Watermark study has, not surprisingly, been largely foliocentric. While folio watermarks are usually integral and centered proudly on the page, quarto marks skulk in gutters, divided and partly obscured. In quartos printed on runs of one or two main stocks of paper, the task of identifying twins and classifying stocks may nevertheless be relatively simple. But paper study becomes considerably more daunting in quartos (and smaller format books) printed on mixed stocks and remnants. An extreme example of proliferating papers is found in the first edition of Robert Crowley's Piers Plowman (1550), where at least twenty-five paper stocks were used to print a quarto of thiry-one gatherings, with some gatherings appearing on as many as four different papers. The problem has thus been how to identify and discriminate fifty separate watermarks, and pair them with their twins. Since to obtain a full picture of a given quarto watermark it is usually necessary to find two or three exemplars which are positioned slightly differently with respect to gutter and type page, beta-radiographs or even Dylux reproductions were out of the question. (250-300 exposures would have been necessary for just this one edition.) And while digital photography and computer imaging hold promise, low tech methods are still necessary for the study of watermarks in books of smaller formats.
My solution has been to adopt the technique of chain space measurement developed by David Vander Meulen ("The Identification of Paper without Watermarks," Studies in Bibliography 37), and adapt it to the identification of papers with watermarks. By combining chain space models with careful sketches of the visible portions of quarto watermarks, I am able confidently to make identifications, pair twins, and distinguish similar sets of marks. I was able, for example, to identify a situation where one of a pair of watermarks had been replaced by a new version of the same mark, while its twin remained unchanged. (For a moment I wondered if watermarks were triplets .) A key point is that the state of chain lines remains relatively stable, even as watermarks go through a number of successive states due to damage or wear. Chain line models can thus be particularly valuable in establishing the identity of watermarks in variant states.
In this paper, I will briefly outline the problem, demonstrate my method with examples from Crowley's Piers , and, in conclusion, argue that chain space models should become a standard feature of bibliographical descriptions of paper, even as high tech solutions make watermark reproduction more practical and affordable.