GEOARCHAEOLOGICAL ANALYSIS OF ANTHROPOGENIC SEDIMENTS FROM THE NEOLITHIC SETTLEMENT OF MANDRA IN THESSALY, GREECE

• S. Kyrillidou (116 Konstantinoupoleos St., 546 44 Thessaloniki, Greece)

This paper presents the results of a multi-proxy geoarchaeological approach applied to a wide range of archaeological sediments from a Late-Neolithic settlement located in Thessaly, Greece. The objective was to test the potential of sediment as a cultural material per se and determine to what extent, if any, certain human activities within a settlement could be extrapolated merely on the grounds of residues incorporated into the sediment. Provided that specific human activities leave exclusive signatures on sediments, the ultimate objective was to provide links between field descriptions and laboratory characterisations in order to facilitate the interpretation of deposits during future excavations.

A total of 63 archive bulk samples was examined coming from as wide and representative a range of archaeological contexts as possible, including open spaces of various uses, buildings, fire installations, and secondary contexts of deposition, such as pits and ditches. Within each context, a sequence of deposits was analysed comprising accumulated deposits from the actual use of a space, deliberately, or non-deliberately, laid activity surfaces, occupation sub-surfaces, and fills. The sediments were classified archaeologically according to basic context and deposit types and they were analysed chemically by X-ray fluorescence (XRF), mineralogically by X-ray diffraction (XRD), and in terms of particle size distribution using laser granulometry. Thin section analysis and scanning electron microscopy (SEM) supplemented by energy dispersive X-ray analysis (EDXRA) were applied selectively to help with the interpretation of the results. The results were analysed statistically using principal component analysis (PCA) to determine how variables coming from different analytical techniques correlate with each other. In addition, discriminant analysis was used to test the validity of archaeologically-defined context and deposit types on the basis of lab-obtained sedimentological variables.

The results showed an extremely high correct proportion, with 87% and 92% probability of sediments' classification by context and deposit type, respectively. Natural soils and sediments sampled from the area not affected by human occupation showed distinctively different values from all types of occupation deposits. Within occupation deposits, differences among different contexts and/or within sequences of contexts do exist, but sometimes they are small. These differences can be best explained by: (a) Ca, or calcite, as a potential indicator of plastering (b) mica in correlation with fine sand (100-200ìm), or alternatively the coarse fraction (>10ìm), as being indicative of building materials, such as mudbrick fragments (c) P in correlation with Cu, Pb, and perhaps Zn, as being indicative of domestic-type deposits rich in organic content (d) clay (<2ìm) and silt (2-60ìm), or alternatively the fine fraction (<10ìm), with which most of the key elements seem to relate.

The results point out that there is great potential in analysing archaeological sediments for identifying specific human activities within settlements. Sediment can be at least as informative as any other class of traditionally-analysed archaeological materials, such as pottery, lithics, bones, and plant macros. The latter are mere single inclusions embedded in larger complicated matrices, those of sediments, which are basically all a settlement is made of.