Collections, traditionally, have been built in two ways:

• Zealous collectors build comprehensive collections in a specialized field or document an era by collecting all known knowedge related to that period. In either case, when they can no longer manage the size of the collection, or when they cannot exercise reasonable control over the collection, they turn it over to a repository for safekeeping.
  
  
• Librarians and bibliographers assume responsibility for different subject areas and studiously collect the current literature in those fields, with an eye toward having collections available to subsequent generations.

Many of the great collections we find at the Library of Congress, the University of Michigan, or the Newberry Library were born of the zeal of an individual collector. Consider the Peter Force collections at the Library of Congress. Like the Jefferson Library, the Peter Force Library was purchased for $100,000 by an act of Congress after the Jefferson collection was partially destroyed in 1851. Peter Force, a Washington printer and publisher, had assembled what was surely the largest private collection of printed and manuscript sources on American history in the United States. Force acquired incunabula, manuscripts, pamphlets, personal papers, and maps—all documenting early American history. When Congress purchased the library in 1867, the Library of Congress became the owner of a comprehensive, well-documented collection that had been intelligently assembled.

A more limited example is the complete set of McGuffey Readers, assembled by Maude Blair, a Detroit school teacher. Miss Blair donated the 195 separate editions of the readers, primers, and spellers to the Library of Congress in 1937. She had diligently sought each of these editions and bought them in varying conditions, but they constitute a complete set.

Many university and independent research libraries have similar stories to tell. Their most prized historical collections have been given or sold by individuals who cared deeply about a particular body of knowledge. Often, the descriptions of the materials are as valuable to the archival institution as the materials themselves. More often than not, the descriptions prepared by the donors are the only finding aids for these rare and special materials.

Today, libraries and archives typically acquire special collections by purchasing them, and the asking prices have escalated dramatically. Two high-profile collections offered to the Library of Congress—the Martin Luther King papers and the Leonard Bernstein collection—illustrate this point. Both families view the collections as financial assets and the conditions set for deposit are closely connected to how the families expect to continue to derive income from the collections. Increasingly, libraries find that significant numbers of families and private collectors view collections in a similar way, and few libraries have budgets to accommodate their requests.

Even when the collections are "free," they are not without problems. The University of Michigan recently accessioned the papers of its last president, James Duderstadt. Dr. Duderstadt is an engineer and a technology enthusiast. Not surprisingly, his papers came to the archives in the form of multiple hard drives. Since a hard drive contains only an alphabetical list of software applications and files, it is very difficult for the archivists at Michigan to appraise the contents of the collection. Many trivial e-mail reminders sit next to files documenting policy decisions—all as undifferentiated bits.

These examples illustrate two types of challenges that libraries and archives face today in building special collections. Both challenges—the trend toward information as a commodity and the explosion of research material in digital form—have important (and not unrelated) implications for scholarly access in the future. This presentation will focus on the second challenge—the development of digital collections—and some of the fundamental issues that managers of those collections are facing. In considering these issues, it will become clear that the survival of the digital scholarly record will depend on new forms of partnership between scholars and the librarians who have traditionally served as the custodians of scholarly output.

Digital collections take several forms:

• Reformatted collections. Many research libraries have converted important portions of their special collections to digital form. This reformatting has increased access to materials that previously were not readily available. The digital collections have extended the reach of libraries to profoundly different audiences. Instead of making collections available only to their students and faculty, research institutions have made their materials accessible to the K-12 audience, liberal arts colleges, and international educational institutions, as well as to independent scholars and the inquisitive public. Research institutions have had an opportunity to take their materials out from their locked cases and present them to much wider audiences. But in so doing, they have had to find ways to fund the digitizing and to place special collections in context for new audiences.
  
  
• New scholarly resources. An excellent example of what I mean by new scholarly resources is ACLS's new program to stimulate the production of electronic texts in history [the ACLS History E-Book Project]. The project encourages well-known historians to take advantage of technology to create a new type of monograph—one that allows the author to combine different kinds of materials to create an educational experience for the reader. The possibilities for bringing many types of historical evidence to bear on a single topic are exciting indeed, but we have not yet determined how these "experiences" will be preserved and made available to subsequent generations of students and scholars. Whose responsibility will it be to manage these electronic resources and maintain them over time?
  
  
• "Custom" materials. Many faculty members are creating their own resources for use by their students. They are, in essence, publishing "books" for a particular group. The library may not even know of the existence of these materials if they have not been acquired or processed. Do we care if they pass into oblivion more quickly than they were created?

To answer these questions, we must understand the evolving role of the library. Libraries have traditionally been physical places—massive limestone or brick structures with commanding arches and windows. They have inspired a great many students to pursue scholarly careers. They have been the site of significant scientific and humanistic discoveries. This is not simply a nostalgic view—read any number of scholars' biographies and find references to the influence of the library as a symbol of open access to knowledge. It is a powerful story. But, how does digital technology, with its ability to transcend time and place, change the role of the library, and by implication, the relationship of the library to the scholar?

Perhaps the single most important fact is that in the electronic world, libraries do not own collections, especially journal collections. Access to the content of journals is provided to authorized users for finite periods by legal contracts. This arrangement raises serious questions about how journals will be retained for access over time. Libraries cannot preserve what they do not own, and publishers have traditionally not been in the preservation business. In some disciplines, scholars themselves have taken control of the scholarly output in their fields. They have created repositories into which authors deposit both their refereed and non-refereed works, but there are usually no plans for long-term retention of such collections. Typically, libraries make decisions about preservation after they acquire published materials. In the dynamic digital world, in which the very notion of "publishing" is in flux, who is responsible for preservation of these materials? Much work has yet to be done to establish the conditions and responsibilities for preservation in the digital environment. In this new world, we must think about access and preservation separately. While we have lauded the possibilities for extending access, we are much less certain what it takes to ensure that digital information will be accessible five, ten, or twenty years from now.

Separating access and preservation is difficult after we worked so hard to combine these activities. When the National Endowment for the Humanities announced its Brittle Books program, there was much fanfare among both scholars and librarians alike, and there was great excitement about joining in the cause of preservation and access. NEH recognized that millions of nineteenth-century books were crumbling on library shelves. Under this new program, though, materials that could no longer be used because of their poor physical condition were microfilmed and given new life for scholarly use while also being preserved.

To be sure, not everyone was thrilled by microfilm. Digital technology was just emerging, and the futurists were telling us that soon anything could be digitized, that there would be universal access to everything. Wasn't it a waste of money to settle for microform, especially since we knew that scholars intensely disliked using microfilm, and students avoided it all together?

Even those who showed little enthusiasm for microfilm recognized that the brittle books problem was so urgent that it demanded immediate attention, and that the best available technology—microfilm—would have to do. We took comfort in the connection between preservation and access while we waited for other technologies to develop. Now digital technology is very much with us. NEH as well as other funding agencies are bombarded with proposals to digitize important collections. The reason: to provide access.

Indeed, the greatest advantage of digital technology is that it has dramatically increased access to information. Libraries, archives, historical societies, and museums are all digitizing collections to make them widely available on the World Wide Web. And so is everyone else—from faculty in today's most prestigious institutions to elementary school children. In a society that praises "new voices," the Internet is a wonderful forum for all who have a contribution to make.

Still, the challenge of preservation in this new context is even greater than it was for brittle books, because unless the preservation of digital information is considered at the point of creation, there is little chance that such information will remain available in the future. Microfilming promised a life span of at least an additional 100 years for books and journals printed on acidic paper. Today, no such promise exists for preserving digital information. Moreover, the storage media for digital information present their own preservation problems. Much digital information is housed on magnetic tapes and disks, and we know from experience that magnetic media are short-lived. A whole generation of folklorists is now agonizing about ethnographic tapes collected in the 1960s and 1970s that are disintegrating on the shelves of library and folklore institutes around the country. As the faculty who recorded the voices and music of distinct cultures approach retirement age, they realize that the information they gathered may never be used for the creation of future knowledge, but will instead turn to flakes of celluloid if action is not taken to rescue this information soon.

The challenges posed by analog audiovisual materials presage some of the challenges presented by digital technology. With each evolution in media, from print to analog audiovisual to digital, we have had less time to decide what to save and to implement effective preservation. Establishing common systems of cataloging, searching, and retrieval has become more complex and urgent, as has addressing copyright issues. Librarians alone cannot resolve the issues of preservation and access in the digital realm. Cooperation between librarians and scholars will become increasingly necessary in managing the collections upon which future research depends.

Scholars could make valuable contributions in the following areas.

• Selection for preservation. Although printed materials remain important in our libraries and archives, the documentation of the twentieth and twenty-first centuries exists in many different formats. Television broadcasts, recorded sound, film, and photographic images hold increasing importance for scholars, but libraries have not perfected procedures for incorporating these materials into their collections. Moreover, the cost of acquiring and maintaining research materials in numerous storage formats has become prohibitive for many institutions. Even more important, many Web-based resources are fundamentally different from regular published materials. They cannot be "acquired" by libraries, but faculty creating or identifying these materials can help alert the librarians to Web resources that should be preserved for future research and instructional purposes.
  
  
• Collection building. Scholars and librarians have become too isolated from one another. Librarians, trying heroically to tame new technology so that it can be applied to scholarly and instructional purposes, need to turn their attention to the truly important question of what constitutes a research collection in the twenty-first century. They can answer that question only in consultation with faculty. Scholars must realize that they also have responsibilities for defining and nurturing collections for subsequent students and scholars. Imagine, fifty years from now, trying to explain the development of the World Wide Web or the history of changes in scholarly communication in the twentieth century. Where will our students go to find the source materials for such subjects?
  
  
• Establishment of electronic archives. Librarians have carried the responsibility for preserving the intellectual record. Such preservation will be enormously difficult in the future unless librarians, scholars, and publishers work together to create trustworthy archival repositories of digital information.

The collaboration I have called for is urgently needed. Digital information resources are short-lived. Creators of the material and scholars in the field are best qualified to identify important materials, and librarians are the best hope for the long-term survival of such materials. I hope my remarks will stimulate this desperately needed collaboration.