Abstract
Carbonate deposits in aqueducts, known as calcareous sinter, can reach a considerable thickness and are commonly annually laminated. Environmental and palaeoclimate proxies measured in calcareous sinter samples, such as stable oxygen and carbon isotopes and trace elements, can provide important high resolution information on hydrological conditions, temperature, and local precipitation in the catchment area of an aqueduct. In order to utilize the proxy data for palaeoclimate reconstruction, the sinter deposits must be dated by annual laminae counting and more precisely by Th-230/U-dating. The major problem of Th-230/U-dating of calcareous sinter in aqueducts is the relatively large amount of initial detrital Th contained in the deposits, leading to apparently wrong Th-230/U-ages. Here we present the first systematic approach to date sinter deposits from various locations in western and eastern Europe using the Th-230/U-method.
Calcareous sinter samples from eight Roman aqueducts in France, Germany, Greece, and Turkey were investigated. Preliminary screening for the U and Th content of the samples by LA-MC-ICPMS revealed that the U-238/Th-232 isotope ratio of the samples is highly variable, with values ranging from 0.03 to >11. The four samples with the largest U-238/Th-232 ratios were selected for MC-ICPMS Th-236/U-dating, even though they still contain substantial amounts of detrital Th-232. The uncorrected Th-230/U-ages are between 2.9 and 461 a and thus all significantly older than required by the Roman origin of the samples (ca. 2 ka). Application of the conventional correction for detrital contamination, assuming a bulk Earth Th-232/U-238 weight ratio of 3.8 (+/-50%) and Th-230, U-234, and U-238 in secular equilibrium, results in younger ages that are in agreement within error with a Roman origin. However, the resulting errors are very large, and the ages are therefore not very reliable. Application of three-dimensional isochrons yields both younger ages and smaller age uncertainties than the conventional detrital correction. However, the isochron age uncertainties are still between 0.8 and 1 ka and thus not as precise as required for paleoclimate and archaeological studies. Precise and accurate Th-230/U-dating of these relatively young Roman sinter deposits is therefore challenging.
The observed high MSWD values (Mean Square of Weighted Deviates) and low Probabilities of Fit of all isochrons show that the isochron assumptions are not fulfilled for our samples. However, it is possible to estimate the degree of scatter around the isochron produced by different processes and simulate the different scenarios in Monte-Carlo simulations. These simulations show that the scatter observed in our isochron data cannot be explained by different ages and initial (U-234/U-238) activity ratios of different sub samples, but also requires variable (Th-230/Th-232) activity ratios of the detrital component. (C) 2015 Elsevier B.V. All rights reserved.