PDF is a useful format. It’s an ISO standard. There’s reliable free software for reading it. It’s widely used and difficult to modify by accident. It can serve as a container for text, illustrations, audio, and video.

If used carelessly, though, it has its risks for long-term preservation. A PDF file isn’t necessarily self-contained. It might depend on external fonts or even require whole files of content. To avoid the risk of dependencies that might cause future problems, you can use PDF/A, a restricted subset of PDF. Good software for generating PDF usually includes a PDF/A option.

On Macintosh OS/X doing this is a little weird. You start by selecting “Print” from your application’s File menu. There will be a “PDF” button in the Print dialog. Clicking on it brings up a menu. Choose not “Save as PDF…” but “Save as Adobe PDF.” This launches an application to save the PDF. You’ll see a dialog like the following:

Macintosh PDF save dialog

In the “Adobe PDF settings” select one of the PDF/A options.

If you have Acrobat, conversion to PDF/A is much simpler.

Correction (29-Jan-2012): If you don’t have Acrobat installed on your computer, you don’t get the “Save as Adobe PDF” option. Sorry about that.

PDF/A documents are even harder to edit than regular PDF’s, so you should defer converting a document to PDF/A if you’re still tweaking it.

There are two versions of PDF/A, called PDF/A-1 and PDF/A-2. A-1 is based on PDF 1.4, and A-2 is based on 1.7. PDF 1.7 is fully backward compatible with 1.4, so you’re safe using either. There are two compliance levels, a and b, with a being the stricter. From a preservation standpoint, there isn’t much difference. Converting to Level b may be easier, since Level a requires you, or the generating software, to produce a tagged structure. The structure helps to make the document self-explanatory but may not be necessary.

There are quite a few restrictions which a file must satisfy to conform to PDF/A, including these:

  • All fonts must be embedded in the document, and there are some restrictions on font implementation. Conversion software will generally refuse to put fonts flagged as non-embeddable into PDF/A.
  • Color profiles must be used to guarantee device-independent color.
  • The document may not be encrypted.
  • There may not be audio, video, or JavaScript.
  • XMP metadata must be included.
  • Forms are allowed with restrictions.

Clearly there are legitimate things to do in PDF that can’t be done in PDF/A. Audio and video can’t be included. The requirements for fonts and profiles can make PDF/A files bigger than they’d otherwise be. But if it’s important to give files the best chance for long-term readability and the restrictions don’t impact necessary content, PDF/A is a good choice.

Suggestions: Use PDF/A for documents that you expect to be needed for a long time. If you’re using PDF as a general-purpose container, though, it may not be a good choice.

Useful links: