Editorial Standards — The Rulebook You Can Hold Us To
Editorial standards
This page is the rulebook. It is written so that you can hold us to it, and so that anyone writing for this site knows what will be rejected.
1. We teach a method. We do not rank brands.
We write about what a material class does and does not do, and about how to interrogate a claim made about it. We do not publish comparisons of one named supplier against another, and we do not publish a verdict on which product a buyer should choose. That decision depends on your spec, your volume and your risk tolerance, none of which we know.
What you will find instead: the question to ask, the document to ask for, and the part of the answer that usually turns out to be missing.
2. What a certificate does not cover is the part worth reading
Every certificate entry on this site carries a not covered section, and it is a required field — an entry cannot be published without it. It must cite the clause or scope statement it relies on. “The standard is silent on this” is a claim about a document, and a claim about a document has to point at the document.
We are strict here because we have been wrong here. In an internal review on 22 August 2026 we found entries where “not covered” had been used as a polite way of saying we did not read that far. That is the exact assumption this site exists to argue against, made by us. The field rule changed as a result.
3. A number without a source does not get published
Not softened, not hedged, not labelled “approximate” — not published.
- Composition shares are optional fields. If we cannot find a published figure for the remainder of a blend, that share stays empty. We do not compute a residual so that the column adds up to 100%.
- We publish no test result of our own, because we run no tests. Where a figure comes from a supplier’s own specification, it is labelled as the supplier’s statement, and a supplier’s statement is not a measurement.
- Where no independent test record exists, the entry says “No independent test record on file.” That sentence is doing real work. It means we looked and found nothing, and that you should not read the absence as reassurance.
4. Every record carries a date and a reachability status
Standards get revised, regulations move, and companies stop trading. A reference that was accurate when written and never touched again is not a reference; it is an archive pretending to be one.
Each record on this site carries three provenance values:
| Value | What it tells you |
|---|---|
Last verified |
The date a person last opened the sources and checked the entry against them |
Verified note |
What changed, or what could not be confirmed, at that check |
Source reachable |
Whether the cited source currently loads: reachable, blocked by anti-bot, dead, or unchecked |
blocked by anti-bot is listed separately from dead on purpose. Several official registries refuse automated requests; that is not the same as the document having disappeared, and collapsing the two would let us quietly drop inconvenient sources.
5. What we are not
- We have no laboratory. No bench test, no ISO/IEC 17025 accreditation, no in-house measurement of any kind.
- We receive no production or transaction data from any factory.
- We are not independent of all suppliers. We are editorially independent and commercially funded — see the interest disclosure.
Any earlier version of this site that suggested otherwise was wrong. It has been corrected rather than reworded.
6. Writing about named companies
A named company gets the same treatment as any other subject: we report what a public record says, we date it, and we cite it. We do not publish an adverse conclusion about a named business that we cannot point to a record for, and we do not infer one from a material’s general weaknesses. Where a company’s legal or trading status is relevant — an insolvency filing, a discontinued product — we take it from the register that holds it and we say which register and when.
Corporate status changes fast. If you are reading an entry whose last verified date is old, treat the status line as stale and check the register yourself.
7. Corrections
We publish corrections rather than edits in silence.
- If a claim on this site is wrong, tell us and quote the line. We would rather be told bluntly.
- A material correction is made in the entry, and the
verified noterecords what changed and when. - If we get a correction wrong, that is recorded too. It has happened: in August 2026 we “corrected” seven files to say an EU proposal had been withdrawn when it had not been, and had to reverse it.
- ⛔ Nothing is ever corrected because a company asked us to soften it. A correction needs a document, not a request.
8. Who writes this
Material Primer is produced by a small team using AI assistance for research and drafting. Every entry is reviewed against the rules on this page before it is published. Where an account on this site is not a human, it is labelled as such. We do not create fictional readers, reviews or testimonials.
Last reviewed: 23 August 2026.