The second temper variable(s), generic temper source, is used to characterize the geographic and tectonic origin of the temper grains observed. If a project has a large number of sherds with mixed tempers—say sand and crushed rock—then two generic temper source variables are used, one for sand, and one for crushed rock. A given sherd is attributed to a generic source based on binocular microscopic observation of the sand grains known to define particular geographic and tectonic settings and/or the distinctive crushed rock types seen in the paste.

The third temper variable, specific temper source, was used to characterize the petrofacies of origin for the observed temper grains. A given sherd was attributed to a specific source based on the binocular microscopic observation of the distinctive suite of rock fragments and monomineralic grains in the abundances known to define a particular petrofacies . In this way, we used the precise data gained from the thin section studies and semiquantitative abundance assessments as a means of assigning tempers to petrofacies without thin-sectioning each sherd.

The difference between the "generic" and "specific" temper source attributes lies in the finer level of spatial resolution implied by the petrofacies. In practice, the information represented by the "generic" attribute is redundant with the information represented by the "specific" attribute when the sand temper observed in a given sherd permits its assignment to a petrofacies. It is sometimes the case, however, that the temper does not exhibit any or all sand grains necessary for a specific assignment. For example, it is often difficult to characterize the specific temper source of small or badly burned sherds. However, the temper grains that can be observed in small or badly burned sherds are often sufficient to categorize the generic origin of the sand. Important compositional and limited provenance information would be lost in those cases if only an attribute recording the petrofacies of origin was required during analysis, because all generic petrofacies assignments would, by definition, have to be recorded as "indeterminate."

Once the sherds have been counted, their data is submitted for discriminant analysis as “unweighted” or unknown samples. The functions calculated for the sand samples are used to classify the sherds—the sherds themselves do not contribute to the classification scheme.

After statistical analysis is complete, the predicted group memberships are compared to those predicted by both the petrographer and the ceramicist, and a final petrofacies determination is made. In most cases, there is agreement between the three measures; however, we have found that there will be a small number of samples in every project on which no-one can agree. This is to be expected in a natural system where boundaries between petrofacies are gradational rather than abrupt. These are characterized as far as possible without bloodshed, since trained petrographers and ceramicists with temper identification skills are hard to find.

Once the final sherd characterization has established, the data are ready for archaeological interpretation by the ceramicist, using the stratified sherd sample to represent the overall data set.

Home | About Us | Compliance Services | Analysis Services
Publications | Selected Projects

© 2004 Desert Archaeology Inc.