How can a vial be 99% pure by HPLC and still contain under 7 mg of a nominal 10 mg?

I sent two vials from two different batches, both labelled 10 mg semaglutide, for independent testing and asked for purity and quantification on each. The results came back:

• Vial A: purity 99.1%, content 10.1 mg
• Vial B: purity 99.2%, content 6.8 mg

Vial B has the higher purity and a third less peptide in it. I have now read four different explanations of this online and they contradict each other. One says purity and content are "completely unrelated numbers", which cannot literally be true since they are measurements of the same physical object. One says the missing mass is water. One says it is salt. One says the vendor just underfilled it and the purity is a distraction.

What I want is the actual mass balance. If I put vial B on a microbalance and weighed the cake, what would it weigh, and where does every milligram of that go? I would like to be able to reconstruct the 6.8 mg figure from a set of numbers that a lab could actually measure, rather than taking it on faith. And then I want to know which of those numbers a COA is obliged to tell me and which it typically hides.

To sharpen one thing in the accepted answer: purity and content are not merely different, they respond to opposite pressures in manufacturing, which is why the correlation between them is often mildly negative rather than positive.

Chromatographic purity is bought with purification. Every additional prep-HPLC pass throws away material, especially the shoulders of the main peak where the closest-eluting related substances live. A crude peptide at 85% can be taken to 99.5% but the yield falls, sometimes to under half. So the batches with the most impressive purity numbers are frequently the batches with the least material — and if the filling operation is volumetric or sloppy rather than gravimetric against an assayed concentration, that low yield propagates straight into low fill.

The version of this that actually loses people money is subtler. A manufacturer working to a purity spec and no content spec will optimise the number that gets checked. If buyers demand 99% and never ask for mg, you get 99% and no mg. This is not a conspiracy, it is a specification design failure: you get what you measure.

Two practical consequences:

• Do not use purity as a proxy for anything else. It is not a proxy for content, for sterility, for endotoxin, for correct stereochemistry, or for how the vendor runs their operation. It is one number about one property.
• When you commission testing, buy content before you buy a second purity run. If you have one budget and one vial, quantification tells you more than another purity figure. Purity varies between batches by a fraction of a percent; content varies by tens of percent.

Janoshik, Medutest and PeptideMeter all price these separately and will run both on one sample if the vial has enough material. Ask what minimum mass they need before you send a single 10 mg vial and expect three assays out of it.

One correction to a widespread misreading of the table above, since it causes people to reject good batches.

Water and counter-ion mass are not adulteration. They are an unavoidable physical property of a lyophilised peptide salt. A peptide with four or five ionisable groups cannot exist as a dry solid without counter-ions; it would not be electrically neutral. And a lyophilised cake will pull water out of the air through the stopper over months of storage no matter how well it was dried at the point of manufacture.

Typical, entirely normal ranges for a GLP-1-class acylated peptide:

So a peptide content figure of 86 to 88% is a normal, healthy result. A peptide content figure of 86% is not a purity of 86% and it is not 14% contamination. The single most common misreading I see is someone finding "peptide content 87.6%" on a COA, comparing it against a 99% purity claim, and concluding the vendor lied about purity. Two different measurements, both true.

The number that tells you whether you were shorted is neither of those. It is mg per vial.

Adding the failure mode that produces a false low content reading, because I have watched people condemn a batch over one.

Quantification requires that essentially all the peptide in the vial ends up in the injected solution. That means the analyst reconstitutes in a known volume, dissolves completely, and either injects from the vial itself or transfers quantitatively with rinses. Every step leaks material:

• Peptide adsorbs to glass and to polypropylene, more at low concentration and at pH near the isoelectric point. A dilute transfer through two tubes can lose several percent.
• Incompletely dissolved cake on the shoulder of the vial reads as missing peptide. Acylated analogues can take several minutes and gentle swirling; if the analyst injects at two minutes they get a low number.
• Filtering through the wrong membrane loses peptide to the filter. PVDF and regenerated cellulose are usually fine, nylon is not.

A competent lab controls all of this and their content figures are repeatable to within a couple of percent. An incompetent one produces low results that look exactly like an underfilled vial. The way you tell the difference from outside is replication: two vials from the same batch, or the same vial split, or the same batch at two labs. A single content number 32% below label from one lab on one vial is a strong signal, but it is one measurement.

If you are commissioning a test that might end a vendor relationship, pay for the duplicate.