Parallel-Chain Provenance with Structural Merge-Absence: A Cross-Domain Substrate for Multi-Authority Audit Trails

Version 1. 5 (Rev 1. 5) — Changes vs. v1. 4 (Rev 1. 4): §2. 8 Para 1 expanded with Vos et al. (2024) 36 as named exemplar for the §5 primitive's containment-bit component, with party-of-origin suppression per Vos §4. 3. Bibliography entry 36 appended after 35. No other §2. 8 changes — Paras 2–7 (32–35 cross-chain contrast cluster, ratified in Rev 1. 4 / Stream H disposition) remain unchanged. Cluster scope is effectively (a) Vos-only by structural-territory fit: Vos sits in PSI/PPRL-family territory (25 cluster), structurally distinct from the cross-chain confidentiality territory of 32–35. Wright 2025 (arXiv: 2506. 13246) was reviewed during the same Rev 1. 5 prior-art audit and excluded from §2. 8 on combined cross-family Q4=NO + Craig Wright UK High Court 2024 credibility grounds (workshop audit-trail entry). Cross-family verification: Vos 2024 returned Q2 = NOT COVERED and Q4 = NO unanimous 4/4 with full paper access (SciSpace + Elicit + Grok ×2 runs) ; Wright 2025 returned Q4 = NO 3/3 with full paper access. No substrate code or test count changes. --- Abstract: We present a data-structural substrate for tracking parallel independent assessments of the same external object such that (i) each authority's chain remains structurally non-fusible, (ii) cross-chain references can be expressed without content disclosure across authorities, and (iii) each chain's history is cryptographically verifiable. The substrate (named provenanceₑngineᵥ2) provides identity-by-history at the primitive level: two records with identical fold-sequences are substrate-identical, while two records with diverging histories become substrate-distinct. We demonstrate applicability across eight independent domains — patent prosecution, verifiable credentials, software bill-of-materials, AI model lineage, NGS variant interpretation, psychiatric assessment, credit rating actions, and bank credit intelligence — yielding 1322 passing test cases (1322/1322) across the eight wrapper implementations. Seven of the eight domains exhibit merge-absence (a structural property formalized as §15. 3): two authorities assessing the same object produce parallel records that cannot be fused at the substrate level; only the verifiable credentials domain exercises operations-with-merge, demonstrating the substrate's ability to surface scope-limited operations explicitly. The eighth domain (bank credit intelligence) introduces a novel confidentiality-preserving cross-reference primitive that supports existence queries across parallel chains without content disclosure. We position this work as defensive prior art for the broader pattern, established by Bitcoin OpenTimestamps anchors and a Zenodo DOI on 2026-05-11. The substrate itself remains a structural-mathematical contribution; we make no clinical, financial-advisory, or medical-device claim, and the domain-specific implementations are synthetic-fixture demonstrations unless otherwise noted. Scope of the novelty claim. Provenance representation 10, 23, secure audit logging 20, 21, transparency logs 22, distributed-data convergence 19, software supply-chain attestation 11–13, and privacy-preserving record linkage / private set intersection 25 are all mature literatures. We do not claim to introduce any of them. The narrower contribution defended here is the combination of (i) substrate-level merge-absence as a primitive API property, (ii) identity-by-history at the substrate level, and (iii) regulator-scoped, content-excluding cross-reference queries embedded in non-fusible multi-authority chains — instantiated and unit-tested across eight independent domains. We further note that the phrase "parallel provenance" has prior use in archival description 26, where it denotes plural contextual narratives over records; the present paper instead studies parallel-chain provenance as an operational data-structural property. Keywords: provenance tracking, parallel audit trails, merge-absence, multi-authority assessment, confidentiality-preserving cross-reference, data-structure pattern, cross-domain methodology