Phosphate Analysis: accuracy and validity of the methodology
Portable spot test methods target extractable active phosphates. In the field, portable spot test approaches can sufficiently distinguish between different levels of phosphate content to infer phosphate patterns from the results (Holliday and Gartner 2007: 313). Improvements on the field-based procedure have created a semi-quantitative procedure that measures the results of phosphate extraction through colorimetric determination (Holliday and Gartner 2007: 313-314). The accuracy of spot test methods can be increased through performing the test in a controlled environment, and creating a modified chart of phosphate levels by recalibrating the phosphate levels set out by Eidt and Wood (1974) to break down every level into quarters (i.e. 1.0, 1.25, 1.5, 1.75, 2) from zero to six (Ullrich 2007).
The ability of portable field techniques to correctly identify archaeological sites and features has been tested through the examination of known archaeological sites and areas thought to be devoid of settlement. Soil from these areas were tested and followed by archaeological excavation to determine what features were influencing the identified phosphate patterning (Bjelajac et al. 1996). Soils retrieved from various depths, as well as from features, have been tested to determine the most accurate testing method. These tests were undertaken to assess the usefulness, as well as the precision, of portable field methods of phosphate analysis (Table 4).
Bjelajac et al. (1996) tested the accuracy of the Eidt and Wood spot test methodology for identifying buried archaeological features. The study showed a 97% accuracy rate on invariant classification for the Eidt and Wood methodology, as all but one sample came back with a reading that corresponded appropriately to the known boundaries of the site. The one sample that did not correlate was from a highly disturbed area where a lot of wind and water erosion had taken place. The ways in which different portable spot test techniques compare were examined through the use of four different spot test methods on the same set of soil samples by Terry et al.(2000: 157). The Mehlich II acid extraction test (Mehlich 1984), the Olsen bicarbonate extraction (Olsen and Sommers 1982), perchloric acid digestion for total-phosphorus (Terry et al. 2000: 157), and the Eidt ring test (Eidt and Woods 1974) were performed on the same soil samples. The mean result for all tests was the same. The Mehlich method produced higher phosphate levels than the other spot test methods. The Terry et al. (2000) and Olsen and Sommers (1982) results were comparable in most cases, but were less consistent than the other two methods. The Eidt ring test method proved less sensitive than the other methods, but more reliable. The phosphate ratios between soil samples remained the same as in the other tests. This study showed that while these tests vary in the amount of phosphates that are measured, the resultant patterns of relative amounts (ratios) are the same (Terry et al. 2000: 158, 160).
Many tests have been undertaken to determine the validity of portable spot test methods compared to more intensive laboratory tests that assign a parts per million numerical value to phosphate content to determine how the two methods compare. The ways in which these studies can aid in site interpretation have also been analyzed. Initially, Eidt did much of the work to counteract skepticism regarding the validity of phosphate testing in archaeology, especially in the correlation between spot test and laboratory test methods. He established that the average inorganic phosphate content of sedimentary rock is roughly 200 parts per million (ppm) (Eidt 1984: 27).
The comparison of laboratory testing methods with portable field testing methods shows that spot test levels are not as exact as laboratory methods, but do fall into definable ranges that are comparable to the laboratory tests (Hamond 1983; Smyth et al. 1995). The quantitative laboratory levels that correspond to spot test levels in one example should not be projected onto other sites. The levels obtained through these studies cannot correspond to any sample set other than their own. The tests do indicate, however, that correlation between laboratory test results and portable field test results occur in the vast majority of cases. The level of definition obtained from portable field techniques is thus sufficient for the accurate determination of phosphate patterning over archaeological sites.
The main drawback to phosphate analysis protocols in archaeology is the inability of the technique to determine the exact activity and/or organic component that caused specific phosphate increases (Wilson et al. 2005: 1097). However, this issue can be addressed through analysis of the character of specific phosphate patterns and creation of a catalog of phosphate patterns tested by excavation. Activity areas with dissimilar uses can be identified through variations in the specific details of phosphate patterning. This allows for the breakdown of an archaeological site into zones of distinct purpose. Phosphate patterning around hearth features, for example, generally consists of an area of relatively high phosphate content with inclusions of smaller areas of relatively very high phosphate content. This pattern is characteristic of hearths; it results from the general scatter of ash and organic debris associated with hearth-based activities and the specific incorporation of high phosphate content food scraps generated from food production and consumption activities. Phosphate analysis is paired with other geo-chemical analyses to help determine what activities increased phosphate content, such as multi-elemental analysis through ICP/AES and similar techniques (Macphail et al. 2004; Middleton 2004; Middleton and Price 1996; Wells et al. 2000; Wilson et al. 2005). Excavation is often the only way to positively identify the specific function of designated activity areas (Parnell et al.2002), which is why the standardisation of phosphate interpretation and the creation of a catalogue of phosphate patterns tested through excavation is important to forwarding use of the technique.
Limiting factors have also included misunderstandings of the ways in which the technique can be integrated into a well rounded archaeological research model. The trend for slower, more expensive, quantitative laboratory methods rather than accurate, non-invasive time- and cost-effective spot test methodologies has restricted the use of phosphate analysis. Semi-quantitative spot test methodologies have proved successful on archaeological sites for the identification of features and phosphate patterns, with a 97 percent accuracy rate (Bjelajac et al. 1996: 246).