The second stage of the migration prepares the raw data for output in the TCFS format. This step includes interpreting the directory listings and decoding the user ID numbers to user name mappings. More importantly, we had to find some way to take the 36-bit words from the ITS file system and translate them into something coherent that can be stored in 8-bit bytes on the systems we are currently using for data storage. ITS predates the presently ubiquitous 8-bit byte, so we faced some unusual challenges maintaining the semantics of the original data.