A wealth of information concerning the artefact assemblages of Late Antiquity is available to us in the forms of excavated material and contemporary texts and images. Research comprising part of the Visualisation of the Late Antique City project at the University of Kent seeks to identify type assemblages associated with specific activities and types of space within the city. In order to do this it is necessary to apply a range of analytical techniques, some of them familiar statistical approaches and others more specialised, to the available evidence. This paper examines the potential for applying correspondence analysis and network analysis to large datasets comprised of evidence from multiple source types, as well as the obstacles to such application. This will allow us to make reasoned suggestions about the groups of objects likely to appear in settings selected for a visualisation scene. The paper also considers how study of the small-scale spatial distribution of objects can complement research, in the rare cases where exceptional site formation processes preserve assemblages in their primary ancient context.