This is a progress report on a project attempting to design a system of compacting text for storage appropriate to disc oriented demand searching. After noting a number of previously designed methods of compression, it offers a tentative solution which couples a dictionary of most frequent character-strings with a set of variable-length codes. The dictionary is not restricted to natural language words or to content-bearing strings. A suite of IBM 360 assembly language programs has been constructed to count frequencies and to compute a set of parameters which will mitigate constructing a dictionary. A program to sort and print the file resulting from the other programs has also been developed. (WBC)