Skip to main content

LIMA: Watermark Databases

There are a number of printed catalogues of watermarks. These are arranged by watermark type (i.e. flag, pot, wyvern, etc.), and provide reproductions of various watermarks of that type along with information about the paper on which the individual watermark is found - its place of origin and date of use (usually based on the date of publication of one or several printed books in which it is found). The most important of these is C. M. Briquet's Les Filigranes (1907, repr. Amsterdam, 1968). Briquet spent many years in archives across France, tracing watermarks from paper dated up to 1600.

Two smaller collections that cover a later period are:

  • W. A. Churchill, Watermarks in paper in Holland, England, France, etc, in the XVII and XVIII centuries and their Interconnection (Amsterdam, 1935)
  • Edward Heawood, Watermarks, mainly of the 17th and 18th centuries (Hilversum, 1950)

These printed catalogues have now been supplemented with online resources. The WWW Watermark Archive Initiative was set up to arrive at standards for internet-based watermark resources, in line with the IPH Standard. Existing databases include:

Searching Watermark Databases

An additional issue regarding the identification of a watermark is the discrepancy between raw visual data and the linguistic data upon which catalogues and databases depend. Printed catalogues are arranged by watermark type, but there may be dozens or hundreds of watermarks reproduced under a single heading.

Online databases such as Gravell are more discriminating, using a combination of type (primary descriptors) and numerous secondary descriptors (ornaments and additions to the basic type). However this is a system of considerable taxonomic complexity (you need to know your trefoils from your fleurons).

This problem would become acute in a database of watermarks of significantly greater scope than any of the existing resources.

What none of these systems allow for is the computerised search of the images themselves, which would clearly be preferable. Such a system is being developed: SHREW is a research programme on shape recognition software based at the University of Northumbria at Newcastle. It aims to determine the best method for reproducing watermark images, and to develop software that will search images. Unfortunately at the time of writing SHREW is on hold due to lack of funding.

The Scope of Watermark Databases

Whilst the scholarship and industry of these databases is very impressive, they should be used cautiously. Firstly, none of the printed or on-line databases are remotely close to being comprehensive. The number of watermarks from the hand-made paper period that do not appear in any database greatly outnumber those that do.

Briquet and the other printed catalogues of watermarks are based on tracings. The images they provide are therefore not always detailed or exact. It is not always possible to be sure that a given watermark is the same as the closest match in a printed catalogue. For this reason online databases are preferable because their images are usually mechanically produced.

It is unusual to be able to find a specific watermark in any of the databases, however similar watermarks can usually be found. It is important to remember, however, that making a comparison between two similar watermarks is an not an identification of the watermark: it is based on similitude, i.e. non-identity. The two stocks of paper do not come from the same mould, it is just that these two different watermarks have a similar design.

Describing a watermark with reference to its similarity to a reproduction in one of the catalogues informs the reader of the type and general appearance of a watermark more accurately and succinctly than a verbal description is likely to do.

Major collections of watermarks like Briquet demonstrate clearly that there were definite trends in watermark design, with a tendency towards increasing elaboration. For example, over the decades pots' crowns tended to gain more jewels, for example, and later crescents, then fleur-de-lis. Finding a close match between a watermark in a document and one in a database allows you to place your watermark at a specific point in this continuum, and therefore makes it possible to estimate a very approximate date for the paper's manufacture.

A closely similar watermark was probably produced in the same geographical region at around the same time as the recorded watermark. The evidence of similitude cannot, however, date a manuscript with any precision - it depends on too many untested assumptions (eg that the two watermarks must have been designed at proximate dates).