A certificate of analysis is evidence only after it has been checked. Before that it is a document asserting that a laboratory tested something, and documents can be edited, reused across lots, or produced wholesale. The reassuring part is that almost every check worth doing needs nothing but the document, the vial, and a few minutes.
This post sets out eight of them and then runs them across all 114 certificates in this library, including the one that failed.
1. Ask the laboratory
This is the check that makes the other seven secondary, and it is the one most often skipped.
A certificate carries a code. Take that code to the issuing laboratory's own verification page and confirm the laboratory says the certificate exists and matches. Every certificate in this library carries its printed code and links to the issuing laboratory's verification page, and that is the point of publishing the code rather than only the document.
What this defeats is the whole class of attack the other checks only make harder. A fabricated document can be internally perfect. It cannot make the laboratory agree that it exists. If a vendor cannot tell you which laboratory issued a certificate, or the laboratory has no way to confirm one, you are holding an unverifiable file whatever else is right about it.
2. Match the lot to the vial in hand
The certificate names a lot. The vial carries a lot. If they differ, the certificate describes other material, however genuine it is.
This is the commonest real-world failure and it is rarely deceptive. A vendor publishes one certificate per product and leaves it up while the lot behind it changes. The document is authentic and belongs to a batch you do not have. This library keeps superseded certificates visible rather than deleting them for exactly this reason, so a buyer holding an older vial can still find its own document; 43 of the 114 here are superseded. How to read a peptide vial label covers where the lot appears.
3. Check the dates run forwards
A certificate has at least two dates: when the sample was received and when the report was issued. Received must come first.
Across all 114 certificates in this library both dates are present and not one has an issue date earlier than its receipt date. The gap runs from 6 to 16 days with a median of 6, and the 16-day lots are the ones carrying a sterility result, because a culture needs its incubation time. That is a pattern a fabricated set tends not to have: real turnaround varies with what was tested, and the variation is explainable.
4. Confirm a method is named for every result
A result without a method is a number without a meaning. "Purity 99.8%" is incomplete; "purity 99.8% by RP-HPLC at 214 nm" is a measurement, because the method and the detection wavelength determine what the figure is a ratio of.
The reporting requirements that accredited laboratories work to are explicit about this, and it is a fast completeness screen. Every certificate in this library names its purity method as reversed-phase HPLC at 214 nm and its identity method as either MALDI-MS or LC-MS/MS. The identity method changed in August 2026, which the dataset post dates, and that change is itself a sign of real documents: a template does not change its methods halfway through.
The posts on HPLC area percent and what LC-MS confirms cover what each method establishes.
5. Check the reported mass against the formula
Here is a check that costs nothing and is surprisingly effective. A certificate reporting an identity by mass spectrometry states a mass. That mass is predictable from the compound's sequence or formula.
Add the residue masses of the sequence, add one water, and compare. The post on calculating molecular weight from a sequence shows the method reproducing eleven of this catalogue's published formulas atom for atom. Two cautions: compare average mass to average and monoisotopic to monoisotopic, because mixing them produces a fraction-of-a-Dalton gap that looks like a finding and is not, and remember that an end modification such as an acetyl group shifts the number legitimately.
A reported mass that cannot be reconciled with the compound's formula by any ordinary modification is the most substantive thing a reader can find without equipment.
6. Look for reporting limits, not just values
Some results are limits rather than measurements, and a real certificate says so.
Every certificate here reports endotoxin as below 0.05 EU/mL, written with a less-than or less-than-or-equal sign on every one of the 114. That is the assay's reporting limit, the lowest value it will state, and the notation is the laboratory being precise about what it knows. A certificate that printed a bare "0" for endotoxin, or a suspiciously specific measured value where a limit belongs, would be claiming more than the assay delivers. The endotoxin post covers why.
7. Check the internal arithmetic
Numbers on a certificate relate to each other, and relationships are harder to fake than values.
Content should sit in a plausible relationship to the label. Across the 96 single-compound vials here the median is three percent over label, and six read under it, each named in the dataset post rather than buried. Retention time should be characteristic of the compound: a dipeptide barely holds a reversed-phase column and comes off early, an acylated compound with a fatty-acid chain comes off late. And on a blend certificate the content line reports one component's mass, not the vial total, so a 20 mg two-compound vial showing roughly 10 mg is correct, which what a blend certificate shows explains.
A set of certificates where every number is independently plausible but none of them relate to each other is a set that was generated rather than measured.
8. Be suspicious of certificates that are too clean
The pattern worth distrusting is not sloppiness. It is uniformity.
Identical purity figures across every compound in a catalogue. No instrument named. No analyst or authorising signature. No reporting limits anywhere. The same issue date on everything. Results for tests that take different lengths of time, all turned around in the same number of days. Real laboratory output is lumpy, because different assays behave differently and different months have different workloads, and the lumps are the evidence.
For comparison, the spread in this library is a third of a percentage point in purity across 114 lots, which is tight, and it is tight for a stated reason: one laboratory, one method, and a vendor that lists lots which passed. That last clause matters and is stated on the dataset post too. A narrow range is not proof of anything on its own; what makes it readable is knowing why it is narrow.
What the checks found here
Running codes and dates across all 114 certificates produced one real discrepancy, and the case for publishing it is stronger than the case for quietly fixing it.
There are 114 certificates and 113 distinct printed codes. The duplicate is COA3362, which appears on two documents issued on 3 April 2026:
| Filed as | Compound | Lot | Printed code | Status |
|---|---|---|---|---|
| COA3362 | AOD-9604, 5 mg | AOD5-0318 | COA3362 | Superseded |
| COA3364 | Thymosin alpha-1, 10 mg | TA110-0318 | COA3362 | Superseded |
The thymosin alpha-1 certificate is filed under COA3364 and the code printed on the document reads COA3362, which belongs to the AOD-9604 certificate. Everything else about both documents is consistent: different compounds, different lots, different content figures, dates in the right order.
Two explanations fit. Either the laboratory's printed document carries a transcription error in its code, or the pipeline that reads these certificates into structured data misread the printed digits. We have not yet established which, the question is with the laboratory, and both lots have since been replaced. It is recorded here because a verification post that reported no findings on its own library would be worth nothing, and because the editorial standards commit to publishing a certificate as it stands rather than as it should read.
The practical note for anyone checking one of those two documents: verify by lot number, which is unambiguous on both, rather than by the printed code.
Frequently asked questions
What is the single most important check?
Confirming the certificate with the laboratory that issued it. Internal consistency can be faked by a careful forger; the issuing laboratory's agreement cannot.
The lot on my vial does not match the certificate online. Is that a red flag?
It usually means the vendor's published certificate is for a newer or older batch than the one you received. Ask for the certificate matching your lot. A vendor who cannot produce one has not documented your material.
Can I detect a fake certificate without any equipment?
Often, yes. Dates in the wrong order, no method named for a result, a reported mass irreconcilable with the compound's formula, and identical figures across unrelated compounds are all visible on the page. None of that is proof, and verification with the laboratory is still the check that settles it.
Why publish a discrepancy in your own certificate library?
Because the alternative is a checklist that has never been applied to anything. The code mismatch above is minor, both lots are superseded, and disclosing it is the only version of this post that is worth reading.
Should every certificate carry a water content or residual solvent figure?
Not necessarily. A certificate reports the tests within its stated scope, and these cover identity, purity, content, endotoxin, heavy metals and sometimes sterility. The posts on residual moisture and what a certificate does not test for cover the absences honestly.
Does a certificate prove the material is safe?
No, and no certificate on this site claims to. It records what a laboratory measured on a sample of a lot. Everything on this catalogue is supplied for in-vitro laboratory research only and is not for human or animal use, which what research use only means sets out.
References
- Pepstral certificate library: 114 certificates issued by Bioviridian Inc., published unaltered with the printed code for each. Code and date consistency across all 114 records was checked on 5 October 2026; the figures in this post are that check's output. pepstral.com/coa.html
- Bioviridian Inc. certificate verification page, where every printed code in this library can be confirmed with the issuing laboratory. bioviridians.com/coa-search.html
- ISO/IEC 17025:2017, General requirements for the competence of testing and calibration laboratories. Clause 7.8 sets out what a report of results must contain, which is the basis for the completeness checks in this post. www.iso.org/standard/66912.html
- United States Pharmacopeia. General Chapter <85> Bacterial Endotoxins Test and General Chapter <1225> Validation of Compendial Procedures, for the reporting-limit and method-identification expectations. www.usp.org
Every product mentioned is sold for laboratory research use only and is not for human or animal use. Nothing on this page describes or recommends use of the material sold here in humans or animals.
