Independent and accredited are not the same thing
The two get mixed up constantly, including by people acting in complete good faith. They answer different questions, and you check them in different ways.
Being good at the job, versus having a stake in the answer
Accreditation — usually a standard called ISO 17025 — says the lab is good at its job: its methods have been checked, its instruments are calibrated, its staff are trained, its paperwork is controlled, and it is regularly tested blind against other labs. A national body audits all of that periodically. It is a real credential.
Independence is a different question: who owns the lab, who paid for the test, who picked the sample, and what it would cost the lab commercially to publish a bad result. No accreditation standard covers any of that, because none was ever meant to.
A lab can be fully accredited and owned outright by the supplier whose material it certifies. Every number it publishes may be correct. It is still not an outside party, and a report calling it one is wrong on that point regardless of how good the testing was.
Checking the credential
This takes a few minutes and is the most reliable check available to anyone without access to a lab.
- Find the accreditation number on the report. If there is not one, the claim cannot be checked, and it should be treated as unchecked.
- Identify who issued it. Each country has its own body — UKAS in the UK, COFRAC in France, DAkkS in Germany, A2LA and ANAB in the US — and each publishes a public, searchable list.
- Search that list for the number. It should lead to the lab named on the report, at the address given.
- Open the scope document — the list of exactly which tests the accreditation covers. Accreditation is never blanket: it applies to named tests on named kinds of material. Check that the test used on your report is on that list.
- Check the dates. Accreditation expires, gets suspended, and is sometimes cut back after an audit.
Step four is the one most people skip, and the one that most often fails. A lab accredited for one technique can perfectly well report a different one that its accreditation does not cover — without misleading anyone. The report simply carries less weight than the logo at the top of it suggests.
Three degrees of distance
Reports rarely say which of these applies. Where it is not stated, assume the second — it is the commercial norm.
Tested by the seller
The company being tested runs the test itself. The work may be faultless and the equipment properly calibrated. It is still marking your own homework, and should be read that way.
Paid for by the seller
An outside lab tests a sample that the seller supplied and paid to have tested. This is the normal arrangement and it is not disqualifying — but the seller chose which sample went in.
Sampled by someone else
The lab, or another outside party, takes the sample from finished stock — the seller does not get to pick which one. Rare, considerably stronger, and almost always mentioned when it happens.
Questions that bring the arrangement into the open
Anyone who knows the answer can give it in one line — and the silence is as informative as the answer.
- Which lab produced this report, and what is its accreditation number?
- Do you own any part of that lab, invest in it, or have an exclusive deal with it?
- Who chose the sample that was tested, and out of what?
- Did the tested vial come from the batch that shipped, or was it made separately?
- Do you publish results when they come back bad, and can I see one?
The last one tells you the most. Any real testing programme, run long enough, produces results that fail. A registry where nothing has ever failed is either very new, or is not showing you everything it has.