Specification in the result column: why that is a meaningful error — the long version

Specification in the result column: why that is a meaningful error — the long version Writing it up because I had to work it out twice and would rather nobody else did.

A documentation question rather than an analytical one.

I have a certificate in front of me that reports a purity figure, names a technique, gives a wavelength, and stops. No gradient, no column, no injection volume, no chromatogram.

What can I legitimately conclude from that document? My instinct is "almost nothing, but not literally nothing", and I would like to know where the people who read these professionally draw the line.

A chromatogram supplied as a small image is legible for peak shape and not for baseline detail. That is enough to sanity-check an integration and not enough to reproduce it.

Agreed, and I will stop repeating the version of this I had been repeating.

The opening post answers the question as asked. The question underneath it is different.

What fields matter on a certificate: lot number matching the vial, test date, the analytical method stated specifically, the measured result as a number, and the acceptance limit stated separately. A certificate missing any of these is weaker.

Worth separating two things that the opening post runs together.

The fields that make one usable: lot identifier matching the container, test date, method identification specific enough to reproduce, the result as a number, and the acceptance criterion stated separately from the result.

The general case is well covered; this is the awkward specific one.

This follows post #5 rather than contradicting it.

An itemised related-substances table with retention times says considerably more about the synthesis than a single total. A total tells you how much is not the main peak and nothing about what it is.

The distinction between release testing and characterisation is one industrial documents make and retail ones usually do not. It is the difference between "we tested every lot for this" and "we established this once".

If anyone can point at the primary source I would be grateful.

A chromatogram supplied as a small image is legible for peak shape and not for baseline detail. That is enough to sanity-check an integration and not enough to reproduce it.

I would rather post the uncertainty than round it away.

"Not less than 98 per cent" is a specification. "98.3 per cent" is a result. A certificate carrying only the first has not told you what was measured.

Not a conclusion. A place to stand while looking for one.

A test date preceding a manufacturing date is a clerical error until somebody explains otherwise, and asking is cheap. Most such things are transcription.

I have written this out at length because the short version keeps being misread.

The arithmetic in post #10 is right; the assumption feeding it is the part to check.

A certificate is a statement by its issuer that a defined test was performed on a defined lot and produced a defined result. That is all it is, and it is not nothing.

Nothing above should be read as advice about what anyone else should do.

When method identifiers are missing: "reversed-phase HPLC" is less specific than "reversed-phase HPLC at 214 nm on a C18 column with a 10 to 40% acetonitrile gradient over 20 minutes". The second lets someone reproduce the analysis. The first does not.

Useful. I had the fact and not the reason, which turns out to be the important half.

Post #12 put the caveat in the right place and I want to underline it.

A chromatogram supplied as a small image is legible for peak shape and not for baseline detail. That is enough to sanity-check an integration and not enough to reproduce it.

That is all I can say without guessing.