To facilitate the interdisciplinary training and to ensure that professional standards were being maintained, the archives staff worked with an archival metadata consultant to produce the Descriptive Policies and Practices Manual (DPPM).¹² This manual provided a much-needed local content standard based on existing national and international content standards and best practices for cataloging unique materials, such as AACR2R and DACS as well as format standards like EAD, MARC, Dublin Core (DC), and Metadata Object Description Schema (MODS).¹³

The DPPM identifies the desired data elements for each level of archival description: collection, series, container or folder, and item. Definition, format guidelines, best practices and standards, and local examples are included for each element. Crosswalks to existing metadata standards’ equivalent fields are also included. Definitions for each element were transcribed from the Society of American Archivists’ (SAA) A Glossary of Archival and Records Terminology; Describing Archives: A Content Standard; the Encoded Archival Description Tag Library, version 2002; or were defined locally.¹⁴ Locally defined elements were noted as such.

Each data element described in the manual was categorized as either mandatory or optional. Mandatory fields had to contain the appropriate information or a locally defined default statement and could not be left blank. Optional elements were left to the processor or archivist to determine whether the use of the element was appropriate and to provide the information. The work that went into defining the necessary descriptive elements and mapping them to existing format standards, such as EAD, MARC, DC, and MODS, would prove to be integral to the success of the first implementation of SCA’s item-level descriptive cataloging and online access to digitized collection materials in PEAK Digital, the library’s digital repository.

While the DPPM was being written, plans were made to select a new standards-based archival collection management system. SCA needed a system that would serve both as a collections management system and as a vehicle to output metadata. The system needed to be able to map to MODS, EAD, DC, MARC, and potentially other metadata standards. It needed to be scalable, standards-based, and interoperable. The SCA chose Re:discovery Proficio, a standards-based system that would serve as a management tool for both print and digital formats and as the metadata provider for a public access tool. Each of the data elements described in the DPPM corresponded directly with a field in a Re:discovery record, depending on the applicable level of description.

The DPPM provided standards for field use so that productivity tools inherent in database management systems could be used to minimize data entry time. These tools made it possible to copy and clone records as well as set up templates so that fields with homogeneous metadata (subject access terms, physical characteristics, etc.) could be automatically populated, much in the same way that records in an ILS can be generated. These tools facilitated the kind of efficiency required to make the item-level cataloging of unique materials a reality. This is the case because, especially in lower levels of description, individual records within any particular collection or series are almost entirely homogeneous in many respects and can be batch-generated.

Cataloging staff added information on the basis of the unique nature of the object in hand. The resulting record has both unique information about that object and contextualizing information that make sorting and grouping this item by numerous attributes possible. The addition of basic, controlled subject-access terms (descriptors) to each item-level record, each of which is presented individually within the digital repository with proper attribution as to collection and series, frees the record from the constraints of being discoverable only within the context of its collection.

Rather than being an access or discovery system in itself, the collections management system is a source of data for other systems. For that reason, the system had to have the ability to export content in multiple formats. The DPPM provided the standard for content description and provided the basis for developing export template mapping. Beginning with the database fields, which, although based on MARC and MODS, were schema agnostic, the management system allowed staff to output data in several metadata schemas and formats, including MARC, MARC Extensible Markup Language (MARCXML), MODS, Metadata and Encoding Transmission Standard (METS), DC, and Resource Description Framework (RDF).¹⁵

Further, Technical Services cataloging staff configured the MARC mapping so that collection-level records could be exported to MARC–based systems. This effort was especially important because it helped cataloging staff learn about the back end of the software and allowed the special collections staff to become familiar with library metadata approaches. Since the original MARC mapping, additional metadata maps have been created, including METS, MODS, DC, EAD, and Text Encoding Initiative (TEIP4).¹⁶ These maps are used to export data to various access systems as needed.

Training was essential to combining the units and merging workflows. Archival processing is an organic process that is dependent on continuing discussion between everyone who has a stake in the outcome. Therefore all staff involved in archival processing were trained in the basics of contemporary archival theory and practice. Experience suggests that a better product would result from all staff understanding how the work they did fit into the larger picture, even if the staff person was not directly involved in all stages of processing. The archival training program set up by the Archives and Technical Services librarians included both theoretical and practical aspects of the archival profession. A series of two-hour training sessions was offered to all staff who would be involved in archival processing. The program was an opportunity to develop new skill sets while building on standard archival skills and to promote team building.

The reorganization of library resources to improve workflow in the Special Collections Unit was an enterprise-wide integration of the needs of special collections into the greater resources of the library. The APU was created by separating the archives processing staff from archives reference services staff and transferring responsibility for archival processing to the Technical Services Unit. Other staff reassignments followed on the basis of a library-wide assessment of changing workflows and needs. Organizational changes were not limited to the Archives and Technical Services units. As workflows and procedures shifted, Stacks Maintenance staff members managed the Special Collections and Archives shelves. Physical space for archives processing was created in the Technical Services area. The area occupies 34,665 square feet; approximately 14,598 square feet (approximately 42 percent) were converted for use by the APU. Several modifications to the physical space were made to accommodate this unit’s special needs. Security practices were reviewed and determined to be adequate to properly protect the library’s collections during processing.

Workload responsibilities and priorities within the Technical Services Unit were modified to support archives processing. Inventory and database maintenance projects schedules were extended to provide more time. The item-level cataloging of archives materials in the collection management system was added to the responsibilities of three catalogers. Catalogers are also responsible for assigning subject headings to collection-level records.

Both the Monographs and Serials units were already cataloging special collections books and serials, and that work was integrated into the new structure. Student workers from the Technical Services Unit, who label books and process government documents for remote storage, were assigned basic tasks for archives processing. End Processing staff members were assigned three hours per week to assist Special Collections in various lower-level tasks, such as box building. In addition, an experienced catalog technician, working with the head of Technical Services, was assigned to work on the back-end configuration of the collections management system to align it to meet the standards for MARC and export through MODS. Four catalogers now spend at least some portion of their time processing archival materials. One spends a minimum of twenty hours per week on some phase of archival cataloging (e.g., authority work or importing spreadsheet data). Three other catalogers spend three to four hours per week on item-level processing.

The team structure has proved especially beneficial in maintaining communication across the newly reorganized units, necessary because the culture of archival processing changed from a one person–one collection relationship to a many people–one collection relationship. The higher, intellectual levels of description, collection, and series are still performed by either professional archivists or highly trained staff members under the supervision of trained archivists, but many people touch the collection at different stages of work. Student processors and hourly workers are assigned lower-level tasks as needed. All stages of processing are highly important and contribute to the quality of the description of archival collections, and that concept always is conveyed to all staff.

Processing backlogs exist in most archival collections, and the local situation—with a significant backlog of legacy collections having little or no processing—was no different. Since one goal of the APU is to gain complete physical control over all collection materials, these materials were integrated into the workflow in multiple ways. One approach to dealing with some of these legacy collections has been to use scanning requests from users as processing opportunities. Another approach is to gain basic physical control over boxes and use that information to inform processing decisions.

For example, the university’s extensive historical photograph collection has a high demand for reproduction requests, yet relatively few of the images have been scanned, and they lack subject access beyond the particular folder in which they reside (buildings, sports, etc.). Under the new workflow, a photograph requested for scanning is first sent to the APU. The staff enters the photograph into the collections management system and is given the next number in the numerical sequence for that collection. The photograph then has an identification number that can be used as the file name when the photograph is scanned. The photograph is then sent to the Digital Production Services Unit for scanning following the scanning standards set up for archival materials.

Early in the process, the APG decided to implement standard subject access points within the collections management system to organize and describe photographs and other images. In this way photographs can be cataloged in the order that scanning requests are presented. Because the access point is a subject term rather than an intellectual series, when this record is exported to an access system a user is not required to look in the artificial Buildings series to locate images of university buildings. Instead, an access system will aggregate all cataloged images with the term “buildings,” regardless of their place within a physical arrangement scheme.

In addition to being more flexible in terms of workflow, the addition of multiple subject access terms to each image enables the user to find an image that may not have “buildings” as its primary subject matter or content. A subject heading may denote the presence of a university building in the image, thus providing the user with the ability to retrieve more useful resources and to recontextualize the object depending on his or her needs. This access to the photograph does not require the user to understand where that particular record resided in a largely arbitrary organization. In this way, processing workflow is not only more efficient, it also supports more flexible discovery for the user.

Not all collections are as homogenous and easy to process as the photograph collection. Many collections, especially personal papers, contain random boxes of material that have little or no intrinsic order. Since these materials were already part of the collection, they could not be accessioned again, yet needed to be accounted for and managed prior to processing.

A separate database, called the Unprocessed Materials Database (UMDB), was set up in the collections management system to handle accessioned but unprocessed material. The UMDB allows the APU to gain physical and basic intellectual control over newly accessioned material and create a preliminary collection inventory. Once an accession is complete, the processor creates a collection-level record in the UMDB with the accession number used as the collection number and a brief descriptive title that mirrors the origin of the material. These are the only two fields that are filled at this level.

The processor creates a record for each box or container record under the collection record in the UMDB, and numbers all boxes consecutively, without regard to intellectual order. Processors are instructed to take no more than two minutes to look in the box or container and come up with a few words about its contents. If conservation or preservation issues are present, they are noted in the description field. The processor then adds a barcode to the box or container and scans the barcode into the barcode field. Accessioned boxes are stored in barcode order in the unprocessed materials area. This preliminary inventory is later used to create a processing plan. When a box is taken for processing, its box record is cancelled in the UMDB. When all boxes have been processed, the collection record is cancelled from the UMDB. This process allows for basic descriptive access and physical control over all special collections materials, processed and unprocessed, while not taking an inordinate amount of time.

Processing on demand and the unprocessed materials database solve two specific challenges facing the APU, but they do not address the issue of providing highly granular access to large groups of items. Creating metadata for thousands of unique collection objects in a manuscript or archival collection has long been considered cost-prohibitive. In the pre–automated environment this was usually the case, and item-level cataloging was reserved for only the most valuable collections. Changing user behaviors and demand for item-level access in the digital environment continued to indicate that developing some means of automated, mass metadata creation that could satisfy user demand for highly granular access was imperative.

Archival and manuscript collections arrangement and descriptive cataloging have always been highly labor-intensive, frequently involving extensive physical arrangement (though not generally description) down to the item level.¹⁷ The archival community has made progress in emphasizing production and throughput, establishing format and content standards for archival description, such as EAD and DACS, and in using collections management software options and online content delivery systems. Still, the professional archival community’s focus has remained on the collection as the primary unit of description and access. This pattern persists despite evidence that users of archives are confused by both archival terminology and the ways that metadata about primary resources is made available.¹⁸

The library’s solution to this dilemma was to use the power of database tools to mass create metadata for groups of homogenous content (e.g., items in an archival series that would allow access systems to aggregate these items into contextualizing groups). The implementation of an item-level archival cataloging content standard (the DPPM) required some in-house adjustment, since neither AACR2rev nor DACS specifically focuses on the cataloging of item-level unpublished content. Rather than try to choose one schema that could meet all possible needs, the item-level field content in the collections management system was created according to a flexible local content standard that could be mapped to multiple metadata schemas. In this way the APU was strategically placed to support multiple metadata consumers. Understanding metadata normalization and crosswalks also allowed the unit to play a major role in the development of a batch ingest process for item-level records into the consortial digital repository supported by the Colorado Alliance of Research Libraries (www.coalliance.org).

The next major decision that affected metadata was that the item records would not use precoordinated Library of Congress Subject Headings (LCSH) but would instead employ individual LCSH terms as descriptors. This decision was made because most Web-based access tools are not designed to handle subdivided headings, preferring to use subject descriptors instead. The use of this widely implemented source of authority records still allows for the aggregation of like records within consortial content delivery systems and aggregators like the University of Michigan’s OAIster (www.oaister.org). This has the added benefit of not requiring either the processors of these records or the end-user to understand the precoordinated structure of LCSH.

• The increase in the number of staff performing item-level cataloging resulted in a significant increase in production.
• The quality of the metadata produced was enhanced because of the participation by experienced, trained catalogers. Catalogers are trained in a standards-based approach to metadata production, while archivists typically are not.
• The entire staff who worked on the project was energized by the opportunity to work on something interesting and different. This project extended their skill set beyond the norm.
• The formation of the Archives Policy Group established a regular channel of communication for the professional librarians that was extremely effective when curators and processors needed to be on the same track. The unit and team meetings also advanced the effort toward constant communication. The benefits extend beyond this project by providing an opportunity for staff to engage in library planning and policy development.
• The need to involve other units earlier in the process, most notably Stacks Maintenance and End Processing, was necessary to streamline workflow. These units were invaluable in contributing their expertise and their resources.
• One unanticipated benefit of integration was a noticeable improvement in morale for the catalogers and archival processors. Copy catalogers had an opportunity to perform original cataloging in the collections management system. The catalogers juggled archives with monographic cataloging and learned to fit this new form of cataloging into their everyday work. For the catalogers, varying their routine with new projects and learning new skills was seen as a professional benefit that is rewarded in performance reviews. Archival processors had assistance in producing item-level records at a level of detail that is rare in the archival world.

• The collection management system selected was not able to handle authority records as expected. Without “see” and “see also” references, the processors needed to enter two forms of the same name in a record (e.g., the earlier name of a building and the later name).
• Because there was no funding for new positions, new technologies workflow processes had to be developed by temporary hourly staff. The use of temporary staff, usually graduate students, has proved somewhat problematic because of high rates of staff turnover, resulting in the loss of specialized technical skills. This loss meant an increase in training needs.
• A comprehensive staff training plan needs to be in place to develop broader skill sets for staff so that they understand multiple metadata standards. For example, MARC catalogers needed to understand MODS and needed to be able to catalog in MODS or DC. Likewise, staff members who were familiar with collections in an EAD–centric way needed more training so that they could be comfortable breaking collections apart, thus advancing to the goal of not limiting users by provenance.