelters at Manod, but it was clearly going to be problematic in a nineteenth century museum building such as the NG-L. The questions of whether or not some other level with a similarly narrow range would be just as good, whether other artifacts needed such close control of variation as panel paintings, or even whether panel paintings were especially 'delicate', were left to one side.

There were probably two reasons for this. First, it was believed that control of indoor humidity to within 55 to 60%RH to duplicate the quarry storage conditions would be practical by retrofitting the National Gallery with air conditioning. The second reason was that, in the words of Keeley and Rawlins, 'Great importance has always been attached to the advantages of removing the glass from paintings, thus eliminating the irritating reflections.' [21: p.195] There were, however, urgent practical objections to removing glazing from the paintings. In the first place, there was still the problem that the outdoor urban air was highly polluted. This was at a time in Britain prior to the passage of the Clean Air Act of 1956. Average levels of carbonaceous matter London during the inter-war years were around 2 pounds per million cubic yards of air (ca. 1200 µg/m3), and the indoor air in the NG-L's galleries contained about two thirds of the outdoor level — the air in the picture galleries was noticeably foggy on occasion.[21] Glazing and backing the paintings protected the paintings from soiling by these pollutants, and also from harmful accumulations of acid by-products of burnt coal, a fact established almost 100 years previously by a committee including Michael Faraday.[[28]] In the second place, glazing and backing produced a microclimate within the frame which reduced the impact of oscillations in ambient relative humidity — the very oscillations which the National Gallery now understood to be so harmful. It was already established that a well-sealed enclosure tended to reduce outdoor humidity fluctuation, that additional hygroscopic material increased the buffering power, and the use of silica gel as a desiccant and as a buffer had been proposed.[[29]] Thus, since it was believed that air conditioning would kill two birds with one stone — allow both efficient central filtration of air to prevent acid attack and soiling, and narrow humidity control to prevent mechanical damage without the need for microclimates — the National Gallery was eager to install HVAC so that they could pursue their agenda of Glassabnahme for the paintings.

3.2 Development in the US

The position was rather different in North America. Mechanical refrigeration plant was becoming somewhat more efficient and reliable, and freon (CFC) refrigerants were developed. The bimetallic strip thermostat was in use, as was the hair humidistat. Both provided mechanical action which could make or break electrical contacts to switch machinery on or off and actuate automatic dampers. In 1934 Cherne and Nelson published a practical (though energy inefficient) plan for central supply to different rooms in a building, together with an appreciation of the some of the practical factors involved in installation (fixed versus recurrent costs, noise abatement, equipment location).[[30]] Air conditioning plant was still expensive, but, for the first time, mechanical control of summer humidity and temperature was attempted in museums.

By 1935 a 1,000 ton (i.e., cooling power equivalent to 1,000 tons of ice) compressor was installed at the US Archives in Washington, DC in an attempt to control the indoor environment to 45%RH and 80°F during the summer — a task described by contemporaries as 'heroic' — although the stabilization of humidity depended on limiting infiltration and recirculating as much air as possible; smaller systems were installed in the Library of Congress, Washington, DC, and in the Columbia Library and the Frick Museum in New York.[[31]] During the immediate pre-WWII period, stabilization to levels below the high summer humidity without increasing the indoor temperature was, given sufficient funds, an option. Improvements in electric motor and fan technology increased supply pressures. The higher pressures allowed the use of narrower supply and return ducts, thus reducing the volume of the building consumed by air handling services when compared with the 19th century convective methods; the introduction of 'turning vanes' allowed tighter bends in ductwork, again reducing the amount of space required for supply and return ductwork. Both these improvements made retrofitting a building with HVAC a more practical proposition. The developments in HVAC technology did not spread rapidly to museums outside the US (and were exceptional within the US) during the 1930s and '40s, partly owing to the pre-WWII depression, and afterwards owing to the pressing demands for reconstruction in Europe.

By 1941 there were some dozen museums in the US with air conditioning installed, and Kooistra, considering the practicalities of humidity control in museums, wrote:

'The optimum condition of moisture content has been thought to be 55 to 60% relative humidity, independent of the temperature held. Most of the existing installations giving satisfaction, however, are designed for a 55% relative humidity in the summer, and 35 to 45% in the fall, winter and spring... A constancy of conditions of air, without rapid fluctuation of temperature or humidity, is of primary importance in this field, continuous uninterrupted operation of the equipment throughout the 24 hours of every day is desirable.' [[32]]

Kooistra was writing as an HVAC systems engineer for the Carrier Corp., rather than a materials scientist, and his criteria of 'satisfaction' may have been based on the realities of plant operation in the US climate (i.e. avoiding condensation in the winter) rather than the optimum humidity for particular materials; indeed, he explicitly eschews zone control. Nonetheless, his remarks are probably indicative of the reality of mechanical humidity control in American museums (where it existed) and indicate a split between British and North American standards, based primarily on differences in climate, but also partly on research from the US National Bureau of Standards which showed better permanence of library collections at reduced humidities.

Thus, by the end of the 1940's the mind-set for next 40 years was largely established. The promising beginnings in the 1920's and 1930's (indications that wood's reaction to humidity rather than simple bending of a painting's substrate was the mechanism of paint failure, and the notion of 'time lag' in moisture response as significant in preventing damage) were overlaid by the 'scientific' conclusions of the Weaver Report and the desire to reproduce exactly the conditions experienced during WWII storage of British collections using air conditioning so that paintings could be displayed without glass. It is notable that, in both museums and the building industry, the end of the WWII led to a simple rules-based approach to design problems: rather than the multiple tables and qualifiers which had been used before, the emphasis was on a single way of doing things and simple (often over-simplified) design criteria. The development of practical air conditioning equipment had made possible what never before been achieved — the control of indoor humidity during the summer without increasing the temperature of the ambient environment, with the additional benefit of central filtration of the supply air (the reason why air washers were originally introduced into museums). However, the problem of condensation during winter humidification and heating remained.

3.3 The spread of air conditioning and the 'era of Manod'

Around 1950 the first mass-produced room air-conditioning units (designed to hang out of a window) became available and sold extremely widely in the US. This introduction completely redefined expectations of indoor comfort during hot humid American summers. New, fully air-conditioned buildings began to appear (the UN Building and the Lever Building in New York) and by middle 1960's the integration of air conditioning into the architectural scheme of US public buildings was well-established (if poorly executed). 'Dual-duct' conditioning (separate air supply of warm and cool air, mixed at the local supply register for the room) was introduced which provided control sufficient for human comfort at a comparatively high energy cost.

In 1956 Plenderleith's Conservation of Antiquities and Works of Art, the 'official textbook [of the British Museums Association] on the conservation of museum objects' and the standard text for the next two decades, was published. In the introductory section, after touching on the probability of mold growth at high humidities, Plenderleith rehearses the story of the technician in the Manod. He then goes on to assert, in a section entitled 'The lower permissible limit of relative humidity':

'Taking into consideration the susceptibility of all organic materials to damage by desiccation, the lower safety limit should be fixed at 50 per cent. R.H... If, on taking observations, the figure is observed to fall, say, to 45 per cent., no great harm would be likely to result, but if such a figure persisted over a period, this might well be dangerous...' [23: p.9, Plenderleith's emphasis]

It is difficult to see how consideration of 'all organic materials' can be made to square with this statement. Further reading of Plenderleith's book gives one example of this 'dangerous' desiccation — he asserts that parchment becomes inflexible below 40%RH and therefore embrittled inks may flake off. The insistence on a lower limit of 50%RH contrasts with Plenderleith's earlier work in 1934, prior to WWII, where the subject of the lower humidity limit for organic objects was dealt with perfunctorily:

'In general, the temperature may be allowed to vary from 50 to 75°F [for comfort], and relative humidity from 40 to 60 per cent.' [[33], our emphasis]

He expands on this with reference to the necessity of the upper limit for the prevention of molds, citing Groom's recent work on the growth of mildew in antique bookbindings [[34]], but gives no explanation for the lower figure. With hindsight it is clear that in 1956 Plenderleith was generalizing, as the Weaver report had done, from the particular experience of wartime storage of panel paintings to all organic collections.

In the late 1950's, the International Council of Museums (ICOM) sent out a questionnaire to museums about their preferred environmental levels. In 1960, a whole issue of Museum was devoted to the publication of the results by Plenderleith and Philpott, although most of the article was, in fact, a long appreciation of the available control and measurement equipment.[[35]] Once again, when it came to the relative humidity, in a section titled 'Ideal climatic conditions: (a) The zone of safety', Plenderleith's position that the lower limit should be 50%RH was reasserted, the upper limit set at 60%RH to prevent the growth of mold, and the story of the Manod technician was retold as proof of the superiority of the 50 to 60%RH range. Although the survey showed little consistency in the preferred levels of RH in different institutions (except that they were grouped between 40 and 70%RH), the article notes that 'the scientists' preference... [is] for approximately 60 per cent.' This 'preference' was, at best, a codification of the pre-WWII custom and practice already described. At worst, it was an uncritical reproduction of the work of MacIntyre and McCabe.

Thus, a standard was set, at least in Europe. In the US, things were slightly different. In 1964 Buck argued that, owing to the differences in climate between Europe and North America, it was much more difficult to maintain mid-range indoor humidity during the severe winters from Boston northwards, than it had been to maintain such conditions in the British WWII cave storage.[[36]] Buck emphasized the dangers of winter condensation and proposed reduced winter temperatures and a lower humidity limit of 45 rather than 50%RH for organic material, based on his estimate of the moisture content to which antique wooden objects were acclimated before the introduction of central heating. In particular, Buck had a realistic appreciation of the limitations of air conditioning in the US climate:

'Most damaging are the long seasonal cycles which occur indoors in our climate. In our northern states the relative humidity in buildings without climate control will range as high as 70 percent in the summer and drop to 10 or 15 percent in the winter. During the four or more summer months even bulky materials such as wood will adjust their moisture content to near equilibrium with the high humidity. When winter heating is begun, a long drying cycle starts. The cycle is inevitable, but the range of humidity variation must be reduced. How far it is possible and practical to reduce the seasonal range depends on climate, the insulating efficiency of the building, and the capacity of the air conditioning equipment. From our experience, it seems evident that the benefits of humidity control are not clearly observable... unless the seasonal range can be reduced to 20 percent or less; that is, allowing a ten percent above an ideal specification in summer, and a drop of 10 percent below it during winter... but it appears that many air conditioned museums find it difficult, if not impossible, to restrict the annual humidity range to less than 20 percent.'[35: p.56]

The only evidence Buck cited for harm at low humidity was a study by the US Navy for mothballing materiel after WWII, which found 'incipient damage' in most kinds of wood kept for long periods at 30%RH or below. Buck's concern was to reduce variation as far as was practical, given the US climate and available control technology, but he could not escape the idea that the 'ideal' RH was 55% and so set limits of 45 and 65%RH for mixed organic collections. He produced a table of different organic materials which was more detailed than Plenderleith and Philpott's blanket 50 to 60%RH, and included important qualifications about where levels were critical (e.g. which objects were especially sensitive to mold growth or prone to cracking).[[37]] This table was widely copied by north American conservators [[38], [39]], but avoidance of condensation forced the lower limit for mixed organic collections gradually down over time to 40%RH, or even to 35%RH in Canada.[[40]]

It is important to be clear that the British 50 to 60%RH standard was rooted in a somewhat questionable analysis of the needs of panel paintings (based, as it was, on the variation of moisture content of wood in response to humidity fluctuation in an uncontrolled indoor London environment). The success of the Manod storage which had so strongly influenced Plenderleith was undeniable, but little work was done to test the often reiterated belief that humidity levels outside the 50 to 60%RH range were damaging to all organic materials. The US levels allowed lower levels overall and greater variation overall — this was at least partly a matter of practicality; but given that Buck's lower limit of 45%RH was accompanied by an explicit warning that it would cause problems in many US buildings during winter heating, it seems clear that practicality was not the prime motivation. In fact, Buck's version seems to display a clearer sense of priorities in mixed collections — of when RH control is critical and when it isn't — than Plenderleith's. Nonetheless, both works were based largely on empirical observation and on literature which dealt with organic objects exposed to uncontrolled annual high humidities during summer.

The subsequent conservation literature uncritically reproduced either Plenderleith or Buck's opinions (sometimes mixing and matching) and questions of whether this level of control was really feasible at a building level (given that automatic humidity control was mostly performed by enthalpy control, or by mechanical elements, the particular levels were probably not sustained year round in practice), and whether such a tight level of control was absolutely necessary for the majority of organic materials were ignored. Although there was some work on dimensional movement in response to changes in humidity, [[41], [42]] there was little attempt to relate this movement to actual damage (cracking, delamination, and so on). However, some measure of humidity control was established in a number of museums, and this must be regarded as an excellent thing.

The narrow mid-range humidity range cited as 'optimum' for organic collections (based on the mechanical needs of unglazed paintings on wood) could not be achieved in most buildings without central air conditioning, as both Buck and Plenderleith pointed out. In most of the Northern hemisphere, not just America, maintaining mid-point indoor humidity levels during the winter heating season leads inevitably to condensation — either occupancy must be reduced during the winter so that indoor temperatures can be allowed to fall, or, according to the building literature, vapor barriers and double- or triple-glazing must be installed. In the 1970's a study at the Royal Ontario Museum, Toronto, showed that even relatively modern buildings could not necessarily sustain the mid-point specification through harsh winters.[[43]] Thus, the argument that if close control to mid-point humidity did no good, then at least it did no harm, could not be sustained if the question was viewed from the point of view of the building, rather than the collections. A number of historic buildings — some built as museums, others adapted to the purpose — were gutted (original plaster and wall coverings stripped away) to allow the installation of vapor barriers and insulation. Many of the attempts to control condensation by intervention on historic buildings were only partially successful: structural damage from rots, corrosion and spalling resulted. The reasons are not far to seek, but require a digression back to the 1930's.

4. Origins of Insulation and Vapor Barriers for Moisture Control.

4.1 Control of condensation on building surfaces

The addition of insulation, often a feature of modern retrofits of historic buildings, arose to meet a need for higher humidities in industrial buildings such as printing works and textile mills whose buildings were often of light-weight frame construction. The primary problem was condensation ('sweating') on cold surfaces — the indoor was humidified, thus raising the dew point, but light-weight walls were poor thermal insulators and remained cold; energy conservation was not an issue at this time. In one of the first articles in the ASHVE literature, the insulating properties of cork and various hair felts were studied. Barrett measured the surface temperatures of walls insulated with these materials, with the intent of keeping all surfaces above the dew point temperature. With regard to moisture he noted:

'After arriving at the correct thickness to prevent sweating on the outer surface of the insulation the next important point is to choose an insulation that is as nearly air tight as possible. If any air reaches the cold surface through cracks or joints in the insulation moisture will be deposited on the surface. The insulation will then become gradually water soaked from the inside and rendered worthless as an insulator. Cork appears to be the most suitable material for the purpose, as it lends itself readily to the attainment of an air tight job. All joints and cracks can be effectively sealed with a brine putty. When other materials are used, special effort should be employed to protect them from moisture and to obtain a minimum of air infiltration.' [[44]]

This conclusion agrees closely with our modern understanding, in that it identifies air movement as the moisture transport mechanism, describes steps to prevent the condensation of water transported by air movement, and makes no mention at all of diffusion.

A further step was taken in 1930 by Close.[8] He recommended applying insulation toward the outside of the building as providing the necessary vapor protection to the insulation, 'for no insulation will function satisfactorily if it is not properly vapor-proofed. All commercial insulating materials are more or less porous and, without adequate surface protection, moisture laden air will penetrate the insulation until the dew-point temperature is reached at which point condensation will take place.' Here, moisture movement is seen as air movement through pores in the insulation material, rather than the joints described by Barrett. The insulation was seen as preventing condensation on the cold surface of the cavity, and vapor-proofing of the insulation was necessary to prevent moisture accumulation reducing the insulating efficiency. The solutions proposed include bituminous emulsions and paints, cement plasters, sheet metal and mopped roofing felts (the preferred 'vapor-proofing course'). Close concludes: 'It is usually advisable to install the insulation as far from the indoor surface as possible.' [45]

4.2 Vapor barriers for condensation control — the 'scientific' basis

The research which led to the widespread adoption of vapor barriers was carried out by two workers. The first was Rowley, Director of the Engineering Experiment Station at the University of Minnesota. Rowley was influential among mechanical engineers for his measurements of thermal resistance of building materials in the 1930's, and also by virtue of his position as President of ASHVE in 1936. He appears to be the one of the first researchers to be concerned with condensation on the inside surface of the building assembly rather than on surfaces within an insulated cavity. (An earlier researcher was L.V. Teesdale of the U.S. Forest Products Laboratory.[[46]]) In 1938 and 1939, he performed experiments to measure the accumulation of moisture in insulation by diffusion through representative wall and attic assemblies.[[47], [48]] Rowley was addressing the need for insulation in typical US residential buildings (lightweight wood-frame construction) to provide adequate thermal resistance during cold winters. His tests were performed in a climate controlled chamber, hermetically divided by the test assembly with a large vapor pressure differential between the two halves of the chamber: one side maintained at a low dew point to simulate external winter conditions, the other at an elevated dew point to simulate indoor heating and humidification; movement through the test assemblies by mechanisms other than diffusion was absolutely excluded. His findings, presented as grams of water accumulating per square foot of test assembly per 24 hours (g/ft2/day), were entirely consistent with his experimental design: wall assemblies with vapor barriers had much less moisture accumulation than those without. For example, at an 'outdoor' temperature of -19°F/-28°C and 'indoor' conditions of 70°F/21°C and 40%RH, an unfinished construction accumulated 2.15 g/ft2/day, while a wall with an asphalt vapor barrier accumulated only 0.13 g/ft2/day. Rowley concluded that vapor-proof construction was necessary; there was no mention that the amounts of moisture accumulation he measured were, in either case, negligible.[[49]]

Rowley also studied attics, again for typical residential construction, and his work appears to have led to the adoption of attic ventilation requirements. The simulated attics in his experiments had no gaps on the 'indoor' side (i.e., vapor movement from the warm side to the 'attic' was by diffusion) variable roof ventilation on the 'outdoor' side, no vapor barrier, and high vapor pressure differential between the 'indoor' and 'outdoor' conditions. Ventilation was expressed as the square inches of vent opening per square foot of ceiling area, although it is unclear what significant driving force for exchange through the ventilation opening there could have been in his setup. He found accumulations of 3.00 g/ft2/day with no ventilation, 0.13 grams with a 1/8 in2/ft2 vent, and 0.00 g/ft2/day with 1/4 in2/ft2 vent. Rowley concluded that attic ventilation was beneficial in preventing moisture accumulation during winter, but went on to assert that this benefit would apply to attics with vapor barriers, even though his experimental attics had not included vapor barriers. The difference between the ventilated and the unventilated attics was significant, but the amount of accumulation by diffusion through the ceiling was still trivial.[[50]]

Beginning in the 1920's, the Federal Housing Administration (FHA) issued Property Standards and Minimum Construction Requirements for Dwellings, a legally enforceable code of building practice. Rowley's results were quickly integrated into US residential construction. The January 1942 version is the first which contains requirements for attic ventilation and vapor barriers. Section 301-L requires attic ventilation in the amount of 1/300 of the horizontally projected roof area, an amount which is still commonly required; section 402-B.4 calls for a vapor barrier on the warm side of the wall whenever insulation is added to the stud spaces of conventional construction; section 402-B.5 states that 'vapor barriers shall be installed as continuous as practicable... any perforations of the barrier required for the installation of electrical outlet boxes or the like should be made so that there is a minimum of free opening.'

Where the 1/300 figure for attic ventilation comes from is unclear — the document cites no references or research and Rowley's ventilated test attics had openings equivalent to 1/574 and 1/1152 of ceiling area. The emphasis on continuity in the vapor barrier in Section 402-B.5 is an obvious attempt to limit air transport mechanisms through the gaps in the furring, rather than just diffusion of water vapor into insulation — in contrast to the carefully planned 19th century ventilation schemes, the interior is to be sealed tight with the vapor barrier as the sole means of controlling transfer of air and water vapor through the building. A more complete account of the FHA document and subsequent research is presented elsewhere by Rose.[[51]]

Immediately following WWII, the FHA was replaced by the Housing and Home Finance Agency (HHFA). In the late 1940's Britton, a senior Research Engineer for HHFA, conducted tests at Penn State University, using a setup similar to Rowley's from eight years before; again, the tests were aimed at the performance of typical US lightweight wood-frame residential construction.[[52]] However, Britton's experimental design was less fastidious than that of Rowley, especially in that cables were routed through the simulated interior wall and ceiling surfaces, and access openings were created through the interior finishes. As might be expected, moisture accumulated much more by air flow at those openings, small as they were, than it accumulated by diffusion into the test assemblies. The accumulations at openings did not cause Britton to reevaluate his focus on diffusion: they were excused as operational problems, even though the problems would obviously apply to real life building construction. In July 1949 Britton stated his conclusions in a four-page technical bulletin [[53]] : ventilation of 1/300 net free area ratio deserved to be adopted, and vapor barriers should be applied to the warm side of the wall. The conclusions were accompanied by a allowances for different climates and construction types, but the actual data collected by the research were more ambiguous than Britton's conclusions suggested: the unvented attic performed better than the excessively vented attic, and some walls with cold-side vapor barriers performed well. We do not know whether Britton would have addressed these problems at a later date because the experimental program was discontinued in 1948 for lack of funds.

In 1949 the HHFA published Condensation Control, the pivotal document in building moisture control, with rich illustrations of venting louvers and vapor barrier application.[[54]] It offered strong recommendations on the three areas — attic ventilation, crawl space ventilation and vapor barriers — based on Britton's conclusions, rather than his data. The recommendations gave adjustments for climate and building type, but later users tended to ignore the adjustments and qualifications and reduce the recommendations to their barest form:

One can imagine the pressure for the development of standards for residential construction at the outset of the postwar housing boom. What is remarkable is that, in the months immediately following the publication of HHFA's Condensation Control, the regulations were uncritically and simplistically adopted not just by residential contractors, but by the entire construction industry. The fourth edition of Architectural Graphic Standards of 1951 [[55]] includes, for the first time, the mention of attic ventilation, crawl space ventilation and vapor barriers in a three-page spread derived verbatim from Condensation Control. The same recommendations were subsequently reproduced in ASHVE Fundamentals (1952?) then in model building codes (ICBO and BOCA) which referenced the ASHVE guidelines. None of these publications indicate that the work on which the recommendations were based was directed specifically at lightweight wood-frame residential buildings, and was therefore of questionable relevance to other types of construction; and no one seems to have worried that the conclusions of Rowley and Britton's work were not supported by their experimental data. By the early 1950's moisture control in US buildings had been reduced from an empirical climate-specific gray-area, to the industry-wide blanket application of the attic venting and vapor permanence requirements cited above. It is difficult to explain the avidity with which the new condensation control guidelines were adopted; one cannot help but be struck by pamphlets such as The Menace of Moisture (1954), which shows on the cover a housewife cowering under the fierce gaze of a looming Bolshevik-like figure, and exactly reproduces the new guidelines. These rules, accompanied by varying and various qualifications, represented the reductio ad absurdum of previous practice regarding moisture control in buildings. Nevertheless, they formed the basis of moisture control design and practice for the next several decades.

Canadian practice differed from US practice. Neal Hutcheon of the National Research Council of Canada postulated in 1950 that the moisture accumulation he noted in Quonset huts in Saskatchewan was in quantities that greatly exceeded amounts that a diffusion model could explain, and proposed that air movement as the principal transport mechanism.[[56]] While US moisture control focused on diffusion control with vapor barriers, Canadian moisture control since 1950 focused on air movement within the building envelope. The Canadian concern was supported in the US with work by Sherwood (US Forest Products Laboratory, 1979) and others, who noted that moisture quantities transported by diffusion were minuscule, while transport by air convection would be typically 100 times greater. However, concern for building pressures did not effectively appear in the US literature until the 1980's. [[57]]

5. The 'Museum Environment': 1978 to 1990

After the OPEC crisis in the 1970's, variable air volume (VAV) systems (air supplied at a single temperature, the volume of supply air varying according to the heat loss or gain of a room) gradually replaced the less efficient dual duct HVAC system in the US and Europe. In much of industrialized Europe the use of air conditioning for human comfort was restricted to lightweight concrete buildings with glass curtain walls. Industrial application of air conditioning for humidity-critical manufacturing processes continued to depend on brute force: 20 to 40% of the building volume might be consumed by ductwork and airlocks while the building form, following the current iteration of the industrial vernacular, was often poorly adapted to humidity control. New museums fell mid-way between these two stools: the Plenderleith specification demanded industrial-level critical humidity control, but architectural attention was focused on fancy building form which was poorly adapted to the stringent humidity requirements. The major change over this period was the spread of HVAC to reduce interior humidity to a level below that of the outdoor summer air to most new public buildings.

In 1978 the first edition of Museum Environment was published.[[58]] The author, Gary Thomson, was Scientific Advisor to the NG-L. As far as humidity was concerned, the book was in many ways a recapitulation (albeit considerably expanded and modernized) of the 1960 issue of Museum. The new element was a much improved understanding of the way in which microclimates could be achieved in well-sealed cases by the use of inert buffering agents such as silica gel. Again, much of the attention was devoted to measurement and control techniques, rather less to the rationale for the specifications. Unlike the authors of the Museum report, however, Thomson was concerned to make the shallow foundation on which the mid-point humidity specification rested. In the introduction explicit. Thomson observes:

'No one who reads this book will fail to end with a realisation of our general ignorance. We have a very uneven knowledge of how things in a museum change [i.e. deteriorate] and what causes these changes, and yet we have to erect this framework of preventive conservation before rather than after our research has reached a dignified level of completion.' [57, p.ii]

In the section on relative humidity, after citing recent extensions to the Tate and National Galleries in London (designed for 55±4%RH and 20 ±1½°C/68±3°F), and advising on threshold switches at 65 and 45%RH, Thomson continues

'There is impressive general evidence, for example in the records of the National Gallery, London, that transferring paintings to an air conditioned environment very greatly reduced the need for treatment of detached paint... But the question of how constant RH needs to be to be to ensure that no physical deterioration will occur remains at present unanswered. The standard specification of ±4 or ±5%RH control is based more on what we can reasonably expect the [air conditioning] equipment to do than on any deep knowledge of the effect of small variations on the exhibit... Choice of RH level depends on several factors but cannot go too far from 50 or 55%RH. An exception may be found in the very low winter temperatures of Canada and north-eastern Europe where attempts at humidification to this level may endanger the building.' [57: pp.114-5]

Here Thomson insists that the humidity cannot go too far from 50 or 55%RH (except where considerations of avoiding condensation, as in the Canadian winter climate, dictate) and he later follows Plenderleith in asserting the humidity below 40%RH is unwise for artifacts such as parchment. It is not clear what this amounts to in terms of a specification — is practicality, or the condition of paintings and parchment with brittle inks, the main determinant? How far was 'too far'? No reference was given for 'the impressive general evidence', but presumably this refers to the WWII cave storage and subsequent air conditioning of collections from the NG-L. However, Thomson later showed an appreciation of the difference between humidity control in museums on the one hand, and historic house and churches on the other. He demonstrated a sound understanding of the limitations of air conditioning plant in practice, asserting that the specifications should be understood as 97½% limits rather than absolute limits, i.e. that it was reasonable to expect that the specifications would be exceeded for about 2½% (ca. 9 days) of the average year.

Thomson also spelled out (echoing Buck and McCabe) that installing machinery designed to meet these limits was not the same as maintaining the machinery so that it could continue to function properly: '...we enter a period where many museums, although claiming to be air-conditioned, may actually be running wildly outside their specifications.' [57: p.248] The problem here was that the automatic controls for HVAC systems had not changed substantially since McCabe's day — they were dumb sensors which reacted mechanically to open or close electrical contacts, and there was no way of determining whether they were operating properly other than setting up an independent audit of actual conditions, typically by the use of continuous monitoring by mechanical recording thermohygrograph (electrical recording systems are described by Arhrens in 1934 [[59]] and Rawlins in 1943 [25], but do not seem to have been in general use). Thomson did not reproduce the earlier detailed appreciation of the limitations of instrumental measurement of humidity given by Plenderleith and Philpott (1960) and MacIntyre (1935).

The impact of The Museum Environment is difficult to over-estimate. It is a rare library (public, museum, or university) which does not have a copy. Thomson's recommendations for humidity were widely quoted, but without the accompanying qualifications. Similarly, his confidence in the sling psychrometer as a primary calibration tool was widely accepted — the sling psychrometer was the best tool available at the time for the money, but this does not make it an adequate primary calibration tool. It was known that the mechanical recording thermohygrograph (a standard measurement technique until the 1980's, and still widely used) was not very satisfactory. In fact, the uncertainty of a mechanical recording thermohygrograph adjusted by reference to a sling psychrometer is around ±6%RH (95% confidence interval) in the range 30 to 70%RH [[60]] and has been shown to reach ±11%RH in practice. [[61]] This accuracy of measurement is clearly inadequate to demonstrate that humidity never falls below 50% or rises above 60%RH.[[62]] Thus, there arose a 'reality gap' between what air-conditioned museums knew to be true (which was vague, when the errors of measurement were taken into account) and what they asserted publicly.

During the late 1970's through the 1980's, questions were increasingly being asked (if not in print) about the benefit of mid-point mechanical control. It was costly and difficult to design for. What, people began to ask, was the relationship between benefit, in terms of measurable reduction in damage, and the cost of mid-point central air conditioning? Could a wider range be tolerated? All the available evidence pointed to the fact that the standard had been set by consideration of the mechanical properties of wood. However, it was also known that mid to high levels of relative humidity had an impact not just on mold growth, but also on several of the chemical properties of materials — rates of photo-oxidation, corrosion, and hydrolysis — which could strongly affect the long-term survival of thin organic materials such as paper, textiles, and photographs. Specifications for archival storage of paper and film in libraries had already moved towards humidities below mid-point, as with the US Archives building in 1935. On the other hand, a number of institutions (generally well-established ethnography, natural and social history museums rather than art museums) had already opted out of the debate. They noted that, given the normal level of control to the human comfort zone (in the range 35 to 70%RH), the majority of their organic collections were holding up surprisingly well (pace Plenderleith) and that their biggest problems revolved around control of insects and vermin on the one hand, and dust and particulates on the other. In North America, as we have noted, accepted museum specifications remained generally wider: typically in the 40 to 60%RH range, or even 35 to 60%RH in Canada — the reduction in lower limit reflecting the difficulty of controlling window and condensation, and condensation in walls, progressively more severe winter climates. In fact, the introduction of air conditioning, whilst beneficial in reducing high summer humidity, added to these condensation problems since the local overheating of windows by perimeter radiant heating was now abandoned.

As mechanical systems increased in sophistication, there arose a general feeling that if ±5%RH was good then ±3, or even ±2%RH, must inevitably be better; all the more so since the ±5%RH variation was based, as Thomson freely admitted, on the performance of mechanical systems. Any deviation from mid-point humidity became a cause for alarm and again, because little quantitative research on the effects of different levels of humidity variation had been carried out, it was felt sensible to play safe by keeping as exactly to the rules as possible. (In fact, there is hardly a museum building in the world — even today — which can guarantee ±3%RH at a gallery level, let alone a crowded gallery during a 'blockbuster' exhibition.) Also, it was assumed that if close control (±2%RH) provided no benefit for the objects than wider control (±5%RH), then at least it did no harm; therefore, why not have the tightest level of control that was possible? In fact, we believe that the inward spiral of humidity tolerances proceeded from a fundamental miscommunication between museum staff (conservators, curators) on the one hand, and mechanical engineers on the other: museum staff were setting tolerances for the ambient condition in the whole gallery space, but mechanical engineers were designing for the air conditions in the return air duct. It is important to be clear about the difference between these situations and to illustrate them we shall take the case of a single room: museum staff were insisting that at any instant the humidity measured at any location the room should not deviate more than ±5%RH from set point, whereas mechanical engineers were designing so that the humidity at a single point in the return air duct did not vary more than ±5%RH from the set point. Fluctuations in the room in general would almost inevitably be wider than for the return air condition. Conservators, puzzled that they were observing wider fluctuation than they had requested, responded by narrowing the tolerance of the design specification (which the engineers still understood as a return air specification) to the point where the room in general began to behave as requested. Thus, the widely heard observation that 'if I ask for ±2, I might get ±5' was correct, but only because two elements of the design team were talking about different things. The confusion had two elements: first, and most importantly, the basis of specification (whole space rather than return air duct) was never made explicit; second, design for the return air condition is deeply embedded in the culture and training of mechanical designers -- so deeply embedded that, even today, it is extraordinarily difficult to find a mechanical systems designer who can be made to think in terms of the whole space condition, rather return air. On this second point, it is worth observing that verification of performance on a whole space basis is difficult (many measurement points must be used), whereas verification of the return air condition is simple (only one measurement point is required).

Publicly, museums insisted on narrow specifications, privately they would shrug off the uncertainty of environmental monitoring and control with 'how accurate do you want it to be?' The penalty for failing to maintain in public that one's institution did meet these standards was that important loans of objects or grants of money could be refused on the grounds of environmental inadequacy. National face could be at stake, as instanced by Ashley-Smith's account of fifteen Soviet curators who 'strenuously denied that the extremely low external temperatures in a typical mid-continental winter would cause low RH in their heated galleries.' [[63]]

The difficulties of experimental investigation to resolve the question of how much humidity can fluctuate without causing damage should not be minimized. Pollution and light-aging studies could be performed at extreme levels and the results scaled down to apply, at least approximately, to the everyday world; damage which might accumulate by oscillation of relative humidity below the fracture threshold is more difficult to simulate in a convenient time frame. Also, although the reaction of individual materials which comprised many organic objects could be studied, there appeared to be no way of understanding their individual contribution to the overall response of a composite object to humidity. In 1982, in a six-day meeting published as Science & Technology in the Service of Conservation, Colville et al gave an account of a finite element analysis (FEA) study of the response of oil paintings on canvas to relative humidity.[[64]] This computer simulation predicted patterns of damage which were remarkably similar to damage seen in some real paintings. What was striking about the results, as Michalski has observed [[65]: p.238], was not the form of the computer model — although novel to many conservators, the FEA method was already well-established in engineering fields — but the underlying understanding of the component materials of a painting which produced such a close correlation between theory and observation. This improved understanding was due to Mecklenburg [[66]], and the conference signaled the beginning of a new phase of experimental investigation into the response of organic materials.

In 1986 the second edition of Thomson's Museum Environment appeared.[[67]] There was little change in the discussion of humidity, and his earlier warnings of the limited scientific basis of the recommendations was reproduced verbatim. This time, however, an appendix was included which laid out the difference in control appropriate to museums on the one hand, and historic houses on the other. Thus, the two-tier system which had already emerged was formalized in Thomson's class 1 and class 2 environments —

The problem became that class 1 was definitely seen as 'better', whereas class 2 became synonymous with 'second class'. Achieving recognition as an 'important new museum' required HVAC designed to maintain the class 1 environment of 50 or 55 ±5%RH. The question of whether the class 1 environment was actually better — better in terms of a measurable reduction in damage to many classes of organic artifacts, when compared to the class 2 specification — was entirely ignored (at least in print). Again, Thomson's work was widely quoted, but the references were almost entirely confined to the numbers in the appendix and, again, his qualifications were not reproduced.

In 1987 Stolow, formerly at the Canadian Conservation Institute (CCI), published his Conservation and Exhibitions, pointing up the contrast between British and North American standards for humidity control.[[68]] Whilst Thomson was recommending 50 or 55 ±5%RH, Stolow contended that

'The optimum relative humidity condition for exhibition and storage is a constant condition year round with a set point between 40% and 55%RH, and with a daily fluctuation not to exceed ±3%... The minimum acceptable relative humidity condition is a set point that is varied by 2% per month from 40% to 55% over a six month period from winter to summer. Daily permissible fluctuation is ±3%RH from the set point.' [67, p.252, Stolow's emphasis]

Like Thomson, Stolow could produce little evidence except custom and practice to back his conviction of 'optimum' environment, particularly as regards the effect of variation in humidity about a midpoint level on composite organic artifacts to back his requirement for ±3%RH. However, the contrast between Thomson class 1 (British specification) and Stolow's optimum (North American specification) should not be over-emphasized. Thomson follows Plenderleith and is concerned more by level (mid point or higher), whereas Stolow follows Buck while emphasizing tight tolerance (±3%RH) because he allows the set point to drift over the year to reduce the danger of condensation during the harsh Canadian winter climate. Thomson's specification is practical (just) for air conditioned galleries in a climate similar to southern England, and allows pictures to be displayed without glazing and backing; Stolow's level of variation cannot be attained, we would argue, without microclimate (case-level) control.

6. Recent Developments

6.1 Control of moisture in the building envelope

Currently [i.e., in 1997], the moisture loads on building envelopes are under serious review, with a very active review of humidity loads in museum and preservation environments in particular. The review of moisture loads is part of a larger attempt to move beyond the rules-of-thumb approach to a more analytic approach, which uses better input values and tests design against well-defined criteria. The traditional analytic tool, described in ASHRAE Fundamentals and elsewhere, is a steady state simultaneous analysis of temperature and vapor pressure gradients through an assembly, using fixed indoor and outdoor design conditions. This is termed the 'dew-point method' in the US and the 'Glaser' method in Europe. Recent developments, recorded through the International Energy Agency (IEA Annex 24 on Heat Air and Moisture Transport through Insulated Envelope Assemblies), include quite sophisticated transient modeling tools. What remains missing from a modeling analysis of heat and moisture transport is appropriate input values and criteria for interpretation of the output. What are the correct input moisture loads for envelope design? What, short of waterlogging, constitutes a criterion of unacceptable building envelope performance? These questions are the focus of an upcoming effort titled Moisture Engineering by the Building Environment and Thermal Envelope Council (BETEC). This effort should lead to enhanced climate compatibility of various envelope designs, and an understanding of moisture storage in structures as a buffering technique for short cold spells. The success of these ambitions remains to be seen. The International Society of Indoor Air Quality (ISIAQ) is sponsoring a Task Group on Museum Environments. The aim of this group is to develop guidelines for 'preservation environments' which specifically include museum and library environments. The temperature and humidity guidelines will correlate closely with BETEC's effort to refine and formalize the moisture engineering approach.

6.2 Humidity levels for organic collections in museums

In 1993 Michalski, working at the Canadian Conservation Institute (CCI), published 'Relative humidity: a discussion of correct/incorrect values' — probably the most widely cited article in recent preventive conservation literature.[2] Although it presented no firm recommendations for humidity control, it provided an excellent and accessible summary of what was known, what was not known, and what could reasonably be inferred. Michalski attempted to determine the effect of repeated small variations in humidity below the critical fracture threshold would be. Here he proceeded by inference and his conclusions are not unassailable, but it seemed that at variations ±15%RH around the mid-point, the majority of historic artifacts accumulated damage only very slowly and hourly variation was irrelevant for thick objects (as Buck had also noted, and Brewster had implied). Michalski also observed some pieces of antique furniture subjected to annual humidity cycling: sustained drops of -25%RH from mid-point during the winter heating period did not cause cracking, whereas -40%RH drops from mid point caused failure in some, but not all, pieces.

'The extension of the small fluctuation criterion to all artifacts has no merit except convenience. Drops of -40%RH do not constitute an emergency for loose skins, fur, textiles, costumes, metals, botanical specimens, or most archival material... Leather bindings [of books] on non-acidic paper are indeed a mechanical issue, but they can only be considered low to moderately vulnerable. Brittle inks on parchment are highly vulnerable. For collections dominated by rigid organic materials (wood and paint), we must accept that data supports common sense, not magic numbers. Safe RH is a broad valley... Overall high risk begins outside the range 25%-75%RH. Slight mechanical damage will accumulate on highly vulnerable assemblies at ±20%RH but this is virtually eliminated by ±10%RH in wood, ±5%RH in paint.' [2: p.628]

The upper limit for Michalski's area of high risk does not depart significantly from Thomson's class 2 environment and has the same purpose in mind: the avoidance of mold growth. The difference lies in the lower limit at which Michalski was prepared to place major risk — always the main difference between European and North American specifications. Michalski also reiterated an important point — visible damage is most likely during drops in humidity, particularly if these drops are preceded by prolonged high humidity levels. Unfortunately, the significance of a drop varies according to the object. For lightly varnished furniture, with the doors and drawers kept closed, the half-time for response is the order of a week.[2: fig. 5] That is, if the humidity drops rapidly, it takes about a week for the object to reach half of its final response to the drop, about four weeks for 95% of the response. If the furniture is stuffed with clothing (natural textiles), the half-time is increased to about two months — at this point a sustained drop of a week becomes largely irrelevant, and a sustained drop of two months would produce only half of the full response. This may help to explain why furniture in use has survived so much better than furniture in many museums — when doors or drawers are left open for display purposes, when the furniture is emptied of contents, it reacts much more quickly to changes in humidity than would normally be the case.

In 1994 Erhardt and Mecklenburg published their 'Relative humidity re-examined'.[[69]] This paper, building on experimental work performed at CAL, provided a graphic introduction to the difficulty of specifying an 'optimum' relative humidity if one considers chemical decay (usually accelerates at mid to high RH) as well as mechanical damage (most noticeable at low RH).

'The RH settings most common in museums, those in the range 40-60%, minimize biological attack, mechanical damage and the efflorescence of common salt [NaCl]. It is interesting to note that these are the most visible, often the fastest, forms of damage seen in museums (other than for materials known to require separate treatment, such as corroded metals, weeping glass and mineral hydrates). It is easy to see how RH values around 50% have become so widely accepted. It is only when less obvious forms of damage are considered, such as the slow but continuous [chemical] degradation of organic materials, that lower values of humidity seem more desirable. In fact, the reduction of mechanical damage is the only major factor that would seem to argue against all but the lowest values of relative humidity, those below 25-30%RH. This conflict — mechanical versus chemical degradation, form versus content — is the main consideration in choosing a suitable RH, and one for which there is no obvious resolution.' [68, p.37]

Perhaps the most interesting part of the paper is the discussion of the possibility of calculation of allowable RH fluctuation. Erhardt and Mecklenburg proceed on the assumption that a fully restrained material represents the most vulnerable state, and then go on to calculate whether a particular variation in humidity will produce a stress which exceeds the elastic limit for the material, i.e., in their model the mechanical damage occurs when the elastic response of a material is exceeded, even though no visible damage such as cracking may have occurred. This is a rather elegant model because it allows us to deal with plastic deformation (compression shrinkage and ductile failure) at high humidities as well as mechanical failure (cracking and brittle failure) at low humidities. It is also an inherently conservative approach in that: 1) it assumes full restraint and maximum response whereas, in most composite objects, the materials will be responding in the same direction; and 2) classifies 'damage' as any excursion beyond the elastic limit. In short, this approach appears promising for arriving at a rational 'safe zone' for mechanical damage to organic materials. This method has already been pursued to the point where a number of very interesting charts have been produced, showing the safe and unsafe zones for a variety of fully restrained materials under different conditions of temperature and humidity. In particular, these diagrams clearly illustrate that the safe range of humidity fluctuation is small for organic materials acclimated to extreme high and low humidities, but considerably wider for objects acclimated at mid-range (45 to 55%RH). The dangers of moving an organic object acclimated to a high humidity to an 'ideal' mid-range humidity are referred to by many of the authors cited above, but the work at CAL gives us a model for evaluating these dangers in detail. What is surprising is how narrow the 'zone of safety' suggested by CAL is for materials acclimated to extreme humidities (both high and low), and how broad in the mid-range. A detailed summary of the work at CAL work is beyond the scope of this paper: the reader is referred to their specific publications for more information.[63, 68, [70], [71], [72], [73]]

Conclusions drawn from the work at CAL are not without their critics.[[74]] In particular, attention has centered on whether the wider 'safe zone' for mid-range humidity fluctuation will allow cost savings in both the capital and running costs for mechanical control. We feel that this consideration, while certainly of great interest, is irrelevant in the context of our paper. What is important is that the data from CAL allow us to construct a coherent model of damage to composite laminar organic artifacts as a result of variations in relative humidity and temperature. The predictions of this model can be compared to the response of organic objects in the 'real world'. If necessary, such a model can be refined in the light of practical experience. But without such a model — without the understanding generated by the tension between theory and observation — it is impossible to improve our understanding of the mechanical damage to organic artifacts any further. Without such an increase in understanding we cannot meaningfully resolve such questions as whether a particular humidity and temperature (which is known to be safe for a building) is safe for an artifact.

7. Conclusions

As we have shown, traditional museum humidity specifications have developed empirically as control of the indoor environment increased in sophistication during the twentieth century. It is not correct to dismiss these observations; with the benefit of hindsight we can see how they came about — the low relative humidities achieved during improved winter heating were destructive to objects which were initially acclimated to high humidity, and annually re-acclimated to high level by uncontrolled summer humidity. The humidity had to be stabilized and empirical observation showed that stabilizing humidity at levels around mid range during winter heating prevented the more obvious damage (indeed, this was the only practical option for most institutions prior to WWII). These observations were codified into a conviction of the superiority of close control around a set point of 55 or 60%RH for paintings, a conviction which was reinforced by the success of the WWII cave storage of British collections. The Thomson class 1 museum environment (55 ±5%RH) derives ultimately from the NG-L's post-WWII desire to improve indoor climate control for antique paintings on wood acclimated to high humidities — a class of material which was 'delicate' when subjected to large annual humidity variations — so that they could be displayed without glazing. We know that 50 or 55 ±5%RH 'works', but this does not make it an 'optimum' level. It is questionable whether the notion of inherently 'delicate' organic objects (i.e. requiring control to ±5%RH) is still relevant in collections where mechanical control has prevented prolonged high or low indoor humidities and acclimated organic materials to mid-range humidities. If such objects do exist, and evidence that they do indeed, the fact remains that they are less likely to sustain further mechanical damage during humidity fluctuation at mid-point than at the same fluctuation at high or low humidities.

Despite the attention given in this article to moisture damage from elevated indoor humidities during winter heating, it must be remembered that most moisture problems in historic buildings are the result of roof leaks, damaged gutters and down spouts, poor surface drainage of rainwater, and wet basements and crawl spaces. That said, the work of Erhardt, , McCormick-Goodhart, Tumosa and Mecklenburg at CAL, together with the work of Michalski and others at the CCI, suggests the following broad picture for organic artifacts acclimated to mid range humidity: Variation of RH of ±10%RH (whether daily, weekly, monthly or yearly) about a central level between 45 and 55%RH presents a low risk of mechanical damage for almost all organic objects (including paintings and furniture); variation ±15%RH about a central level is low risk for most, but not all composite organic objects; variation ±20%RH about a central level is dangerous to some composite objects; variation ±40%RH about a central level is destructive to most organic objects if the extreme levels are maintained long enough for the object to react. These figures do not address the susceptibility of any materials to chemical decay, which can only be done on a material-by-material basis. The lower variations do not represent 'ideal' conditions in that, for some organic materials, considerations of increased chemical deterioration at mid-range humidities may make a lower set point and variation desirable[2, 68]. Response times will vary according to the materials the object is composed of, its mode of manufacture, and the way it is displayed. If the less robust and more quickly responding organic objects (such as oil paintings on canvas) are enclosed in a suitable microclimate with vapor-impermeable materials then ambient room conditions can be allowed to vary more widely than the traditional 'optimum' recommendations would suggest. Tight control of humidity variation is most beneficial during extreme high and low humidity conditions. As conditions are allowed to vary more widely, so the inaccuracy of mechanical recording thermohygrographs becomes more of a concern. Ironically, a recording thermohygrograph performs best at room temperature and relatively constant RH. As temperature and RH are allowed to vary more widely, so its performance becomes progressively less satisfactory.

In a traditional museum setting it is natural to display 'portable' objects in vitrines, as much to deter theft as otherwise. If the suitable control can be achieved at display case level (by the use of micro-climates) then the argument for tight control of ambient gallery conditions becomes much less convincing. Similarly, there are cogent reasons for glazing paintings (whether on wood, canvas, or paper) apart from the control of the local humidity or dust accumulation; for example as a protection against sticky fingers or spills (individual metal-plastic-film cartons of juice with straws stuck in them will project fluid a surprisingly long way when squeezed and the public do bring them into galleries). Unfortunately, this does little to solve the particular problems of historic houses. In a historic house setting, or in a social history diorama, display cases are considered undesirable because they reduce the immersive quality of the historical experience. Glazing and backing of paintings, on the other hand, can probably be tolerated — it is anachronistic in most historic settings, but the public are used to glass in front of paintings. Given that paintings are the most delicate class of objects on display in this context, but that they can be glazed and backed to produce a suitable microclimate [[75]], the tighter control levels can be relaxed for the ambient room environment. The problem of non-portable artifacts, however, still remains — both in a traditional museum and in the historic setting. Most non-portable artifacts (furniture, for instance) are, almost by definition, too large for passive microclimate control to be practical.

The problem for these non-portable classes of object, and for portable artifacts on open display, in historic houses reduces to the following two questions:

  1. What level of risk is inherent in displaying social history material (textiles and furniture, tools and books) in a climate which exceeds the limits set by traditional museum specifications?
  2. What risk does a particular indoor environment pose to the building structure, given the local exterior environment?

Given the answers to these two questions, and understanding risk as the probability of damage, one can take a decision about whether one can preserve the object without destroying the building. This is essentially the position enshrined in the New Orleans Charter. Neither Stolow nor Thomson were in a position to provide answers to question 1, but the new direction which historic materials research had taken since 1982, mostly published in the 1990's, is beginning to provide some rational answers for the artifact side of the problem.

Can we draw conclusions about allowable indoor humidities in existing historic buildings? Variations in outdoor climate and building assemblies and displayed collections make any estimate, however rough, of allowable wintertime humidities entirely speculative. Several points may be noted however:

7.1 Role of instrumental measurement in preservation

The limitations of continuous in situ measurement of the variables involved in the preservation of historic buildings and their collections are often not appreciated. In particular, variables dealing with water are difficult to measure: relative humidity is elusive because temperature variations remote from the humidity sensor affect relative humidity (e.g. humidity is decreased by radiant heating from direct sunlight or incandescent light sources, increased in proximity to cold building elements) and therefore measurements may be highly location specific; the moisture content of materials can be measured (although not wholly non-destructively) by embedded resistive elements, but their life is limited and, again, the measured level may be highly location specific; rates of moisture transfer by diffusion or convection elude convenient field measurement at the moment, but may be crudely inferred in some circumstances by analysis of moisture content and thermohygrometric data.

The analysis of accumulated digital data (from digital dataloggers) is easier than the analysis of large quantities of analog data (e.g. recording thermohygrograph charts), but instrumental monitoring as an effective preservation management tool is still in its infancy.[[76]] We have noted an increasing feeling amongst historic preservation institutions that if some instrumental measurement is good, then more must be better; and that instrumental measurement is, in and of itself, a prophylactic against damage. More data is not necessarily better. We would draw a distinction between investigative monitoring (temporary installation, high sensor density, portable units, high data volume, short-term objective) to discover the source of a particular acute problem and generate appropriate solutions, and confirmatory monitoring (low sensor density, low data volume, permanently installed units, long term objective) aimed at ensuring that general, chronic problems of indoor climate are being controlled to the desired limits. Confirmatory monitoring may detect a general problem (RH too high, or condensation on windows), but it is often the case that additional investigative monitoring is needed to assign causes and generate sensible solutions. Confirmatory monitoring is not a substitute for a regular program of careful visual inspection.

7.2 Environment for buildings

Prevention of condensation is still non-trivial and poses a threat to historic buildings which are humidified and heated during cold winters. Historic buildings of massive porous masonry construction that are unheated (or heated only to a limited extent) will tend to stabilize RH above midpoint by absorbing and desorbing moisture from the masonry. This property was known in the 1930's and has been the subject of some investigation recently.[[77]] Churches are one obvious example, but this mode of stabilization will also apply to rooms with a large moisture reservoir in organic materials (extensive wooden paneling, large areas of tapestry, high density of books). However, the stabilizing effect can be dominated by prolonged heating to comfort levels during the winter, particularly where vapor barriers turn the indoor of the building into an hygrometrically inert box. In historic buildings convective transport (stack effect, draughts) is the major means of moisture movement.

Buildings that use 'compact' rather than 'cavity' construction have fewer moisture concerns. 'Compact' assemblies include solid masonry, solid concrete, and wall and roof systems composed of closed cell foam insulation. 'Cavity' assemblies are buildings of wood frame or steel frame or cavity brick, which contain voids, typically filled with insulation. Controlling moisture within building cavities requires a level of design that is not necessary where the cavities are absent. Cavity construction is very sensitive to building pressures (particularly the positive pressure required for efficient central filtration) and to the continuity of air barriers (which are invariably discontinuous in practice) whereas compact construction is generally sensitive to ageing and failure of water-proofing agents at joints. Ensuring good moisture performance of cavity assemblies requires an identification and characterization of the paths of air leakage to ensure that the escaping air encounters no cold surfaces along the path.

Determining the maximum allowable indoor dew point for cavity construction is still problematic. The allowable maximum indoor humidity during wintertime in an historic building with compact construction can be quite accurately determined by finding the lowest temperature of the surfaces that face the indoor, and setting this as the maximum dew point for a given ambient indoor temperature. The hand-held infrared pyrometer is a useful instrument for the survey of surface temperatures.

Window condensation remains a de facto indicator of excessive indoor humidity, but the utility of window condensation as a diagnostic tool is building specific. On the one hand, use of sealed multiple glazing units greatly reduces the likelihood of window condensation; on the other hand, the dismantling of many perimeter radiant heating units in favor of forced air systems generally increases the likelihood of window condensation. Storm windows can be particularly troublesome: in northern climates the seal between the prime and the building must be airtight, while the cavity between the storm and the prime needs rain-proof ventilation to the outdoors. The reverse situation (tight sealing of the storm, ventilation of the prime to the indoor) is a sure recipe for condensation in northern winters.

7.3 Technology of humidity and moisture control

It should not be presumed that structural retrofit is capable of rendering historic buildings in any particular climate safe for an elevated indoor dew point. In particular, the use of vapor barriers is likely to be unsuccessful in the long run. In historic buildings, it is wise to increase indoor humidity only incrementally during the winter, while watching for signs of moisture damage. Conversely, for organic objects acclimated to high humidities, it is wise to decrease humidity only slowly, while watching for signs of damage.

Humidistatic heating (reduced temperature to reduce low humidity extremes during winter [[78], [79]]) should be considered as a winter strategy, particularly since visitors are often heavily clothed against the cold, and is especially suitable for buildings which are closed during the winter. Flexible control systems which track indoor and dew point and surface temperature, and outdoor dew point, rather than simply indoor humidity, would also be beneficial. The further development of non-steady state control models is entirely practical from the artifact point of view, in that the processing power to operate the model in real time is cheaply available and the work at CAL and elsewhere provides enough data to produce a predictive model of damage to representative organic artifacts, albeit a rather conservative one if we define damage as CAL has done. What is lacking is the organization of data about the reaction of building construction to differences in internal and external conditions into a form which could provide the other half of such a model.

The forthcoming AHSRAE Fundamentals chapter on Museums & Art Galleries is being totally revised and it is hoped that the new form of the chapter will provide clear guidance to both conservators and mechanical designers.

8. Bibliography and Notes

[12]. Association for Preservation Technology & American Institute for Conservation, The New Orleans Charter for the Joint Preservation of Historic Structures and Artifacts, APT/AIC, Fredericksburg /Washington, DC (1991).

[3]. Michalski, S. 'Relative humidity: a discussion of correct/incorrect values' ICOM Committee for Conservation 10th Triennial Meeting Washington, DC, USA. Preprints, Vol. II, ICOM Committee for Conservation, Paris (1993), 624-629.

[4] Banham, R The Architecture of the Well-tempered Environment, 2nd edn., University of Chicago Press, Chicago (1986).

[5]. Cook, J. & T. Hinchcliffe, 'Designing the well-tempered institution of 1873', Architectural Research Quarterly, 1 (Winter, 1995), 70-78.

[6]. Lull, R. 'The new Peabody Museum. Part I - building and equipment', Museum Work, 7 (1924), 109-117.

[7]. Lewis, S.R. 'Heating and humidifying in museums', Heating Piping and Air Conditioning, (November, 1931), 910-915.

[8]. Lyle, J.I. 'Relative humidity' ASHVE Transactions, 18 (1912).

[9]. Close, P.D. 'Preventing condensation on interior building surfaces' ASH&VE Transactions, 36 No. 854, New York (1930).

[10]. Coleman, L. Manual for Small Museums, GP Puttnam's & Sons, ? (1927), 319.

[11]. Meller, H. 'Air pollution from the Engineer's standpoint', ASHVE Journal Section Heating Piping and Air Conditioning, (January, 1931), 70-75.

[12]. Smallcombe, W. 'Standard air conditions and dust problems' Museums Journal, (April, 1934), 39-40.

[13]. Topical Discussion, 'The air washer as a means of humidifying the air for ventilation', ASHVE Transactions, 16, New York (1910).

[14]. On the development of air conditioning generally see: Banham, R. loc. cit [3], chapter 9; Cowan, H.J. Science and Building — Structural and Environmental Design in the Nineteenth and Twentieth Centuries, John Wiley & Sons, New York (1978), 215-250; Thévenot, R. A History of Refrigeration Throughout the World, trans. from French by J.C. Fidler, International Institute of Refrigeration, Paris (1979), 356-367; for an early European view see Debesson, G. 'Theory and practice of heating and ventilating in France' ASHVE Transactions, 20 (1914);

[15]. McCabe, J. 'Humidification and ventilation in art museums', Museum News (September 1, 1931), 7-8.

[16]. Some Notes on Atmospheric Humidity in Relation to Works of Art, Courtauld Institute of Art, London (1934 or 5). There is a fairly dismissive contemporary review of this publication: A.P. Laurie 'Atmospheric humidity and works of art', Museums Journal, 35 (May, 1935), 51-52.

[17]. MacIntyre, J. & H. Buckley, 'The fading of water-colour pictures', The Burlington Magazine, 57 (July, 1930), 31-38.

[18]. MacIntyre, J. 'Air conditioning for Mantegna's cartoons at Hampton Court Palace', Technical Studies in the Field of the Fine Arts, 2 (April, 1934), 171-184.

[19]. Rosenberg, G. 'Antiquities and humidity', Museums Journal, 33 (December, 1933), 307-350. This article recommends humidities between 45 and 65%RH for mixed collections of organic 'antiquities'. It is, perhaps, the first extended treatise on the humidities necessary for a wide range of museum collections.

[20]. MacIntyre, J. 'Some comments on antiquities and humidity', Museums Journal, 33 (January, 1934), 350-351; G. Rosenberg, 'Reply to some comments on antiquities and humidity', Museums Journal, 33 (February, 1934), 419-420; J. MacIntyre, 'Comments on antiquities and humidity', Museums Journal, 33 (March, 1934), 459-460.

[21]. Brewster, D. 'Air conditioning as applied to furniture, fixtures and other interior woodwork', ASHVE Journal. Section Heating Piping and Air Conditioning, (January, 1931), 65-69.

[22]. Keeley, T. & F. Rawlins, 'Air conditioning at the National Gallery, London. Its influence upon the preservation and preservation of pictures', Museum (UNESCO, Paris), 4 (1951), 194-200.

[23]. Coremans, P. 'Air conditioning in museums', Museums Journal, 36 (November, 1936), 341.

[24] Plenderleith, H. Conservation of Antiquities and Works of Art, OUP, New York (1956)

[25] Weaver, J., G. Stout & P. Coremans, 'The Weaver report on the cleaning of pictures in the National Gallery', Museum (UNESCO, Paris), 3 (1950), 113-135.

[26]. Rawlins, F. 'The National Gallery in war-time', Nature, 151 No. 3822 (1943), 123-128.

[27]. Sikander Ali, S. 'Chemistry in the Service of Archaeology', Museums Journal of Pakistan, 12 (1959), 61.

[28]. Eastlake, C., W. Russell, & M. Faraday, Report of the Commission to Inquire into State of Pictures in the National Gallery, London (1853?).

[29]. Rawlins, F 'The control of temperature and humidity in relation to works of art', Museums Journal, 41 (March, 1942), 279- 283. See also: Some Notes..., loc. cit. [15], 35-42; MacIntyre, loc. cit. [17].

[30]. Cherne, R. & C. Nelson 'Preliminary planning for air-conditioning in the design of modern buildings', Architectural Record, 75 (June, 1934), 538-548.

[31]. Review article 'Air conditioning of museums', Refrigerating Engineering, (August, 1935), 85, 105.

[32]. Kooistra, J. 'Highlights of air conditioning for museums', Museum News, (May 15, 1941), 7-8.

[33]. Plenderleith, H.J. The Preservation of Antiquities, The Museums Association, London (1934), 1-2.

[34]. Groom, P. & T. Panisset 'Studies in penicillium chrysogeum thom. in relation to temperature and relative humidity of the air', Annals of Applied Biology, 20 (1933), 633-660.

[35]. Plenderleith, H. & P. Philpott, 'Climatology and conservation in museums', Museum (UNESCO, Paris), 13 (1960), 202- 289.

[36]. Buck, R. 'Museum News technical supplement no. 6. Part I: A specification for museum air conditioning,' Museum News, 43 (December, 1964), 53-60.

[37]. ASHRAE 1964 Applications was published the same year as Buck's paper. It includes a chapter on libraries and museums which recommends design conditions of 72°F/22°C and 35 to 40%RH in Winter, and 78°F/26°C and 45-50%RH in Summer. Subsequent editions of ASHRAE Applications (1974, 1978, 1982, 1991) lowered the design temperatures, increased the recommended humidity for paintings to the 'ideal' 50 to 55%RH, decreased the recommended humidity for archival storage in libraries to 35%RH, and increasingly appealed to curators to define the 'best' conditions for specific artworks. The concentration in all editions except the first is on libraries.

[38]. Leisher, W. Humidity - Temperature Requirements for Museum Collections, American Association of Museums, Washington, DC (1977), 1-5.

[39]. Johnson, E. & J. Horgan Museums collection storage, UNESCO, Paris (1979), 31. The humidity specification in this document is written by Leisher and conforms to his earlier work (loc. cit. [37]) by following Buck.

[40]. Barclay, R.L. Care of Musical Instruments in Canadian Collections, CCI Technical Bulletin No. 4, Canadian Conservation Institute, Ottawa (1978), 21.

[41]. Buck, R. 'The use of moisture barriers on panel paintings', Studies in Conservation, 6 (1961), 9-19.

[42]. Stevens, W. 'Rates of change in the dimensions and moisture contents of wooden panels resulting from changes in the ambient air conditions', Studies in Conservation, 6 (1961), 21-25.

[43]. In Search of the Black Box. A Report on the Proceedings of a Workshop on Microclimates Held at the Royal Ontario Museum, February 1978, Royal Ontario Museum, Toronto (1978 or 9).

[44]. Barrett, L.L. 'Insulation of cold surfaces to prevent sweating', ASHVE Transactions, 19 No. d661 (1923).

[45]. Wingspread, a Frank Lloyd Wright-designed home of 1934 in Racine, Wisconsin has been studied by Rose. The building has a roof deck with mineral wool and cellulose-product insulation applied directly to the underside of the sheathing, covered with a sprayed bitumen emulsion. This construction, which contemporary requirements for attic ventilation would forbid, is entirely consistent with the practice of the time, and has performed well.

[46]. Teesdale, L.V. 'Condensation in Walls and Attics', U.S. Forest Products Bulletin, Madison, (no date).

[47]. Rowley, F.B., A.B. Algren, & C.E. Lund 'Condensation within walls', ASHVE Transactions, 44, no. 1077 (1938).

[48]. Rowley, F.B., A.B. Algren, and C.E. Lund 'Condensation of Moisture and its Relation to Building construction and Operation', ASHVE Transactions, 45, no. 1115 (1939).

[49]. Assume wooden sheathing at an initial moisture content of 10% of oven-dry weight. Under worst-case conditions (40%RH and 70°F/21°C indoors, -19°F/-28°C outdoors, and no vapor barrier) Rowley's measurements suggest that after 60 days (two months) the moisture content of the sheathing to would rise to 25% of oven-dry weight, i.e. not to a dangerous level at the temperatures in the attic space.

[50]. It is interesting to note that, in a published discussion section at the end his 1939 paper, Rowley was asked by E.C. Lloyd if air pressures acted to force air into his test buildings or outwards; his response indicated that he did not comprehend the nature of the question. Rowley's experiments, after all, were concerned solely with diffusion. Lloyd was the first, but certainly not the last, to highlight the practical weaknesses of a pure diffusion model.

[51]. Rose, W.B. 'History of attic ventilation regulation and research' , Proceedings of Thermal Envelopes VI, Oak Ridge National Laboratory, Oak Ridge (1995).

[52]. Britton, R. Condensation in Walls and Roofs, Housing and Home Finance Agency Technical Papers #1, 2, 3 (July, 1948); Britton, R. Condensation in Walls and Roofs, Housing and Home Finance Agency Technical Paper #8 (April, 1948); Britton, R. Condensation in Wood Frame Walls Under Variable State Conditions of Exposure, Housing and Home Finance Agency Technical Paper #12 (June, 1949).

[53]. Britton, R., Condensation Control in Dwelling Construciton Good Practice Recommendations. Technical Bulletin #10, Housing and Home Finance Agency, Washington DC (May-July, 1949).

[54]. Condensation Control in Modern Buildings, Housing and Home Finance Agency (HHFA), Washington, DC (August, 1949).

[55]. Ramsey, C. & H. Sleeper, Architectural Graphic Standards, John Wiley and Sons, New York (1951), 328-330.

[56]. Hutcheon, N., Report on Quonset Building, Saskatchewan Hospital, North Battleford, typewritten report to NRC (Canada) (1950); Control of Water Vapour in Dwellings, Technical Paper no. 19 of the Division of Building Research, National Research Council of Canada, NRC No. 3343 (1954).

[57]. Lstiburek, J. & J. Carmody The Moisture Control Handbook, Oak Ridge National Laboratory, (1992).

[58]. Thomson, G. The Museum Environment, 1st edition, Butterworths, London (1978).

[59]. Ahrens, W. 'Controle et réglage de la température et de l'humidité dans les musées', Mouseion, 25/26 (1934), 125-131.

[60]. Brown, J. 'Hygrometric measurement in museums: calibration, accuracy, and the specification of relative humidity', in A. Roy & P. Smith (ed.s), Preventive Conservation: Practice, Theory and Research. Preprints of the Contributions to the Ottawa Congress, International Institute for the Conservation of Artistic and Historic Works (IIC), London (1994), 39-43.

[61]. Daniels, V. & S. Wilthew, 'An investigation into the use of cobalt salt impregnated papers for the measurement of relative humidity', Studies in Conservation, 28 (1983), 80-84.

[62]. Mechanical recording thermohygrographs calibrated at a single-point by reference to a sling psychrometer are capable of better performance over narrow humidity bands in the midrange (50 ± 5%RH) at constant room temperature — the conditions that Thomson was familiar with. In this narrow range the error depends almost entirely on the skill with which the psychrometer is used. However, even with the most scrupulous use, sling psychrometer error is probably not less than ± 4%RH (95% confidence limit) at room temperature and the error becomes greater as temperatures decrease. Uncertainty of hair elements will increase sharply if humidity oscillates outside the limits 30 to 70%RH between calibrations because of the inherently poor stability of the hair element, or if temperature varies widely. Mechanical recording thermohygrographs are not suitable instruments for the investigative study of extreme climates; their utility is generally confined to confirmatory mode of monitoring discussed in § 7.1 in reasonably stable internal environments. See j Brown (reference in note 59) for a full discusion.

[63]. Ashley-Smith, J., N. Umney, & D. Ford, 'Let's be honest - realistic environmental parameters for loaned objects', in A. Roy & P. Smith (ed.s), Preventive Conservation: Practice, Theory and Research. Preprints of the Contributions to the Ottawa Congress, International Institute for the Conservation of Artistic and Historic Works (IIC), London (1994), 28-31.

[64]. Colville, J., W. Kilpatrick & M. Mecklenburg, 'A finite element analysis of multi-layered orthotropic membranes with applications to oil paintings on fabric', in Science and Technology in the Service of Conservation. Preprints of the Washington Congress, International Institute for the Conservation of Artistic and Historic Works (IIC), London (1982), 150-156.

[65]. Michalski, S. 'Paintings - their response to temperature, relative humidity, shock and vibration', in M.F. Mecklenburg (ed.), Art in Transit. Studies in the Transportation of Paintings, National Gallery of Art, Washington, DC (1991), 223-248.

[66]. Mecklenburg, M. 'Some aspects of the mechanical behavior of fabric supported paintings', unpublished manuscript, Washington (1982).

[67]. Thomson, G. The Museum Environment, 2nd edition, Butterworths, London (1986).

[68]. Stolow, N. Conservation and Exhibitions, Butterworths, London (1987).

[69]. Erhardt, D. &. M Mecklenburg 'Relative Humidity re-examined', in A. Roy & P. Smith (ed.s), Preventive Conservation: Practice, Theory and Research. Preprints of the Contributions to the Ottawa Congress, International Institute for the Conservation of Artistic and Historic Works (IIC), London (1994), 28-31.

[70]. Mecklenburg, M. & C. Tumosa, 'Mechanical behavior of paintings subjected to changes in temperature and relative humidity', in M.F. Mecklenburg (ed.), Art in Transit. Studies in the Transportation of Paintings, National Gallery of Art, Washington, DC (1991), 173-203.

[71]. Mecklenburg, MF, M. McCormick-Goodhart & C. Tumosa, 'Investigation into the deterioration of paintings and photographs using computerized modelling of stress development', Journal of the American Institute of Conservation, 33 (1994), 153-170.

[72]. Mecklenburg, M., C. Tumosa & D. Erhardt, 'Structural response of wood panel paintings to changes in ambient relative humidity', in Proceedings of the Painted Wood Symposium, Williamsburg, VA (in press) (1996).

[73]. Richard, M., M. Mecklenburg, C. Tumosa & T Olstad, 'New vs. old wood: differences and similarities in physical, mechanical and chemical properties', ICOM Committee for Conservation 11th Triennial Meeting. Edinburgh, UK. Preprints. (in press) (1996).

[74]. AIC News, 20 (September, 1995).

[75]. Hackney, S. 'Framing for conservation at the Tate Gallery', The Conservator, 14 (1990), 44-52.

[76]. Brown, J. 'What can psychrometric data tell us?' in R.E. Child (ed.) Electronic Environmental Monitoring in Museums. Proceedings of a Conference Held at the Welsh Folk Museum, National Museum of Wales on 22nd April 1992, Archetype Press, Denbigh (1993), 41-47.

[77]. Eshøj, B. & T. Padfield 'The use of porous building materials to provide a stable relative humidity', ICOM Committee for Conservation 10th Triennial Meeting Washington, DC, USA. Preprints, Vol. II, ICOM Committee for Conservation, Paris (1993), 605-609.

[78]. Lafontaine, R. & S. Michalski 'The control of RH recent developments', ICOM Committee for Conservation 7th Triennial Meeting (Copenhagen). Preprints, Vol. II, ICOM Committee for Conservation, Paris (1984), 84.17.33-36.

[79]. Staniforth, S., B. Hayes & L. Bullock 'Appropriate technologies for relative humidity control for the museum collections in historic houses', Preventive Conservation: Practice, Theory and Research. Preprints of the Contributions to the Ottawa Congress, International Institute for the Conservation of Artistic and Historic Works (IIC), London (1994), 123-128.


[Search all CoOL documents]


URL: http://cool.conservation-us.org/byauth/brownjp/humidity1997.html
Timestamp: Sunday, 23-Nov-2008 15:20:04 PST
Retrieved: Saturday, 20-Dec-2014 15:42:50 GMT