Context: The history of European forest dynamic remains fragmental. In France, the Daubrée statistics (1908) and agricultural statistics (1892, 1929) formed fundamental material to fill this gap. Aims: Release, test, and summarize the digitalized dataset. Analyze long-term forest changes in forest area, composition, and structure. Methods: Primary data on forest area across NUTS-3 geographic units, split by forest management and ownership categories and dominating tree species (Daubrée), were digitized and cross-compared. Centennial changes in forest attributes were assessed from modern forest inventory data. Results: Cross-comparison revealed: (1) strong temporal consistency in forest changes over time, (2) systematic and interpretable biases in ownership/management categories between Daubrée and agricultural statistics. Strong shift from coppices to high forests, increased prevalence of private ownership, and constant proportion of broadleaf- and conifer-dominated forests were highlighted, with increased tree species diversity at country scale. Conclusion: Ancient statistics are shown to play a major role in retrospective land-use and forest policy analysis.