Source-coded specimen-level dataset and source-audit records for rotational stiffness degradation in traditional timber mortise-and-tenon joints

Source-coded specimen-level dataset and source-audit records for rotational stiffness degradation in traditional timber mortise–tenon joints This record archives the primary-cohort dataset and the associated source-audit records used in a study of rotational stiffness degradation in traditional timber mortise–tenon joints (manuscript submitted for publication). All values were extracted from published quasi-static cyclic tests; no new experiments are reported. Content. The database holds 76 unique source-coded records from 12 published studies. The analysed primary cohort comprises 33 records from 7 studies (18 records with all four target rotations form the complete-case subset). For each record the files give the source location, the extraction status (reported, digitised, calculated or unreported), the combined first-cycle secant stiffness K at 0.020, 0.030, 0.050, 0.075 and 0.100 rad, the normalised stiffness K_norm = K(θ)/K(0.020 rad), the stiffness-loss index D_θ = 1 − K_norm, joint type, reinforcement and initial-condition coding, timber species and density, and the tenon geometry ratios R_L, R_W and R_H. Values that are not available were left empty and were not imputed, interpolated beyond the plotted curves or extrapolated. Files. 1. PPCE_V23_FinalAudit_20261012_corrected.xlsx: final 33-record analysis dataset (Sheet1) and geometry revision log. 2. PPCE_V22_Huan_Y1Y4_FinalAudit_corrected.xlsx: full 76-record specimen and analysis databases, rotation and condition audits, and source QC. 3. PPCE_V15_FinalAudit_Reconciled_20261012_corrected.xlsx: version-by-version audit workbook with the source-level audits and the final reconciliation. Historical tabs are kept unchanged for traceability; the reconciliation tab states which earlier summary values are superseded. The 14 additional records obtained from the database search of 18 October 2026, the search strings and the selection rules are not in this record; they are given as Supplementary data to the manuscript. Limitations. Counts are unique source-coded records, not a verified census of independent physical test units. In particular, Huan et al. (2019) reports 12 model units but lists 15 labels, and whether the three baseline labels (B0, T0, Y0) are distinct physical units is unresolved. Digitised values are curve-read estimates, not transcriptions of tabulated values. Geometry ratios not tabulated by a source were read from its figures and are operational definitions. Version 2 (31 October 2026): correction. For Huan et al. (2019) labels B0, T0 and Y0, the values at 0.075 rad (K = 5.18, 17.48 and 10.88 kN m/rad) and the K_norm and D_θ values derived from them had been extrapolated beyond the end of the source curves (about 0.0714 rad). They have been removed and the cells left empty; no value was replaced. At 0.075 rad the primary-cohort pool therefore has n = 19 (previously 22), with median D_θ = 37.4%. All other values are unchanged. Each file contains a Correction_20261031 sheet listing the cleared cells and their former values. Summary tabs from earlier versions that quote n = 22 at 0.075 rad are superseded.