How to Read a Certificate of Analysis

This is a guide to reading a certificate of analysis for a research peptide. It is written to be used against any vendor. If you apply it to us and we come up short, we would rather you found out here.

Nothing below is proprietary. It is how analytical chemistry documentation works, described in ordinary language, because the single largest advantage a dishonest supplier has is that most of their buyers have never been taught to read the document they are being shown.

What a certificate of analysis is

A certificate of analysis is a laboratory’s report on one sample that arrived at that laboratory. That is the whole of it. It is not a certificate about a product, a brand, a batch, or a vial. It is a record of what an instrument measured when a particular sample was placed in it on a particular day.

Every failure mode in this article follows from that one sentence. The document is honest about a sample. The question is always whether the sample has anything to do with what you were sold.

The consequence: a genuine COA can be worthless

Consider the most common scenario in this market, which requires no forgery at all. A vendor selects material, mails a sample to a laboratory, and receives a genuine report showing excellent results. The laboratory did nothing wrong. The report is authentic and would survive any authenticity check you could run. It then travels for two years attached to product it never described.

This is why “is the COA real?” is the wrong first question. The right questions are: which specific production run does it describe, and can I confirm that the vial in my hand came from that run?

The eight things a real certificate contains

Work down this list on any certificate you are shown. Missing items are not always fatal, but each one is a question the document declined to answer.

The testing laboratory, identified. Legal name, physical address, and a way to contact them that is not the vendor. “Third-party tested” with no laboratory named is not a claim, it is a phrase.
A report or reference number. Unique to that report, and ideally retrievable from the laboratory. Certificates without a report number cannot be looked up, which is frequently the point.
Sample identification exactly as received. What the lab was told the sample was, and any internal sample ID assigned on receipt. A real lab reports the sample name it was given, because it has no way of knowing independently what the powder is.
The lot or batch number. This is the binding between the paper and the physical material. See below.
Dates. Date received, date tested, date of report. Three separate dates on a real certificate, and they run in that order.
The method, per test. Not “HPLC” but the method: reversed-phase, detection wavelength, gradient, and the basis of quantitation. “LC-MS” alone tells you nothing about whether identity was actually confirmed.
Results as numbers, with units and limits. A measured value, its unit, the specification it is being compared against, and for trace analysis the limit of detection or quantitation. See the section on ND below.
Raw data, not just conclusions. For purity, that means the chromatogram itself with axes, retention times, and the peak integration table. A summary table with no chromatogram is a claim about data rather than data.

The lot number is the whole argument

The lot number is the only thing connecting a piece of paper to a physical object. Everything else on the certificate is describing chemistry; the lot number is describing provenance, and provenance is what is actually in dispute.

Three checks, in order of how much they tell you:

Is there a lot number on the certificate at all? A certificate that names only a product — “BPC-157 5 mg” — and no lot describes an unbounded quantity of material across an unbounded period. It can never expire and can never be contradicted.
Is the lot number printed on the vial? Printed, as part of the label artwork or as a machine-applied code. Handwritten lot numbers are a long-standing grey-market signal: they are applied after the fact, by hand, to whatever vial happens to need one. A lot number that exists only on the website and not on the container is not traceability, it is a caption.
Do the two match, character for character? Including any suffix. Vendors under pressure sometimes publish a certificate for lot A-2291 and ship vials marked A-2291-B, which is a different production run and a different set of results.

If a vendor cannot show you a certificate for your lot number specifically, they have not tested your material. They have tested some material.

Reading the purity result

Purity is usually reported as a percentage from reversed-phase HPLC with UV detection, calculated by area normalization: the integrated area of the main peak divided by the total integrated area of all peaks, expressed as a percentage.

Two properties of that calculation matter enormously and are almost never explained to buyers.

It is a ratio, so it says nothing about how much you got

A percentage has no mass in it. A vial can be 99.4% pure and contain 7.1 mg where the label claims 10 mg. Purity and fill are independent, and a purity figure cannot be made to imply a quantity no matter how high it is. This is why fill accuracy — the vial weighed, gravimetrically, and the milligrams reported — is a separate line on a serious panel and absent from most certificates in this market.

It only sees what absorbs UV light

This is the single most useful thing on this page. Area normalization at a UV wavelength measures the relative proportions of UV-absorbing species. Water does not meaningfully absorb there. Nor does acetate, nor chloride, nor most inorganic salt.

So chromatographic purity and net peptide content are different numbers, and the gap between them is not small:

Lyophilized peptide is hygroscopic and routinely carries a few percent water by weight.
The counterion — trifluoroacetate from purification, or acetate after exchange — adds several percent more, and for some sequences considerably more.
Residual solvents from synthesis and cleavage contribute further.

The result is that a lot which is legitimately 99% pure by HPLC can be meaningfully less than 99% peptide by mass. Neither number is a lie. They are answers to different questions, and the vendor is quoting the flattering one.

To close that gap you need three additional tests: water content by Karl Fischer, counterion and residual TFA by ion chromatography, and residual solvents by GC headspace. If a certificate omits all three, you cannot convert its purity figure into an amount of peptide. Ask for them. The response is informative either way.

Reading the chromatogram itself

If a chromatogram is included, look at it rather than at the number printed above it.

Are both axes labelled? Time in minutes, response in mAU or similar. An unlabelled trace is decoration.
Is there an integration table? Retention time, area, and area percent for every integrated peak — not just the main one. A single-row table means the other peaks were not reported, and the other peaks are the impurity profile.
Where does the baseline sit and where was integration cut off? Purity can be inflated by starting integration late, ending it early, or setting a threshold that excludes small peaks. A trace that begins abruptly just before the main peak is worth a question.
Is the main peak symmetric and fully resolved? A shoulder on a peak is a co-eluting species being counted as product.
Does the gradient and run time make sense for the molecule? A five-minute run on a 39-residue lipidated peptide is not resolving much of anything.

Reading the identity result

Purity tells you the material is one thing. Identity tells you which thing. They are frequently confused and a certificate can carry a superb purity figure for entirely the wrong compound.

Identity is normally confirmed by mass spectrometry, and a real identity result shows the measured mass against the theoretical mass for the stated sequence — monoisotopic or average, stated as which. Look for the observed value, not just the word “conforms.”

Points worth knowing:

Mass alone does not distinguish between sequences with the same composition, or between certain isomers. For an unusual or heavily modified peptide, mass confirmation plus a matching retention time against a reference is stronger than mass alone.
Some certificates report identity by comparing a retention time to a previous in-house run. That confirms consistency with a prior batch, not identity with the named compound.
“Conforms to standard” with no standard named, and no number, is not an identity result.

Why “99.9% on every product” is a tell

Scan a vendor’s certificates side by side across their catalogue. If every compound, every size, and every lot reports the same purity — or a set of figures clustered implausibly tightly at the top of the range — that is a pattern real manufacturing does not produce.

Purity is a function of the sequence and the run. Long peptides accumulate more deletion sequences than short ones. Difficult couplings, aggregation-prone stretches, methionine and cysteine oxidation, and the operator’s luck on a given day all move the number. A catalogue of a fifteen-residue peptide, a seven-residue peptide, a thirty-nine-residue lipidated peptide and a copper complex should not report identical purity, because those four molecules do not purify identically.

Related signals in the same family:

Suspiciously round figures. Instruments produce values like 98.7% and 99.23%. A catalogue of 99.0%, 99.5%, 100% was typed, not measured.
Purity above 99.9% as routine. Achievable occasionally on a short, well-behaved sequence. Not routinely, and not across a catalogue.
The specification and the result being the same number. If the spec is “≥ 99%” and every lot returns exactly “99%,” the result column is copying the spec column.
No variation over time. Pull three lots of the same product from different months. Identical numbers across all three is not consistency, it is one document with the date changed.

Why heavy metals get skipped

Elemental impurity testing is the most commonly omitted panel item, and the omission is economic rather than scientific.

Metals reach a peptide from coupling and deprotection reagents, from catalysts, and from the surfaces of reactors, tubing and filters. Detecting them means ICP-MS — a separate instrument from the HPLC and LC-MS that produce the rest of the certificate, requiring separate sample digestion, separate calibration standards, and a separate line on the invoice. Purity and identity can be run on one system in one sitting. Metals cannot. So they are dropped, and the certificate looks complete because most readers do not know what is missing.

When elemental data is present, read how it is reported:

“Complies” or “Conforms” is a conclusion, not a measurement. It hides both the value and the limit it was compared against.
ND — “not detected” — is not a result. Not detected at what limit? By what instrument? Every method has a detection floor, and “not detected” with no limit of detection stated means the floor could be anywhere. A real trace result reads < 0.5 ppm (LOD 0.1 ppm), which is a measurement. ND is an assertion.
A single “heavy metals” row is an obsolete style. Modern elemental impurity work under USP <232> reports element by element, because the limits differ per element by orders of magnitude.

What endotoxin means

Endotoxin is lipopolysaccharide from the outer membrane of Gram-negative bacteria. It is measured by the LAL assay and reported in endotoxin units, EU.

The important property is that it is orthogonal to chemical purity. Endotoxin is not a synthesis by-product. It arrives from water, from glassware, from the room, from a filter, or from a stopper — and it survives conditions that destroy bacteria, so an absence of viable organisms says nothing about it. A chemically immaculate peptide can carry a substantial endotoxin load, and no amount of HPLC will reveal it, because lipopolysaccharide does not present as a synthesis-related impurity on a reversed-phase column.

Reading the number:

EU/vial and EU/mg are not interchangeable. A per-milligram figure can only be converted to per-vial if you know the actual fill in milligrams — which is the other measurement most certificates omit. Per-vial is the more useful reporting basis and the harder one to present favourably.
“Passes” is not a figure. Ask for the value and the specification.
Check which LAL method was used — gel clot, turbidimetric, or kinetic chromogenic. Gel clot is a threshold test that returns pass or fail at a chosen sensitivity. Kinetic chromogenic returns a number.
An endotoxin figure describes the material as shipped. It says nothing about the material after it has been through your bench with your water and your glassware.

An endotoxin result is a manufacturing hygiene measurement. It is not a safety representation, and no research reagent supplier — including us — can turn it into one.

Reused, generic and altered certificates

These are the recurring patterns. None require expertise to check; most require ten minutes and two open tabs.

Reuse

The same certificate across multiple products. Download several and compare the sample identification field. Identical report numbers or identical sample IDs across different compounds means one document is doing all the work.
The same certificate across time. Pull the certificate for a product now and compare to a copy from six months ago. Same report number and same date means no new lot has been tested regardless of how many lots have been sold.
An identical chromatogram on two different lots. This is conclusive. Two real injections never produce identical baseline noise. If two traces overlay perfectly, including the wobble between peaks, it is one image used twice.
A certificate that outlives its stock. A product that has restocked three times against one certificate has two untested restocks.

Genericness

No laboratory named, or a laboratory with no address, no website, and no findable accreditation scope.
No report number, so nothing can be looked up.
No lot number, so nothing is bound to anything.
“Third-party tested” as a badge on a product page with no document behind it at all. This is the most common case by a wide margin.
An accreditation logo on the certificate for a body the laboratory is not listed with. Accreditation bodies publish searchable directories. The search takes a minute.

Alteration

Alteration is less common than reuse because reuse is easier, but it happens, and it leaves marks.

Try to select the text. In a natively generated PDF, everything is selectable text. If the result values are not selectable but the surrounding labels are, a number has been pasted over as an image.
Look at the numerals against their neighbours. Substituted digits commonly differ subtly in weight, spacing, or baseline alignment from the rest of the line. Zoom to 400% and compare the result field to the header text.
Check for a rectangular patch. Around an altered field there is often a faint block of slightly different background tone, or a compression artefact halo not present elsewhere on the page.
Read the PDF metadata. Document properties will show the producing application. A certificate authored in a laboratory information system and last modified in an image editor is answering the question for you.
Check the dates for coherence. Report date before receipt date, or a test date before the stated manufacturing date, means the document was assembled rather than generated.
An image, not a document. A certificate distributed only as a JPEG or a screenshot has had every one of the above checks removed from you. That is sometimes carelessness and sometimes the intent.

Verification at source beats forensics

All of the above is what you do when the only artifact available is a file the vendor gave you. It is much weaker than the alternative.

Some analytical laboratories serving this market publish results to a public lookup portal: you enter the report or lot number and the record loads from the laboratory’s own database, with the laboratory’s timestamp, on the laboratory’s domain. A vendor cannot edit that. If a vendor uses a lab with a lookup portal, use the portal and skip the forensics entirely. If a vendor is unwilling to tell you which laboratory tested a lot, you have your answer without opening anything.

A ten-minute checklist

Find the lot number on the vial. Is it printed, not handwritten?
Find a certificate for that lot number. Not for the product — for that lot.
Is a laboratory named, with an address and a report number?
Can the report be retrieved from the laboratory rather than from the vendor?
Is there a chromatogram with labelled axes and a full integration table?
Is fill accuracy reported as measured milligrams?
Is endotoxin reported as a number in EU per vial?
Are elemental impurities reported per element as values, not as “complies” or ND?
Are water content, counterion and residual solvents reported, so purity can be converted to peptide mass?
Pull two other lots of the same product. Do the numbers differ the way real production differs?

A vendor failing items 1 through 4 has a documentation problem you cannot work around. A vendor passing 1 through 4 and failing 6 through 9 is testing honestly but partially, which is a different and much more recoverable situation — and worth asking about directly.

One thing this cannot tell you

No certificate, however complete, describes the vial in your hand. It describes a sample. Documentation reduces the distance between the two; it never removes it. Lot-level binding, a printed lot number, two independent laboratories, and results held in a system the seller does not control are all ways of making that distance smaller and harder to exploit. They are not a guarantee, and anyone telling you their paperwork constitutes one has misunderstood their own paperwork.

About us, briefly

Puritas synthesizes, purifies, fills and releases research peptides in its own Florida laboratory. Every lot goes to two independent laboratories, one accredited to ISO/IEC 17025 and one publishing to a public lookup portal, and the panel includes the items most often omitted — measured fill in milligrams, endotoxin in EU per vial, and elemental impurities as values. Held and failed lots are published and never deleted. We do not host our own certificates.

You are welcome to run the checklist above on us. Lot Ledger · Quality & Testing · The Lab

Research use only. Not for human or veterinary use. See Research-Use Terms.