Sources of uncertainties in OSL dating of archaeological mortars: The case study of the Roman amphitheatre "Palais-Gallien" in Bordeaux

Archaeological mortars are more convenient and much more representative for the chronology of buildings than brick or wood constructions that can be re-used from older buildings. Before dating unknown samples of mortars, further investigation of OSL from mortars is required and the most efficient methodology needs to be established. In this study we compared the ages obtained by OSL dating of quartz extracted from mortars of the Roman amphitheatre Palais-Gallien in Bordeaux with independent age information. Resetting of the OSL signal occurred during the preparation of mortar when grains of sand (quartz) were extracted and mixed with lime and water. The mortar was subsequently hidden from light by embedding within the structure which is the event to be dated. Various factors contribute to uncertainties in the age determination. The frequency of measured equivalent doses reveals a large scattering. Optical bleaching of certain grains can be partial due to the short duration of the exposure to light. We worked with the single grain technique in order to find and select the grains that were sufficiently exposed to daylight. To determine the average equivalent dose, we tried three different approaches: a calculation of an arithmetic mean and one following either the central age model or the 3-parameter minimum age model, the latter turned out to be the only relevant way to evaluate the experimental data. The proportion of grains included in the calculation of the average equivalent dose represents 2.7–4.7 % of the overall analysed grains. The results obtained for the three out of four samples are approaching the expected age, however, the minimum doses and the corresponding ages are significantly over-estimated in case of two samples. The studied material is very coarse, which causes heterogeneity of irradiation at the single grain scale, and contributes to a dispersion of equivalent doses. Different analytical methods (scanning electron microscopy with energy dispersive X-ray spectroscopy cartography, Beta-radiography imaging, inductively coupled plasma mass spectrometry) were employed to demonstrate the presence of this phenomenon. Despite the extremely large proportion of high equivalent doses in equivalent dose distributions, there is an apparent presence of well-bleached grains at the beginning of equivalent dose distributions. The study shows the potential of dating mortars by single grain OSL. - Highlights: • Fewer than 8% of all measured grains exhibited an OSL signal. • The heterogeneity of β-irradiation arises principally from variations of K-content. • Beta-imaging confirmed inhomogeneous distribution of beta sources in studied samples. • The samples are highly affected by heterogeneous bleaching. • Standard age models are not adapted for the studied series of samples