The design of an English-like query language and an interactive computer environment for searching the full text of the U.S. patent collection are discussed. Special attention is paid to achieving a transparent user interface, to providing extremely broad search capabilities (including nested substitution classes, Kleene star events, and domain restrictions) and to specifying the query language by means of a formal grammar and an associated parsing mechanism which particularly facilitates error detection and reporting. A description of an initial working prototype system on the IBM 360/370 is also included. (Author)