Recovery Record of the Harvard Khipu Database (KDB) Legacy Archive: Manifest, Tooling, and Validation

The original Harvard Khipu Database website (khipukamayuq.fas.harvard.edu) isoffline; its holdings (hand-curated per-khipu recording sheets, catalogs, and a2012 SQL dump) survive only as Internet Archive Wayback Machine captures. In July 2026 we enumerated and recovered all 345 uniquely archived files of its DataTablesdirectory. This deposit is the recovery record: a per-file manifest (original URL,snapshot timestamp, size, SHA-256), the idempotent recovery tool, and two validationdocuments showing the recovered sheets parse cleanly (26,108 + 6,686 cord rows, zeroerrors) and agree with the modern Open Khipu Repository where they overlap (274/274and 66/66 khipus matched via alias-aware resolution; 99.99% per-cord value agreementfor the later-vintage sheets, 6,685 linked cords). The recovered archive preserves material the modern OKR does not hold: 14 dropped khipu assessments(2,323 cords; independent recordings of khipus the OKR merged under compounddesignations, where the OKR kept the other recording), one khipu for which the OKRholds no data (AB001, 156 cords; catalogued by the OKR as KH0701 since December2025 but without cord records), and Gary Urton's per-khipu classification table. The recovered files themselves are not republished here; they are theintellectual work of the Harvard Khipu Database Project (Gary Urton, PI; CarrieBrezine, database manager) and are being offered to the Open Khipu Repositorymaintainers. This record allows any future holder to verify content identity bychecksum.