I. Activities to date

Began to plan out the sequence of transfers to LSFF

From the beginning it has been clear that materials from some parts of the collections would be in the front of the queue to move to the LSFF, from necessity. These "early movers" include the Music Library, the Divinity Library, Manuscripts and Archives, and the (new) Arts Library (Art & Architecture, Drama, and art materials from Sterling and Mudd). The catalog records for these collections are either fully converted to electronic form or soon will be. It is probable that the transfer of materials to the LSFF from these four parts of the collection will take eighteen months to two years. We are identifying additional "early movers" as opportunities arise.

Surveyed YUL selectors

Early in the summer the Selection Group designed a survey to elicit ideas from library selectors and curators about categories of material to transfer to the LSFF. The librarians were asked to suggest criteria against which to select 20 percent of their collections or areas of responsibility for the LSFF. One hundred percent of the selectors responded, and there were discernible patterns indicating that the Orbis database would be a useful tool for identifying a significant number of titles eligible for transfer. Sub-classifications, location (e.g., Mudd Library), date of publication, and language were also suggested as useful indicators obtainable from online records. The availability of other formats, such as electronic versions, multiple copies and editions, form or genre (e.g., auction catalogs), or presently non-browsable items (e.g., archives collections, restricted access volumes) would be additional significant considerations for transfer to LSFF. Transfer decisions that might require detailed review of content presented particular challenges, including materials that might no longer be relevant to the curriculum, indexed periodicals that are no longer published, secondary works, etc.

While marking the beginning of the wider discussion among selectors, the survey also began to indicate the range and depth of issues that continue to surround the selection process.

Held two information forums for selectors

These took place in August and September and were designed to help inform and prepare selectors for faculty meetings by offering brief formal presentations and opportunities for extensive questions and discussion. The presentations focussed on 1) possible strategies for using Orbis and the recon process to help identify appropriate materials for transfer to the LSFF, 2) some of the extra-LSFF circumstances that will determine the schedule by which material will be transferred, and 3) the Systems support that will be needed if we are to use automation to best advantage. Paul Conway reported on the results of a related study conducted by the Preservation Department designed to help define use and circulation patterns, preservation needs of collections, and the scope of preservation challenges involved in transfer from libraries to the LSFF (see attachment A). In addition, there were numerous questions and extended discussion about such matters as the use of language as a criterion for selection and the ways in which the need for preservation might govern decisions about transfer to the LSFF.

The survey responses and the participation in the forums underscored the probability that there would be no single selection method that could be successfully applied across all areas of the collections.

Encouraged and coordinated selectors' initial consultations with their faculty

To date the following schools, departments, and programs have been contacted by the relevant selectors and a process of identifying candidates for LSFF has begun: Art & Architecture, Divinity, Drama, Forestry & Environmental Studies, School of Management; Chemistry and other science departments, Classics, East Asian Languages & Literatures, Economics, English, Film Studies, French, German, History, History of Art, Italian, Judaic Studies, Linguistics, Near Eastern Languages & Civilizations, Philosophy, Religious Studies, Spanish & Portuguese. The Canadian Studies Committee and the Council on East Asian Studies have also been visited; members of the Selection Group have participated in some of these meetings by invitation. Additionally, selectors have met in several small groups to coordinate the responsibility for, in one case, older science and medicine material and, in another, all facets of religious studies.

These early (and continuing) conversations with the faculty have elicited many varied responses, in some cases skeptical, in others enthusiastic. There has been near-universal concern about the state of the Orbis bibliographic records, and almost as universal agreement that scholarly journals should remain on campus. Otherwise, the faculty's recommendations cannot be generalized.

Devised and tested selection reports from Orbis

The Selection Group developed a method to use existing Orbis records, and those to be created during recon, to facilitate the process of selecting materials for the LSFF. Individual selectors, in consultation with the relevant faculty, graduate students, and library colleagues, would determine criteria that would be discipline-specific and could be translated into formulae for reports to be run against both the existing Orbis database and the regular loads of records from the recon vendor (OCLC?). Some possible criteria include publication date, Library of Congress or Yale classification, and format. Ideally these criteria could be chosen and cast in terms of fairly limited Orbis data fields; in some cases, recurring patterns or standard reports might suffice. The reports (whether output as paper or loaded into an electronic database) would then serve as the lists of titles to be transferred. This method has the advantage of using automation to manage the otherwise overwhelming volume of selection; it also can be implemented both on the present database and on the records recon will add. Its principal disadvantage, at least in our initial thinking, is the "systems premium" it commands in both up-front resources and continuing support.

To test the method, Suzanna Lengyel ran several reports. The first series identified the number of titles in the Sterling and Mudd general collections (including Judaica) with imprint dates of pre-1800, pre-1850, and pre-1900. Because of the nature of the formula we developed, these reports excluded circulation records, thus ensuring that most of the identified titles would fit one of the overriding criteria for LSFF transfer: appropriate representation in the online catalog. The numbers were disappointing: 10,046 pre-1800 titles, 28,849 pre-1850 titles, and 86,331 pre-1900 titles exist in appropriately cataloged form in the current Orbis database. Clearly an across-the-board date cut-off criterion--even if it won widespread agreement--would not yield much volume. Succeeding reports run against the English literature classes in Sterling yielded the same sort of results: 343 pre-1800 titles with full Orbis records, for instance. The only report that seemed even slightly promising in terms of numbers was developed to catch English fiction published between 1900 and 1980 and not used after 1990. This report yielded slightly fewer than 4000 titles.

These tests signal several things. First, there is nothing wrong with the methodology: Orbis reports can be developed to produce the information we need to select materials for the LSFF. Second, however, and not surprisingly, the bulk of materials most suitable for transfer are precisely the materials without adequate (or any) representation in the online catalog.

Considered and tested other selection methods (e.g., the modified "Carpenter method"):

At Harvard, one librarian (Ken Carpenter), not a subject specialist, selects for the Harvard Depository from the Widener Library stacks. His method (described to us during his visit to Yale earlier this year) is to target a specific area of the stacks and to hand-pick the books for transfer to the Harvard Depository, bringing into play his long experience with the library system and the academic curriculum, and his bibliographic knowledge. The advantage to this approach is the ability to view the books in context rather than relying on lists from Orbis reports; the disadvantages include labor-intensiveness and inability to use automation effectively.

In response both to Ken Carpenter's visit and to the frustrations experienced with the results of trial Orbis reports, we took up Audrey Novak's suggestion of using lists generated from the online shelflist to help select materials directly from the shelves. Sue Roberts volunteered to be the test case for this method, choosing part of the Yale French history class as her sample. The online shelflist includes only records for materials in Orbis, so printing out pieces of it serves to identify books represented in Orbis and thus possible candidates for transfer to LSFF. This trial is still underway, but preliminary results underscore the obvious: most of what is in Orbis is there because it is used and thus not likely to be transferred, and most of what we would wish to transfer is not yet in Orbis.

Offered funding to Systems Office to offset "systems premium" posed by selection for the LSFF

By agreement, by common sense, and by definition, the contents of the LSFF must be represented in Orbis by a cataloging record of appropriate level. Because the low-use material planned to fill the LSFF is exactly the material without pre-existing Orbis records, the obvious path to newly-added records is the retrospective conversion project. One cannot easily select using the "Carpenter," or hands-on, approach in the stacks, because it is not possible to discern whether a book on the shelf has a cataloged presence in Orbis. Hence, the Selection Group sought funding to finance a Systems Office "premium" to buy the programming time required to create reports to aid in selection by running specific criteria against data on the recon tapes that will be loaded into Orbis at a rate of 50,000 records per month. The Group was successful in its pursuit of funding and set aside $25,000 for the Systems Office.

Unexpectedly, funding did not elicit the results expected by the Selection Group. It became apparent that there is no mechanism in place to put a Systems premium to work to buy temporary Systems project support. The full loads of the staff could not be expanded, at any price, to accept the additional burden of selection reports. This turn of events presented a quandary to the members of the Selection Group, as well as to selection staff with responsibilities that require regular, sporadic, or project-based support from Systems.

Helped lead the system-wide discussion of treatment of special material

Planning for the LSFF has generated discussion of related issues, one of them the possibility of changing the way some "special" (or "semi-rare") materials are housed, treated, and used. The Sterling and Mudd stacks offer many examples of items particularly at risk from the ravages of time, mishandling, mutilation, and theft: early (17th- and 18th-century) imprints, crumbling 19th-century paper, books with plates, ephemera, pamphlets, limited editions, folios, controversial material, fine printing. Most of these still circulate. Selecting for the LSFF offers an opportunity to identify these materials and to rethink earlier decisions about their location, decisions influenced by the difficulty of providing secure shelving and reading room service. The Collection Development Council asked a small group to investigate the scope of the problem and to write a proposal, the full text of which may be found in its report, "Reassessing the Circulation of Special Materials in the Yale Library." The Selection Group was well represented in this effort and endorsed the general proposal that these materials need special treatment and that the LSFF be considered as a highly suitable location for many of them.

II. Plans for the Next Six Months (to the end of FY 98)

Plan for coordination, leadership, funding of ongoing selection work

Once the Selection Group has finished its work of mapping out general LSFF selection strategies, and once the selectors have conferred with their faculty, devised selection plans, and ordered up Orbis reports, what selection work will remain to be done, and how will it be coordinated? Should there be a "meta-selector" to lead the necessary work across the system? If so, how will it be funded? We plan to explore these and other related questions and determine how best to coordinate the ongoing LSFF selection work.

Begin to develop Orbis reports

In order to be ready with report requests once recon has added significantly to the database, we plan to develop a template which each selector will use to specify reports helpful to selection. We expect to be able to standardize this process (and the resulting reports) to some extent. This effort will require, first, identifying appropriate Orbis data elements and, second, meeting and working with all selectors individually in order to fill out the templates. It will also entail cooperation, and time, from the Systems Office.

Develop method to track early selection decisions

We will continue to work with colleagues from Systems, Acquisitions, and Cataloging to develop a process to track early individual selection decisions. Even now, before the LSFF is ready to receive the materials selected or even, perhaps, before there is a suitable record in Orbis, selectors need to be able to identify material suitable for the LSFF from the stream of current acquisitions and from other collection development and management activities such as preservation decision-making. Such a process must identify the selected items as LSFF designates so they can be retrieved and transferred when the LSFF is open, without the selector having to see the item again.

Continue to identify appropriate categories of materials for early transfer

Given the incompleteness of Orbis, and the length of time before recon will repair that incompleteness, we will continue to identify categories of materials that are suitable for transfer regardless of the state of Orbis.

III. Conclusion and Recommendations

1. Accelerating the recon process.

On its present course, it will be at least two to three years before the recon process has generated enough records to make selection from the bulk of the collection possible; even then it may well be fitful. Transferring materials from those parts of the collection whose records are in Orbis and which urgently need to move (Music, Divinity, Arts, Manuscripts and Archives) will take one and a half to two years, at the most, of the LSFF's capacity to process materials, if the process moves at the lower proposed rate (250,000 volumes per year). Unfortunately, after those initial transfers, the anticipated pace of recon makes it presently impossible to conceive a method to find and move most of the very materials most suitable for the LSFF.

Therefore, we recommend that the Library investigate the possibility of accelerating the recon process. Might the RAC or LMC or LMT discuss the feasibility of such a thing? What are the costs of completing the work in, say, half the time? We ask these questions in a spirit of positive inquiry, believing that not only the selection of materials for LSFF but also numerous services and procedures would be made far more effective by the completion -- or at least a critical mass -- of converted catalog records.

2. Place of serials in the recon process

Because a single serial title can represent many volumes, individual serials, if selected for transfer to the LSFF, will liberate much greater space per title in the on-campus, browsable collections than will individual monographs. While neither the selector survey nor consultations with faculty found widespread agreement on the wisdom of transferring scholarly journals to the LSFF, serials' relatively great impact on space might argue for some as candidates for early transfer, and, accordingly, as candidates for early recon. There may be certain types of serials (non-periodical serials such as annual reports and directories, for example) that could be moved without detriment.

Therefore, we recommend that the particularly nettlesome issues surrounding the recon and transfer of serials be addressed as expeditiously as possible. These issues include the role of outside vendors in converting or revising records, the level and consistency of holdings information, decisions about splitting single titles between locations, and the role of alternative access and alternative format in facilitating browsing the contents of serials shelved off-campus and in streamlining retrieval.

3. Loading the LSFF

A. We recommend that in the first two to three years of LSFF, before a critical mass of converted bibliographic records exists, the facility be loaded more slowly than initially planned (i.e., perhaps even below the 250,000 per annum rate).

B. As this report has suggested, the first items sent to LSFF will be entirely on a need-to-move basis, and some 1.5 years' worth of those have already been identified (see Attachment B). While the online database is accumulating a critical mass of converted records, we recommend that the next wave of transfer be "opportunistic," involving those collections or libraries that have been able to identify and prepare suitable materials more rapidly than others. We anticipate gaining at least another year's worth of material from this approach.

C. If it is not possible to agree on a transfer rate lower than 250,000 volumes per year, we recommend exploring other methods to load the LSFF during the wait for converted records, such as moving blocks of material currently not available through Orbis and, in some cases, not browsable. Such a method might require devising an acceptable means of creating brief records or shelving the material so that it could be retrieved by call number.

D. Later years will see a gradual leveling or balancing between subjects and collections housed on campus and in the LSFF. Recon records will routinely and rapidly enter ORBIS and will be used for systematic selection decision-making in a way that cannot be done now. That is, LSFF selection work will be more opportunistic in the immediate future and move towards being more systematic in the later years.

4. Treatment of "special" materials

We recommend that the LSFF planning committees develop a service model that will accommodate "special" (a.k.a. "semi-rare") materials identified by selectors and curators. This model should provide for the supervised use of non-circulating materials in one or more reading rooms.

5. Systems Support

We recommend that the Systems Review process look to build in a flexibility that the Library can use in its data-heavy environment. This could be a network composed of external consultants who are known to the Library and who know its data needs, drawn from the New Haven area or from other NOTIS sites, or it could be composed of internal consultants from ITS or other parts of the University. Temporary hiring of freelance programmers has long been established as an effective approach to overcome the limitations presented by overloaded staff, and this should be tried at YUL.

Respectfully submitted,

Paul Conway
Max Marmor
Ann Okerson
Margaret Powell, Coordinator
Tom Schneiter
Paul Stuehrenberg

© 2006 Yale University Library
This file last modified 09/21/06

Send comments to ann.okerson@yale.edu