This datasets was collected for a case study that sought to reconstruct the spatio-temporal spread of the American chestnut blight in Tennessee by collecting, reconciling, and analyzing heterogeneous archival and special collections materials from 1904 to 1950, the major period of infection and tree loss. Using the spread of the American chestnut blight in Tennessee as a case study, the project explored the potential for archival resources to serve as proxy data sources for the compilation of historical climate datasets.

The spread of the American chestnut blight in Tennessee was recreated using data obtained from the archival record. Digitized primary source materials were gathered through keyword searches in Chronicling America, Google Books, the National Agricultural Library Digital Collections, Newspapers.com Southeast Edition, Open Parks Network, ProQuest Historical Newspapers, the Tennessean Archives, Tennessee Virtual Archive, University of Tennessee Digital Collections, University of Memphis Digital Commons, Archives of Appalachia, and the digital collections listed on the Tennessee Archives Directory webpage. The initial keywords used to locate relevant resources were selected according to the Library of Congress authorities file for the American chestnut tree. These topical terms include the subject heading American chestnut and the tracing and reference terms American chestnut tree, Castanea dentata, and Castanea. Other keywords were identified from the metadata and text of relevant primary sources. These include chestnut, chestnut blight, American chestnut blight, and Chestnut bark disease. When necessary, search parameters limited results to only artifacts published in Tennessee between 1904 and 1950.

Once relevant sources were compiled through keyword searches, the resulting set was further filtered according to several criteria: a source had to mention a tree or stand at a specific date, location, and condition (alive, blighted, or dead). Additionally, a source had to indicate that the tree it mentioned was an American chestnut and not an Asian chestnut, European chestnut, horse chestnut, chestnut oak, sweet chestnut, or water chestnut. The identified trees and stands were then geocoded with Excel’s MapCite add-in, and the documents were classified in a spreadsheet according to the following variables: date, location, latitude, longitude, chestnut status, document type, document title, search terms, source, website, and URL. Evidence of pre- and post-blight tree stands found in the archival record thus informs a spatial-temporal recreation of the blight at a local scale.