By Robert Waller and Stefan Michalski

Consider a day in the life of a museum conservator—we'll call her Carrie—working in the field of preventive conservation (PC).

At 8:00 am on Monday, Carrie arrives at her office. Before her coat is off she notices several small plastic bags on her desk left by the weekend staff. Each packet contains one or more insects, accompanied by a form showing where they were found. The samples in one of the packets—gathered near a temporary exhibit case containing material loaned from a smaller institution—are live beetle larvae, but not the genus of Dermestid commonly seen. Carrie makes a note to identify the pest and examine that display case. Among Carrie's emails is a request from a curator to raise for three months the temperature in the textile collection room from the normal 17°C to 22°C; an elderly scholar will be visiting and working in the collection. Carrie knows it is possible to predict increased rates of thermal degradation of various fibers and the increment in risk due to pests given a change in temperature, and she recalls that she can call a Canadian colleague who put the prediction model into a spreadsheet. Even so, she will need to speak with the curator and collection manager to determine the benefits of the proposed work and whether alternative locations for the work are possible. Then she will need to consider and communicate the transport and temporary housing risks associated with relocating part of the collection.

Another email message advises of the need to replace the building's main water supply pipe. Carrie replies asking for a meeting with security and facilities staff and lists issues to consider, including flood risk, sprinkler function and fire protection, insurance continuity, and humidification system operation. The phone rings with news that a condensate water leak from an air-conditioning unit has been discovered in the rare book vault. While there is no need to muster the water emergency response team that Carrie assembled last year, she dispatches a conservation technician to assess that situation, ensure cleanup is occurring, take relative humidity readings, and report back. As she sends an email to facilities management asking for a report on the leak's cause and the proposed action to fix the problem, the museum director's personal secretary calls to say that the director would like to see her.

"You recall our exhibit of our six Turner watercolors earlier this year," says the director. "Well, I had dinner with Mr. Smith on Saturday. He's prepared to solicit his corporate board for donations for our new capital project and feels that having those watercolors displayed in his boardroom for six months would help open the pockets of his board members. I know you were keen to get those paintings back in the dark when the exhibit was over, but you must understand how important the capital project is and how much better environmental conditions will be once it's complete. I'd like you to advise us how those Turners can be safely put on display in Smith's boardroom." He turns back to his desk, then pauses. "Oh, and Carrie, I'd like your idea by noon tomorrow."

conservation image
 

Back at her desk, Carrie considers the fact that the policy-permitted light dose for the Turner watercolors for the next five years has been used up. She's certain the model that the policy is based upon is reliable, but she also knows that the sensitivity of the paintings was conservatively estimated. Still, what about the much more serious risks of transporting the watercolors and leaving them to the vagaries of a working boardroom—chairs on wheels, food and drinks, preoccupied occupants, nighttime cleaning staff, and ordinary office locks? How can she convey to her director this mix of certain and uncertain concerns? Her information should support a defensible decision on the loan in a way that identifies and prioritizes each risk, offers ways to mitigate those risks, and gives the cost of those options—but by tomorrow?

As she handles these issues, Carrie knows she must also contribute to the museum's planning cycles and set priorities for improving care among ten major collection units, not to mention the museum's heritage building. A conservation survey exists, but it is more than five years old. Several departments are unhappy with the low priority of their particular collections. Besides, Carrie knows that new knowledge has emerged since the survey, knowledge not reflected in current priorities.

What she needs is a system to do all this—or at least to help her do it.

Challenges for the Conservator

While many professions have become exceedingly narrow, preventive conservation has evolved to become one of the most interdisciplinary fields. It uses knowledge from materials science, building science, chemistry, physics, biology, engineering, systems science, and management, as well as a host of technical fields. In addition, it requires an understanding of multiple value systems within many cultures and an appreciation of how cultural properties deliver value to those groups. Carrie exercises a Renaissance breadth of knowledge in just one working day and, in addition, taps into her personal network of international experts. Not every collection can be so lucky. Today, perhaps none can.

We can consider this breadth of required knowledge for the preventive conservator the ''encyclopedia'' challenge. Depressingly obvious to Carrie is a further reality of encyclopedias: they need constant revision by experts, and she has no time to read the latest one.

Even if the encyclopedia on collection risks can be organized into a few large themes, each will contain hundreds of independent subentries. In a single decision dilemma, such as the Turner watercolors, Carrie contemplates a multiplicity of entries from her risk encyclopedia: from light damage on several colors, to a dozen or so sources of physical damage, to spillage of alcohol, to fire, and theft. Traditionally, it has been difficult enough to develop sound advice on how to control any one risk, but balancing different risks—finding a single scale to apply to all of them—has not been normal practice. In fact, attempts to do so have been dismissed as "trying to compare apples and oranges."

Even with all the right facts from the encyclopedia at hand (and the problem of comparing apples and oranges somehow solved), the Turner watercolors dilemma becomes a formidable array of contributing factors. Calculated risks and benefits, in turn, need to be balanced against the larger, long-term risks to works in the museum—which is itself a huge array of related factors. If one or two entries in the risk encyclopedia are updated, then all interacting risk estimates need to be recalculated. So even if one reads the whole encyclopedia and also finds a way to compare apples and oranges, redoing this each time the encyclopedia is updated, let alone for every new minicrisis, is impossible.

If preventive conservation is so huge and complex, how has it been done up until now?

consesrvation image
 

Writing a definition of preventive conservation and its activities, and then writing its history, tends to provoke disputes over the meaning of the term and over the roles of conservators, scientists, curators, housekeepers, and many others in its development. What seems without dispute is that human beings have always kept and cared for those things that are labeled precious. Preservation is an ancient human activity.

Modern philosopher and historian John Ralston Saul proposes that six core qualities allow us to act as reasonable, decent human beings: common sense, memory, intuition, imagination, reason, and ethics. Saul argues that over the last few centuries, and especially in the twentieth, reason has assumed a dangerous dictatorship over the other qualities. This dictatorship of reason has fed, and has been encouraged by, narrow professional specialization, specialized technical languages, and the growth of large institutions.

conservation image
 

If we apply Saul's ideas to a history of preservation, it would seem fair to say that from ancient times, memory and common sense provided the preservation encyclopedia, while a mix of common sense, intuition, and reason handled the apples and oranges problem, as well as the complexity problem. Imagination helped with innovative solutions, and ethics tried to keep it all, well, ethical. Although a large amount of practical preventive conservation activity still relies on a balance of these six human qualities, ''modern'' preservation, or preventive conservation, is very much about the rise of reason, through scientific knowledge and professional expertise. Carrie is stuck trying to use an enormous and highly fractured modern encyclopedia, with no tools other than her own common sense and intuition to solve the problems of apples and oranges and of complexity.

Toward a Risk Management Future

Current practice in preventive conservation involves standard survey methods based on checklists, supplemented by professional judgments and focused investigations. From a decision making perspective, this is a ''satisficing'' procedure in which the conservator steps through checklists, skipping over topics that she is satisfied are not critically important to preservation. This rational approach, like all approaches to rationality, relies upon bounded rationality. A fully rational approach would require that all possibly relevant information be obtained and used in arriving at a decision or evaluation. The cost of—and time required for—a fully rational decision is infinite. Instead, a conservator evaluates, using heuristics (simple rules of thumb or intuition), whether there is an advantage in seeking more information before deciding whether a particular issue, such as a risk to a collection, is significant or not.

Checklists give the illusion that the apples-and-oranges problem has been solved. But it has not. For example, there will be questions on fire control and questions on humidity control, among many others, but the relative significance of each item within a group of items, and of each group within the whole survey, remains unknown. Without a formal structure or a common assessment scale for all items, judgments are impossible to evaluate objectively. They can be inconsistent, biased, or both.

German psychologist Dietrich Dorner, in his 1996 book The Logic of Failure, wrote, ''Methodism is likely to flourish in those situations that provide feedback on the consequences of our actions only rarely or only after a long time. In particular, if our plans apply to a field in which we rarely act, our planning gradually degenerates into the application of ritual." Unfortunately, we too often see ritual in our profession. One example is a tenacious holding to magic numbers for humidity and temperature standards, even in situations where their application is not just wasteful but counterproductive. The major problem is the sparse, slow, or even absent feedback to our actions. Did reducing humidity fluctuations by expensive machinery improve preservation? If so, by how much? And if by that much, was it worth it? The impressive machinery and flat-line humidity data must not be confused with an ultimate goal. They are simply hopeful intermediaries that are easily observed, as are improved facilities in general. They might contribute to the preservation goal but they are not the goal.

The escape from ritual comes from stepping back to conduct a full analysis that embraces real uncertainties in the system—in other words, taking a risk management approach. Preservation is then the cost-effective reduction of the total of all predicted risks. There are three key elements:

  • a common scale for magnitude of all risks, such as fractional loss of value per century, so that we can compare apples and oranges;
  • a prediction of the magnitude of each risk if nothing is changed; and,
  • a prediction of how these magnitudes will change if certain improvements are made. This reduction in risk is a prediction of expected results, which—uncertain though it is—is the only reasonable guide we can use for our preservation decisions.

A subtle distinction, which we note here briefly, is that uncertainty enters risk assessments in two ways:

Ignorance: Information may be lacking, either in the survey resources or in the encyclopedia of our knowledge. We can reduce this uncertainty, but we would be wise as a profession to focus on significant uncertainties rather than on insignificant ones.

Variability: Some risks occur ''by chance,'' so it is uncertain if Carrie's museum will ever experience them, just as we will never know if we will win a lottery. It is certain, however, that the chances are predictable and changeable—having two tickets doubles our chances at the lottery. In most cases, the expected effect of these by-chance risks can be influenced.

In many parts of the world, new ideas for approaches to preventive conservation are taking hold. Many, if not most, of these involve risk-based strategies. The Risk Map of Cultural Heritage in Italy being developed by the Istituto Centrale per il Restauro is just one example of a national-scale, risk-based PC project. Higher-level degrees in conservation being offered by schools such as the Royal College of Art, University College London, and the Institute of Conservation at the University of Göteborg increasingly involve risk-based preventive conservation research topics.

An Expert System

So what about the system that Carrie needs? This system would maintain an up-to-date encyclopedia of knowledge, take advantage of the common risk scale to compare apples and oranges, use a sufficient array of details, and make complex calculations instantaneously. Further, it would allow Carrie to change one or two details (i.e., enter possible improvements) and then recalculate to see the predicted results of the proposed improvement. It would also allow Carrie to explore all the known facts of the Turner watercolor dilemma and evaluate different options. In fact, the system could automatically suggest those factors that contribute most to total risk and that might be addressed first for the most cost-effective risk reduction.

The Canadian Museum of Nature and the Canadian Conservation Institute are working in partnership on such a system. This long-term project builds on the systematic risk assessment approaches that the authors developed over the last decade using computerization, in the manner of what were previously called ''expert support systems'' but which may now be considered ''knowledge-base systems. ''Internally, the system will build on the checklist approach, but it will allow each question to come with comprehensive lists of options. To each option will be attached either a summary expert opinion of its significance or, where possible, a predictive mathematical model. The system will integrate the effects of buildings, collection rooms, cabinets and boxes, jars or bags. It will also incorporate the influence of museum policy and procedures in reducing risk. The system will ask users questions about their situation. Where are you? What kind of collection do you have? What kind of building is it in? Beyond the first few questions, further questions will be presented according to the likely importance of further information. For a paper-based collection, for example, questions related to water and relative humidity will likely be important. For a gemstone collection, in contrast, questions related to security would be more important and more carefully explored.

It is a large task, but patterns are evident in risk analysis employed in other fields, including the application of the key elements we listed earlier: a common scale for magnitude of all risks, a prediction of the magnitude of each risk if nothing is changed, and a prediction of how these magnitudes will change if certain improvements are made.

During the last two years, we widely explored the field of risk analysis, looking into systems applicable to ecosystems, engineered systems, and marine and aviation transport safety, among others. We are also researching available geo-referenced sources for natural hazard potential and for climate norms; once users enter their geographic location into the system, the system will provide the flood risk, the weather, the frequency of damp days, and so on. Similarly, once users enter their building type or wall fabric, the system will know such things as the fire rating or the time it takes for an expert thief to break through. There is a great deal of relevant knowledge already computer accessible.

There is also a great deal known but not readily available. We therefore need a system that is social, open, and evolving over time. It must be an accessible framework in which PC scientists and practitioners in all disciplines can combine their knowledge, observations, concerns, and best advice. By pooling expertise, the system will both leverage and ratify each expert's advice and will make the advice available according to a user's needs.

conservation image conservation image
Enlarge Enlarge

Preventive conservation advice offered to a user will be guided by the assessed importance of risks. To some extent, suggesting a range of answers to questions is educational in itself. For example, responding that archives are in: (1) piles on the floor, (2) folders on shelves, or (3) acid-free folders in archive boxes on shelves, then immediately seeing the impact on estimated risks, should be educational.

Basic risk mitigation advice may be offered from a system database. At the very least, the system would generate relevant search terms for any other searchable knowledge bases of conservation publications, via the World Wide Web. Wherever the system identifies large uncertainties in risk caused by missing or uncertain information, it will have identified areas of priority for conservation research.

The Best Available Prediction

The potential for preventive conservation to progress is limited by our ability to surmount three major challenges. The encyclopedia challenge and the complexity challenge are exactly what computers are good at. The apples-and-oranges challenge can be solved by adopting a common scale for all risks, as is normal in any risk-management business. The authors have promoted the scale ''fractional loss of value per unit time,'' as described elsewhere.

John Ralston Saul reminds us to be humble, even cautious, about the power of reason alone. The Terminator movies remind us to be cautious about reason and computers combined. We need the equivalents of common sense and intuition in our computer system. Part of those equivalents will be the pooled expert judgments at many levels within the system. These will be the heuristics that constitute the system's bounded rationality. Many of the system's calculations will use tables of expert judgments. The system's outputs, ultimately, will be pieces of advice written by practitioners in the field, as well as by the authors. The system will, in fact, amplify common sense and intuition and imagination by sharing the best examples of each. Ethics will be left to the users themselves, such as Carrie and her director. That will not change. But they will know the implications on preservation much, much better.

The risk management approach does not need a computer system—it will just be much easier with one. What it does need is recognition that effective preservation cannot be measured by easy observables as soon as the money is spent, or even in our lifetimes. It can only be measured by the best available prediction of those effects, however imperfect those predictions might be.

Robert Waller is chief of the Conservation Section of the Canadian Museum of Nature in Ottawa. Stefan Michalski is the manager of the Preventive Conservation Services Division of the Canadian Conservation Institute, also in Ottawa.