Archaeometry · 2022 · 17 citations · 28 references
HumanitiesArchaeologyEarth SciencesCeramic AnalysisElemental AnalysisGeochemistryAnalytical GeochemistryCeramic ClassificationCeramic Provenance AnalysisElemental Composition AnalysisNeutron Activation AnalysisArchaeological Materials AnalysisArchaeometric Analysis Of CeramicsChemical Reference GroupingElemental GeochemistryGeochemical Provenance AnalysisInstrumental Neutron Activation AnalysisNeutron ActivationStatistical Discriminant Analysis
Abstract For the first time in Russia, instrumental neutron activation analysis (INAA) and related analytical techniques were used to determine the distribution patterns of elemental composition in archeological pottery in terms of classification and provenance of medieval pottery. A total of 48 fragments have been discovered in three locations; Bolgar, the Moscow Kremlin (MosKremlin), and Selitrennoe settlement (Selitrennoe). The assemblages were subjected to INAA in order to determine the concentrations of 37 elements in mg/kg. The data were analyzed using geochemical, bivariate, and multivariate statistical techniques. To classify the fragments according to their origin hierarchical clustering (HCA), linear discriminant (LDA), and principal component analyses (PCA) were used. Based on the INAA results, the first attempt was made to apply machine learning methods to the study of artifacts in Russia. The results show that they are reliable and powerful in extracting information about critical elements for identifying ceramic fragments and using them as chemical fingerprints. Chromium was the most useful element and was used in conjunction with other elements as a fingerprint to distinguish sherds. Chemical and statistical analyses help to establish of reference groups for medieval archeological pottery, which will be used in the future to classify and identify various unknown sherds. These reference groups serve as baseline data for determining the fragment's place of manufacture and are considered a reasonable judgment based on experimental data.
28
R: A Language and Environment for Statistical Computing
R Core Team · 2000 · 353K citations · Full text
Natural Sciences, Computer Science, Statistical Computing +10
An analysis of variance test for normality (complete samples)
Samuel S. Shapiro, M. B. Wilk · Biometrika · 1965 · 19.3K citations
Martin Bulmer · Sociology · 1978 · 1.4K citations