Two bottles sit on the same shelf. One says the extract inside is standardized to 24 percent flavonol glycosides. The other says it is a 10:1 extract. Both numbers look like specifications. Neither one is.
Here is the finding that should change how you read them. In 2016, a research team bought the same eight herbs from three different commercial suppliers each, extracted every sample under identical conditions in its own laboratory, and sent the extracts to an outside analytical company for standard marker testing. In licorice, the marker compound came back at 3.2 percent of sample weight from one supplier, 1.37 percent from the second, and 0.14 percent from the third. Same plant. Same extraction method. A spread of more than twenty-fold.
That is not a story about one bad supplier. It is a story about how little sits behind the numbers on an extract label.
This article is about extracts as a product form: what the figures claim, who verifies them, and how a product can satisfy a published standard and still be deliberately faked. The individual herbs, and the science of confirming a plant is what it says it is, belong to other articles here.
“Standardized” has no legal definition in the United States
Start with the agency most consumers would expect to settle the question. The NIH Office of Dietary Supplements publishes a background page on botanical dietary supplements, last updated December 11, 2020. It describes standardization as a process manufacturers may use so that all of their batches come out similar, and explains that it involves identifying and measuring specific chemicals known as markers.
Then it says the part that no label implies. On standardization, the ODS page states: “In fact, there is no legal or regulatory definition of the term in the United States.”
The same page adds a second limitation that matters more than the first. Ideally, it says, the chemicals chosen for standardization would also be the ones responsible for the botanical’s effect in the body. But the constituents responsible for the effects of most botanicals are not known.
Put those two sentences together and a label reading “standardized” says something narrower than it sounds. It says the company measured a chemical and adjusted the batch toward a target. It does not say which chemical, by what method, at what target, or whether anyone outside the company confirmed it.
What federal law requires a supplement maker to test
Supplement manufacturing is not unregulated. It is regulated in a specific way that does not reach the claims on the front of the bottle.
The operative testing rule is 21 CFR 111.75, part of the good manufacturing practice regulations for dietary supplements. Read in plain language, it requires four things.
First, for each incoming component that is itself a dietary ingredient, the firm must run at least one appropriate test or examination to verify that ingredient’s identity. One test. Identity only. Not strength, not the extract ratio, not the marker percentage.
Second, for components that are not dietary ingredients, such as fillers and capsule shells, the firm may rely on a supplier’s certificate of analysis rather than its own test, provided it first qualified that supplier, the certificate names the methods, limits and actual results, and the firm rechecks periodically.
Third, for finished product, the rule requires the firm to verify that a batch meets product specifications for identity, purity, strength and composition, along with limits on contamination that could adulterate it. That check can cover every finished batch, or only a subset of batches chosen through what the regulation calls a sound statistical sampling plan.
Fourth, and this is the hinge the whole article turns on: those are the manufacturer’s own product specifications. US law requires a product to match the number the company wrote down for itself. It does not require the number to be meaningful, the ratio to be real, or the marker to be the thing that does anything in a human body.
So “10:1” and “standardized to 5 percent of compound X” are not regulated terms being enforced by an agency. They are manufacturing conventions, written by the seller, for the seller’s own specification sheet.
A marker compound is not an active ingredient
The conceptual heart of extract labeling is the gap between a marker and an active constituent. A marker is a chemical with a known structure that can be measured reliably. An active constituent is a chemical that produces the effect. The industry measures the first because it can, and treats it as a stand-in for the second.
The 2016 study described above, published in PLoS One, tested whether that stand-in holds. Eight botanicals, three commercial sources each, one extraction protocol, marker levels measured commercially, and bioactivity measured in four matched laboratory assays covering bacteria, yeast, a herpes virus and immune cell signaling.
The markers moved wildly between suppliers of the same herb. St. John’s wort’s hyperforin came in at 0.059 percent, 0.026 percent and 1.47 percent, a spread of about twenty-five-fold. Black pepper’s piperine measured 0.082 percent, 0.008 percent, and nothing detectable at all in the third sample. Uva-ursi’s arbutin ran from 0.351 to 1.33 percent.
The correlation with activity was worse. Eucalyptol, the marker for eucalyptus, tracked antibacterial activity reasonably well. Arbutin and coumarin showed weaker relationships the authors described as questionable. For the rest, including hyperforin, piperine, echinacea’s three measured markers and astragalus’s astragaloside I, marker level did not predict measured activity. Pooled across the whole set, the difference between marker level and bioactivity reached statistical significance at p less than 0.04, and the difference between the three commercial sources at p less than 0.0019.
One clear hit out of eight is the number worth carrying. Two more were ambiguous. Five showed nothing.
Two caveats. Five of the six authors worked at a naturopathic medical college, two of them also at Arizona State University’s Biodesign Institute, so this is not an industry result, and the finding cuts against the trade’s own quality control practice. And the assays were laboratory assays. The study does not show these extracts work in people. It shows the number on the bottle does not track even a test tube.
How an extract can pass the test and defeat it at the same time
The sharpest material on extracts comes from the Botanical Adulterants Prevention Program, a joint effort of the American Botanical Council, the American Herbal Pharmacopoeia and the University of Mississippi’s National Center for Natural Products Research. Its January 2018 bulletin on ginkgo leaf extract is a case study in how a specification becomes a target for fraud.
Commercial ginkgo leaf extract is sold to a trade specification the bulletin describes as 24 percent flavonol glycosides and 6 percent terpene lactones, known in the trade as 24/6. Flavonols are cheap and widely available. Adding pure rutin or quercetin, or a flavonol-rich extract of Japanese sophora or buckwheat, raises the first number without adding any ginkgo.
The United States Pharmacopeia responded by tightening its monograph. Since 2012, according to the bulletin, the USP method has required that after chemical hydrolysis the kaempferol-to-quercetin peak ratio be not less than 0.7 and the isorhamnetin-to-quercetin ratio not less than 0.1. The powdered extract monograph also sets a limit of 4 percent rutin and 0.5 percent quercetin in unhydrolyzed extract.
And then the bulletin reports what happened next: “some manufacturers have found ways to fraudulently spike ginkgo extracts while complying with the USP criteria.”
Complying with the standard and defeating it are not opposites here. They are the same operation, performed by someone who read the published test method carefully.
Reported adulteration rates span a wide range, and the bulletin is candid that no reliable comprehensive picture exists. Individual studies found 14 percent (three of 22 samples), 44 percent (11 of 25), 48 percent twice (10 of 21 and 14 of 29), and at the top of the range 94 percent, meaning 33 of 35 products with raised rutin or quercetin, depleted ginkgo markers or no ginkgo at all. Within that same set of 35, the raised rutin and quercetin finding alone accounted for 74 percent. One count rose to 76 percent when the isoflavone genistein was treated as an indicator of sophora, though researchers disagree about that: some report genistein as a genuine minor constituent of ginkgo leaf, most ginkgo chemistry studies failed to detect it, and a multi-laboratory study produced contradictory results. Another cited study found adulteration in 27 percent of bulk extracts and 71 percent of finished products.
A 2023 peer-reviewed paper in the Journal of Natural Products, from the same program’s chief scientist, generalizes the mechanism across botanicals. Adulteration, it explains, usually works by supplying ingredients that appear to comply with specifications for specific marker compounds while exploiting the lack of specificity of the test used to measure them. Most pharmacopeial methods still turn on the presence or absence of one or a few chemical markers, and many of those markers are phenolic compounds that occur in many plants and cannot identify a species.
The paper documents four recurring shapes.
- Same plant, wrong part. Asian ginseng root extract diluted with extract of the leaves, which carry some of the same ginsenosides in different ratios.
- Cheaper plant, same marker. Goldenseal extract standardized to berberine and hydrastine, topped up with berberine from coptis root. Bilberry and elderberry anthocyanin extracts cut with black rice, black bean, mulberry or purple carrot. Cranberry gets the same treatment with hibiscus, whose characteristic anthocyanins were isolated from an extract sold as cranberry.
- Pure compound spiking. Undeclared synthetic curcumin added to turmeric extract, rutin added to ginkgo leaf extract, ellagic acid added to pomegranate peel extract. The paper calls fortification with a pure or highly purified marker one of the hardest kinds of adulteration to detect.
- Dyes that fake the reading. A cocktail of four food dyes added to St. John’s wort extract so that a spectrophotometer returns a hypericin-like number. Brilliant blue absorbs light at 590 nanometers, the same wavelength at which hypericin is measured. The other dyes make the powder look right.
There is a fifth, and it is the one that defeats identity testing outright. Adulterators may dilute with spent material, an ingredient from which part or all of the valuable constituents has already been removed. Spent material still carries every characteristic marker, so it passes an identity test despite containing almost nothing. The paper notes weak fingerprints of this kind have been reported in extracts of milk thistle seed, ginkgo leaf, echinacea root and herb, cranberry fruit and St. John’s wort.
One disclosure is owed. The American Botanical Council is a non-profit funded substantially by the herbal industry, and the prevention program is underwritten by industry sponsors. Its analytical work is rigorous and peer-reviewed. It is also not a regulator and has no enforcement power. It publishes bulletins. That is the extent of it.
What the ratio claims, and what sets it
An extract ratio describes how much raw plant went in for each unit of finished extract. A 10:1 label says ten parts of dried herb produced one part of extract by weight.
The ODS page describes the process plainly: an extract is made when the botanical is soaked in a specific liquid such as water or alcohol, and tinctures are sold at strengths expressed as botanical-to-extract ratios. Nothing in US law defines the ratio, dictates the solvent, or requires the printed ratio to match the batch.
Solvent choice does move the chemistry, though the public evidence is thinner than you would expect. In the 2016 marker study the researchers used different solvent mixtures for different herbs, from about a quarter ethanol for some roots to roughly two-thirds for a bark. The ginkgo bulletin supplies a documented case of that being abused: Chinese authorities acted against companies alleged to have used 3 percent hydrochloric acid in place of costlier organic solvents such as ethanol and acetone, which yielded extracts with higher amounts of free quercetin, kaempferol and isorhamnetin.
That is one case, in one herb, reported by one bulletin. This article could not locate a source systematically comparing extraction solvents for a single botanical, so claims comparing carbon dioxide, ethanol and hexane extracts of a given herb are not established here either way.
The same honesty applies to pharmacopeial monographs. The 2023 Journal of Natural Products paper describes the USP, the European Pharmacopoeia and the Chinese pharmacopeia as official compendia carrying authentication methods, and the American Herbal Pharmacopoeia as an unofficial one. USP’s public page on standards in development lists monographs under way for 20 botanicals and 11 non-botanical ingredients and says only that its scientists are working on them. This article could not establish how USP or AHP monographs interact with US supplement law, and does not assert that they bind anyone.
The evidence attaches to a named extract, not to a plant
There is a reason the research literature keeps naming specific preparations.
The 2014 Cochrane review of echinacea for colds pooled nothing in its main analysis, and said why. Twenty-four double-blind trials with 4,631 participants produced 33 comparisons using a variety of preparations based on different species and different parts of the plant, and the reviewers refrained from pooling because of strong clinical heterogeneity. The review opens by noting that most consumers and physicians are unaware that products sold under one botanical name differ appreciably in composition, largely because of variable plant material, extraction methods and added components.
The Ginkgo Evaluation of Memory study is the mirror image. It tested one characterized preparation at one dose: 120 milligrams of a ginkgo extract twice daily, in 3,069 community volunteers aged 75 and over, across five US academic medical centers, with a median follow-up of 6.1 years. The result was null, with an all-cause dementia hazard ratio of 1.12 and a 95 percent confidence interval from 0.94 to 1.33.
Both of those findings carry the same lesson for a shopper. A bottle labeled with the species name has not been tested. A bottle with a different ratio, a different solvent or a different marker target is not the preparation that was studied, even when the Latin name matches.
Processing is what makes this hard to check later. A 2017 paper in Food Chemistry, from university chemists working with co-authors at Procter and Gamble and at a supplement manufacturer, makes the point directly: milling, drying and extraction strip away the physical characteristics used to tell species apart, and the material then moves through a multilevel supply chain before reaching a finished product. By the time a powder is in a capsule, the only description of it is paperwork.
How to read an extract label
Four questions that tell you what a number is worth
This is a reading framework, not a shopping list. Ask these of any figure printed on an extract.
1. Does the number name a compound, or a category? “Standardized to 24 percent flavonol glycosides” names a measurable class. “Standardized extract” with no analyte named tells you nothing at all, because no law defines the word.
2. Does it name a test method? A percentage without a method is uninterpretable. The ginkgo case shows two labs can both be right and disagree, and the fraud cases show a method can be gamed once it is published.
3. Is the number a ratio or a content figure? A ratio such as 10:1 describes a manufacturing step. A content figure describes what is in the finished powder. Only the second can be checked against an assay, and even then only against the company’s own specification.
4. Does any party outside the seller stand behind it? Under 21 CFR 111.75 the answer is often a supplier certificate the manufacturer itself accepted. A third-party verification mark is a different thing, and other articles here cover what those marks certify.
The decisions that belong to a clinician
ODS states plainly that you should always talk with your health care provider about botanical and other dietary supplements, and warns that “natural” is not a guarantee of safety, noting that concentrated preparations such as peppermint oil can be toxic when used incorrectly. Concentration is exactly what an extract does.
The situations where that conversation stops being general advice and becomes a real decision are worth naming. If you take prescription medicines, the relevant fact from the 2016 marker study is that two bottles of the same labeled herb differed roughly twenty-five-fold in the constituent chosen to represent it, so reasoning about dose from the label is unreliable. The same applies before planned surgery, during pregnancy or breastfeeding, and in liver or kidney disease, where what matters is the chemical load rather than the plant’s reputation. Bring the physical container, not a photograph of the front panel: the ratio, the marker and the plant part are the only description of the product that exists.
The one thing to take from all of this
The number on an extract label describes a manufacturing process. The manufacturer wrote it, the manufacturer’s own specification is what it gets checked against, and it is usually keyed to a chemical nobody has shown is responsible for the effect. Federal rules require one identity test on the incoming ingredient and conformance to a self-set spec on the finished batch. They do not require the ratio to be true, the marker to matter, or the extract to resemble the preparation used in any published trial.
None of that makes extracts worthless. It is a reason to read the figures on the front panel as marketing copy with a decimal point, and to treat the evidence as attaching to the specific named preparation a study tested, when a study exists at all.
Sources
- National Institutes of Health, Office of Dietary Supplements. “Botanical Dietary Supplements: Background Information.” ods.od.nih.gov
- Cornell Law School, Legal Information Institute. “21 CFR 111.75 – What must you do to determine whether specifications are met?.” law.cornell.edu
- American Botanical Council / ABC-AHP-NCNPR Botanical Adulterants Prevention Program (copy hosted by the Center for Science in the Public Interest). “Botanical Adulterants Bulletin: Adulteration of Ginkgo biloba Leaf Extract, January 2018.” cspi.org
- PLoS One (PubMed). “Ruiz GG, et al. "A Lack of Bioactive Predictability for Marker Compounds Commonly Used for Herbal Medicine Standardization." PLoS One, 2016. PubMed. DOI: 10.1371/journal.pone.0159857.” pubmed.ncbi.nlm.nih.gov
- Journal of Natural Products (PubMed). “Gafner S, et al. "Botanical Ingredient Forensics: Detection of Attempts to Deceive Commonly Used Analytical Methods for Authenticating Herbal Dietary and Food Ingredients and Supplements." J Nat Prod, 2023. PubMed. DOI: 10.1021/acs.jnatprod.2c00929.” pubmed.ncbi.nlm.nih.gov
- Cochrane Database of Systematic Reviews (PubMed). “Karsch-Volk M, et al. "Echinacea for preventing and treating the common cold." Cochrane Database Syst Rev, 2014. PubMed. DOI: 10.1002/14651858.CD000530.pub3.” pubmed.ncbi.nlm.nih.gov
- JAMA (PubMed). “DeKosky ST, et al. "Ginkgo biloba for prevention of dementia: a randomized controlled trial." JAMA, 2008. PubMed. DOI: 10.1001/jama.2008.683.” pubmed.ncbi.nlm.nih.gov
- Food Chemistry (PubMed). “Raja HA, et al. "DNA barcoding for identification of consumer-relevant mushrooms: A partial solution for product certification?" Food Chem, 2017;214:383-392. PubMed. DOI: 10.1016/j.foodchem.2016.07.052.” pubmed.ncbi.nlm.nih.gov
- United States Pharmacopeia. “Dietary Supplements and Herbal Medicines: Standards in Development.” usp.org