Two photographs arrive from a customer, both of them HPTLC plates under ultraviolet light, and the bands do not line up. The plate on the left is the reference the lab keeps for that species. The plate on the right is the batch in the warehouse. Same number of bands, roughly, but the intensities are wrong, and one band sits higher than it should.
The email asks the obvious question: is this material adulterated? The honest answer is that the photograph cannot tell you, and that the two most likely explanations point in completely different directions.
Two explanations, and only one is a supply problem
A fingerprint that does not match has two families of cause, and they are not equally common.
The first is analytical. Thin-layer chromatography is sensitive to almost everything: how much was spotted, how the plate was saturated, which batch of plates was used, how far the solvent front ran, how the plate was derivatised and how long it sat before the camera came out. Two laboratories running the same method on the same material can produce plates that look different enough to start an argument, especially at the edges of the plate where the solvent front behaves badly.
The second is material. A different species, a different plant part, a different harvest, a different extraction, or something added to the powder after extraction.
The professional response is to eliminate the first family before accusing anyone of the second, because the first family is cheap to check and the second one ends relationships.

HPTLC is a photograph, not a measurement
The method's strength and its limitation are the same property: it shows you a pattern.
A plate separates a sample into bands by how far each compound travels up the silica in a solvent system. The pattern of bands, their positions and their relative intensities, is a fingerprint. It is excellent at answering whether two materials are the same kind of thing. It is not able to answer how much of a given compound is present, because band intensity under UV is not proportional to mass in any way a buyer can rely on.
This matters because disputes about fingerprints are often really disputes about assay. A buyer sees a weaker band and concludes the material is diluted. The lab sees a weaker band and knows only that less was spotted or that the plate ran hot. Quantitative work needs a column, a detector and a calibration curve, which is a different method with a different cost.
So the first question about any mismatched plate is what the plate was asked to do. If it was asked to confirm identity, a mismatch means the identity is in question. If it was asked to confirm strength, it was asked the wrong question.
The ginkgo case, because it is the clearest one in the trade
No botanical illustrates the split between a number that can be faked and a number that cannot as well as ginkgo, and the reason is chemical.
The reference specification for that extract is written as three figures. Flavone glycosides at 22 to 27%, terpene lactones at 5 to 7%, and ginkgolic acids below 5 ppm. Those three are not three views of the same quality. They are three different defences.
The flavone glycoside figure is measured by ultraviolet absorbance after acid hydrolysis, which converts the glycosides to their aglycone cores and reads the total. It is a robust, standardised, pharmacopoeial method, and it counts anything that hydrolyses to a flavonol and absorbs at that wavelength.
The terpene lactone figure is measured by HPLC with evaporative light scattering or mass detection, because ginkgolides and bilobalide have no useful ultraviolet absorption. It is a harder measurement, on a more expensive instrument, and it counts compounds that occur in ginkgo and essentially nowhere else in the commercial trade.
Now put those two facts together with the economics of the leaf. Ginkgo leaves naturally carry more terpenes relative to flavonol glycosides than the specification requires, so hitting 6% terpene lactones is not the hard part. Hitting 24% flavone glycosides is, because it takes more leaf per kilo of extract, and more leaf is more money.
That is the opening. A supplier who is short on flavone glycosides can add a cheap flavonol source, and the 24% figure will come up to target while the terpene figure stays where it was, because the added material does not contain any. Rutin, extracted from buckwheat, is around 98% quercetin after hydrolysis and has been reported trading at roughly ten dollars a kilogram, against a material whose cost is driven by leaf. It is the obvious thing to add, and it has been added for a long time.
The scale is not marginal. A testing programme that examined eighteen commercial ginkgo supplements sold in North America and Europe between 2015 and 2017 found that three of them contained genuine ginkgo leaf extract. An investigation by a broadcaster and a university pharmacy school reported that a large majority of sampled products carried high levels of rutin or quercetin. These are not obscure finds in a niche journal. They describe a category with a structural problem.

What catches it, and what does not
The 24% figure does not catch it, because the adulterant is designed to satisfy that figure. The terpene figure does, because the adulterant has none.
A second check exists and is worth knowing about, because it costs nothing once the chromatogram exists. The ratio of quercetin to kaempferol in genuine ginkgo extract sits in a narrow band, and the ratio of the three main aglycones together is also characteristic. Add rutin and you move quercetin without moving kaempferol, so the ratio drifts. A certificate that reports only a total, with no breakdown of the individual aglycones, has thrown away the one piece of arithmetic that would have caught the substitution.
This is the general shape of the problem, and it applies far beyond ginkgo. A specification built around one number invites a supplier to satisfy that one number. A specification built around a number, a ratio and a compound that is expensive or impossible to add invites a supplier to make the real thing.
Where the adulteration happens
The techniques are few and they repeat across the trade.
- Spiking a marker. Adding a cheap source of the compound the assay measures. Rutin in ginkgo, pure quercetin or catechins in a green tea extract, synthetic curcumin in a turmeric extract.
- Exhausted material. Extract the leaf once, sell the extract, then re-extract the spent leaf and blend the second, weaker liquor back in. The chemistry looks broadly right. The potency is not.
- Species substitution. A cheaper relative with a similar profile. American ginseng for Asian ginseng, or one Panax species for another. This one is hard to see on a plate and easy to see with DNA.
- Plant part substitution. Leaf sold as root, stem sold as rhizome. The marker may be present at a lower level, which reads as a weak batch rather than as the wrong material.
- Colour. A dye added to make a dull extract look like the photograph in the brochure. Brilliant and easy to detect, and still done.
Each of these fails a different test. Marker spiking fails a ratio check. Exhausted material fails a ratio of marker to a compound that degrades during extraction. Species substitution fails DNA. Part substitution fails a marker profile specific to the part. Colour fails a simple dye screen.
There is no single test that catches all five, which is why a specification that names one assay is not an identity specification at all.

What a buyer can do in forty-eight hours
When a plate does not match and the shipment is on the water, four things can be done quickly.
- Ask for the same sample on three developing systems. If the material is what it claims, the pattern will be consistent across all three. If a band is an artefact of one solvent system, it disappears in the others.
- Ask for the raw chromatogram, not the peak table. A peak table is a summary prepared by the supplier. A chromatogram shows retention times, peak shape and the small unlabelled peaks that a summary leaves out.
- Ask for the individual components behind the total. Not 24%, but the quercetin, kaempferol and isorhamnetin figures, or the ginkgolide and bilobalide figures. The ratio is the tell.
- Ask for the retained sample of the batch that was approved. If the approved sample and the delivered batch differ, the conversation is no longer about analytical method.
None of this requires a new instrument. All of it requires a supplier who kept records.
The slow documents, which settle it properly
The definitive answers take longer and cost more, and they are worth knowing about because they change what a buyer asks for at the start of the next project.
Reference material from a pharmacopoeia or an authenticated herbarium specimen is the anchor. Without it, every comparison is against the supplier's own standard, which is circular.
DNA barcoding identifies the species from a small sample and is the only method that reliably catches a substitution between closely related plants. It says nothing about potency, and it cannot tell leaf from root in a milled powder unless the markers are part-specific.
Stable isotope work can suggest a geographical origin, and is used where a regional claim carries commercial weight.
A chain of custody from a named growing area, with harvest records and a lot number that follows the material into the drum, is slower than any of these and often more persuasive, because it addresses the motive rather than the symptom.
What to put in the specification so this stops happening
What prevents most of it is a clause that asks for a profile and cannot be answered with one number.
- The assay, with its method named. Not 24%, but 24% by ultraviolet after acid hydrolysis.
- A second marker that the cheap adulterant cannot supply. Terpene lactones in ginkgo, the specific catechins in a green tea extract, the ginkgolides individually rather than as a total.
- A ratio with a range. Quercetin to kaempferol, or the three main aglycones together. A range is what a natural material produces, and it is what a spike cannot reproduce.
- The impurity that the process is supposed to remove. Ginkgolic acids below 5 ppm in ginkgo is not a purity flourish. It is evidence that a purification step happened.
- A fingerprint, held on file, from an approved batch. The reference against which every future shipment is compared.
- A retained sample obligation, with a stated retention period. Because the argument in eighteen months will be about a batch nobody kept.
A specification written this way costs a little more to set up and considerably less to enforce, because it stops asking whether a supplier can hit a number and starts asking whether the material is what it says it is.
The botanical extract production lines page describes how identity and traceability are handled on the line, and the dietary supplement pages cover what a certificate has to show for each dosage form. When a batch is in question and the paperwork needs a second pair of eyes, the contact page reaches the people who run the extraction.
Ding Shaohua, Foreign Trade Manager, takes enquiries for this line. The same person answers on WhatsApp, TikTok, LinkedIn and Facebook.
FAQ
Does a different HPTLC fingerprint always mean the wrong species?
No. It frequently means a different plate, a different saturation, a different amount spotted or a different derivatisation. Two laboratories can produce visibly different plates from the same drum of material. The way to separate the two explanations is to re-run both samples side by side on the same plate, in the same session, with the same operator. If the bands still differ in position and not only in intensity, the material is worth investigating.
Is HPTLC quantitative?
Not in any sense a buyer can rely on. Band intensity under ultraviolet or visible light after derivatisation is not proportional to mass across a useful range, and the method is designed for pattern comparison rather than for measurement. Quantitative assays need a chromatographic column, a detector and a calibration curve. HPTLC answers whether two things are the same kind of material. HPLC answers how much of a compound is present.
Why is ginkgo adulterated more often than other extracts?
Because its specification contains one figure that is expensive to reach honestly and easy to reach dishonestly, and another figure that resists cheating but is not usually the one buyers check. The flavone glycoside target requires leaf, and leaf costs money. Rutin from buckwheat supplies the same aglycone for a fraction of the price. The terpene lactone figure would expose the substitution, which is why a complete specification asks for both and why a quotation that mentions only the first is worth a second question.
Can a supplier fake a terpene lactone result?
It is much harder than faking a flavonoid total, because the measurement requires an instrument that most adulterants cannot satisfy and because the compounds are structurally specific to the plant. It is not impossible to misrepresent a result, which is why the raw chromatogram matters more than the summary figure. What cannot easily be done is to make a rutin-spiked powder produce a genuine ginkgolide peak.
