The authors have previously developed a model of hypertext, called Trellis, based on the formalism of Petri nets. As in other hypertext models, Trellis represents information fragments and the relationships that tie the fragments together. An important premise underlying Trellis is the claim that the actions of browsing a hypertext should be formally represented. A formal description of browsing not only permits analysis and verification of the dynamic properties of a document, but also standardizes the semantics of browsing by providing a succinct, unambiguous, and implementation-independent definition of behavior. Within its notational framework, Trellis provides the ability to synchronize simultaneous traversals of separate paths through the hypertext and the ability to incorporate access considerations into the hypertext, specifyingnodes that can be proven to be visible only to certain classes of browsers. Trellis is a timed model, permitting specification of event delays and durations as well as system-generated invocations. The investigators conduct two main projects under this research. Both are based on the Trellis hypertext work already accomplished. The first is the development of abstract representation methods that permit the mapping of a single unified hypertext specification into documents suitable for each of many different hypertext systems. This, in turn, can reesult in interchange methods for different hypertext systems based on translation into and out of the unified model. The second is an authoring language "laboratory" to allow the rapid specification and evaluation of hypertext authoring languages. The projects will produce results of both theoretical and practical interest.