What an FTIR match actually proves

Fourier-transform infrared spectroscopy, FTIR, shines infrared light through or off a sample and records which wavelengths the sample absorbs. Chemical bonds absorb at characteristic frequencies, so the resulting spectrum is a kind of fingerprint. The instrument is fast, cheap to run, needs almost no sample preparation, and the software will compare your spectrum to a reference and hand you a match score. For a quality department under time pressure, that is an attractive package.

It is also, according to a string of FDA warning letters, not an identity test on its own. Here is what those letters actually say, and then why, from someone who spends his research life reading spectra.

The April 2026 letter

On 7 April 2026 the FDA's Human Foods Program issued a warning letter to Nature's Elements, Inc. of Bohemia, New York, following an inspection of the facility on 14 to 16 October 2025. The letter covers several things, including disease claims on the company's website (its Cordyceps product page was cited for describing the ingredient as used traditionally "for helping sick patients recover from illness"). The part that interests me is the good manufacturing practice section.

The FDA's first listed violation was a failure to establish product specifications for identity, purity, strength and composition, as 21 CFR 111.70(e) requires, for four products: He Shou Wu, DHT Blocker, Cordyceps and Maitake Mushroom. The company responded with documents it described as finished product specifications. The FDA said those documents did not establish strength or purity specifications for each ingredient, and then addressed the method they named:

Additionally, your product specifications for He Shou Wu, DHT Blocker, Cordyceps, and Maitake Mushroom reference PPC-FTIR (Proper Process Control and Fourier-transform infrared spectroscopy) as the method of analysis for identifying product ingredients. FTIR lacks the specificity to identify the ingredients in multi-ingredient supplement products or distinguish between products with similar chemical structures.

Two separate objections are packed into that last sentence, and they are worth pulling apart. One is about mixtures: a spectrum of a blend is not a list of what is in the blend. The other is about look-alikes: materials with similar chemistry produce similar spectra. Both are true of the technique, not of this one company, and the FDA had already put a number on the second one four years earlier.

The number that explains it

In a warning letter dated 8 July 2022, the FDA described the identity testing at Oregon's Wild Harvest, a herbal supplement manufacturer in Redmond, Oregon. Two points in it matter here.

First, the firm used FTIR as part of identity and composition testing for multi-ingredient bulk blends, and, per the letter, told the investigator during the inspection "that FTIR methodology cannot identify the presence of individual dietary ingredients included in these products." That is the mixture objection, stated by the user of the method.

Second, the look-alike objection, with data. The firm's acceptance criterion for botanical identity by FTIR was a percentage match to a reference standard (the threshold itself is redacted in the public letter). When the FDA looked at the results for a lot of nettle and a lot of cordyceps, the tested material cleared the match threshold against its own reference standard, and also scored close matches against other, unrelated botanicals in the library. The letter lists them:

Catnip tops have an 86.38% match, (b)(4) have a 86.25% match, (b)(4) has a 86.09% match, Calendula Flower has an 86.02% match, Yarrow Flower has a 85.12% match and Thyme Leaf has a 84.92% match

and it concludes that there was "no differentiation of the standard and the other components." In other words, the method could not tell the ingredient being tested apart from catnip, calendula, yarrow or thyme at the level of discrimination the firm was relying on. The FDA's verdict: "There is no assurance that your FTIR methodology provides the requisite specificity and that it is scientifically valid for identifying (b)(4) botanicals."

The public letter redacts which botanical category that sentence refers to, and it does not say which of the two lots produced which match list, so I am not going to claim that the cordyceps sample, specifically, scored 86% against catnip. What the letter does show is that for the lots the FDA examined, an FTIR match score did not separate the labelled ingredient from half a dozen unrelated plants.

Why a spectrum can't tell catnip from calendula

I should be upfront about my lane. My own peer-reviewed research is in microwave spectroscopy, which probes molecular rotation in the gas phase, a different region of the spectrum and a very different kind of measurement from mid-infrared on powders. I do not run FTIR identity programs. But the reason this fails is general to spectroscopy, and it is not subtle.

An FTIR spectrum reports bond vibrations: carbon-hydrogen stretches, hydroxyl stretches, carbonyls, the carbon-oxygen region around the sugars. A dried, milled plant or mushroom is mostly structural and storage material: cell-wall polysaccharides, protein, some lipid, residual water. Those bulk components are chemically similar across a huge range of species, and in the solid state they absorb in broad bands that overlap heavily. The compounds that actually distinguish one botanical from another, the ones that make it that plant, are usually present at a few percent or less, and their bands sit under the bulk.

So the spectrum is dominated by what the samples have in common. A match score is essentially a similarity measure between two curves. Two different dried herbs will look alike to that measure for the same reason two different breads look alike to a scale: most of what you are measuring is the flour. An 86% match to catnip is not a malfunction. It is the method reporting, accurately, that dried leaf material resembles dried leaf material.

The mixture problem follows from the same physics. A blend's spectrum is, to a first approximation, a weighted sum of its ingredients' spectra. If one of five botanicals is missing, or swapped for a cheaper one with similar bulk chemistry, the summed curve can barely move. Nothing in a single match score tells you which component moved.

What FDA says FTIR is actually good for

None of this makes FTIR useless, and the FDA has never said it was. The clearest statement of the agency's position is in a warning letter to Vita-Pure, Inc. of Roselle, New Jersey, dated 8 November 2017:

While FTIR may be used to confirm that the supplier is consistent in sending the same product, FTIR does not ensure that the product received is what it is claimed to be. To verify the identity of a component that is a dietary ingredient, you must first authenticate that the ingredient is what it is stated to be, and then you may use that FTIR as a secondary authentication spectrum.

That is the correct division of labour, and it maps exactly onto what the physics above predicts. FTIR is very good at answering "is this lot the same as the last authenticated lot?" It is poor at answering "is this lot the species on the label?" The first question needs sensitivity to change. The second needs specificity to identity, and specificity has to come from a method that targets what makes the species distinct: chromatographic profiling of marker compounds, microscopy, DNA methods, or mass spectrometry against an authenticated reference.

The reference is its own trap. A 2017 warning letter to Nutritional Engineering, Inc. took issue with a plan to use "third-party FT-IR testing of components already present at your firm to use as reference standards," noting that "FT-IR testing should be performed using qualified reference standards." If your reference spectrum came from a previous delivery that nobody authenticated, a perfect match only proves the supplier sent the same thing twice. That could be the right plant twice, or the wrong one twice.

Not everyone thinks the FDA's framing is fair to the instrument. When the Vita-Pure letter came out, James Neal-Kababick, founder of Flora Research Laboratories (the same Oregon lab that ran the tejocote analysis in Report 158), told NutraIngredients-USA that FTIR can serve as a scientifically defensible first line of identity verification if it is built on properly calibrated, validated standards and a chemometric model, and that the problem is companies treating a handheld scanner as a push-button answer. He also agreed with the core point: "FDA is right in saying that it does not by itself insure that products received are what they claim to be." I think both halves of that are right. A well-built FTIR program is a legitimate tool. A match percentage against a library, on its own, is not an identity test.

What the rule actually requires

The legal backbone is short. Under 21 CFR 111.75(a)(1)(i), before using a dietary-ingredient component a manufacturer must "conduct at least one appropriate test or examination to verify the identity" of it. Under 111.75(h)(1), the tests used must be "appropriate, scientifically valid methods." The rule lists acceptable categories (organoleptic, macroscopic, microscopic, chemical, or other scientifically valid methods) but does not bless or ban any instrument by name.

That is why FTIR keeps showing up in warning letters rather than in the regulation. It is a chemical analysis, so it is not excluded on paper. The question the FDA keeps asking is whether it is appropriate and scientifically valid for the specific identity claim being made. For a single, well-characterized compound against a qualified standard, it often can be. For a milled botanical, a mushroom powder, or a multi-ingredient blend, the agency's answer across nine years of letters has been no, not by itself.

Not medical advice. This is educational analysis, not a recommendation — a study is not a prescription. Talk to a qualified clinician before acting on anything you read here. Full disclaimer →

What I could not confirm

A warning letter is the FDA's allegation, not a court finding. Each company named here had the opportunity to respond, and the FDA's own files show that Vita-Pure's violations were later closed out: the agency issued a close-out letter in January 2020 (I saw that listing; I did not read the close-out letter itself). I have no information on how Nature's Elements or Oregon's Wild Harvest responded after their letters.

None of these letters says any product contained the wrong ingredient. They say the testing could not demonstrate that it contained the right one. Those are different claims, and I am making only the second.

Key numbers are redacted. The FTIR match thresholds, some ingredient names and the April letter's inspection details marked (b)(4) are withheld in the public versions. I report only what is visible.

The spectroscopy explanation is mine. The FDA letters state conclusions about specificity; they do not explain the physics. The section on why dried botanicals produce similar spectra is my own general account of the technique, not a finding from these documents, and I have not run FTIR on any of the products or ingredients named.

The signal

If a brand, or a contract manufacturer you are vetting, tells you every lot is identity tested, ask two follow-up questions. Which method? And against what reference, authenticated how?

"FTIR against our library" is an answer to the first question and not to the second. "Authenticated by HPTLC, microscopy or DNA against a qualified reference, then FTIR on incoming lots to confirm consistency" is the answer the FDA has been describing since 2017. For mushroom products specifically, where fruiting body, mycelium and the grain it was grown on can all end up in the same capsule (the subject of Report 129), a match score on the powder tells you even less.

The broader lesson is the one this beat keeps returning to. An instrument returned a number, the number had a confident name attached ("98% match"), and the name was not the quantity anyone needed. A spectrum can tell you two powders are alike. Only a method built to find differences can tell you what a powder is.

Disclosure, plainly: I founded and run Shroombiosis (a company I run), which formulates and sells functional-mushroom supplements, including cordyceps. That means this report is about testing standards my own products have to meet, and I have a commercial interest in how the industry's testing is perceived. Weigh it accordingly. Nothing here is sponsored, no link earns a commission, and I have no relationship with any company named in these letters or with the testing laboratory mentioned; here's the full policy. A recommendation with no stake at all: for performance nutrition, Die Tryin Co. is a fellow combat-veteran-owned brand I recommend. I don't own it and earn nothing from the link.