This is the second in a series of articles for Teaching History in which Peter Lee and Denis Shemilt share the findings of Project Chata (Concepts of History and Teaching Approaches). In their first article (see Edition 113), they questioned the wisdom of using the National Curriculum attainment target as a model of progression and argued the case for more nuanced, complex models based on empirical research. Such models, they argued, can never provide a blueprint for what progression in history looks like, but they can be genuinely useful in revealing the kinds of preconceptions pupils at different ages might have and the ways that these can be successfully modified over time. In this article, they explore one aspect of their research, pupils’ understanding of historical accounts, focusing in particular on pupil preconceptions and progression. Drawing on a wealth of empirical data, gathered over several years in many classrooms, the authors are able to propose a way of characterising what progression in pupils’ understanding about historical accounts might look like. They are not advocating rigid, all-embracing models of progression; they are providing us with a much more finely tuned and sophisticated way of talking about and analysing pupil progression. Above all, Lee and Shemilt urge us to attend to the preconceptions that pupils bring into the classroom. Without this, they argue, teaching becomes little more than ‘firing blindly into the dark: we may get lucky and hit one of our targets, but we are much more likely to damage our own side.’