How do you turn a watermark from a visible image in a sheet of paper into a meaningful piece of information? This is the major problem in the study of watermarks. If it is to be useful to others, the data that is seen when looking at the watermark must be made accessible without recourse to the paper itself; in other words, information must be recorded that differentiates this particular watermark from all others.
There are a number of ways of gathering identifying information about a watermark:
- Watermarks are visual data, not linguistic data, therefore reproducing the image itself will be a much more economical way of retaining information than describing it. The old-fashioned way of doing this was tracing, but this may incur the wrath of a librarian. Irritating librarians is a very bad idea. There are a number of methods for the mechanical reproduction of watermarks.
- The International Society of Paper Historians have developed a set of criteria that can be used to describe paper with considerable precision. This is known as the IPH Standard. The IPH Standard system includes, in section 3.1, a series of entries for watermark data that allow for the identification of watermarks based on several pieces of data that combine to form a set of unambiguous individualising characteristics. These are described in detail on the IPH website linked above, but can be summarised as: a brief but precise description of the watermark type; the watermark's dimensions; the position of chainlines relative to the watermark; and, crucially, the precise location of the watermark on the sheet (relative to chainlines and laid lines). The usefulness of watermark location of the sheet was argued by Paul Needham in, 'Allen H. Stevenson and the Bibliographical Uses of Paper', Studies in Bibliography, 47 (1994), 23-64. The exhaustive set of measurements prescribed in the IPH Standard are followed by leading scholars in the field of literary manuscripts such as Mark Bland although, as Dr Bland has made clear, they are never as satisfactory as a high quality reproduction such as a beta radiograph.
- The most common (but in some ways least satisfactory) way to identify a watermark is to cite the closest match in a database of watermarks. There are a number of printed volumes or on-line resources that contain reproductions of thousands of watermarks. These databases are a very important resource and the result of extensive and painstaking scholarship, but they are not perfect or exhaustive and should be used carefully.
Given the problems that are encountered in turning watermarks into useful data, some scholars have suggested that paper is more easily identified through other physical features that are more readily translated into numerical data.
The watermark is not the only distinctive feature of a sheet of handmade paper. Handmade paper retains an impression of the entire mould in which it was made, including its sieve-like wires. The impression of one set of wires, called chainlines, run parallel to the shorter edges of the sheet. They are therefore often used to determine the format of a printed book (e.g. they will run vertically in a folio, horizontally in a quarto).
Chainlines are typically about 20-25mm apart but their precise distribution will vary from mould to mould. Measuring the spaces between can therefore give an individual "fingerprint" rather like a watermark. It is a procedure that requires extremely careful measurement but can be useful, especially if no watermark is visible. See, for example, David L. Vander Meulen, 'The Identification of Paper Without Watermarks: The Example of Pope's Dunciad', Studies in Bibliography, 37 (1984), 54-81.
David Gants has suggested that chainspaces (i.e. the space between chainlines) can be used much more widely for the identification of paperstocks. See David Gants, 'Identifying and Tracking Paper Stocks in Early Modern London', Papers of the Bibliographical Society of America, 94 (2000), 531-40. He argues that chainspaces, because they form sets of numerical variables, are more suitable for digital processing than watermarks.
Gants's method is to take digital photographs of backlit paper, rapidly producing images in which chainspaces can be easily measured, and compared. The results are encoded in XML and are thus easily compatible with other XML projects.