Application of principal component analysis to μ-PIXE data in lapis lazuli provenance studies

The application of multivariate analysis is extremely useful when dealing with large datasets and can be applied also in provenance studies. In this work, Principal Component Analysis (PCA) was applied for the first time to the study of the provenance of lapis lazuli rocks used for glyptic art in ancient times, on a database including four different sources in Afghanistan, Tajikistan, Siberia and Myanmar. Results of the application of PCA on mu-PIXE data collected in mineralogical phases contained in lapis lazuli rocks, in particular diopside and pyrite, confirmed the role of some trace element contents as characteristics for different provenances, i.e. as provenance markers. As an additional important result, PCA seems to strengthen the role of previously found weaker markers (i.e. trace elements not sufficient alone to distinguish univocally among provenances) such as, for example, the coexistence of high Ti, V, Cr and Mn concentrations in diopside in Afghan lapis lazuli. This can be relevant when other discriminant phases like pyrite cannot be found in the sample or are not suitable for analysis. Moreover, for geological purposes and to try to distinguish among different quarries inside an extraction area, a study on intra-provenance variability was carried out on diopside data for Myanmar and Siberian samples. The separation into three groups for Myanmar samples, previously identified by means of minero-petrographic and elemental analysis, is confirmed and a partial differentiation was identified also for samples coming from the two Siberian areas of Sludyanka River and Malaya Bistraya River.

Automated summary

The study applied Principal Component Analysis (PCA) to analyze the provenance of lapis lazuli rocks. The results confirmed the role of trace element contents as provenance markers. PCA also strengthened the use of weaker markers and allowed for differentiation among different quarries. This study provides valuable insights for the analysis of large datasets and the study of provenance in mineralogy.