Sustainability of Digital Formats: Planning for Library of Congress Collections

PDF (Portable Document Format), developed by Adobe Systems Incorporated, is described by Adobe as a general document representation language. PDF represents formatted, page-oriented documents. These documents may be structured or simple. They may contain text, images, graphics, and other multimedia content, such as video and audio. There is support for annotations, metadata, hypertext links, and bookmarks.

Versions through 1.3 are described here together for convenience.

The PDF Reference editions that specify PDF versions 1.0 through 1.3 are:

• PDF Reference, First Edition, version 1.0 (June 1993), Addison-Wesley, ISBN 0-201-62628-4. Associated with Acrobat 1.0.
• PDF Reference, First Edition Revised, version 1.1 (March 1996), Adobe Systems Incorporated. Associated with Acrobat 2.0.
• PDF Reference, First Edition Revised, version 1.2 (November 1996), Adobe Systems Incorporated. Associated with Acrobat 3.0. Includes Appendix G which discusses how viewers for Acrobat versions handle features introduced in PDF specifications through 1.2. Available via Internet Archive.
• PDF Reference, Second Edition, version 1.3 (July 2000), Addison-Wesley, ISBN 0-201-61588-6. Associated with Acrobat 4.0. Available via Internet Archive.

PDF 1.3 has been very widely used. As of early 2019, it is still the version of PDF built in to MacOS 10.14 (Mojave) and used by the Pages word-processing application.

Identification of chronological versions of PDF can be given in two places in a PDF file. All PDF files should have a version identified in the header with the 5 characters %PDF– followed by a version number. For PDF files conforming to ISO 32000-1:2008 or earlier specifications (i.e. prior to ISO 32000-2:2017), the version number has the form 1.N, where N is a digit between 0 and 7. For example, PDF 1.3 is identified by %PDF–1.3. However, beginning with PDF 1.4, a conforming PDF writer may use the Version entry in the document Catalog to override the version specified in the header. The location of the Catalog within the file is indicated in the Root entry of the file trailer/footer. In early 2019, a test on Mac OS 10.14 (Mojave) was performed: a .pages document created in the Pages word-processing application was imported into the Preview application and exported to PDF. The resulting file had %PDF-1.3 in the header but version 1.4 was specified in the Catalog. This override feature was introduced to facilitate incremental updating of a PDF by simply adding to the end of the file. As a result, it is necessary to locate the Catalog within the file to get the correct version number. Unless the PDF is "linearized," in which case the Catalog is up front, this will require reading the trailer and then using the reference there to locate the Catalog, which will typically be compressed. This has practical implications because format identification tools, including DROID, typically look for particular characters at the beginning of a file (i.e., in the header), to permit identification with minimal effort. DROID can look for characters at the end of the file, but is not able to follow an indirect reference or decompress file contents. When the version number is not the same in the header and the Catalog, there is potential for format identification errors.

The JHOVE PDF module does take account of the situation, stating that for PDF 1.0 - 1.6, "The PDF version is determined by the data specified in the PDF header and the Version key of the document catalog dictionary. In the event that these two values do not match, the Version key is taken as the authoritative value."