Built on the hypothesis that some epistemic beliefs can act as bottlenecks to impede conceptual understanding in history, I aimed to find ways in which these could be identified, interrogated and challenged. Over three cycles of a teaching experiment, which is a specific form of design-based research, a sequence of hypothetical learning trajectories (HLTs) was implemented in three Irish primary classrooms. The learning trajectories designed for this study were based on four concepts. These were: multiple perspectives, historical significance, using evidence and historical argumentation. Retrospective analysis of the first cycle led to the formulation of a new conceptual change model: the Analogical Conceptual Change Model. This was used to underpin the design of the HLTs used in Cycle 2 and 3 of the study. Thematic analysis of each cycle led to the development of a local instruction theory for using historical evidence in the primary classroom. Local instruction theories can be considered as learning routes relating to exemplary instructional activities that can be used to teach for conceptual understanding. They are developed from the design and testing of HLTs. Findings showed that the ideas students hold about the nature of history are complex, rooted in everyday epistemologies and shift back and forth depending on topic and context. Findings also showed that the efforts of the teaching experiment to integrate current research into a HLT that interrogates children's initial understandings assisted them in moving towards more sophisticated forms of reasoning about historical evidence and the nature of history.