It is widely recognised that natural disasters emerge in the interplay between extreme geophysical events and the human communities affected by them. Whilst detailed natural scientific knowledge of a given event is critical in understanding its impacts, an equally thorough understanding of the affected communities, their economies, ecologies, religious structures, and how all of these have developed over time is arguably as important. Many extreme events leave methodologically convenient traces in the geological and archaeological records in the form of discrete stratigraphic layers often associated with both accurate and precise dates. This paper focuses on volcanic eruptions and draws on matched case studies to illustrate the usefulness of a two-step, quasi case–control comparative method for examining vulnerability and impacts in the near- and far-fields of these eruptions. Although issues of data resolution often plague the study of past disasters, these limitations are counterbalanced by the access to unique long-term information on societies and their material expressions of livelihood, as well as a similarly long-term perspective on the critical magnitude/frequency relationship of the geophysical trigger(s) in question. By drawing together aspects of contemporary Disaster Risk Reduction research, archaeology, and volcanology, this paper sketches out a methodological roadmap for a science of past disasters that aims to be relevant for not only understanding vulnerabilities and impacts in the deep past, but for also better understanding vulnerability in the present. Copyright Springer Science+Business Media Dordrecht 2014
