Computers today give us the impression that the internet makes all data available, and that an archive is a store of everything that has ever happened. Archivists are imagined as selecting the best items from a wealth of resources, so that we frequently hear that collections of various types have been "curated" by some unseen expert, with the authority to pick the best materials for users, from song lists to clothing lines to bread. Even Derrida wrote of the power that comes from controlling the archive.1 But none of this has anything to do with how historical archives work. It is not uncommon for historians to find that what they wanted to study is impossible, because no documentary evidence can be found in any collection. With new interests and new populations becoming the subject of humanities research (which is the normal evolution of academic scholarship), frustration grows as archives increasingly do not have what researchers need, and what they have come to expect from internet research. In this situation, many special collections have taken it upon themselves to increase their holdings to reflect their communities better, and in some cases, to address injustices of the past. Gaps in the archival record are understood to be a result of oppression, so the answer is to fill that emptiness. To expand collections, however, takes a great deal of creativity from both librarians and researchers, because in some cases the desired documents were never created or retained in the past. Dealing with absences has become a new model in serving patrons, even in a time when it feels like we are awash in information. The responses of wealthy American institutions to what they lack in the stacks will be important in understanding what needs to be done, mostly in terms of the absences that persist, to truly increase representation and understanding of peoples around the world. Librarians' thoughts on the destruction of materials in Sarajevo, show us some of their approaches to the issues of serving scholars when their subject cannot be covered by the usual tactics of research. In this catestrophic case, valuable resources provided information, and now they are gone, so people need new avenues to bypass the missing records and do research.

A second reason to pay attention to librarians' thoughts on Sarajevo is the more general issue of loss inherent in the collecting process mentioned earlier. Access may be increasing, but little written culture from anything beyond the recent past is truly being discovered, though improvements in access give the impression of history continually coming to light. Instead, each shelf weeding or leaky pipe at a library favors new replacements or space reassignments. Local archives lose their funding. National catalogs and interlibrary loan encourage librarians to worry less about preserving printed items that exist in other copies nearby, even digital ones, which means that physical holdings of historic publications have diminished dramatically in academic libraries, because of conscious decisions by administrators, while the increased focus on special collections has the effect of putting material, paradoxically, at greater risk through handling or disaster. The result is the loss of heritage which technology can only do so much to replace, even as digital access increases the demand for more. And combatants everywhere actively work to destroy the memory of their enemies.2 Richard Cox has estimated that the number of historic items destroyed in conflicts of the twentieth century surpassed the number saved. The cultural destruction has continued in Iraq and Syria, with little effective response from UNESCO or the International Committees of the Blue Shield.3 Looking at how libraries respond to this issue of loss can tell us how best to prepare for the realities of the future in which more and more written culture will be lost.

If the goal is to recover as much as possible about the people whose lives appear to have left no record for us to study, in an effort to expand the usefulness of special collections and to reduce their authority over what gets to be researched,4 then it must follow that we also need to acknowledge that the future will have new interests and methods of research. Just as much of the Classical world only survived in Arabic or Carolingian translations, the history that gets written by today's winners relies on the practices of last year's winners. When the United States looks very different, the country's historians will depend to some extent on the work being done in special collections now. So, the responses that American librarians have on the fate of Sarajevo's historic libraries can also reveal to us what academic institutions think that it is that are saving for, what the future means for special collections.

This is not to say that History as a field would falter without expanding libraries or changing the nature of archives. As Carolyn Steedman writes, "The practice of history in its modern mode is just one long exercise of the deep satisfaction of finding things."5 In many cases, that means finding the few relevant bits out of oceans of data. Our digital world can make the situation worse than ever. This means that the work of the historian today is a process of winnowing as much as interpretation, but that has always been the case to some extent. And even when dealing with scarcity, researchers still follow the same path. The loss of documentation to the air raids on Tokyo or the rage of the Cultural Revolution has not stopped historians, but their work needs to be built on a foundation of evidence. In his last, unfinished book, History: The Last Things before the Last, Siegfried Kracauer wrote that Historiography "is a distinctly empirical science which explores and interprets given historical reality in exactly the same manner as the photographic media render and penetrate the physical world about us."6 Scholars use computers and scientific analysis, they focus on micro history on the small scale or look at grand sweeping historical vistas.7 The work involves artistry and technical expertise to create an digestible image of complexity, just as a photographer creates meaning out of the many possibilities of the visual. So, this all means that while special collections do need to consider the needs of current and future users. They do need to protect what is under their stewardship. But they also need an understanding of their role in research and cultural heritage that goes beyond providing service to their immediate users or to abstract ideas of history or justice. If academic libraries are to offset the general trends of fewer physical documents in fewer places involving greater risk, special collections in the United States need to consider what disaster means for communities and how to protect the written culture of the world. Preliminary results of my research on special collections suggests that this is only happening in a few cases where libraries have a sense of mission that includes responding to specific destruction as part of their jobs. Otherwise, umbrage and outrage seem to do little to defend culture or recover from disasters.