A certificate and a laboratory result answer bounded questions

7.11 An assay tests a sample against references, not an abstract map

Stable isotope analysis offers a possible corroboration route for geographic provenance. The opened review abstract describes region-specific isotopic signatures in natural fibres and examines fractionation alongside environmental and processing effects.[S33] That scope supports a qualified hypothesis: measurements on a sample may be compared with suitable reference material to assess whether the result is consistent with a proposed origin. It does not support a universal country detector, a claimed accuracy for Cameroon, or exclusion of another country without a relevant validation study.

The reference library is therefore part of the method, not an administrative accessory. GFT's proposed qualification study would assemble known-origin cotton references that represent the intended growing regions, seasons, production conditions and relevant processing stages. Each reference needs its own provenance evidence and sample custody. A model trained or assessed on raw lint cannot simply be assumed to perform identically on yarn, dyed fabric or a finished garment. The qualification boundary must state which material forms and processes were actually represented.[S33]

Sampling must also represent the decision. A convenient sample taken from a bale selected by the supplier may test that bale, but it does not establish the composition of an entire delivery or the origin of fibres in every downstream garment. The proposed study would separate method validation from routine lot assurance. Method validation asks whether the assay and classification procedure can discriminate the defined origin classes under controlled conditions. Routine assurance asks whether a new, independently linked sample falls within the qualified decision boundary. Both questions can return inconclusive results.[S29][S33]

7.12 Laboratory competence and claim validity require separate checks

ISO describes ISO/IEC 17025 as setting requirements for laboratory competence, impartiality and consistent operation, and notes that accreditation bodies use it as criteria for assessing and accrediting laboratories.[S0701] GFT would check the laboratory, method and relevant scope rather than treating any accreditation mark as universal approval for origin classification. Competence for one test does not establish competence for every analyte, material or interpretation. The report still needs to identify the tested item, method, result and applicable limitations.[S0701][S29]

The sample-to-report chain remains equally important. The review should identify who selected and sealed the sample, which lot it represents, how subsamples were formed, what happened to retain material and how the laboratory received it. A technically sound measurement on an unlabelled bag cannot repair the missing connection to a shipment. Evidence-selection control and assessor independence must again be read separately: an independent laboratory can test a supplier-selected sample, and a buyer-selected sample can be sent to a laboratory whose method is outside the required scope.[S29][S0701]

Blind hold-out testing is a distinct check on the classification procedure. The party interpreting the result should not receive the hold-out origins before issuing its determinations. The qualification owner then compares those determinations with the documented references and reports errors by origin class and material form. A small or unrepresentative reference set cannot be repaired by an apparently precise percentage. GFT would retain confidence intervals and sample counts where the eventual statistical method uses them, with the unit of independence made explicit. At this chapter's research boundary, no such Cameroon performance results are available.[S33]

Exhibit 30 makes an inconclusive outcome a normal terminal state rather than a failed software branch. It also prevents a supportive assay from bypassing documentary custody. The flow is a proposed qualification design based on the isotope review's abstract and laboratory-competence overview, not a description of an assay already performed.[S33][S0701]

A supportive result does not upgrade the claim beyond the reference library or sample frame. If the qualified hypothesis is consistency with a bounded Cameroon reference group, the result cannot establish a particular farm or exclude every alternative origin. An inconclusive result leaves the documentary assessment unchanged and may justify more sampling or a narrower statement. A contradictory result triggers investigation of the sample link, method suitability and documentary claim. It is not, on its own, proof of fraud.[S29][S33]

7.17 Current status and event-time evidence answer different questions

A reviewer needs to know both what was accepted when a transaction occurred and what may be relied upon now. W3C's status-list specification provides a mechanism for publishing suspension or revocation information about verifiable credentials. It also cautions that credential status concerns the credential itself and may not determine the status of an underlying or backing object.[S24] GFT would therefore avoid a single rule that declares every historical physical observation false when a credential is withdrawn, or every current claim safe because an old certificate was once valid.

Suspension, revocation, expiry and unavailable status need distinct reason codes. Suspension may be reversible under the governing policy, while revocation cancels credential validity under that profile. Expiry marks the end of a stated validity interval. An unavailable status service establishes none of those outcomes.[S24] For final claim release, GFT's proposed default is to hold the dependent assertion until fresh evidence resolves the required check. For historical reconstruction, the reviewer retains the earlier acceptance decision, its status observation and the policy version used at that time.

Scheme-specific integrity procedures can add another evidence route. Global Standards states that its Quality Assurance Unit performs random checks of Scope Certificates, Transaction Certificates and Letters of Approval, and that Transaction Certificates linked to fraud will be revoked by the relevant certification bodies under its procedure.[S0702] This example shows why issuance is not the end of assurance. It does not give GFT authority to revoke a document, and it does not make every later discrepancy proof of fraud. The applicable scheme owner and certification body must resolve their own processes.

7.18 Contradiction handling must contain the claim without destroying the record

A contradiction can arise when a transaction certificate names a different seller, a registry shows a suspended scope, a laboratory result conflicts with a qualified origin expectation, or a physical lot does not match the reviewed document. GFT's first response is to identify the exact proposition and affected material set. The claim owner places only the dependent claims on hold, while operations decides separately whether material must be physically quarantined. This avoids both unsafe release and unnecessary implication that every unrelated lot or attribute is defective.[S03][S29][S33]

The investigation then preserves the original evidence, retrieval times, sample custody and release decision. The reviewer requests an explanation through an authorised counterparty channel and checks the issuer's current scope and status. New documents are added with their own versions; they do not overwrite the files that created the original concern. If the contradiction is resolved, the release owner records the evidence that justifies reopening. If it remains unresolved, the claim is narrowed or withdrawn and the affected disclosure is corrected.[S24][S29]

GFT's proposed contradiction playbook assigns four roles. The evidence owner preserves the dossier and exception chronology. The issuer-contact owner seeks clarification without altering the original record. The claim owner decides hold, reopening, narrowing or withdrawal under the applicable policy. The disclosure owner identifies every passport, buyer pack or other statement that used the affected claim and records the correction. Where staffing cannot separate these roles, an independent reviewer should examine the decision before the claim resumes.[S03][S29]

An illustrative case shows the boundary. A transaction document remains authentic, but its current status is suspended after the related goods have entered production. The suspension does not rewrite the earlier receiving observation, and it does not prove the cotton origin was false. It does block new reliance on that credential where the release policy requires current acceptable status. The reviewer maps affected inputs to outputs, examines the reason and effective interval, and decides which downstream claims must pause. Physical genealogy and quantity reconciliation continue as separate evidence, not as automatic substitutes for the suspended certification.[S03][S24]

Closure requires more than a replacement file. The reviewer must show that the new evidence addresses the original contradiction, covers the same claim boundary and has its own acceptable issuer and status. A renewed certificate for a different site, period or material does not close the exception. If the underlying physical evidence remains insufficient, a newly signed assertion cannot repair it. The final record therefore includes the decision rationale and any residual uncertainty, ready for independent reconstruction in the pilot.[S24][S29][S0702]

Across all six sections, the release principle is conservative but workable. Accept evidence for the bounded question it can answer, combine it with the other required evidence, and stop when a necessary proposition remains unsupported. Assurance becomes more useful, not less, when its limits are explicit: a buyer can see why the claim was accepted, what was sampled, which evidence was current and what remains unknown.