| Abstract: | SUMMARIES. At the archaeological site of Gordion, Turkey, all registered ceramics are routinely desalinated. The standard desalination treatment was investigated by the author with a view to minimizing damage to the ceramics, as well as the amount of water used and the time taken for desalination. An equation that standard¬izes conductivity measurements in disparate desalination situations is presented. A series of experiments was performed to clarify the behavior of salts at Gordion. Two variables which directly affect the amount of water needed for desalination were considered: the frequency of water changes and the weight-to- volume ratio. The experimental results were compared to on-site experience, and a revised endpoint for desalination is proposed.
CONCLUSION. Desalination occurs on a predictable schedule across a range of Gordion ceramic types. The desalination schedule is not affected by either of the two variables tested: by whether or not the desali¬nation baths are changed—ever—during the desali¬nation process, or by the water volume used. The actual amount of salts extracted was affected by both variables, with more frequent water changes and larger volumes of water extracting slightly larger amounts of salts. However, the differences in the amounts extracted are not believed to be great enough to impact practical desalination. For all ceramics monitored, both experimentally and on site, 80-90% of salts removed were released in the first four days of monitoring. At a maximum of four or five days the slopes of the kadj per day graphs changed, always at an adjusted conductivity of 2 or below. After that point, even prolonged soaking produced no appreciable increase in the amount of salts removed. For Gordion ceramics, across a range of ceramic types, desalinating past the point at which the desalination graphs level off is unproductive and potentially deleterious. The change in slope of the kadj per day graph (Figure 1) is proposed as the desalination endpoint. While it is preferable to calculate and track the k per day in order fully to document a desalina¬tion, note that the change in slope is identifiable as the point at which the conductivity no longer changes dramatically. Using this method, a conduc¬tivity reading does not need to be converted to kadj; it only needs to be monitored for the magnitude of change. It is not necessary to achieve 'equilib¬rium'--only to observe a notable lessening in the difference between conductivity readings. (The actual Jk observed at this point will vary with vol¬ume of water.) The point at which the change in conductivity slows markedly, regardless of the numerical readings at that point, is the endpoint at which desalination should be terminated. k adj =2 as a numerical endpoint at Gordion It is advantageous to have a numerical endpoint. in order to judge the relative salinity of the object, to estimate where in the process the treatment stands, and to know immediately when the end has been reached. At Gordion. the change of slope consis¬tently occurs at adjusted conductivities of 2 or below. It appears that the target endpoint—the change in slope—corresponds to a numerical endoint of kadj = 2 or below. Only a small percent¬age of Gordion ceramics require four or five days to reach kadj= 2; most of the ceramics monitored reached kadj= 2 after only two or three days. This degree of'àesalination may be adequate. In 1999 and 2000, ceramics desalinated on-site at Gordion were removed at kadj = 2 or below. They will be monitored for future stability. |
|---|
