Is Non-destructive Provenancing of Pottery Possible With Just a Few Discriminative Trace Elements?

Detlef Wilke,Dagmar Rauch,Patrick Rauch

doi:10.1080/20548923.2016.1209030

Detlef Wilke, Dagmar Rauch + Show 1 more

Open Access

https://doi.org/10.1080/20548923.2016.1209030

Copy DOI

Abstract

This study outlines how to perform a fully non-destructive compositional analysis of pottery, but with the limited number of discriminative elements quantifiable by pXRF. The spectrometer was calibrated with fired clay samples spiked with Ti, Fe, Ga, Rb, Sr, Y, Zr and Nb. Methodological outliers resulting from surface modifications and paste heterogeneity were identified by multiple spot analysis per fragment and sorted out. From ten late medieval/early modern potting centres located all over Germany 75 to 360 waster sherds were analysed, resembling plain earthenware, proto-stoneware and glazed stoneware. The 90% quantiles of the assemblages could be distinguished by simple bi- and trivariate scattergrams of the trace elements, although the overall compositional difference was low. As a general trend paste group distinction of similar wares got less pronounced with decreasing geographic distance of the production sites. With the limited total spread of elemental concentrations, the n≤15 dimensional space of discriminative elements is too small to separate hundreds of paste recipes generated over centuries in a densely populated territory. Thus a restriction to the anthropologically reasonable context is required when compositional provenancing of unknown pottery is undertaken, and the successful resolution of production sites needs to be confirmed on a case by case basis. Statement of significance Apart from cultural heritage and art historical conservation aspects, non-destructive XRF analysis allows for high, though non-automatized throughput of even very small ceramic fragments, which strongly improves the significance and validity of location, ware and style specific reference groups. However, pottery also provides considerable methodological constraints to true non-destructive compositional analysis, and the difference in chemical pattern between production centres is generally small. Diligent instrument calibration and a stringent protocol are therefore prerequisites for distinguishing geographically neighbouring pottery produces, and assemblages with adjacent and partially overlapping chemical composition in general.

Highlights

XRF is a surface analytical method which as a precondition requires homogenous sample material of infinite thickness, larger than the penetration depth of the emitted photons of the heaviest element to be quantified in the sample matrix
Methodological outliers resulting from surface modifications and paste heterogeneity were identified by multiple spot analysis per fragment and sorted out
In the case of pottery it is the principle heterogeneity of the fabric, clay and nonplastic minerals, as well as the many intentional and unintentional surface modifications which contradict the requirement of sample homogeneity

Summary

Introduction

XRF is a surface analytical method which as a precondition requires homogenous sample material of infinite thickness, larger than the penetration depth of the emitted photons of the heaviest element to be quantified in the sample matrix. Since it was argued that Compton ratioing is essential for mass correction in non-destructive XRF analysis (correction of “practical” matrix effects, Shackley 2011) the performance of the Compton and the Rayleigh scatter peak for normalizing w/w (LoI), w/v (density, micro-porosity) and micro-crystallinity differences was checked in air dried unfired and in fired clay samples, covering both medium (800–850°C) and high fired pottery (1150 °C) This is important if assemblages containing regular earthenware, high fired earthenware and fully sintered stoneware are analysed, and since medieval waster material typically consists of overfired fragments.

Conclusions

Findings

Conflict of interest