Estimating the approximate firing temperature of burnt archaeological sediments through an unmixing algorithm applied to hysteresis data

Estimating the temperature to which burnt archaeological sediments and soils have been exposed in antiquity is of interest as it may well elucidate the interpretation of specific features. Certain semi-industrial activities, such as metal working or the production of pottery, are often associated with the controlled use of intense high-temperature processes (>500 °C). Exposure to such high-temperatures will, inevitably, lead to the thermal alteration of commonly found iron minerals within the soil. Magnetic measurements made on a series of soil samples subjected to controlled laboratory heating, confirms both the sensitivity of iron minerals to thermal alteration and also suggests a correlation between the maximum exposure temperature and the hysteresis properties. From this data a method for estimating the maximum exposure temperature of burnt archaeological samples, recovered from similar soil types is proposed, based on the application of a linear unmixing model. This model compares hysteresis data from the archaeological samples to an end-member data set created from the laboratory heated soil. The maximum exposure temperature for the archaeological samples is estimated from the relative proportion of known temperature end-members present within the final model describing the experimental hysteresis data. The validity of the model is demonstrated through application to samples recovered from a series of actualistic fire experiments and a range of burnt archaeological features recovered during excavation of a multi-period site at Yarnton, near the city of Oxford, UK. A further application of the method, to determine the fidelity of samples collected for archaeomagnetic dating, is presented from the excavation of a large hearth type features at Whitby, North Yorkshire, UK.