A study on the classification of the composition of ancient glassware based on statistical law analysis

Abstract

Ancient glass is highly susceptible to weathering by the environment in which it is buried, and weathering can cause changes in the proportions of its chemical composition, which can affect the correct determination of its category. In this paper, we solve the problem by studying the chemical composition of ancient glass artefacts to establish a mathematical model. We first used the Spearman correlation test and obtained that the weathering of the surface of the artefacts had no relationship with the color and decoration of the glass, and a negative weak correlation with the type of glass; then we analyzed the statistical pattern of the chemical composition content of the surface of the artefact samples with and without weathering, and conducted the Mann-Whitney test on the chemical composition of the sample data. The mean values of , , , , , and in lead-barium glass were not significantly different before and after weathering, while the mean values of other oxides were significantly different, e.g. the CaO content increased from 1.89% to 4.49% after weathering; finally, the chemical composition of surface weathered glass before weathering was predicted. By analyzing the discrete degree coefficients of each component, we found that the component content is stochastic, and proposed the index of random variation to quantify the stochasticity, and combined with the inevitable variation in the weathering process to establish a mathematical model of the chemical component content before and after weathering for prediction.