The study described in this report investigated the effects of two context variables on small-group learning--namely, group size and group gender composition--within an adult learning framework. In the study, the "revolving circle" method was innovated. In this design, one group member is the center of the cooperative circle and all discussions are around his/her task completion. The method was implemented for a period of 2 months, 4 hours per week in seven different classrooms in Haifa, Israel, taught by four teachers in literature and language arts. Students in each class were randomly assigned to groups of different size (three, four, or five group members) and different gender composition (majority male or female). Two trained female observers watched each classroom for six full periods of 90 minutes each, and coded behavior for 5 minutes using a checklist. Observed behaviors were grouped in six categories: listening and social interaction, group maintenance, interactive summary, giving and requesting information, cooperative learning behaviors, and general learning behaviors. The study found that most of the significant differences occurred in odd-number groups. Groups of three members elicited more integrative summary and general learning behaviors, while groups of five members elicited more cooperative learning behaviors, listening, and social interaction. The study also found that cooperative learning behavior was significantly higher in groups with either gender majority, while giving information was highest in equal gender composition groups. The results of the study can be used in further research on group learning behavior. (KC)