Every quality article on this site so far has asked what a number on a certificate of analysis means — what a purity percentage measures, what a mass match confirms, what net peptide content accounts for. This one asks a more basic question: what is that number a claim about?
Not the vial in your hand. It cannot be — the material that was analyzed was consumed by the analysis. A certificate of analysis (COA) describes a lot, sampled at a moment, under an assumption about storage. Each of those three words carries a failure mode that no amount of chromatogram-reading will catch.
A Lot Is a Population, Not a Product
In pharmaceutical manufacturing, a batch or lot is a defined quantity of material produced in a single process run and expected to be homogeneous within specified limits. That homogeneity expectation is the entire logical foundation of lot-based testing: you test a sample, and the result is extended to the rest of the lot only because the lot is presumed uniform. ICH Q7, the GMP guideline for active pharmaceutical ingredients, requires homogeneity to be demonstrated — blending operations that combine sublots have to be validated to show the combined batch is actually uniform, not simply assumed to be.
For research peptides, the practical consequences are ordinary but easy to miss:
- A lot number is the unit of evidence. A COA for lot A says nothing verifiable about lot B, even for the identical catalog item from the same supplier. Same product, different lot, different evidence.
- A "lot" may be a filling campaign, not a synthesis. One synthesis of BPC-157 can be filled into vials across multiple runs on different days. Whether those are one lot or several is a documentation decision made upstream, and the COA rarely says.
- Repackaging creates a new population. If bulk material is aliquoted into vials by someone other than the original manufacturer, the original COA describes the bulk, not the vials.
Where Vial-to-Vial Variation Actually Comes From
The homogeneity assumption is reasonable for a well-mixed powder. It is less automatic for a lyophilized product, because the drying step happens vial by vial.
Freeze-drying is not uniform across a shelf. The edge vial effect — vials at shelf edges and near the chamber door receive more radiative heat than vials in the interior — is a well-documented source of vial-to-vial variability in primary drying. It is significant enough to be studied on its own terms: work published in the Journal of Pharmaceutical Sciences in 2025 examined the edge vial effect specifically as a risk of exceeding collapse temperature in the outer vials, and a 2022 study in the Journal of Drug Delivery Science and Technology investigated how vial packing density changes the inhomogeneity of primary drying.
The nuance worth carrying: faster sublimation does not straightforwardly mean drier. Final residual moisture depends on formulation, cake structure, and secondary drying conditions as well as heat input, so edge vials are not reliably the wet ones or the dry ones — they are simply the different ones. Since residual moisture is both a stability driver and an input to mass accounting, "different" is enough to matter.
A lot whose vials dried differently is a lot whose homogeneity assumption is weaker than the paperwork implies. (The process itself is covered in the piece on lyophilization.)
The Sampling Problem Nobody Puts on the COA
A COA reports results. It almost never reports how many vials were sampled, or how they were chosen. That omission hides a genuinely unsettled area.
The most widely used sampling heuristic in the industry — take the square root of the number of containers, plus one — has no statistical derivation at all. Tracing published in Pharmaceutical Technology places its origin in the 1920s as a memorizable rule for agricultural regulatory inspectors, formalized in a 1927 AOAC committee report, with contemporaneous warnings that it rested on no theoretical basis. A 2011 analysis in the Quality Assurance Journal (Muralimanohar et al., doi:10.1002/qaj.482) ran it through operating-characteristic curves and concluded the resulting sample size cannot support confident inference about a population defect rate. USP General Chapter ⟨1097⟩ covers bulk powder sampling formally.
Two things follow, and they pull in opposite directions:
One: for a research peptide, the sample size behind a COA is frequently one vial, or a portion of one. That is not scandalous — for a synthesized single-sequence compound from one purification it is often adequate. But it is an inference, not a measurement of your vial.
Two: the vial that was tested no longer exists in usable form. Amino acid analysis requires acid hydrolysis, which destroys the sample. Karl Fischer titration consumes it. LAL endotoxin testing consumes it. Sterility testing consumes it. The characterization methods that produce the most definitive numbers are the ones that guarantee you are holding a different vial than the one described.
Retest Date vs Expiration Date
These are different concepts and the distinction is not cosmetic.
Under the ICH stability framework, a drug substance is typically assigned a retest period rather than an expiration date. When that period elapses, the material is not automatically waste — it is re-examined against specification and may be used if it still conforms. An expiration date, by contrast, is normally applied to a finished drug product and is terminal.
Research peptides sit outside that regulatory system entirely, which is exactly why the vocabulary matters: a date printed on a research vial is usually best read as a retest-style projection, not a hard cliff, and often it is neither — just a default interval applied to every SKU regardless of the molecule.
A further point about how those dates are generated. Shelf-life claims are built from real-time stability data at the labeled storage condition, with accelerated data used as supporting evidence. ICH Q1E constrains how far a claim may be extrapolated beyond the real-time data actually in hand. For peptides, accelerated testing carries a specific trap: raising temperature to speed degradation can push a lyophilized cake past its glass transition, at which point the material is failing by a mechanism (collapse, mobility, moisture redistribution) that does not operate at the storage temperature you are trying to model. Arrhenius-style extrapolation assumes the same reaction is running faster. When the physical state changes, that assumption breaks.
Worth noting for currency: ICH reached Step 2b on a consolidated revision of Q1 on 11 April 2025, intended to supersede Q1A–Q1F and Q5C in a single guideline. It is a draft in consultation, with finalization not expected before late 2026. The legacy Q1A(R2)–Q1E and Q5C guidelines remain the operative standard until then.
The Storage Condition Is Part of the Claim
A specification is not a number. It is a triple: an attribute, a method, and a condition-plus-time under which the attribute holds.
"98.5% by HPLC" is an incomplete statement in the same way "half-life is 7 days" is incomplete without a route and species. The complete form is closer to: this attribute, by this named method, on this lot, on this date — and the shelf-life line adds and is projected to hold for this long under this storage condition.
Once material leaves the tested condition, the projection no longer describes it. The chemistry that runs when it does — hydrolysis, oxidation, deamidation, aggregation — is covered in the article on degradation pathways. The point here is procedural: the storage condition is not advice appended to the claim, it is a premise of the claim.
And no lot system captures transit. A COA is issued at the point of testing. Whether the material then sat on a loading dock, went through freeze-thaw cycles, or was exposed to light is invisible to the document. That gap is structural rather than a vendor failing — but it is a real reason a purity figure is a statement about the past.
What This Looks Like on a Real Document
A short traceability pass, independent of anything in the data itself:
- Does the lot number on the vial match the lot number on the COA? If the COA carries no lot number at all, it is a marketing document.
- Is there a date of analysis, distinct from a date of issue? A reissued PDF can carry a fresh date over old data.
- Does the storage condition appear, and is it specific? "Store cold" is not a condition. A temperature and a light/moisture instruction is.
- Is a retest or expiry date present, and is it plausibly molecule-specific? An identical interval across every product on a catalog is a template default.
- Who performed the testing — the manufacturer, the seller, or a named independent laboratory? Each is a different evidentiary claim.
- For anything not a defined single molecule, ask whether lot-based purity is the right frame at all. For HMG, HCG, and heterogeneous preparations like Thymalin, potency is assigned against a reference standard rather than expressed as a purity percentage — see the HCG vs HMG comparison.
FAQ
If the tested vial was destroyed, is COA testing pointless?
No — it is how essentially all analytical release works. Destructive testing on a sample plus a homogeneity argument is a sound structure. The error is not trusting it; it is reading the result as a direct measurement of the container you received rather than an inference about a population — a weaker but still useful claim.
Two COAs for the same product show different purity figures. Which is right?
Possibly both. If they are different lots, they are measurements of different material and there is nothing to reconcile. If they are the same lot, the likely explanation is method — injection load, detection wavelength, gradient, and integration settings all move the number, which is why purity percentages are only comparable within a method. That mechanism is worked through in reading the chromatogram and mass spectrum.
Does a longer retest date mean a more stable compound?
Not on its own. A date is only as good as the data behind it, and for research-market material there is frequently no real-time stability study specific to that molecule — just a uniform interval applied catalog-wide. A short sequence with an Asp-Gly motif and a 9 kDa disulfide-folded protein like IGF-1 LR3 have different failure modes and no reason to share a shelf-life number. Identical dates across chemically unrelated products tell you about the document template, not the chemistry.
Related reading: net peptide content, endotoxin testing, and the evidence hierarchy. More on documentation at /quality/ and compound profiles at /library/.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.