Separate sample collection dates from test and report dates

Give every date an event role

Preserve each role so sorting the register by date does not create an incorrect picture of when the work happened.

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.

Keep document upload time separately as an administrative event. It may help reconstruct handling, but it does not establish when the sample was collected or measured. If the source prints an unlabeled date, record it as an unresolved role until supporting context is available.

Record evidence before interpreting the calendar

Capture raw date text, stated role, page or table location, source revision and any timezone or time-of-day information. Microsoft: Power Query data types distinguishes date, date/time and date/time/timezone types. Choose the representation supported by the source rather than adding a midnight time or timezone merely to satisfy an import format.

An ambiguous string such as 03/04/2026 also needs a verified calendar convention. Keep the raw string until day-month order is established. A file’s country of origin can guide the question, but a report assembled from several systems may not follow one uniform convention.

Work through a report timeline

Synthetic example: report Q17 revision 2 contains these invented dates. All unambiguous dates below use year-month-day notation in the review worksheet.

The nine-day gap between collection and revision issue does not mean testing lasted nine days. Revision 2 may correct wording without adding a new test. Preserve the test date unless the revised source explicitly changes it.

Check relationships without inventing them

W3C: PROV Data Model distinguishes activities from the entities they use or generate. Apply that distinction by linking a report revision to its issue event and a reported measurement to its test context, rather than assigning every event the newest date.

Flag impossible-looking sequences for clarification, including testing before the reported collection date or a revision dated before its original. Do not automatically reorder the dates. The issue may be a transcription error, a timezone boundary, an incorrect role or an actual source inconsistency. Retain the conflicting evidence until it is resolved.

Verify the export tells the same story

Ask a colleague to identify when the sample was collected, when testing occurred and which report version was reviewed using only the register and linked evidence. Each answer should come from a role-specific field. Unknown events should remain visibly unknown rather than borrowing another date.

Use the multi-date report to define acceptance criteria in a supplier-report scope discussion. Collection, test and report dates need their own agreed mappings and review. The invoice trial covers invoice dates and fields; it does not promise automatic interpretation of laboratory timelines.