Digitizing and Delivery

Garden and Forest: A Journal of Horticulture, Landscape Art, and Forestry is the first digital reformatting project done by the Library of Congress's Preservation Reformatting Division. It is also the first serial digitized in its entirety by LC. This digital-reformatting project represents the addition of digital technology to the list of options available for crafting preservation strategies for Library collections.

Selection

There has been high interest in digitizing Garden and Forest since the early years of the American Memory program. Its brittle condition, however, prevented it from being inverted for scanning on a flatbed scanner, and its size (21.5 by 30 cm.) made it too big for the book cradles then available for use with overhead capture devices. When the Library of Congress's Preservation Reformatting Division began to plan its first digital reformatting project, Garden and Forest was considered a good candidate for several reasons: its embrittled condition precluded continued use; its content was of high interest; and digitizing the full run of the serial would provide an opportunity for the Library to gain experience with digitizing and interoperability models beyond those employed by the National Digital Library Program (NDLP), including approaches developed in the Preservation community.

Planning and Preparation

Planning and preparation for the project involved a blend of well-established methods from preservation microfilming and digitizing models used by NDLP and others. Critical steps in the process included:

• searching the holdings of other institutions to determine the existence of microfilm copies of the journal and evaluating microfilm copies to determine whether they met LC standards
• analyzing the original volumes to determine completeness and the extent of key needs, such as conservation repair and replacement of missing pages
• identifying potential institutions for interlibrary loan requests for replacement pages
• investigating the original publication order of pages (because advertising pages in the LC copy had been bound out of order)
• developing project workplans and tools, such as databases
• developing, circulating, awarding, and managing a new contract for digital imaging
• conferring with the Library recommending officer about the disbinding and long-term storage of the original, and disbinding the bound volumes
• developing a partnership with the University of Michigan in order to explore new conversion and interoperability options
• developing and adapting structural and administrative metadata for the serial, based on Making of America models, and
• planning for access aids to be developed in collaboration with the Arnold Arboretum

Collation

Which included the following:

• reviewing the ten volumes page by page
• creating page-level structural and administrative metadata
• entering metadata and other information into a database
• requesting replacement pages, as needed
• sending pages for conservation repair, as needed
• rearranging pages into the original publication sequence
• preparing targets for insertion into the volumes, as needed, and
• inserting replacement and repaired pages into the volumes

Reformatting

The 8,400 pages of Garden and Forest were captured on a Xerox Docuimage 620s flatbed scanner at 600 dpi bitonal. The resulting images were 7216 by 5088 pixels in size. The 480 pages with printed halftones were descreened and rescreened using Scantool software. Later in the project, after the results of the halftone-processing technology were evaluated, it was decided to also create 400 dpi grayscale images for the 480 halftone pages. (See the Illustrated Book Study for further discussion of related imaging issues.) Craig Jensen, of Acme Bookbinding, provided the scanning and image processing services.

The text was converted and encoded by the University of Michigan Digital Library Production Service(DLPS), under the direction of John Price-Wilkin and Christina Powell. Machine-readable text was produced by Optical Character Recognition (OCR) using Prime Recognition. This text was left uncorrected ("dirty OCR") and minimally encoded with the Making of America (MOA) SGML Document Type Definition (moa.dtd) used at the University of Michigan for their MOA texts. This DTD is conformant with the Text Encoding Initiative (TEI). Texts were encoded according to the recommendations for Level 1 in the TEI in Libraries Guidelines.

Reformatting work performed by PRD staff included:

• preparing and shipping volumes to the contractor
• evaluating technical issues with service-providers during conversion steps
• developing quality-review methodology and standards
• training staff to perform quality review, use new hardware and software, and keep appropriate records
• performing quality review on delivered image files and reworked files
• assessing contractor and staff performance in maintaining preservation standards
• making copies of complete image set to send for text conversion and encoding at the University of Michigan
• preparing technical requirements for text conversion and encoding
• preparing document headers for text files and sending to DLPS
• adapting tools for the quality review of text files
• performing quality review on text files

Backup and Archiving of Digital Files

Digital image files were delivered to the Library on CDRs and copied to one of the Library's servers after quality review was completed. Reworked and replacement files were integrated to make a complete and final image set on the server. Administrative metadata regarding stages of archiving were entered into the project database. The delivery media, CDRs, were housed and stored according to preservation requirements for handling, storage, and environmental conditions.

Delivery and Interoperability

Plans for online delivery were developed in collaboration with NDLP and Information Technology Services (ITS) at LC, and the University of Michigan DLPS with the goal of building a foundation for the interoperability of Garden and Forest with other Making of America digital materials. This involved developing a model for phased delivery that allows progressive additions of features and functionality to be provided by different parties, over time. Also, a collaborative effort with the Arnold Arboretum of Harvard University led to plans for the phased addition of access and interpretive aids, such as background essays and a comprehensive subject index. Phase 1 of delivery, in December 1999, included:

• working with ITS and APLO to register a persistent identifier for digital reproduction
• working with catalogers in the Serial Record Division to add persistent identifier to existing LC bibliographic record
• mounting G&F image and text files at the University of Michigan
• indexing text files for searching through the OpenText engine at the University of Michigan
• modifying MOA middleware and web pages at the University of Michigan
• creating the G&F web site and web pages at LC
• creating background essay at the Arnold Arboretum
• writing and editing web page contents at LC
• testing online implementation by University of Michigan and LC

The features of Phase 1 delivery include the Garden and Forest website at LC; image and text files mounted at University of Michigan; Michigan search and browse forms mounted at LC but pointing to the Michigan search engine; and search, retrieval, and display at Michigan. During Phase 1, searching of Garden and Forest is separate from searching Michigan MOA materials.

Future phases of delivery will include cross-collection searching of Garden and Forest and Michigan MOA materials; article-level indexing of Garden and Forest; addition of interpretive "value-added" enhancements developed by the Arnold Arboretum, such as historical background essays; and addition of a subject index created from the print volume indices, developed collaboratively by the Arnold Arboretum and LC. Every phase of delivery will add more access aids for users and create greater integration of Garden and Forest and the Making of America digital library.