Have designed and implemented a new database system at the University of Alaska Museum, I have some thoughts on the process: First of all, forget about scanning existing documents as a data entry tool. Scanning takes as long (or longer) than physical data entry and introduces multiple sources of error. Even the best text recognition software is not 100% accurate and introduces random errors that are difficult to locate and correct. As to abbreviations, codes and other forms of cryptic human activities. With todays database designs, it is no longer necessary to use esoteric codes for data entry. Use the enter word, phrase or data group in the entry. With proper dbs design, full entries are as easy to handle as codes and eliminate the necessity of translation by experienced user as well as the general; public. Codes are a plague that must be eliminated from dsbs's everywhere! I used 4th Dimension in a Macintosh system, with a server and four data entry clients. It has worked phenominally well for our 1.5 million piece collections, and has fully adapted to 70 years of intermiottent curation and data management (or lack thereof). Up-front dbs design is the key to success in terms of ease and accuracy of data entry and ease of use at the general user end. Mike Lewis Research Assistant University of Alaska Museum