Compositional variability of archaeological ceramics in the eastern Mediterranean and implications for the design of provenance studies Anno Hein ⁎, Vassilis Kilikoglou Institute of Nanoscience and Nanotechnology, NCSR ‘Demokritos’, Aghia Paraskevi, 15310 Athens, Greece abstract article info Article history: Received 10 August 2015 Received in revised form 18 December 2015 Accepted 12 March 2017 Available online 27 March 2017 The so-called ‘chemical fingerprints’ of production sites that are determined in provenance studies of archaeolog- ical ceramics comprise not only an estimate of the intrinsic chemical compositions but also an estimate of their variability. The compositional variability of ceramics from a specific production site is affected by the natural var- iability of the raw materials used, variation in the ceramic production process and potential post-depositional al- teration. In order to characterise the production site as whole, average concentrations and their variations are estimated on the basis of a necessarily limited number of samples selected for analysis. The sampling strategy therefore has a significant impact on the results. The compositional variability is interfered from uncertainties in- troduced during the analysis subject to the analytical method chosen. This paper provides an overview of the sources of variability that influence such analyses. Case studies challenging the classical unbiased provenancing approach are presented using examples from the eastern Mediterranean region. © 2017 Elsevier Ltd. All rights reserved. Keywords: Archaeological ceramics Chemical fingerprint Pottery production Provenance Variability 1. Introduction Provenance studies of archaeological ceramics represent one of the most popular subjects in the field of science-based archaeology. During the past few decades, several mineralogical and chemical techniques, and combinations from both, have been developed with the aim of un- derstanding fabric or paste variability, grouping like materials and assigning a provenance to the resulting groups (Day et al., 1999; Maniatis et al., 1984). The first methodology to be investigated and de- veloped for such provenance studies was based on elemental composi- tion, which has been a well-established approach in archaeological science since the end of the1950s (Richards and Hartley, 1960; Sayre and Dodson, 1957), and the eastern Mediterranean has often been the focus of research and provided the basis for new developments in terms of analysis and data interpretation (Jones, 1986). The basic tenet of provenance studies is that ceramics from a specific production site present a chemical composition distinct from ceramics produced elsewhere, as a result of the use of different raw materials and/or different methods of clay paste modification. This composition can be used to define a local chemical pattern or ‘chemical fingerprint’ of a production site or even a single workshop. In an unbiased approach, an arbitrary number of patterns can be attributed to a particular site and the assignment of provenance depends on a straightforward compari- son of the resulting patterns. In practice, however, a series of constraints has to be considered, such as the natural inhomogeneity of raw material sources. The ‘provenience postulate’ assumes that chemical differences with- in a single source of material must be less than the chemical differences between different sources (Weigand et al., 1977). However, analytical studies of clay deposits have shown that, in particular, deposits from the same geological context sometimes present very similar chemical compositions, even though they can be considered different in terms of geographical distance. Furthermore, the natural range of element concentrations appears to be restricted in terms of correlations and in terms of absolute values (Hein et al., 1999). Case studies occasionally re- port that specific accessory minerals, not homogeneously distributed in the clay, can affect the element composition within a single clay source. The ‘human factor’ must also be considered a source of variability. It can be assumed that craftspeople within the same cultural context but oper- ating at different sites tried to select raw materials with similar physical properties for the production of similar vessels. In many cases, such raw materials may belong to similar geological contexts and thus decrease the discriminative power of trace element analysis. Furthermore, mod- ification of the clay paste can affect the chemical composition and its variability, for example if different raw materials are mixed together. Fi- nally, the chemical composition of ceramic objects can be altered by en- vironmental factors during burial. Another type of constraint concerns the sampling strategy and ana- lytical method used in provenance studies. The choice of method affects the subgroup of element concentrations that are determined, the ana- lytical precision and, when the results are compared with reference data, the analytical accuracy. This has become a very important issue, Journal of Archaeological Science: Reports 16 (2017) 564–572 ⁎ Corresponding author. E-mail address: a.hein@inn.demokritos.gr (A. Hein). http://dx.doi.org/10.1016/j.jasrep.2017.03.020 2352-409X/© 2017 Elsevier Ltd. All rights reserved. Contents lists available at ScienceDirect Journal of Archaeological Science: Reports journal homepage: www.elsevier.com/locate/jasrep