Iterative reconstructions involve several types of data. We want files containing these data to be transportable across machine architectures, and to be easily readable by MPI-enabled parallel computers. For these and other reasons, we have chosen to use the Heirarchichal Data Format version 5 (HDF 52.2) as our basic file format. Routines to work with these files are provided in the C subroutine library dm_fileio.c in CVS diffmic/c/util as described in Sec. 2.4, and the IDL routines described in Sec. 2.5. Examples of using these routines are shown in the files dm_test_fileio.c and dm_test_fileio.pro in the respective test directories. You can also view the contents of an HDF 5 file using a variety of other programs, including h5dump (a utility program provided as part of the HDF 5 distribution) and a Java viewer (hdfview, available at http://hdf.ncsa.uiuc.edu/hdf-java-html/hdfview/).