Ab~ractAlthough peer-based work is encouraged by theories in developmental psychology and although classroom interventions suggest it is effective, there are grounds for recognising that young pupils find collaborative learning hard to sustain. Discontinuities in collaborative skill during development have been suggested as one interpretation. Theory and research have neglected situational continuities that the teacher may provide in management of formal and informal collaborations. This experimental study, with the collaboration of the science faculty in one urban secondary college, investigated the effect of two role attribution strategies on communication in peer groups of different gender composition in three parallel Year 8 science classes. The groups were set a problem that required them to design an experiment to compare the thermal insulating properties of two different materials. This presents the data collected and key findings, and reviews the findings from previous parallel studies that have employed the same research design in different school settings. The results confirm the effectiveness of social role attribution strategies in teacher management of communication in peer-based work.In science education, a decade of clinical interview based research pioneered by Osborne and Freyberg (1985) into "children's science," has led to theoretical advances in understanding the nature of students' constructed conceptual understanding. The broad tenets of constructivism as a model have attracted considerable support amongst advocates of progressive classroom reform (Ritchie, 1998). However, the process of coming to know in the social setting of the classroom has not been so well researched. If the constructivist model of learning is to be elaborated and usefully modelled for classroom teachers then further empirical data is required. Learning theories that fall within the constructivist school require methodologically that teachers provide for the negotiation of social context and academic context to enable meaning making. From the late sixties science curriculum materials such as the Australian Science Education Project (1974), influenced by Piagetian theory, advocated peer-group settings for science learning. Piaget (1928, p. 204) argued that through "the shock of their thought coming into contact with that of others" children greatly benefit from having to coordinate their thinking with that of their peers. Many sociocultural theorists after Vygotsky (1978) view learning as inherently social because it involves appropriating modes of social discourse; that is gaining access to representational systems that permit distinctive ways of organising and communicating human experience. So educational practice, it is argued, must offer opportunities to explore such discursive systems through interacting with similarly motivated peers: in fact to collaborate. The perception that active individual learning in science classes occurs in a social and cultural context underpins the situative perspective that takes the ...