proteindiffraction.org contains archives where the top-level downloadable file is a tar file, and the individual files are compressed using gzip or bzip2. This way of storing files is not anticipated by the external data pointer definitions, which assume that compression is performed on the overall archive only (e.g. tar.gz or tar.bz2). I propose adding a new column to array_data, _array_data.external_file_compression with enumerated values corresponding to common compression formats. Where the value for this attribute is not 'None', after a file has been downloaded and extracted it will be decompressed according to the specified format.