Reading a COA

What a COA is and is not

A certificate of analysis reports what a laboratory measured on a specific batch at a specific time. That is a genuinely useful thing. What it is not: a statement about the vial in your hand, unless the vial in your hand came from that batch and nothing happened to it in between.

The distinction that matters most on this site is vendor-supplied versus independent. A vendor-supplied COA is produced or commissioned by the seller. An independent result is one where someone sent a vial they actually received to Janoshik, Medutest, PeptideMeter or VendorInvestigate and paid for it themselves. Both belong in a discussion; only one of them survives the evidence standard in c/vendorvetting on its own.

The four fields, and what happens when one is missing

A cropped screenshot that shows a big number and nothing else is not a COA, and in c/coa it gets a removal and a request to repost the whole page.

Area% is not mass%

This is the single most consequential misunderstanding in the whole space, so it gets its own section.

In a UV-detected HPLC run, the detector produces a trace with peaks. "Purity 98.6%" almost always means the main peak accounts for 98.6% of the total absorbance under all the peaks the detector saw. That is area percent. It is a ratio of one chromatographic species to the others, and it is the right number for the question "how much of the peptide-like material here is the peptide I wanted".

What it does not tell you:

• How much peptide is in the vial by mass. Anything that does not absorb at the detection wavelength is invisible — water, salts, most counterion, some excipients.
• Whether the main peak is the right molecule. Area% is blind to identity. A closely related impurity eluting at the same time is counted as product.

The mass question is answered by net peptide content, which is a genuinely different assay and appears on far fewer certificates. This is how a vial can be "99% pure" and contain noticeably less than the labelled milligrams: purity is a ratio, not a mass.

The counterion

Synthetic peptides are usually isolated as a salt, most often an acetate or a trifluoroacetate. That counterion is part of the mass on the scale. Add residual water — a few percent is normal even in a well-dried lyophilisate — and the labelled weight of a vial can be meaningfully more than the peptide it contains, while every purity figure on the certificate remains perfectly honest.

This is not a scandal, it is chemistry, and reputable documentation states it. It becomes a problem when a vendor quotes only the flattering number and lets you infer the other one.

Water content

Residual moisture is measured by Karl Fischer titration or loss on drying, and it is the field most likely to be absent. It is worth caring about for two reasons: it eats into your net content, and a high figure suggests the lyophilisation cycle or the packaging is not doing its job, which has implications for stability that no purity number will show you.

Across the vendor directory, water content appearing consistently on a supplier's certificates is one of the better proxies for QC culture that members have found. It is not required, nobody asks for it, and the ones who report it anyway tend to be the ones who get the other things right too.

Identity, and why LC-MS matters

UV chromatography separates. Mass spectrometry identifies. An LC-MS result that finds the expected molecular mass — and ideally a fragmentation pattern consistent with the expected sequence — answers a question UV cannot: is this the right molecule at all?

For well-known compounds from established suppliers, identity failures are rare and purity variation is common, so the UV number is what people watch. For anything unusual, anything renamed, or anything where the sequence is written four different ways across four vendor sites, identity is the first question and purity is the second. c/researchpeptides exists largely because of this.

Two labs, two numbers, both right

Inter-lab variance on this kind of assay routinely runs one to two percentage points. Different column chemistry, different gradient, different integration decisions on a wandering baseline. If Janoshik says 98.1% and Medutest says 96.9% on the same vial, those results agree; they do not contradict each other, and building a public accusation on that gap is how c/scamalerts ends up having to retract things.

What is not normal variance: a four-point gap, a result below 95% against a 99% claim, or a consistent negative delta across multiple batches. Those are findings. The vendor test logs deliberately show the claimed figure next to the measured one so the delta is visible, because the delta predicts your next order and the absolute number does not.

Tells that a COA is decorative

• The same batch number on two unrelated vendors. Either a shared upstream source, which is normal and fine, or a shared template, which is not.
• An identical chromatogram for two different compounds. Traces are not interchangeable. This one is always fatal.
• The same signature image across suppliers. Someone bought a PDF, not a test.
• PDF metadata that does not fit. A producer string naming a browser or an office suite, or a creation timestamp after the date printed on the page. This is the cheapest check available and it catches a surprising amount.
• "≥98%" with no measured value. That is a specification, not a result. Real reports state what was found.
• A purity figure to three decimal places off a visibly noisy baseline. Precision that the data cannot support is a tell in itself.

The two-minute checklist

1. Batch number present, and does it match the vial or the box?
2. Date present, and is it before your vial shipped?
3. Method stated properly, not just "HPLC"?
4. Which quantity is the headline number — purity, or net content?
5. Is water content reported? (Bonus if yes.)
6. Is there any identity confirmation, or is this UV only?
7. Has anyone independently tested this batch, and how big is the delta?

If the answer to the last one is no, and it matters to you, the fix is boringly simple and costs less than a wasted vial: send it to one of the four services and post the result in c/labresults. The community's whole evidence base is people who did that.