State of Analytical Evidence in Research Peptides

A certificate stating 99.7% purity is evidence that a figure was published — not that any vial is pure. Nothing here is a claim about the quality of any vendor’s goods, including our own.

Every analytical signal in every certificate is coded on three levels: absent, named — the technique appears — or reported, meaning a result is shown. Most of what follows lives in the distance between the second and the third.

The corpus holds 9,646 certificates collected from the public pages of 278 eligible vendors in the United States and Europe, signed by 41 identifiable laboratories. 8,936 of them (92.6%) carry a laboratory we could attribute.

The cleanest test is to hold the vendor constant. Among the 25 vendors that use both the dominant issuer and at least one other laboratory, catalogue and collection are the same; only the signature changes. Chromatography appears in 3.4% of the dominant issuer’s certificates and 96.6% of the others’ — a 93.2-point gap that no plausible measurement error could produce.

This is a controlled comparison, not a randomised one: a vendor may route one class of product, or a fuller test panel, to one laboratory. That would not create a gap this size, but it means the share attributable purely to the issuer sits somewhat below the headline figure. And the absence of a chromatogram is a property of the report, not proof the analysis was never run.

Read one issuer at a time, the pattern is a set of report templates. The table below measures what each laboratory’s published certificates show — which reflects what vendors ordered and what the report format displays, not what the laboratory is able to do. A vendor that orders a purity test alone gets a purity-only certificate from any issuer.

For most signals the two numbers sit close: a certificate that names chromatography usually shows its result. Purity is the exception. 87.1% of certificates state a purity, but 74.9% report a measurement. The difference is 739 certificates stating a specification — >99%, a threshold claimed rather than a number an instrument produced — and 440 that name purity with no value at all.

“Publishes nothing” means nothing recoverable by this collection method; evidence behind a login or sent on request is not counted. Among the vendors that do publish, the screen is gentle until it reaches the requirements that cost money.

Earlier versions of this study published five comparisons between Europe and the United States. Four do not survive — not because the data changed, but because the study now measures how often its own detectors miss what they look for.

This study has been audited eight times and 27 defects have been found and corrected. Publishing them is not a disclaimer; it is the reason the remaining figures can be trusted.

Certificates were collected from vendors’ public pages, de-duplicated by content, read from native text or OCR, and coded by a single pipeline against a written coding manual. Where code and manual disagree, the manual governs. Sensitivity was measured by sampling negatives, not only positives, which is why two signals are published as floors.

Laboratory accreditation is recorded against ISO/IEC 17025; elemental impurities follow ICH Q3D(R2); endotoxin and microbiological limits follow the United States Pharmacopeia. Guidance on certificate content is published by the FDA and the EMA.

Reuse is free under CC BY 4.0: the data, the figures and the text. Credit is all that is asked, and every downloaded figure already carries it. A backlink is not required.