Mameli Valentina, Sanna Angotzi Marco, Farinini Emanuele, Leardi Riccardo, Lugliè Carlo, Cannas Carla
Department of Chemical and Geological Sciences, University of Cagliari, Monserrato CA, Italy.
Consorzio Interuniversitario Nazionale per la Scienza e Tecnologia dei Materiali (INSTM), Cagliari Unit, Italy.
Heliyon. 2023 Feb 23;9(3):e13958. doi: 10.1016/j.heliyon.2023.e13958. eCollection 2023 Mar.
In this work, a detailed study of Monte Arci obsidian sub-sources using the increasingly accessible technique of pXRF is presented based upon a large dataset of 68 geological samples, for the development of X-ray fluorescence-based analytical standardless procedure. In addition, a non-conventional (for obsidian provenance study) direct application of multivariate analysis on XRF spectra (continuous variables), rather than absolute concentrations or intensity ratios (discrete variables) is proposed. Results from different softwares and data analysis approaches (bi-/trivariate multivariate) were compared. In a blind test, the bi-/trivariate approach led to the correct assignment for the main SA, SB, and SC sub-sources, taking into account averaged values of intensity ratios with their standard deviation obtained from three independent measurements. A high intra-source variability for the SB subgroups was detected (almost 13% of error in the assignment, 9 samples out of 68). A non-conventional application of multivariate analysis was carried out directly on the XRF spectra and correct assignments were obtained for SA, SB1, SC groups, while 71% of the SB2 samples were correctly identified. The non-destructive analysis on 14 archaeological samples from Su Carroppu (Carbonia, southwestern Sardinia) rockshelter and from the Middle Neolithic (MN) 422 structure of the open-air dwelling site at Cuccuru is Arrius (Cabras, central-western Sardinia) permitted to test the method and hypothesise their provenance. The comparison with visual characterization or previous analyses by Particle Induced X-Ray Emission (PIXE) permitted to verify the correct provenance assignment of all artifacts for the bi-/trivariate method, and for 12/14 samples in the case of the multivariate one. The standardless analytical approach proposed in this work can represent a more general method exploitable for other obsidian sources, other glassy materials, besides other materials of archaeological interest.
在这项工作中，基于68个地质样本的大型数据集，利用日益普及的pXRF技术对蒙特阿尔奇黑曜石子源进行了详细研究，以开发基于X射线荧光的无标准分析程序。此外，还提出了一种非传统的（用于黑曜石产地研究）对XRF光谱（连续变量）直接应用多元分析的方法，而不是对绝对浓度或强度比（离散变量）进行分析。比较了不同软件和数据分析方法（双变量/三变量、多变量）的结果。在一次盲测中，双变量/三变量方法考虑了从三次独立测量中获得的强度比平均值及其标准偏差，对主要的SA、SB和SC子源进行了正确的归类。检测到SB亚组的源内变异性很高（归类错误率近13%，68个样本中有9个）。对XRF光谱直接进行了多元分析的非传统应用，对SA、SB1、SC组进行了正确的归类，而71%的SB2样本被正确识别。对来自苏卡罗普（卡尔博尼亚，撒丁岛西南部）岩棚的14个考古样本以及来自库库鲁伊（卡布拉斯，撒丁岛中西部）露天居住遗址的新石器时代中期（MN）422结构的样本进行无损分析，从而能够测试该方法并推测其产地。与视觉特征或先前通过粒子诱导X射线发射（PIXE）进行的分析相比较，证实了双变量/三变量方法对所有文物的产地归类正确，而多变量方法对14个样本中的12个归类正确。这项工作中提出的无标准分析方法可以代表一种更通用的方法，可用于其他黑曜石来源、其他玻璃质材料以及其他具有考古价值的材料。
