Record sample lineage without inventing relationships

A similar identifier is not a lineage record

Preserve the reported links and unresolved gaps so the register does not create a chain of custody through naming assumptions.

Our public material-register example shows a synthetic report beside downloadable output. Use it to inspect how values and source references travel together. The self-service ParseMill trial is for invoice PDFs; supplier test reports and engineering review need a separate fit discussion.

Lineage describes how samples relate; it does not make their results interchangeable. A prepared or composite sample can have its own methods, dates and reported properties. Preserve those distinctions even when the relationship to a parent is well documented.

Create an evidence-backed relationship table

Record parent identifier, child identifier, reported relationship, source reference, source wording, effective event and review state. W3C: PROV Data Model describes derivation relationships between entities and the activities involved. This is a useful model for documenting reported preparation steps, while leaving unsupported relationships absent.

Use one row per claimed relationship. A composite may have several reported contributors, so do not force it into a single-parent column. Record quantities or proportions only when supplied and relevant; unknown proportions must remain unknown rather than being divided equally for convenience.

Work through a small lineage map

Synthetic example: an invented preparation note P6 identifies two contributors to C1 and one prepared portion from S2. It says nothing about S2-B.

C1 has two recorded contributor relationships, not two independent copies of the same result. S2-B remains unconnected pending evidence. The map should not imply that any measurement on C1 describes each contributor separately.

Check identifiers before joining records

Microsoft: Merge queries overview explains merges based on selected matching columns. Before using such a merge, verify that identifiers are unique within the relevant source or project and that the selected key includes the necessary context. Sample 1 can occur in several unrelated reports.

Keep relationship evidence separate from a software match. A successful join can attach a note to a row, but it cannot prove the note’s scientific interpretation. Check for reversed parent-child direction, duplicated edges and relationships copied from an obsolete preparation revision. Escalate conflicting records rather than selecting whichever produces a simpler tree.

Verify the map can be reconstructed

Ask another reviewer to trace C1 to both contributors and locate the evidence for each edge. They should also identify why S2-B has no accepted parent. Reconcile the relationship table to the source notes, and list any unrepresented or ambiguous references for clarification.

In a supplier-report scope discussion, define which identifiers and preparation notes the register must preserve. Build and review lineage as a separate step, with source support for every relationship. The invoice workspace does not infer sample lineage or propagate technical conclusions.