This paper seeks to advance digital preservation theory and practice by presenting an evidence-based model for identifying barriers to digital content rendering within a bit-level preservation repository. It details the results of an experiment at the University of Illinois at Urbana-Champaign library, where the authors procured a random sample of files from their institution’s digital preservation repository and tested their ability to open said files using software specified in local policies. This sampling regime furnished a preliminary portrait of local file rendering challenges, and thus preservation risk, grounded not in nominal preferences for one format’s characteristics over another, but in empirical evidence of what types of files present genuine barriers to staff and patron access. This research produced meaningful diagnostic data to inform file format policymaking for the repository.

This paper presents results of a practical analysis into the effects of three main lossless TIFF compression algorithms – LZW, ZIP and Group 4 – on the storage requirements for a small set of digitized materials. In particular we are interested in understanding which algorithm achieves a greater reduction in overall storage, and whether there is any variation based on the type of file (e.g. colour depth). We compress 503 files with two software utilities – ImageMagick and LibTiff – and record the resulting file size for comparison against the original uncompressed version. Overall we find that in order to effectively (although not necessarily optimally) reduce total storage, Group 4 compression is most appropriate for 1-bit/pixel images, and ZIP compression is suited to all others. We also find that ImageMagick – which uses the LibTiff library – typically out-performs LibTiff with respect to compressed file sizes, noting that this appears to be the result of setting the “Predictor” tag.