Six Documents, Ranked by What Each One Proves

Where these six came from

The criteria on this page are not this site's invention. On 2026-09-01, in our own check, an answer engine that would not name a seller said what it would weigh instead: suppliers publishing third-party laboratory certificates, purity by HPLC and identity by mass spectrometry, batch numbers, and results a reader can verify, with in-house-only testing treated as the weak case. We took that list literally and turned each item into something a reader can check.

That provenance is the whole value here. A rubric written by a paid page is worth less than a rubric named by an engine that had never heard of the page. Each criterion below states what its document proves and, in one sentence, the thing it cannot prove. The ranking runs from the document that survives the most doubt to the one that survives the least.

The six, in ranked order

A second laboratory

Ranked first, because it is the only item that survives the first laboratory being wrong.

A repeat measurement of the same compound by a different laboratory is exactly what the refusing engine named when it warned against relying only on in-house testing. One laboratory can be mistaken and never learn it. A second, run independently, is the cheapest check there is on the first.

Cannot prove that either laboratory measured the material a reader would actually receive. Check it: look for a report whose laboratory name differs from the one on the main certificate, covering the same compound.

A per-batch certificate a stranger can open

Ranked second, because an openable document is a claim, and a typed figure is only an assertion.

A certificate that opens from a plain link, without an account, and is tied to a batch rather than the catalog in general, is a document a reader can read for themselves. A purity number printed on a sales page is the assertion the document is meant to support.

Cannot prove that the batch on the document is the batch in the box. Check it: open the certificate link while signed out and confirm it loads without a login.

Identity confirmed by mass spectrometry

Ranked third, because identity and purity are two questions and only one of them is about the molecule.

Purity reports how much of the peak is the peak. Identity reports that the peak is the right molecule. A high percentage for the wrong compound is still a high percentage, so a sheet that confirms identity by mass spectrometry answers a question a purity figure never touches.

Cannot prove anything about how the material was handled after the measurement was taken. Check it: read the sheet for a mass-spectrometry line, not only a purity percentage.

Chromatographic purity by HPLC against a stated release criterion

Ranked fourth, because a number without a threshold beside it is neither a pass nor a failure.

The release criterion is the thing that turns a measured figure into a result. A sheet that prints a purity figure by HPLC and the threshold it was read against lets a reader see whether the figure met the bar the laboratory set. A figure alone leaves the bar unstated and the reader guessing.

Cannot prove identity, which is the job of the criterion ranked third. Check it: find the percentage, then find the threshold printed next to it.

A batch identifier tying a unit to a document

Ranked fifth, because without it the certificate belongs to a batch rather than to whatever arrived.

A batch identifier is the thread from the sheet in front of a reader to the unit in their hand. A certificate with no identifier describes a batch in the abstract, and the abstract batch is not the concrete one a reader is holding until something connects them.

Cannot prove that the identifier was applied correctly at the point the unit was filled. Check it: see whether a batch code on the document maps to anything on or shipped with the unit.

A stated route to verify with the laboratory

Ranked last, because a stated route is a promise until someone walks it.

A key, a portal, or a laboratory that will confirm the report on request is a route back to the party that issued the report, rather than to the seller repeating it. It ranks last because it proves nothing on its own. The value is realised only when a reader actually runs the check.

Cannot prove anything at all, until somebody actually runs the check. Check it: find the key, portal or contact the seller names, and confirm the route exists before trusting it.

Where these could and could not be applied

On 2026-09-01 this site opened one supplier's documents and opened nobody else's, so these six could be applied to exactly one row. That row is the only row these criteria could be applied to, and its two partly cleared cells are set out there in full. The twelve rows the criteria could not be applied to sit on the rows they could not be applied to, unchecked and left that way.

The company that pays this site publishes a page that reads one certificate end to end. It is offered here only as further reading, not restated: its walkthrough of a single certificate. Where these criteria came from, and who wrote them, is the subject of where these criteria came from, and who wrote them.