AN INTEGRATED APPROACH TO RISK ASSESSMENTS AND CONDITION SURVEYS
6 INTEGRATING RISK ASSESSMENTS AND CONDITION SURVEYS
The integration of the collection condition survey with the risk assessment can be done by adding a list of probable causes of damage, in addition to damage type, to the condition database. Murray has successfully attributed damage to causes in condition survey design (Edwards and Murray 2002), and the notion is advocated by Ashley-Smith (1997) when practicing conservators are involved. Both the causes and effects of deterioration to an object are based on “risks,” and therefore the relationship between hazard and damage can be expressed this way. Establishing a one-to-many database relationship between each hazard and the objects surveyed (fig. 6) means that the list of probable causes is the same list as the risks looked for in the risk assessment. As a result, the visible impact of each risk can be viewed in terms of collection condition.
This link acts as the connecting pathway between separate assessments, similar to Marr's (1982) theory of perception. The format of risk assessment can be based on Waller's (1999) equation of risk—P × FS × E × LV—in which scores allow locations, collections, and risks to be compared and given priorities, where P is probability of damage, FS is the fraction of the collection susceptible to damage, E is the extent of damage, and LV the expected loss of value. Waller's “Loss of Value” (Waller 1999) is, of course, expected loss from which the condition is the result.
This simple method allows probable causes of damage to be related to actual risks, so an immediate idea of which deterministic risks are having an impact on the collection can be established. It is not always possible to precisely determine causes of deterioration, and there may well be more than one (Koestler et al. 1994; Ashley-Smith 1999, 2000), so several causes can be added to the object section. The method consequently addresses the problem associated with cause-effect relationships mentioned in table 1. Risks need to be designated one category in assessment to reduce potential for the same risk to be recorded twice (Waller 2003). This connection to various possible causes of deterioration offers an insight into synergistic effects of risks to collections.
Using a one-to-many relationship, the connection is logically valid. The fact that different causes may be attributable to a given effect is acknowledged in this relationship. The problems associated with inference of cause from effect, mentioned in table 1, are reduced.
Taylor and Watkinson (2003) point out that the difference in deterioration mechanisms for different materials can be helpful. Causes attributed erroneously for one material would not be recorded for other materials. For example, possible causes of discolored paper may be high temperature, high RH, external pollutants, inherent instability, or a combination of these. Metalwork may indicate that corrosion is present, and high RH the cause, stone may show salt efflorescence, and wood may show dimensional change or mold. The real causes are drawn out because of these differences in effect. Therefore, the impact of recording any erroneous causes will be limited to particular types of material, where real causes would be consistent among different materials. The attribution of these causes is something that would be consistent in a risk assessment but not in a condition survey.
For preventive conservation, however, not all damage provides useful information. As mentioned earlier, damage that took place before an object entered a museum does little to indicate the effect of its present environment. Damage that clearly relates to past environments, such as broken archaeological ceramics, should be omitted, as it does not determine if an institution is “succeeding in its basic duty to preserve collections” (Keene 2002, 139). Condition scoring is similar to that of Keene's “stability,” although it is not “predicting the rate at which an object is likely to deteriorate” (Keene 2002, 147). “Stability” is the forecast of risk to single objects rather than to the whole collection from risk exposure, and it overlaps with the risk assessment. Waller (2003) points out problems with this approach, using an example of contamination from old treatment methods not indicating contamination from future treatment methods. Such damage may indicate a greater instability of treated objects to environmental risks but cannot help forecast future treatments. It is intended that only risk assessments look at forecast damage, and condition surveys look at existing effects. Only damage that can be observed, rather than used for forecasting, should be recorded. “Present” condition is suitable for many type 3 risks, and “recent” damage is suitable for risks such as handling and pests. Evidence of damage from known hazards is important for the overall assessment. Inherent object instability, such as salt efflorescence from buried ceramics or yellowing from poorly processed cellulose nitrate negatives, should be identified because it has implications for preventive conservation decisions. The worse the kind of damage identified, the more indicative of a threat to the collection. For example, high RH might facilitate corrosion that could be a threat, in the case of hematite, or comparatively benign, in the case of patina.
Surveying condition must take into account recent movements of objects and factors that might create discrepancies between assessments of risk and condition. Existing damage is sometimes hard to attribute to particular risks. An example is UV and visible radiation, where fading could have taken place at any time. However, the circumstances around an object, such as being in a box or on open display and faded only in parts where the object is exposed to display lighting, will allow one to make a judgment. There may be occasions when a risk is not present at all (Waller 2003), which will inform the condition survey in terms of identifying past damage. As mentioned earlier, some damage types are easy to omit, and odd occurrences are reduced by using a large sample of objects. Having to consider several possible causes of deterioration means that any correlation of risk and condition is not direct, but this correlation is not always the case in reality either (Ashley-Smith 2000), and the consideration of several possible causes makes the determination of synergis-tic effects easier.
Discrepancy between risk and condition does not necessarily mean that one or the other is correct, but it allows expectation and observation to be integrated to focus on any uncertainty, whether that is due to assessment or object behavior. Cognitively, addressing such discrepancies increases performance in real-life reasoning situations (Dunbar 1993).
The relationship between risk and condition does change with the type of risk, but all risks have a connection to condition at some level. All interpretation is clearly based on the relationship between hazard and damage. The strongest relationship is, of course, with deterministic risks, but all kinds of recent damage can help refine present or future assessments. Table 4 illustrates the different relationships that different types of risk have with condition. The relationship is stronger between type 3 risks and condition surveys, but risk assessments are well suited to type 1 risks. Corroboration between a risk assessment and condition survey indicates both “exposure” and “consequence” of risk. Lack of corroboration may be due to a number of reasons, such as damage that has been overor underestimated, risk that has been overor underestimated, a lack of visible effects, or damage occurring outside expected bands of preservation or not occurring in conditions expected to promote deterioration.