Literature for Children seeks to make accessible volumes or children’s literature, published in the United States and Great Britain between 1850 and 1923. Titles published during this period are now in the public domain and can be reproduced and access freely by anyone. This effort is enabled through funding from the National Endowment for the Humanities, with initial grant projects awarded to the University of Florida. Literature for Children draws from the children’s literature collections of the state’s public university libraries and from the collections of their partners in K-12 and higher education. These volumes form part of the Baldwin Library of Historical Children’s Literature, housed in the Department of Special Collections at the University of Florida. But, the currently available on-line collection includes volumes drawn from the Special Collections of Florida State University, Florida Atlantic University, and the University of South Florida. Together these institutions’ holdings are impressive. The Baldwin Library, alone, holds more than 86,000 titles. Literature for Children includes titles both with and without color. Effort is currently concentrated on digitization of titles with color, since color is not well preserved by traditional reproduction-for-access methods such as microfilm. Literature for Children makes titles freely available via the Internet. Readers may acquire copies on CD-ROM, in PDF format, for a fee – PDF versions can also be downloaded freely from the Internet and saved to CD by readers with their own burner. Additionally, the majority of titles are contributed to and made available through the International Digital Children’s Library and through the Internet Archive where patron’s of the Archive’s bookmobile may reproduce facsimile reprints.

IMAGE CAPTURE AND CONVERSION

Participating libraries are responsible for digitization of selected materials from their own collections. Each library may perform its own digitization, or contract with a vendor or with another SUS library for digitization services. Image capture must adhere to the standards promulgated by the Cornell Department of Preservation and Conservation (see Digital Imaging for Library and Archives, Kenny and Chapman, 1996). A Quality Index of 5 or better for visual images is required.

Three types of images are created for all textual materials in the collection: TIFF, JPEG and PDF. A TIFF and JPEG image is created for every page; related sets of pages (e.g. chapters or articles) are bundled into PDF files. In the first year of the project, participating libraries created TIFF images and submitted them to FCLA, which subsequently created PDF and JPEG derivatives.

TIFF images are created as the direct result of scanning source materials (that is, as the native file format), using a variety of scanning hardware, primarily flat-bed scanners. TIFFs are archived as uncompressed electronic masters. Bit-depth is appropriate to the source and its anticipated use, and may be bitonal, 8-bit grey, 24-bit color, or greater. Color images are created and maintained in the sRGB color-space. Both grey and color images are calibrated and scanned to within the tolerances promulgated by the Library of Congress for the American Memory project. Images created from microfilmed sources reflect the quality of the source microfilm.

TIFF images are used to create JPEG derivatives using Adobe ImageReady Version 2.0 in a batch executable process. The TIFF image is resized setting the width to 600 pixels and the height accordingly. The process then progressively optimizes the image creating an image that displays progressively in a Web browser. The image will display as a series of overlays, enabling viewers to see a low-resolution version of the image before it downloads completely.

Creation of PDF files is a function performed by the locally written Florida Heritage loader software. The loader calls LeadTools custom ActiveX control to open sets of JPEG images, and then uses Thomas Mertz's PDFLib software to build the PDF.

Text-based versions, whether encapsulated with PDF, HTML or other mark-up, are produced either by re-keying from source documents or by optical character recognition (OCR) of TIFF images. A minimal accuracy rate of 99.995% is required.

RESOURCE DESCRIPTION -- CATALOGING

Participating libraries are responsible for creating full MARC catalog records for selected materials from their own collections. Cataloging records are maintained in a union database of all Literature for Children materials at FCLA and are also contributed to the OCLC WorldCat.

Cataloging is expected to adhere to guidelines developed by the Technical Services Planning Committee Cataloging and Access Guidelines for Electronic Resources (CAGER). The guidelines specify that records should represent the electronic versions only, and include specific instructions to:

* Put the date of the original in Fixed Field Date1, the date of digitization in Date2, and use Form of Reproduction "s";
* Include a title (245) subfield h to indicate the resource is electronic;
* Specify the digitizing institution and date of digitization in the imprint (260);
* Include a series statement (830) for the Florida Heritage Project, justified by a general note (500);
* Use an original version note (534) to record the location of and publication information for the source document.

Complete MARC cataloging instructions can be found in the CAGER Guidelines.

RESOURCE DESCRIPTION -- STRUCTURAL METADATA

A file of structural metadata is created for every document to indicate the relationship between the physical units of digitization (TIFF, JPEG and other images) and the logical units of publication (pages, chapters, and other parts). The metadata format used is a modified version of the Elsevier EFFECT format called DataSet.TOC.

For each electronic resource (book volume, journal issue, manuscript, etc.), the DataSet.Toc file:

* identifies and names the image files comprising the resource,
* defines the order of images,
* identifies and names the subsections (such as chapters),
* says which images belong to particular subsections,
* and establishes the order and hierarchy of subsections.

IMAGE LOADING, STORAGE and NAVIGATION

For each volume that is digitized, a directory containing one DataSet.TOC file and a set of images is sent by FTP from the contributing institution to FCLA. The metadata and images are processed by a locally written loader, which first checks that all the image files referenced by the DataSet.TOC are present, copies the images into a Literature for Children directory, and loads the structural metadata into DB2 tables maintained on a Unix server. If instructed, the loader will also create derivative formats such as PDF files.

Once structural metadata is loaded and images are moved to the appropriate directories, access and navigation is provided by another locally written DB2 server program.

Persistent URLs referencing the server application are created by program and inserted into the bibliographic record describing the resource.

RETRIEVAL

The cataloging records describing Literature for Children resources are loaded into a shared central library management system, a locally developed application based on NOTIS, on an IBM mainframe. The records can be searched through the SUS Libraries' online catalog application, WebLUIS. Once records are retrieved, the URLs in the bibliographic record are used as hotlinks to the DB2 server application, which initially presents a Table of Contents display.