Following several studies considering the potential toxicity of food-type containers, we hypothesized elemental analysis would help us test and classify a collection of Ecuadorian ceramic sherds from Andean and Amazonian sites.

μ-XRF spectrometer analyses were carried out on 48 ceramic sherds coming from 4 different archaeological sites. Major elements values were transformed into oxides and data were constrained to 100%, making our results semi-quantitative. A principal component analysis (PCA) was then performed on the additive log-ratio (ALR) transformed data to identify main compositional axes and plot the sherds. Besides, a hierarchical cluster analysis (HCA) was applied on the coordinates of the individuals from the PCA to estimate the chemical similarity between the ceramic samples.

The lead detected on the internal face of the ceramics locally produced in Quito was generally below the limit of quantification, while the lead concentration mean in the Cosanga ceramics was 180±34μg/g. The lead concentration values in the Pucará ceramics and the type labelled Mango Montaño were of the same order. Arsenic, mercury, cobalt, chromium and antimony values proved to be below the quantification limits. The PCA on the ALR transformed data evidenced 2 main axes. The first main axis PC1 made it possible to graphically distinguish the “local Quito” ceramics from the “Cosanga” ceramics. The second main axis PC2 made it possible to refine the distribution of the samples. Three of four samples from Pucará were close to the ceramics locally produced in Quito. The HCA was applied on the first five factors to take into account about 80% of the total variance of the sample set. The dendrogram discriminated two main clusters, demonstrating a significant clustering pattern of certain fragments belonging to the same craft tradition, essentially the Cosanga ceramics vs. the Local Quito ones.

No particular hazard was identified regarding the exposure of past populations to potentially toxic pottery. We managed through elemental analysis followed by a principal component analysis and a graphic representation to clearly identify 2 groups of pottery out of 4 different locations, 3 different periods and 4 cultural traditions corresponding to 4 different populations. We aim in the near future at testing the samples we presented for lead leachability and comparing them with ceramic samples from other locations.