Aim – Paper-based publications were the main repository for phytogeographical information until the end of the 20 ^th century. These texts are still an important reference source for phytogeography and potentially a valuable source of data for research on environmental change. The recent digitization of biodiversity publications, text-mining and mark-up protocols means that these data are now more accessible than ever before. Here I examine the value of legacy literature specifically for studies on phytogeography. Methods – Three contrasting data mobilisation projects are used as case studies for the extraction of phytogeographic data. Two were digitisations and XML mark-up of floras, the Flore d'Afrique Centrale from the 20 ^th century and the Flora of Northumberland and Durham from the 19 ^th century. A third case study used Chenopodium vulvaria L. as a test case, where I attempted to recover as much phytogeographic data as possible for one species, both from literature and from herbarium specimens. Results – A large amount of useful information was extractable from legacy literature. The main limitations are that most localities need georeferencing and that observations are only rarely associated with a precise date. In the case of C. vulvaria literature contributed about 20% of all available observations of the species. Literature becomes a progressively more important source of data the further back in time one looks. However, useful observations become much rarer earlier than about 1850. Main conclusions – Sourcing phytogeographic data from legacy literature is valuable. It contains observations and links to other data that are unavailable from any other source. Nevertheless, its extraction takes a substantial investment in time. Before commencing on such a project it is important to prioritise work and understand the limitations of such data, particularly with regard to georeferencing.

BOTANICAL LITERATURE; GEOREFERENCING; OBSERVATIONS; PHYTOGEOGRAPHY; XML MARK-UP