Limits, Methods and Gaps: Interrogating the Contaminant Pages of a CBD Certificate of Analysis

A certificate of analysis arrives as a single PDF, but it is really four or five documents bound together. The cannabinoid page answers the commercial question — how much CBD, how much THC, what the ratio looks like — and it is read by everyone from the procurement manager to the managing director. The pages that follow, covering residual solvents, pesticides, heavy metals and mycotoxins, answer the compliance question, and on many buying desks they are skimmed at best. That imbalance is understandable; potency drives margin. It is also where a considerable amount of avoidable exposure sits.

Two pages, two different conversations

Potency is a number you can sell. Contaminants are a number you may have to defend, months later, to a retailer's technical team, to an auditor, or to a trading standards officer who has picked up a sample from a shop shelf in the Midlands. The evidence required at that point is not the potency figure. It is a batch-specific contaminant result, produced by a named method, at a limit of quantification low enough to be meaningful.

Trade buyers who treat the contaminant section as an appendix tend to discover its importance at the least convenient moment. Those who treat it as part of the purchase specification find that it changes which suppliers reply to their enquiries at all — often a useful filter in itself.

Panel by panel

Residual solvents

Extraction route dictates the likely residue profile. Supercritical CO2 leaves little behind, but it is frequently followed by ethanol winterisation, so ethanol is the usual candidate. Hydrocarbon extraction raises questions about butane, propane and pentane. Ethanol-only routes may show traces of the solvent itself together with whatever was used in clean-up.

Ask which solvent classes the panel covers and whether the classification follows the ICH Q3C framework of class 1, class 2 and class 3 solvents. A panel that reports ethanol alone tells you the laboratory ran a narrow method; it does not tell you the oil is clean.

Pesticides

This is the panel most often run in a shortened form. Multi-residue methods vary considerably in scope, and a 60-analyte screen and a 200-plus-analyte screen both produce a tidy table of "not detected" results. The questions to ask are which analyte list the method covers, whether it includes the compounds most associated with hemp cultivation, and whether the reporting limit for each analyte sits below the applicable maximum residue level.

Heavy metals

Lead, cadmium, arsenic and mercury are the standard four. Less commonly reported are nickel and chromium, which can appear where stainless steel equipment has been abraded and which a number of technical buyers now request as a matter of course. As with pesticides, the value of the result depends entirely on the reporting limit.

Mycotoxins

Aflatoxins B1, B2, G1 and G2, together with ochratoxin A, are the expected entries. Hemp biomass stored warm or damp is a plausible host, and the mycotoxin panel is the one most likely to be missing altogether from certificates issued for oils that were never intended for the UK market. Its absence is not proof of contamination; it is proof of an incomplete file.

The two columns that decide whether a result means anything

A row reading "not detected" is reassuring only when read alongside the method and its limit of quantification. If the LOQ for a given pesticide sits above the relevant maximum level, then "not detected" means the method could not have found a non-compliant sample even had one been submitted. The result is not false, but neither is it evidence.

Two checks follow. First, that the method is identified by a recognised reference — a pharmacopoeial procedure, a published multi-residue method, or an in-house method validated against one. Second, that the laboratory's scope of accreditation, as listed by UKAS, actually includes that method for that matrix. A laboratory may be accredited for cannabinoid potency and not for mycotoxins, and the certificate will not always make the distinction obvious.

It is also worth noting that the maximum levels applied to CBD products in the UK derive from assimilated food law and are expressed per kilogram of product. A result reported on the extract rather than the finished oil requires arithmetic before it can be compared. Buyers who skip that step occasionally celebrate a compliant number that is nothing of the sort.

Does the certificate belong to the lot?

Three details connect a certificate to a consignment: the sample identification, the date of receipt by the laboratory, and the batch or lot code, which should match the supplier's paperwork rather than merely resemble it. Certificates carrying no sample identifier, or dated before the material was manufactured, are not batch evidence however impressive the letterhead.

Where a supplier buys from a third-party processor, ask whether the certificate relates to the oil as received or as further diluted or blended. Blending changes contaminant concentration as surely as it changes potency, and a panel run on the concentrate does not describe the finished article.

Writing the specification into the contract

Contaminant requirements belong in the technical schedule, not in an email chain. A workable clause identifies the analytes, the maximum levels, the method family, and the maximum acceptable limit of quantification expressed as a proportion of the limit. It states the sampling point, requires a batch-specific certificate issued by an accredited laboratory, and gives the buyer a right to have a retained sample analysed independently, with the cost falling on whichever party's result proves wrong.

Add a notification window for out-of-specification results — short enough to matter, long enough to be realistic — and define what happens to material already dispatched. Silence in the contract on that point is routinely read as permission.

A short review sequence

Read the analyte list before the results. Check the LOQ against the limit. Confirm that the accreditation scope covers that method for an oil matrix. Match sample identifier and date to the delivery documents. Only then look at the numbers.

None of this requires a laboratory background. It requires a buyer who treats the second half of the certificate as part of the product, and a supplier who expects to be asked.