A study investigated communication patterns among undergraduate student groups using email for instructional discussion at the University of Minnesota. A sample of three 6-person groups was studied over a 7-week period, during which the groups were given three topics to discuss sequentially for instructional purposes. Students used email outside of class time and were required to contribute at least two posts (or entries) for each discussion topic. The task assigned to the students was to: (1) generate ideas and arguments on a given topic; and (2) learn how to improve their on-line conversation skills. There was no tangible goal for the group to produce; instead, groups were asked to focus on the process of discussing. Transcripts of the discussions were analyzed inductively using a process of repeated summarizing to reduce the data and look at patterns. Each discussion was treated as a separate case and dynamics were noted at the micro-level for all cases. Then cross-group dynamics and cross-discussion dynamics were analyzed. Finally, categories were generated for further research: (1) discussion momentum (the use of an "anchor" post as a touchstone); (2) question-asking strategies; (3) use of humor or sarcasm; (4) norms for naming participants; (5) style of writing and competence; (6) beginnings and endings; (7) disruptive events; (8) software differences; (9) role functions (for instance, the emergence of a leader); (10) agreement and conflict; (11) nonparticipants; and (12) the relationship between computer-mediated and face-to-face interaction. (Contains instructions for email discussion used in the course.) (TB)