This paper considers a variety of methodological questions arising from the study of readers' annotations in books. It is particularly concerned with issues concerning the categorisation and systematic study of annotations, and the importance of recognising the overlap between the collection and the interpretation of primary data/evidence. It concludes by suggestioning that an inevitable compromise must be made if the study of readers' annotations is to develop beyond localised surveys and case-studies, and that the recognition of the philossophical challenges which are inherent in such an enterprise may be more valuable than attempts at comprehensively resolving them.