Blind reproducibility audit of 50 passive daytime radiative cooling (PDRC) claims — frozen benchmark (M25–M31)

Complete frozen evidence base for the accompanying manuscript. Two auditors independently applied a frozen reproducibility protocol (RR2) to the same 50 PDRC claims. This dataset contains all five analysis layers (both auditors' ledgers, the inter-rater reconciliation, the evidence-accessibility challenge, and the bibliographic QC), the final frozen register, and a SHA-256 provenance manifest. No auditor decision was modified; bibliographic repairs are confined to the record layer. Cohen's κ is descriptive only; the pooled reproducibility fraction (0.28) is descriptive of this stratified sample and is not an estimate of field prevalence.