The Digital Scriptorium intends to establish the technical and organizational framework for a visual union catalog of medieval manuscripts. It was created by The Bancroft Library, University of California, Berkeley, and the Rare Book and Manuscript Library of Columbia University. See <http://sunsite.Berkeley.EDU/Scriptorium/> for a detailed description of the project. The Digital Scriptorium is designed to solve two primary problems for medievalists: (1) locating the corpus of manuscript materials required for carrying out a given research or pedagogical project; (2) attempting to tie those materials to a concrete time and place. For this there is simply no substitute for visual inspection. Since most manuscripts of cultural interest (literary, legal, scientific, religious, philosophical, etc.) are not dated or localized, they provide little evidence for knowledge of a specific text in a specific place and time, i.e., for precise cultural or literary history. Thus a long-felt need of medieval scholars is a repertory of dated and datable manuscripts that can be used to establish a taxonomy by which undated and unlocalized manuscripts can be tied to specific geographical and chronological coordinates. The classical method of paleographical training is to study a series of photographic or printed facsimiles, accompanied by transcriptions, in order to learn the characteristics of the various scripts as well as how to read them. Facsimiles are used because few university libraries have large collections of original manuscripts. Heretofore these problems, finding source materials and, once found, examining, dating, and localizing them, have been attacked using paper-based reference tools (e.g., the several Catalogues des manuscrits datés). The computer and the web make possible a better solution to the long-standing scholarly problems adumbrated above; and it was the perception of that solution that lay behind the genesis of the Digital Scriptorium project. As we began to conceptualize a digital database of medieval manuscripts, we identified the following significant criteria:

• 1. Heretofore these problems, finding source materials and, once found, examining, dating, and localizing them, have been attacked using paper-based reference tools (e.g., the several Catalogues des manuscrits datés). The computer and the web make possible a better solution to the long-standing scholarly problems adumbrated above; and it was the perception of that solution that lay behind the genesis of the Digital Scriptorium project. As we began to conceptualize a digital database of medieval manuscripts, we identified the following significant criteria:
• 2. Standards: One of the curses of the electronic age is the lack of standards and the concomitant inability to exchange information transparently among different systems. One of the great achievements of the international library community has been the creation of the MARC format as the standard means of exchanging bibliographic information. With medieval manuscript collections found all over Eastern and Western Europe and in both Americas, the same sort of standardization (of various kinds) is required.
• 3. Extensible and updatable: Given the immense variation in technological sophistication among libraries holding medieval manuscripts, it was evident that any union list would have to be created incrementally, with the more technologically advanced institutions taking the lead and providing the tools so that other institutions could add their own materials when it became technically and economically feasible for them to do so. Similarly, it is necessary to assume that manuscripts described at the beginning of the project might need to have their descriptions updated on the basis of information added later by other institutions.
• 4. Distributed: The logistics of centralizing all of the descriptive information about medieval manuscripts in a single repository are formidable. What is needed is a framework and a set of standards that will make it possible for institutions that hold medieval manuscripts to make information about them available as part of a distributed process, just as information about local holdings of monographs and serials is made known through the various national and international bibliographical utilities.
• 5. Digitized facsimiles: Text descriptions of medieval manuscripts are inadequate, and published catalogs, for cost reasons, can only provided black-and-white facsimiles of a limited number of manuscripts. While there is a good consensus on how a manuscript ought to be described, the descriptions omit features that cannot be described textually yet which are of great importance for comparative purposes. An example: A great deal of effort has gone into the nomenclature of medieval scripts; yet even scripts with exactly the same name are often quite different visually. The user can extract information from an image that would take a cataloguer days to encode or cannot be encoded at all.
• 6. Commercial software: Projects like this one cannot afford the maintenance costs of locally developed software. With the use of industry standard encoding methods (i.e., SGML), the Digital Scriptorium can use commercial software for both encoding and display.
• 7. Revenue: A project like this must generate income in order to make it viable in the long term. Participating institutions should be recompensed for the use of their images in the catalog; while the project ought to generate enough funding so that it can offer seed money to institutions as an inducement to participate.

Thus our goal is to establish an internationally-accepted framework for the eventual creation of a comprehensive database of medieval manuscripts, in short, a world union catalog, access to which would be available through a fee-based system. In order for this project to be successful it must command the participation of the community of scholars and librarians who use and catalog medieval manuscripts as well the cooperation of libraries holding significant quantities of medieval manuscripts. It must be collaborative in the best sense of the word. This is why we enlisted the support of distinguished scholars and manuscript cataloguers from the U.S. and Europe to design a data model that responds to the needs of medievalists as well as to the exigencies described above. Our modus operandi has been explicitly modeled on the process that led to the successful establishment of the Encoded Archival Description as an international standard for the description of archival collections.