The Panels: What Each One Covers, and What It Does Not

The governing principle, stated once so it can be applied everywhere

Every analytical panel answers a question of the form: how much of THESE analytes, by THIS method, is in this sample. An analyte that is not in the method — no reference standard, no calibration curve, no place in the target list — cannot be quantified and will usually not be reported at all. This is not a shortcoming of any particular laboratory; it is how quantitative analysis works. A calibration curve is built by running known concentrations of an authentic reference standard and fitting the instrument response, and without the standard there is no curve and no defensible number. The consequence runs through every section below and it is the single most useful thing to understand about a COA: the panels define the universe the report can describe, and everything outside that universe is silently absent rather than reported as unknown. A clean full panel means the listed analytes were within limits. It does not mean the sample was clean.

Sources: International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023 · Auwarter V 2009*

Cannabinoid potency — and the core lesson of the K2 era

A potency panel is usually liquid chromatography with ultraviolet or diode-array detection, sometimes with mass-spectrometric detection, calibrated against authentic reference standards for a defined set of cannabinoids: commonly delta-9-THC, THCA, delta-8-THC, CBD, CBDA, CBG, CBGA, CBN, CBC, THCV and CBDV, with the exact set varying by laboratory and jurisdiction. It reports how much of each of THOSE analytes was present. It cannot report a cannabinoid for which the laboratory has no reference standard, and the analytical history of the synthetic-cannabinoid era is the clearest demonstration of why that matters. Auwarter and colleagues, analysing the herbal blends sold as Spice, found that the products contained synthetic cannabinoid receptor agonists that no routine screen was looking for — compounds absent from every target list precisely because they were novel, so that material could pass through testing, screening and clinical assessment while carrying pharmacologically potent compounds nobody had a method for. The lesson generalises exactly to the current novel-cannabinoid market: a product marketed under a cannabinoid name that has no commercially available certified reference standard cannot have been quantified, whatever the report says. If a name appears on a label and a number appears on a COA next to it, the question is which reference standard established that number, and a laboratory can answer it. The red-flags page makes this a checklist item.

Sources: Auwarter V 2009* · International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023 · Babalonis S 2021*

Terpene panels contested

Terpene panels are typically gas chromatography with mass-spectrometric or flame-ionisation detection, against a selected list of analytes — commonly somewhere between ten and forty compounds, depending on the laboratory. Three limitations are worth knowing. First, the list is a selection, and a terpene present in the sample but absent from the list is not reported; the rare and thinly documented compounds discussed on the terpene shelf, such as faurinone and the curcumene isomers, are exactly the ones least likely to be on any panel and least likely to have a certified standard available. Second, there is no meaningful total: a terpene total on a report is the sum of the listed analytes, not the sample's total volatile content, and two laboratories with different lists will report different totals for identical material. Third, terpene results are the most sample-handling-sensitive numbers on any COA, because the analytes are volatile by definition and losses during grinding, weighing, storage and transfer are real and systematic; the most volatile compounds, alpha-pinene and myrcene above all, are the ones most easily lost between the field and the injector. Treat a terpene figure as a lower bound on what was in the plant and as a statement about the laboratory's list as much as about the sample. Note also that isomers matter and are frequently not resolved: an ocimene figure without an isomer assignment, or a curcumene figure without one, has discarded information, and chiral compounds such as limonene, linalool and alpha-pinene are reported as a single combined figure unless a chiral column was used.

Contested — caveat. Panel sizes and analyte lists are laboratory-specific and this section describes the general convention rather than any particular laboratory's method. Terpene totals are not comparable between laboratories with different lists.

Sources: Compiled from published cannabis 2026* · International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023

Residual solvents — USP 467 and the ICH classes

Residual solvents are the volatile organic compounds left in a product from processing. The relevant framework is not a cannabis-specific invention: it is USP General Chapter 467, Residual Solvents, which operationalises the ICH Q3C guideline, and it exists because the pharmaceutical industry had to solve this problem first. ICH Q3C sorts solvents into three classes by toxicological concern, and the class of a finding is the first thing to read. Class 1 solvents are those to be avoided: known human carcinogens or compounds of severe environmental or toxicological concern, including benzene and carbon tetrachloride, with limits set in the low parts-per-million and in some cases lower. Class 2 solvents are to be limited: compounds with known but less severe toxicity, each with its own permitted daily exposure and concentration limit. Class 3 solvents are those of low toxic potential, where a generous default limit applies. What a CLASS 1 FINDING MEANS is therefore specific and serious: a class 1 solvent is not supposed to be in the product at all, so its presence is evidence about the process rather than a quality-margin question. It indicates either that a class 1 solvent was used, or that a solvent used was contaminated with one, or that one was generated. Two further points. The residual-solvent panel, like every panel, covers a defined list: a solvent not on the list is not looked for. And a residual-solvent result is only meaningful for the matrix tested, so an input COA showing clean solvents does not describe a finished product that went through further processing.

• Class 1: avoid. Known carcinogens and severe hazards. A finding is a process finding, not a margin question.
• Class 2: limit. Each has a permitted daily exposure and a concentration limit.
• Class 3: low toxic potential, generous default limit.
• The panel covers a list. Off-list solvents are not looked for.
• Test the finished matrix, not only the input.

Sources: United States Pharmacopeial Convention 2026 · International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2021

Heavy metals — and why four is not enough for converted products contested

The standard heavy-metals panel in this industry covers four elements: lead, cadmium, arsenic and mercury. The framework behind the approach is the USP elemental-impurities chapters, and the four are chosen because cannabis is an efficient accumulator of soil metals and because these four are the classic toxicological priorities. For plant material grown in soil, a clean four-element result is genuinely informative. For a converted or synthesised cannabinoid product it is much less so, and this is a structural gap rather than a technicality. Chemical conversion processes can introduce process metals — metals that were part of the chemistry rather than part of the soil — and a four-element panel does not look for them, because the panel was designed for an agricultural contamination model and not for a manufacturing one. This shelf does not name the metals in question, because naming which metals to expect from which process is route information and this shelf does not carry route information. The actionable point for a buyer or a formulator does not require the names: for a product whose cannabinoid was made rather than extracted, ask whether the metals panel was an expanded elemental screen or the standard four, and understand that a standard four-element pass is silent about process residues. An expanded multi-element screen is the right question to ask, and a laboratory can say whether it offers one within its accreditation scope.

Contested — caveat. Panel composition varies by jurisdiction and some programmes require more than four elements. The general point stands: verify which elements were actually determined rather than assuming a metals panel is comprehensive.

Sources: United States Pharmacopeial Convention 2026 · Babalonis S 2021* · Compiled from United States state cannabis 2026*

Pesticides — a target list, so an off-list compound is invisible

Pesticide screening is multi-residue analysis by liquid or gas chromatography with tandem mass spectrometry against a defined target list of active ingredients, with an action limit for each. Jurisdictional lists range from a few dozen compounds to several hundred and they do not agree with each other. The blind spot is structural and unavoidable: multi-residue methods identify and quantify against reference standards for listed compounds, so a pesticide that is not on the list is not detected, not reported, and not implied by a clean result. Three practical corollaries. A pesticide panel pass means the listed residues were below their action limits in that sample. It does not mean no pesticide was used. And a product tested against a short jurisdictional list and then sold into a jurisdiction with a longer one has not been tested to the destination standard. For anyone specifying testing, the list is the specification — ask which list, count the analytes, and check that the compounds of concern for the growing region are on it.

Sources: Compiled from United States state cannabis 2026* · International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023

Mycotoxins and microbials contested

Mycotoxin panels typically determine the four aflatoxins — B1, B2, G1 and G2, usually with a limit on total aflatoxins and sometimes a separate tighter limit on B1 — together with ochratoxin A. These are fungal metabolites, so they are markers of a fungal history: mycotoxins can persist after the organism that produced them is gone, which means a clean microbial result and a mycotoxin finding are not contradictory, and remediation that kills mould does not remove the toxin it already made. Microbial panels combine non-specific counts with specified organisms: total aerobic microbial count, total yeast and mould count, bile-tolerant Gram-negative bacteria, and specified pathogens including Escherichia coli, Salmonella species and, in many jurisdictions, named Aspergillus species — commonly A. flavus, A. fumigatus, A. niger and A. terreus, which are singled out because of inhalational risk in immunocompromised users of inhaled products. Two limitations to carry forward. Culture-based and molecular methods answer subtly different questions — a molecular method may detect nucleic acid from organisms that are no longer viable, and a culture method may miss organisms that are viable but not culturable under the chosen conditions — so the method reference matters for interpretation. And microbial results have the shortest useful life of any panel, because a population can grow after testing if the water activity permits it, which is why the next section exists.

Contested — caveat. Specified organisms, action limits and permitted methods differ substantially between jurisdictions and are revised often. The organism lists here are the common pattern, not a universal requirement.

Sources: Compiled from United States state cannabis 2026* · Compiled from published cannabis 2026*

Water activity and moisture — the two numbers that predict the future contested

Every other panel is retrospective: it says what was in the sample when it was tested. Water activity is the one number on a flower COA that is predictive, because it says whether microbial growth can occur in the package going forward. Water activity is the ratio of the vapour pressure of water in the material to that of pure water at the same temperature, expressed from 0 to 1, and it measures AVAILABLE water rather than total water. Below roughly 0.60 to 0.65 most moulds cannot grow; a common regulatory ceiling for cannabis flower is 0.65. Moisture content is the separate, related measurement of total water as a percentage of mass, commonly capped somewhere in the 12 to 15 percent region, and it matters for a second reason as well: the dry-weight basis conversion on the total-THC page needs it. The two numbers are not substitutes. Material can be at an acceptable moisture content with unevenly distributed water and a local water activity that supports growth, and material can be dry by water activity while being too dry for quality. For a buyer, a flower COA with a passing microbial panel and no water-activity figure has told you about the past and nothing about the shelf.

Contested — caveat. The 0.65 water-activity ceiling and the 12 to 15 percent moisture range are common regulatory values compiled across jurisdictions, not a single universal standard. Verify the limit in force.

Sources: Compiled from United States state cannabis 2026* · United States Department of Agriculture 2021*

LOD, LOQ, and why non-detect is a statement about the method

Three result qualifiers appear on COAs and they are routinely misread as one thing. The limit of detection, LOD, is the lowest concentration at which the analyte can be reliably distinguished from a blank — the method can tell something is there but cannot say how much, and the ICH validation guideline is the standard source for how it is established. The limit of quantitation, LOQ, is the lowest concentration that can be measured with acceptable precision and accuracy — the floor for a reportable number, and always above the LOD. ND, not detected, means the analyte was not detected AT OR ABOVE THE DETECTION LIMIT OF THIS METHOD. It does not mean zero, it does not mean absent, and it does not mean safe. A result of ND on a method with a detection limit of 1 part per million and a result of ND on a method with a detection limit of 1 part per billion are different statements by a factor of a thousand, and the COA distinguishes them only if it prints the limits. This is precisely why the LOD and LOQ columns are on the anatomy checklist, and why a report that gives ND without limits is much weaker than it looks. The corollary for the middle ground: a result reported as detected but below the quantitation limit is a real detection, and treating it as a non-detect is an error in the direction of false reassurance. In practice: read ND as "below this method's floor", find the floor, and decide whether that floor is low enough for the decision you are making.

• LOD: the analyte can be distinguished from a blank. Presence, not amount.
• LOQ: the lowest concentration reportable as a number with stated precision. Always above the LOD.
• ND: not detected at or above this method's detection limit. Not zero, not absent.
• Detected below LOQ: a real detection that cannot be quantified. Not the same as ND.
• An ND with no printed detection limit is an uninterpretable result.

Sources: International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023 · International Organization for Standardization 2017

The unidentified peak: the most important thing on a converted-cannabinoid COA, and usually absent contested

This section is the analytical heart of the shelf. A chromatogram separates a sample into peaks; a quantitative method assigns the peaks it was calibrated for and reports concentrations for those. Anything else that came off the column is still there in the trace, as a peak with a retention time and an area and no name. On a report for extracted plant material, unassigned peaks are usually minor plant constituents and are of modest interest. On a report for a product whose cannabinoid was made by chemical conversion rather than extracted, the unassigned peaks are the single most informative feature of the entire document — because the conversion literature establishes that acid-catalysed cyclisation and isomerisation of cannabidiol yield a MIXTURE of products rather than one clean compound, and because those additional products are, by definition, compounds for which no reference standard exists and no pharmacology has been done. Marzullo and colleagues document in the chemical literature that such cyclisation gives multiple products; this shelf cites that work for that fact alone and reproduces nothing of any procedure. Babalonis and colleagues, reviewing delta-8-THC products, identify exactly this as the safety gap: widely available products made by conversion, sold without characterisation of the accompanying reaction products. So the questions that matter on such a report are: how much of the total chromatographic area was assigned to identified analytes, how much was not, and what is the largest single unassigned peak. And the reason this section is difficult to act on is that MOST COAS DO NOT REPORT ANY OF IT. A potency report prints a table of named cannabinoids and their concentrations. It does not print the chromatogram, it does not print total area, it does not print an unassigned area percent, and there is usually no field in which an unidentified major component could appear. A product can therefore show a clean, plausible, internally consistent potency table while a substantial fraction of the material is uncharacterised, and nothing on the certificate would indicate it. What to ask for, in order: the chromatograms rather than only the summary table; a statement of total identified versus unidentified area; a full-scan mass-spectrometric screen rather than a targeted potency method; and, for anything inhaled, characterisation of the finished formulation rather than of the input. A laboratory operating in good faith can answer all four. A refusal to answer is itself information.

• Ask for the chromatogram, not only the results table.
• Ask for identified versus unidentified area percent.
• A targeted potency method cannot see an unknown; a full-scan screen can at least flag one.
• For converted-cannabinoid products, an unassigned major peak is the finding that matters most and is the one least likely to be printed.
• No reference standard means no quantitation; no pharmacology on the by-products means no basis for a safety claim about them.

Contested — caveat. The claim that conversion chemistry yields mixtures is established in the chemical literature and is cited by reference only. The specific identity, quantity and toxicology of by-products are product- and process-specific and are not characterised here; that absence of characterisation is the point of the section, not a gap in it.

Sources: Marzullo P 2020* · Babalonis S 2021* · Auwarter V 2009* · International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2023