In the case of Majolica products of Barcelona and València there does not exist a general accepted typological catalogue. Traditionally, its classification is based on art history criteria mainly related to decoration. For this reason, it was necessary to elaborate a typology in order to make ArchAIDE a functional system on the automatic classification of this pottery. From the Universitat de Barcelona, we are working with the collections of different institutions, such as the Museu del Disseny de Barcelona, with the aim to obtain and draw examples of all the different types that were manufactured in Barcelona and València between the 13th and 18th centuries (bowls, plates, jugs...). All drawings, also those coming from paper and digital publications, requires to be digitized one by one to be introduced in a geometric 2D description of each ceramic class in the database. To determine these classes is necessary standardize the drawings at the same scale. Up to now, it has been possible to digitize 468 different drawings.
The typology that we are elaborating uses the Compositional EDMA (Euclidean Distance Matrix Analysis) geometric morphometrics approach to approximate the form and/or shape of ceramics, capturing their geometry. The form of and object, or a subject, is related to its appearance and structure, and morphometry is the measurement of these external and perceptible characteristics, related to the object's appearance as well as to its physical and diachronic constitution. Numerical data are used through mathematical relationships whose coefficients are then used as variable in the statistical analysis. EDMA is a landmark-based method easily linked to Compositional Data analysis.
Coordinates of 8 landmarks (1), or fixed points of precise location, were recorded for the case of ring-based majolica bowls. The two-dimensional coordinates were recorded from the drawings of this particular type with the help of vector graphics editor. This coordinate-free landmark method enables the study of form avoiding the nuisance parameters of translation and rotation, and the study of shape regardless of size. Landmarks provide configurations of points in the space. The analysis is performed on the matrix of the Euclidean distances among points (2). Logratio transformations enables compositonal EDMA and thus shape analysis. After cluster analysis on the CLR transformed data, the shape data, performed with R, using the square euclidean distance and the centroide agglomerative algorithm, it is possible to define some preliminary groups thanks to the study of the resulting dendrogram (3).
Our results are still preliminary but promising because through this method it is also possible to establish the appearance of the mean shape form for a defined group through multidimensional scaling of its matrix of mean distances (4).
