In this phase of the project, two are the priorities and one unlocks the other: firstly, it is necessary to get enough data to train the algorithm and check its functioning, then.
Currently, a trial database is going to be created from the documents held at the Minutier des notaires de Paris (National Archives).
The archives of the notaries practising in Paris exist from the end of the Fifteenth century. About a hundred notaries were available to draw up familiar or economic deeds, or any other private deed with authentic value. Some of them produced several tens of deeds every month and kept the minutes on paper in their archives, bound in registers or collected in set of loose sheets.
The notary’s minutes suit perfectly the required purpose, given the multiplicity of daily or weekly deeds drawn up by the same person, that means a signifcant consumption of paper. Also, when the minutes are kept as loose sheets, the shoot is simplified. These documents allow a real richness and variety of watermarks, so that it is possible to collect easily and quickly significant series of the same pattern (several tens).
The data collection has sistematically been concerning the minutes of 1500, 1525, 1550, 1575, 1600, 1625, 1650, 1675 et 1700. The following step will be the addition of the watermarks listed by Charles Moïse Briquet (1839-1918), the well known author of Les filigranes (1907), in the reference bases for the automated recognition. All the years from 1500 to 1700 have been analysed, except for 1625 (in progress) and 1575 (to do in autumn 2019, after finishing the works on Briquet’s watermarks).
The watermark professionals are less familiar to the ways of production and the customs of handmade paper from the Sixteenth century, in France furthermore. Besides, if some repertories of watermarks and texts about paper in general do exist, no specific manual has been written. The phase of the collection of a trial database, passing through the analysis of a significant quantity of sheets of paper and watermarks concerning a precise period and context of use, allows even now to deduce some interesting conclusions.