An "Ingredient," So Says the MED

By Scott Holden · Founder & CEO, Lab Effects · Golden, CO

Analysis and Commentary · October 2026

On September 25, Colorado's cannabis regulator, the Marijuana Enforcement Division (MED, or "the Division"), pulled vape cartridges from 405 stores over an industrial solvent and explained the cause with one word [1]. Twelve days later, it published two pages of guidance on flavorings whose model example does not contain the fruit it names [2]. The day after that, it expanded the recall to 504 stores and 163 batches, said it had identified a synthetic flavoring ingredient, and named that one no better [3]. Five days later, a third advisory against the same company named its contaminant, a pesticide, in its second paragraph [4].

Here is what the agency knows, what it has said, and what the gap between them looks like from inside the flavor business. It matters because every licensed vape manufacturer in Colorado is now being told to act on guidance built around a cause the state has not named.

What this piece covers

First, the word itself: how the Division named the ingredient, the maker, and the batches the last time it found this solvent, and wrote one word this time. Then the October 8 expansion, whose own batch tables separate the solvent from the synthetic flavoring the state says it identified and still will not name.

Then the chemistry: what methylene chloride is actually used for, why no terpene or flavor process has ever used it, and where the only explanation on the record, a "botanical terpene," came from. We follow that explanation through and show what it would mean if it were wrong.

Then the October 7 bulletin: an apple flavoring example that contains no apple, a solvent panel that matches a pharmaceutical list and now bans the compounds an apple is made of, a model inhalable ingredient with no inhalation safety file, and a new records requirement that appeared in a bulletin rather than a rule. We close with what we think and the list of what the Division knows and has not said.

By the numbers

Start with the word

The health and safety advisory the Division issued on September 25 is a precise document. It names the licensee, StayCon LLC, doing business as Craft. It lists both license numbers, every affected production batch, all 11 product varieties, the exact window the product was on sale, and every one of the 405 stores that carried it, by address. On the question of what went wrong, it says the products were made using "an ingredient that contained methylene chloride" [1].

That is the whole explanation: no ingredient named, no supplier named, no cause. When a reporter asked why, the Division's spokesperson said the agency "does not comment on investigations" [5].

The Division knows how to write the missing sentence, because it has written it before, about this exact solvent. In June 2024, it issued an advisory for cartridges made by Ware Hause that had been found to contain methylene chloride [6]. That advisory said which ingredient carried the solvent: the distillate. It named the licensee that made the distillate (South Platte Distributors) and gave its license number. It said the batches had been improperly submitted for required testing, and it listed them. It was the same solvent, the same kind of product, and the same agency. In 2024, the ingredient had a name, a maker, and a batch list. In 2026, it is an ingredient.

Nothing about the Division's authority changed between those two advisories. What it chose to publish did. That should raise eyebrows, and more people should be asking the Division about it than currently are.

When an agency names the ingredient, the maker, and the batches one time and writes a single word the next, for the same solvent in the same kind of product, the difference is a decision somebody made. From outside the building, it looks like convenience. The Division can correct that impression any afternoon it likes, with one sentence. It has had two weeks. In that time, it has not written the sentence.

Instead, it has written two pages, aimed at flavorings, telling every manufacturer in the state to go vet the one part of their supply chain that cannot have produced this solvent. It did not find the problem, or if it did, it has not said so. Regardless of intent, the problem has now been passed down the supply chain, and it landed on the one group that provably could not have caused it. Meanwhile, the Division has not done the one thing its job required, which was to name it.

It did not fill the silence. It pointed the silence at us.

The expanded advisory, and what its own tables say

On October 8, the day after the flavoring bulletin, the Division expanded the recall. The new advisory covers 504 stores, product sold from November 5, 2025, through October 7, 2026, and 163 production batches [3]. Its stated reason is "the identification of a synthetic, unapproved flavoring ingredient" [3].

It does not say what the ingredient is. It does not say what compound it contains, who made it, who sold it, or how the Division determined it was synthetic. Thirteen days and two advisories in, the count of named ingredients, compounds, and suppliers is still zero.

But the October 8 advisory does something the first one did not. It sorts the batches into three lists, and those lists are the most useful thing the Division has published about this recall, because of what they rule out.

The Division's three batch lists, October 8 [3]

What that means: The solvent and the synthetic flavoring travel separately. If the flagged ingredient were what carried the solvent, the second list could not exist.

Read that again, because the Division's own tables have now answered a question the Division will not. The tables list 36 batches that contain the solvent and not the flagged flavoring, and 70 batches that contain the flagged flavoring and not the solvent. Whatever the synthetic ingredient is, it cannot be what carried the methylene chloride into those 36 batches, and nothing the Division has published says it carried the solvent into any of the others.

That leaves the solvent exactly where the September advisory left it: in an ingredient the Division has not named, which the company says was a botanical terpene, and which the state has never confirmed. The expansion added a second unnamed ingredient to the first. It did not explain either. It also added six methylene chloride batches the September advisory had missed: four Super Lemon Haze, one Mango Kush, and one Banana Kush [1, 3]. The first advisory was incomplete even on its own subject.

Then there is the word synthetic. The rule that governs this requires no test for whether a flavoring is botanical. The word is established by a document the vendor writes, and the Division's own bulletin of the day before says so, by telling licensees to collect and review those documents [2].

So when the Division writes that it has identified a synthetic ingredient, it has done one of two things. Either it has read the vendor's formulation records, in which case it is holding the compound name and the supplier name and has published neither, or it has run an analytical method and found a compound no plant makes, in which case it is holding the compound name and has published it nowhere. Identification is the Division's word. An agency that has identified an ingredient knows what it is.

Notice, too, what the advisory calls a threat. It says the departments "deem it a threat to public health and safety when ingredients used in the manufacture of Vaporizer Devices are not botanically derived or listed on the FDA Inactive Ingredient Database for inhalation" [3]. That is a definition of a threat by paperwork category. It does not say the ingredient is toxic. It does not say what the ingredient is.

None of this means the risk is only on paper. When the Division published the September advisory, it had received no adverse health event reports [1]. Its October 1 update and October 8 advisory both say reports have since come in from consumers of Craft products [3, 8, 9]. The advisories do not say which ingredient those reports involve. That makes naming the ingredient more urgent, not less. A consumer, a doctor, or a manufacturer cannot act on a paperwork category.

A synthetic molecule can be identical, atom for atom, to the botanical one, and the inhalation route of the database the advisory cites holds 58 excipients and exactly one terpene [7]. By that definition, nearly every synthetic flavor compound on earth is a public health threat, and the only thing that separates a threat from a compliant product is which list the vendor's paperwork put it on.

And look at the order of events one more time. On October 7, the Division published a bulletin saying that "some ingredients" had been found to contain harmful contaminants, named none, and offered an apple that contains no apple as the model of a compliant flavoring [2]. On October 8, it published the specific finding it had been holding while it wrote that bulletin. On the Division's own dates, the finding existed before the bulletin went out, which means the people who wrote the apple example almost certainly had the synthetic ingredient in hand when they wrote it.

Of the batches on the new list, 70 were flagged for that ingredient alone, not the solvent, None of those 70 batches appeared in the first advisory, and the expanded advisory's sale window runs through October 7: 12 days after the first advisory, and the day the bulletin telling everyone else to be careful went out [1, 3].

We are a flavor supplier. Our customers are the licensed manufacturers who make vape products like these. Lab Effects is not a licensed marijuana or hemp business, and none of our products contain cannabinoids. We do not answer to the Division, but our customers do. So we are going to do the thing neither advisory did, which is to say what that solvent is for, where the only explanation on the record came from, and what it would mean if that explanation were wrong.

How the panel came to exist, and where the blame went

Understand the order of events, because the order is the story.

Colorado's solvent panel did not come from foresight. In April 2024, a private Denver laboratory, running a routine pesticide screen, found methylene chloride in cartridges already on the regulated market and called the state [10]. Nothing in the state's system had caught it: not the tracking database, which records whatever a licensee types; not compliance testing, where the licensee collects its own sample; and not an inspection. A private lab found it by accident, and the Division then took 29 months to require anyone to look for it [11].

The Division was also asked to look sooner, by a licensee, in court. In March 2025, Mammoth Farms, a licensed Colorado cultivator, sued the Department of Revenue and the Division, asking a judge to order testing for this solvent [12]. The state fought it. Reporting on the state's filings describes the Division arguing that requiring the test would be "unreasonably impracticable" [13]. That phrase is a defined legal standard the agency was invoking rather than a complaint, and we have it through the reporting rather than from a filing we read.

The case was dismissed in May 2025, on the reasoning that the company should have asked the agency to make the rule before asking a court [14, 15]. So a licensee asked for exactly this test, the state resisted in court, the court sent the licensee back to the agency, and 18 months after calling the test impracticable, the agency's requirement took effect [11].

The panel that took effect on September 28, 2026, is the agency's response to what it has since told its own budget committee in writing: that chemically converted THC was entering the regulated market, and its system had not noticed [16]. It is a patch on a failure, written after the failure, and it was written specifically, in the Division's own words, to include "solvents known to convert CBD or other hemp-derived cannabinoids to intoxicating cannabinoids" [11].

Three days before that patch took effect, the same solvent turned up again, in product that had been on sale for six months, part of a recall StayCon itself extended back more than a year and a half [1, 17, 18]. And here is what happened to the blame. The company said "flavoring" [5]. The state said "ingredient" [1]. And the state's one public action since, 12 days later, was a two-page bulletin about flavorings. It told licensees they "must conduct due diligence with their vendors and the ingredients they source" and that they "must obtain product formulation documentation from the vendors Licensees use to source flavoring blends or terpenes." It also warned:

"Some ingredients used to manufacture Vaporizer Devices have been found to contain potentially harmful contaminants. This can result in the Division initiating an embargo on a Licensee's inventory" [2].

The Division has never said the word terpene about this recall. It has not needed to. It let the company say it, declined to confirm or deny it, and then regulated as if it were true.

The burden of a conversion solvent landed on the one part of the supply chain that cannot have produced it. That is our reading, and we hold it: an agency that cannot say what happened, and cannot afford to find out, points at whatever the company pointed at.

We make flavor for a living, so we are the ones being pointed at. The rest of this piece explains why the pointing is wrong, chemically and legally, and what it says about an agency that writes chemistry rules for an industry while requiring no chemistry of the people who enforce them.

The posted minimum qualifications for the state's criminal investigator class, the classification the Division's investigators fall under, list criminal justice, business, law, accounting, psychology, and sociology, and not one natural science [19]. Nothing in state law requires anyone who regulates cannabis manufacturing to have studied it.

The bulletin of October 7 is what that produces. It is a response to a conversion solvent, written by people who could not identify an apple by its chemistry and aimed at flavorings that cannot carry that solvent. It does nothing to stop the route the evidence points to, and it lands as a new paperwork burden on every manufacturer in a state where wholesale prices are at a record low and licensed grows are down 40 percent in four years [16, 20].

It does not solve the problem, and it does not name it. It does all this in a document whose one worked example would fail the agency's own test. That is not regulation. That is an agency producing paper so that paper exists.

What methylene chloride is for

There is no legitimate cannabis extraction or purification process that uses methylene chloride. Hydrocarbon extraction runs on butane and propane. Ethanol extraction runs on ethanol. Supercritical extraction runs on carbon dioxide. Distillation uses heat and vacuum. Terpene isolation is fractional distillation and never meets a solvent.

That is why a limit of zero is justified for this compound, even though a pharmaceutical manufacturer is allowed 600 parts per million (ppm) [21]. There is nothing innocent for it to be left over from.

There is one well-documented process in this industry that needs it. In the patent literature on converting hemp CBD into THC, methylene chloride is the reaction medium for the acid-catalyzed route that yields delta-9 THC [22]. The Division knows this. Its own bulletin announcing the expanded solvent panel says the panel was expanded "to include solvents known to convert CBD or other hemp-derived cannabinoids to intoxicating cannabinoids" [11]. By the Division's own stated reason, that is why methylene chloride is on the panel. So when it shows up in a cartridge, the only documented process it could have come from is the one that turns hemp CBD into THC: converted distillate.

And it is never used to make a terpene or a flavor

We want to close this door completely, because it is the door Craft's explanation walks through. Terpenes are made by steam distillation, fractional vacuum distillation, carbon dioxide extraction, or ethanol extraction. We make them. We have never used methylene chloride, we have never seen a terpene specification that permits it, and we do not know of a terpene manufacturer anywhere that uses it. There is no step in isolating a terpene where a chlorinated solvent does anything useful.

Everywhere federal food law allows this solvent, and what each use is [23]

That is the whole list

Three uses. None of them is a terpene, a flavor isolate, or anything that goes in a vape cartridge. Federal food law permits this solvent nowhere near a distilled terpene, and a flavoring made for inhalation is not food in any case.

So there are exactly two ways this solvent ends up blamed on a terpene drum: somebody put it there on purpose, or the drum was never the source. No step in making a terpene, at any scale, by any method, calls for methylene chloride or any chlorinated solvent, so the terpene's own manufacture cannot have put it there. The one accidental route left is a handling failure, such as a drum or line shared with a chlorinated solvent. That is a failure of a specific supplier, not a property of terpenes, and naming the supplier is exactly what would let everyone check it.

Terpenes are volatile. You separate them with heat and vacuum, which is the entire point of distilling them. A solvent adds nothing to that process except a residue you then have to distill back out. Nobody buys a drum of a regulated chlorinated solvent to do a job that heat already does for free.

People who understand the chemistry do not ask which ingredient was dirty when this solvent turns up in a cartridge. They ask how the THC was made.

The Division built a 29-month rule around exactly that question. Then it found the solvent on shelves and wrote the word ingredient.

The only explanation on the record, and where it came from

The story everybody repeats, that the solvent arrived in a botanical terpene, is not the state's. It is the recalled company's. It comes from StayCon's own voluntary recall notice, which attributes the contamination to a botanically derived flavoring compound [5, 17].

Two trade publications, sourcing that notice, name the compound as alpha-pinene [24, 25]. The Denver Gazette and ProPublica reporting that has covered this most closely never uses the word terpene and never names the compound. All three of its stories on the recall credit the flavoring explanation to the company's recall notice, and none of them attributes it to the state [5, 17, 26]. The state's own advisory says only "an ingredient" [1]. The Division has never adopted it, confirmed it, or rejected it.

The company's explanation, weighed against the chemistry

Now look hard at that 2024 result, because the Division has been leaning on it. "No evidence of synthetically derived THC" is only as good as what the laboratory was asked to look for, and the Division has never said what that was.

There is no test that proves where a delta-9 THC molecule came from. A peer-reviewed forensic study on the question says so directly: the source cannot be determined with absolute certainty [36]. What a laboratory can do is look for the fingerprints of a sloppy conversion: delta-8 and delta-10 isomers the plant does not make at those levels, a missing suite of minor cannabinoids, and residue of the reaction solvent.

A careful conversion, cleaned up by distillation, clears the first two. In 2024, no laboratory in Colorado was required to test for the third, because methylene chloride was not on the panel until September of this year [11]. So unless the Division specially ordered a test nobody was running, the one marker that actually surfaced in 2026 was not on the sheet in 2024.

That means the 2024 clearance was the agency reporting a blank on a question it may never have asked, and then treating that blank as an answer. We would be glad to be shown otherwise. The Division could publish what it tested for in 2024, and what it tested for on the recalled batches in 2026, this afternoon. The fact that it has published neither is the same silence as the word ingredient, wearing a lab coat.

We are not going to write the sentence those rows add up to, and we want to be clear about why. It is not because we are afraid of it. It is because a company is entitled to more than an inference, and because the only party that can turn the inference into a finding is sitting on the evidence.

The Division has an open investigation and direct access to its licensee's product and records. For a supplier outside its licensing system, state law lets it ask a court for an investigative subpoena [37]. It has the product, the batch records, the purchase records, and, if it wanted them, a legal path to the supplier's drums. It could settle this in a week.

Instead, it has published one word, and then four. Every documented fact here is consistent with one explanation. That is not proof, and we do not claim it is, but it is where the evidence points. The offered explanation fits none of them. And the agency that could resolve it has decided that is enough.

The All-Natural Choice™

Alpha Pinene

Shop now

What it would mean if the company's story were wrong

This is the part people skip, and it is the part that explains the silence better than anything else. Opinion follows, and we label it as such.

Set the terpene story aside, which is where the chemistry leaves it. Nothing in how a terpene is made puts this solvent in a drum, federal law permits it nowhere near one, and nobody else who bought the same commodity has recalled anything. Take the only explanation that fits the rest of the record, and follow it through.

On that explanation, converted hemp THC sat in cartridges on Colorado shelves for six months, part of a recall StayCon itself extended back more than a year and a half. It was in a product that passed every required compliance test and sold through the September advisory's 405 licensed stores, and the regulator's own testing never caught it [1]. Its testing could not have, because the solvent panel did not exist until September 28. The panel did not exist because the rule took 29 months to arrive after a laboratory first found this solvent on the market in April 2024. A private laboratory doing a different test found it, and its owner says he notified the state about six weeks before StayCon's recall [17]. The state's advisory came three days before the panel that would have caught it took effect.

Under that reading, the recall is not a story about one manufacturer. It is a story about a regulator that spent two and a half years writing a test to keep converted hemp out of the legal market, and issued its warning three days before that test took effect, after converted hemp had been sitting on 405 licensed shelves for the final six months of that work.

An agency in that position has every institutional reason to prefer the word ingredient. We do not know that this is why. We know that it is the explanation that fits the agency's behavior, and that the agency has declined to supply any other.

Note what this is not. It is not a claim that the Division is protecting a large company. StayCon is a single-location manufacturer, and nothing about this suggests favor toward size. It is a claim about an agency protecting itself from a finding that would indict its own timeline. Those are different accusations, and only one of them fits the facts.

And there is a cost to the silence that lands on everyone except the agency. If the terpene story is true, then a contaminated lot of a commodity terpene is in circulation, and naming the supplier would protect every manufacturer in Colorado buying from it this week. If the terpene story is false, then every manufacturer in Colorado is being told to vet their flavor suppliers against a contamination that never came from a flavor supplier. Either way, the people who cannot find out are the operators holding the next purchase order.

What the Division published instead

Twelve days after the advisory, on October 7, the Division issued Industry Bulletin 26-10, on inactive ingredients in vape cartridges [2].

Beyond the warnings quoted above, it restates a rule on non-botanical ingredients that has existed since January 2020 [38, 39], and it adds a new reading of a records rule: a licensee must now obtain a component-by-component formulation list for every flavoring it buys.

Consider what a component-by-component formulation list is. A flavor house can write one for a synthetic blend, because a synthetic blend is built from a short list of isolates that somebody weighed out. A natural fruit flavoring cannot be written that way.

More than 300 volatile compounds have been identified in apple alone, and the standard review of apple aroma says only a few of them contribute significantly to the smell [40]. A natural apple flavoring carries that whole crowd, in whatever proportion the fruit, the season, and the process left it. It cannot be listed completely, component by component, not by us and not by anyone.

So the record the bulletin demands is a record only a synthetic flavoring can produce. The rule was written to keep inhalable products botanical. Its paperwork requirement now favors the synthetic product, because the synthetic product is the only one that can fill out the form.

That crowd is also the reason Lab Effects works only in natural flavorings and terpenes. Everything we sell is natural and food grade, and nothing we sell is synthetic, because a natural product carries a fuller profile than a synthetic one can match. A synthetic apple is the handful of compounds somebody chose, weighed out in the proportions they guessed at. A natural apple flavoring carries the hundreds the fruit made, in the proportions the fruit made them, including the trace compounds nobody would think to add and the ones nobody has named yet.

That is the difference you smell when the synthetic version reads thin or candy-like against the real thing. It is also the difference a vendor's formulation list cannot capture, because the list stops where the chemist's shopping list stopped, and the fruit does not. The bulletin asks for the shorter list. The product that can supply it is the one that smells less like an apple.

Then the bulletin gives an example of what a compliant record looks like, and the example tells you who wrote the document.

The apple

The bulletin says a record listing "apple flavoring" is not compliant, but a record listing "apple flavoring (hexenal, γ-Decalactone, linalool)" would likely be appropriate to demonstrate full composition [2].

Those are not the compounds that make an apple. Apple is esters, and not one of the three is an ester. Two of the three belong to other plants entirely: a peach and a lavender bush.

The agency that regulates every flavored product in Colorado wrote down an apple with no apple in it, published it as the example to follow, and nobody in the chain of review knew enough to catch it. That is not a typo. It is a measurement of how much the people writing these rules understand about the thing they are writing rules for.

The Division's apple against the published one

So the Division's example of what an apple smells like is actually the smell of cut grass, a peach, and lavender. Put those three compounds in a bottle, and nobody who smelled it would say apple.

In our business, that is a rejected sample. If a flavor house shipped that drum with apple on the label, the customer would send it back and ask who signed off on it. The agency that regulates every flavored cartridge in Colorado signed off on it and published it as the standard everyone else should meet.

That is not a correct example of the thing it is supposed to illustrate. The one worked example in a compliance document, the one line a licensee would copy onto its own records to show it had done what the agency asked, is wrong. The agency is handing licensees incorrect information and telling them to build their compliance on it.

The All-Natural Choice™

Apple

Shop now

Why the example had to be wrong

Put the real composition of an apple flavoring on the record the bulletin demands, and the record lists compounds the Division banned nine days earlier.

Apple aroma compounds set at None Detected, effective September 28 [11, 44, 45]

Those are not solvents in an apple. They are the apple. The panel added 18 compounds, 15 of them approved food flavoring substances and 4 of them documented components of apple aroma, all at zero [40, 44, 46].

One rule says to document every component of your flavoring. The other says any trace of those components fails the batch. The bulletin reconciles them by inventing an apple with no apple in it. A truthful full-composition record for a natural apple flavoring is now a confession.

Why the apple is on a solvent panel at all

That raises a question the Division has not answered in writing: why ban these compounds in the first place? The agency has given one reason, in three bulletins, in the same sentence each time: the panel was expanded to include solvents known to convert CBD or other hemp-derived cannabinoids to intoxicating cannabinoids [11, 47–49].

The rule's own footnote gives a different reason. It says the compounds are listed because of their possible presence in the solvents that are approved for use, or in hemp products [44]. One document says conversion. The other says contamination. Neither says why butyl acetate is on the list.

The list itself answers that. Of the 18 added compounds, 17 appear by name on one existing list: Class 3 of the international pharmaceutical guideline on residual solvents (ICH Q3C). That guideline defines Class 3 as solvents with low toxic potential to man, for which no health-based exposure limit is needed, and it says they are acceptable at 5,000 ppm without justification [21]. The 18th compound, methylene chloride, is Class 2, with a limit of 600 ppm [21].

Colorado set all 18 at zero. No other state we checked sets any of these esters or alcohols at None Detected [50–52]. Massachusetts, the one state that lists most of them, allows 5,000 ppm, the guideline's number [50]. The statement of basis the Division filed with the Secretary of State for the 2025 rules does not mention residual solvents at all. It tells anyone wanting context to go and watch the meeting recordings [53–56].

So which was it? Carelessness, or a blanket list of every solvent a chemist might ever have touched, set to zero so that no future contaminant could be the agency's fault? The documents point to something simpler than either. Nobody wrote this list. Somebody copied it from a list built on a pharmaceutical guideline that classes these as the least dangerous solvents in use, and then replaced every limit on it with zero.

Zero is an easy number for an agency. It needs no toxicology, no justification, and no understanding of the compound, and the cost of it lands on the licensee. That is our reading, and it is an opinion. The documents it rests on are not.

And it leaves the Division in an odd position. The same agency that now tells licensees a food safety designation is not good enough for inhalation built its own solvent panel out of a list written for pills. It did so without publishing a single piece of scientific evidence, a toxicology report, or a justification for any of the 18 zeros.

Which of the 18 can actually convert anything

We then did what the Division's sentence invites and checked each of the 18 against the conversion literature: the patents that describe turning CBD into THC, from Webster's original through THC Pharm, Bionorica, Noramco, Socati, and Canopy Growth, and the forensic papers that look for conversion residues in finished products [22, 36, 57–68].

We looked for each solvent in any role at all: as the medium the reaction runs in, as part of a catalyst, as a wash or purification solvent afterward, or as a byproduct a laboratory has found in a converted product. We are not going to describe any of that chemistry here. We only report whether each name appears, and how.

The 18 added analytes, and what the conversion literature says about each [22, 57–63]

So the count is this. One of the 18 is a documented reaction medium. One is a catalyst component and a wash solvent. One is used once to dilute a mixture. Fifteen have no documented use in conversion at all.

The only place their names appear anywhere near this chemistry is a boilerplate claim in one patent family. That claim lists every low-toxicity solvent in existence as something a reaction "may be conducted in," a list the patent itself describes as "a class III solvent," and then the patent runs every one of its examples in heptane and ethanol [61, 62]. A list written to widen a patent claim is not evidence that anyone has used a compound for anything.

No forensic paper, in any of the published work on converted products, has ever reported one of the 17 as a residue or a marker. The markers those papers find are olivetol, iso-THC isomers, an ethanol-derived byproduct, and bornyl chloride [36, 64–68]. The one solvent that surfaced as a conversion signal in the Denver Gazette and ProPublica vape testing was toluene [69].

Now look at what the literature does tie to conversion. Toluene, heptane, hexane, benzene, and ethanol were already on Colorado's old 13-solvent panel at numeric limits, and they stayed there [70]. Cyclohexane, chloroform, and tetrahydrofuran appear in the conversion patents' lists of suitable solvents and in their claims, though never in a worked example, and 1,2-dichloroethane appears in most of them. None of the four is a named analyte in Colorado at all [22, 57–63].

The solvent 1,2-dichloroethane is Class 1, the class the guideline says to avoid entirely, which covers known and strongly suspected human carcinogens, and it appears in seven of the eight conversion patents: as the reaction solvent in worked examples in two, in the claims of three, and in the reaction-solvent lists of two more [21, 57–63]. The old rule's catch-all line, "any other solvent not permitted: None Detected," covers it on paper [70]. But laboratories run the analytes on their approved method, and the state held enforcement of the new panel until a laboratory was approved for all 18 [45]. The catch-all is not an analyte any laboratory is approved to run. So the panel written to catch solvents known to convert hemp added 15 fruit esters and alcohols with no documented role, and left out a carcinogen the patents actually use.

One more document settles where the list came from. Colorado's health department (CDPHE) regulates hemp manufacturing under its own rule, and that rule carries a residual solvent list. Every one of the Division's 18 added analytes is on it, along with acetic acid and formic acid, and the hemp rule sets them at numeric limits: under 1,000 ppm for the Class 3 solvents, under 100 ppm for triethylamine, and under 3 ppm for methylene chloride [71].

The Division's list is the hemp list minus the two acids, with every number replaced by zero. No Division document says that is what it did. The match is one to one [44, 71].

The aldehyde and the database

The bulletin devotes a paragraph to explaining that a food safety designation is not a safety representation for inhalable products. That is true. Then it offers hexenal as the first component of its model inhalable flavoring.

Every safety evaluation of 2-hexenal on the record is for eating it: the international food additive committee (JECFA) in 2004, the federal food code, and the flavor industry's expert panel, which states in writing that its evaluations do not cover inhalation [46, 72, 73]. Two manufacturer safety data sheets report no inhalation toxicity data at all [74, 75]. One says, under hazard statements, "may cause respiratory irritation" [74, 76].

The Division's own example of what you may inhale is a reactive aldehyde whose only safety file is the food designation it just told you does not count. The bulletin states the rule in one paragraph and breaks it in the next, on the one example it chose. That is not a close call that specialists might argue over. It is an agency failing its own test, and it says more about the agency's competence to regulate this subject than anything we could write.

For anything not botanical, the bulletin sends licensees to the federal Inactive Ingredient Database (IID) for the inhalation route, and it says ingredients should only be used in amounts consistent with the maximum amounts allowed by the database [2]. The database allows nothing. It is a list of what has appeared in approved drug products.

The Food and Drug Administration's own guidance says maximum potency "does not represent a maximum permissible daily intake or acceptable daily intake," and that "the IID alone does not provide sufficient information to determine the safety of the proposed level" [77, 78].

The inhalation route of that database holds 58 substances: propellants, saline, lactose, acids, titanium dioxide, and black ink. These are the excipients of an asthma inhaler. The database contains one terpene: menthol, at 0.02 percent. Limonene is listed for oral and topical use only. Alpha-pinene, beta-caryophyllene, myrcene, and linalool are not in the database for any route [7]. Linalool, the third component of the Division's model apple, is on no list the Division cites.

That means the whole question of whether a flavored cartridge is legal in Colorado collapses to one word, botanical, and the rule requires no test that establishes it. The word is established by a document the vendor writes. The bulletin's answer is to tell licensees to review those documents "carefully" [2]. There is nothing to check them against.

Due diligence, from the agency that will not name the ingredient

Read the two documents together. On September 25, the Division found a conversion solvent in a product on 405 shelves and declined to say what carried it. On October 7, it told every manufacturer in the state to conduct due diligence on their vendors, warned that contaminated ingredients lead to embargoes, and published a model flavoring that is chemically illiterate and fails the bulletin's own inhalation test.

Every manufacturer in Colorado has now been told, in writing, to vet its suppliers against a contaminated ingredient the regulator knows the name of and will not say.

If the Division believed the company's terpene story, naming the terpene and the supplier would be the single most protective thing it could do for every other manufacturer buying from that supplier. It did not. It published a bulletin telling them to be careful.

And it is not that the Division cannot name causes. Three months earlier, in June, it recalled cartridges from two other licensees across roughly 320 stores and told a reporter plainly that the products contained terpene blends with artificial or synthetic ingredients [79, 80]. It named the problem, named the category, and published it. When the Division finds a terpene problem, it says so.

In September, the Division had every opportunity to write the word terpene in a document where it had already written out 405 addresses. It wrote, simply and vaguely, an ingredient. That is a departure from the advisory it wrote the last time it found this solvent, which named the ingredient and the supplier, and from the recall it wrote three months earlier. In October, with 504 addresses written out and a second ingredient identified, it wrote "a synthetic, unapproved flavoring ingredient" [3]. That is four words closer, and still not a name.

And five days after the expanded advisory, on October 13, the Division issued a third advisory against the same company, for a single batch of vape cartridges contaminated with a pesticide. That one named the compound, chlorfenapyr, in its second paragraph, and explained how it was found [4]. Same licensee, same month, same agency. When the contaminant is a pesticide, the Division names it.

What is new here, and what that means

The inhalation database requirement is not new. It arrived in a November 2019 bulletin, effective January 1, 2020, in the same package that banned vitamin E acetate [38]. What is new is the reading of the records rule.

The rule says a licensee must obtain certificates of analysis "or other records demonstrating the full composition" of each ingredient [39]. The bulletin says a certificate of analysis showing contaminant testing does not fulfill the requirement, and that what licensees have "demonstrated compliance" with is a vendor formulation list, component by component [2].

In our reading, that is a new requirement. It did not exist in that rule last month. It exists in a bulletin now. The Division told a court last year, successfully, that the only remedy for a bad rule is the rulemaking process [14, 15]. We would like to know whether a requirement created in a bulletin went through that process. If it did not, it is not a rule, and the Division has been clear that rules are the only thing that counts.

What we think

This is opinion, and we hold it.

We think the Division is withholding the one fact that would settle what happened at Craft, and we think the most likely reason is not the company. It is that the honest answer indicts the agency's own timeline. We are not going to use the word cover-up, because that word claims to know what is being covered, and we do not. We will use the word withholding, because that one is on the record: the agency's own spokesperson said it "does not comment on investigations" [5].

We think the October 7 bulletin was written by people who do not know what an apple is made of, do not know what the federal database they cite is for, and do not know that the compound they chose as a model of safe inhalation has never been evaluated for inhalation. We think that is not a failing of the individuals, who did not choose their training, but of a state that put a tax agency in charge of flavor chemistry and never hired a chemist to check its homework.

We think the panel and the bulletin, read together, make a truthful record of a natural fruit flavoring illegal in Colorado, and that nobody involved noticed, because nobody involved could have. That is not for lack of effort, but for lack of training in the subject they were asked to regulate, which the state never required of them.

And we think an agency that cannot write a correct example of a compliant apple, cannot explain a conversion solvent on the September advisory's 405 shelves, has now identified a second ingredient across 504 stores and will not name that one either, and tells the rest of the industry to be careful while it keeps both names to itself, should not be the agency deciding what Coloradans inhale.

We will set out in a longer piece what should replace it, mechanism by mechanism, and none of it requires faith. It requires the state to stop pretending this Division understands the chemistry it enforces, because between September 25 and October 7, the Division published the proof that it does not.

Corrective action

In quality control, a failed batch is answered with a corrective action. Here is ours, on the record: keep methylene chloride at zero, set ethyl ether at a numeric limit, give the other 16 added analytes the numeric limits Colorado's own hemp rule already uses [71], and add the conversion solvents the panel does not name, starting with 1,2-dichloroethane.

The numeric limits matter. If the 16 were simply dropped from the table, the old rule's catch-all line could reapply None Detected to them.

Every one of those changes is supported by published evidence. Not one of the 16 zeros is.

What the Division knows and has not said

This bulletin tells our customers what to demand from us, so we are not neutral, and we are not pretending to be. We are not alleging that anyone at the Division acted in bad faith, took anything, or broke the law. We are not alleging a cover-up, because we cannot see inside the building.

What we can see is the list above, which is what the agency with the answers has chosen not to say, and a two-page document in which the agency with the authority named an apple that contains no apple. Everything in this piece that is a fact is cited to the advisories, the bulletin, the rule table, the federal guidance, and the peer-reviewed literature, and we will provide every document to anyone who asks.

One more thing, about what comes next. This piece began as a response to a single bulletin. The work behind it turned up a great deal more about how this industry is regulated, from 12 years of the Division's own enforcement orders, to three state audits of the agency, to the way the state's testing system actually operates. That material is going into a much longer piece we are still finishing.

We published this one first because the recall and the bulletin are live, licensees are being told to act on them now, and the questions above are easier to answer before the next advisory, embargo, or rule arrives than after it.

We are also publishing it as a courtesy, to the industry and to the Division. Everything in it is correctable, and we would rather see it corrected than defended: name the ingredient, fix the example, and take the 16 zeros off the panel.

Every week that passes without those corrections costs licensees money in failed batches, embargoed inventory, destroyed product, and paperwork that proves nothing, and the people paying are the ones who did nothing wrong. Putting this on the record now gives everyone, the Division included, a chance to get ahead of that damage instead of absorbing it.

About the author

Scott Holden is the founder and CEO of Lab Effects, the first and longest-standing terpene company in the cannabis and hemp industry, and of CBD Global, which provides raw material supply, custom formulation, and private-label manufacturing. He has worked in the industry since 2012, with a background in biochemistry and molecular biology.

Disclosure: Lab Effects supplies natural flavorings and terpenes to licensed manufacturers, including in Colorado. It is not licensed by the MED or CDPHE and sells no cannabinoid products, but the guidance discussed here directly affects its business.

Sources

Every factual claim in this piece rests on one of the documents below, numbered in the order they are first cited. We will provide copies of any of them on request.

1. Colorado Department of Revenue (DOR) and Colorado Department of Public Health and Environment (CDPHE). Health and safety advisory: Staycon LLC dba Craft. September 25, 2026. https://drive.google.com/file/d/1Aw1VharZnUsiecFAjrUxjMMGrPhrLUQ_/view
2. Marijuana Enforcement Division. Industry Bulletin 26-10: Inactive ingredients in vaporizer delivery devices. October 7, 2026. https://drive.google.com/file/d/1iX06H6BzrKRFwFv27qxmzy0Ku5hBnOt_/view
3. Colorado DOR and CDPHE. Expanded health and safety advisory: Staycon LLC dba Craft. October 8, 2026. https://drive.google.com/file/d/1xaQQuTQfAf2MJdeO6wHiV4HMwsAvju5y/view
4. Colorado DOR and CDPHE. Health and safety advisory: Staycon LLC dba Craft (chlorfenapyr). October 13, 2026. https://drive.google.com/file/d/1FBZJyY9P2TuoenCS54VgZcRV9RfI4fKf/view
5. Regulators say hazardous marijuana vapes sold at 405 Colorado dispensaries. Denver Gazette. September 26, 2026. https://gazette.com/2026/09/26/colorado-warns-of-hazardous-marijuana-vapes-sold-at-405-dispensaries-6/
6. Colorado DOR and CDPHE. Health and safety advisory: Ware Hause LLC and South Platte Distributors LLC. June 6, 2024. https://drive.google.com/file/d/1rcQ8SVl7CWyI11LnNKLViIZCQJPg0e5d/view
7. US Food and Drug Administration. Inactive Ingredient Database search. https://www.accessdata.fda.gov/scripts/cder/iig/index.cfm
8. Marijuana Enforcement Division. Health and safety advisory update: Adverse health events reported related to Craft vaporizer delivery devices. October 1, 2026. https://drive.google.com/file/d/18juMY301rCChOGypColc3V-8aHE--r7H/view
9. Health problems reported after Colorado cannabis vaporizer recall. MJBizDaily. October 2, 2026. https://mjbizdaily.com/news/health-problems-reported-after-colorado-cannabis-vaporizer-recall/618431
10. Smoke and mirrors: How intoxicating hemp seeped into Colorado's market. ProPublica and Denver Gazette. January 29, 2026. https://www.propublica.org/article/colorado-marijuana-thc-intoxicating-hemp-regulation
11. Marijuana Enforcement Division. Industry Bulletin 26-09: Expanded residual solvent testing effective September 28, 2026. https://drive.google.com/file/d/1ScTLJJybPyfvLmV1By5W9IFg8CEoFy7R/view
12. Colorado cannabis cultivator files lawsuit against state regulators. Denver Gazette. March 2025. https://www.denvergazette.com/?p=11141
13. Judge rejects bid for emergency cannabis testing overhaul in lawsuit alleging Colorado failed to stop synthetic THC scheme. Denver Gazette. May 1, 2025. https://gazette.com/2025/05/01/judge-rejects-bid-for-emergency-cannabis-testing-overhaul-in-lawsuit-alleging-colorado-failed-to-stop-synthetic-thc-scheme/
14. Colorado judge dismisses lawsuit over marijuana testing rules. Denver Gazette. May 20, 2025. https://www.denvergazette.com/2025/05/20/colorado-judge-dismisses-lawsuit-over-marijuana-testing-rules-0b730ee4-6d80-5c1e-b855-42d607d9d2ae/
15. Pot grower's claims against regulators go up in smoke. Courthouse News Service. https://www.courthousenews.com/pot-growers-claims-against-regulators-go-up-in-smoke/
16. Colorado Department of Revenue. FY 2026-27 Joint Budget Committee briefing. https://content.leg.colorado.gov/sites/default/files/images/dor_jbc_briefing.pdf
17. Osher C. Colorado marijuana company recalls dozens of contaminated products amid regulatory delays. Denver Gazette. September 24, 2026. https://www.denvergazette.com/2026/09/24/marijuana-company-recalls-contaminated-products/
18. Colorado cannabis firm recalls vapes over cancer-causing solvent. MJBizDaily. September 25, 2026. https://mjbizdaily.com/news/colorado-cannabis-firm-recalls-vapes-over-cancer-causing-solvent/618301/
19. Colorado Department of Personnel and Administration. Class description and minimum qualifications: A2A criminal investigators. https://dhr.colorado.gov/class-description-and-minimum-qualifications-for-a2a-criminal-investigators
20. Colorado Department of Revenue. Marijuana data: Average market rates and sales reports. https://cdor.colorado.gov/data-and-reports/marijuana-data
21. International Council for Harmonisation. ICH Q3C(R8): Impurities: guideline for residual solvents. https://ema.europa.eu/documents/regulatory-procedural-guideline/ich-guideline-q3c-r8-impurities-guideline-residual-solvents-step-5_en.pdf
22. Webster GRB, Sarna LP, Mechoulam R. Conversion of CBD to delta-8-THC and delta-9-THC. US patent 7,399,872 B2. https://patents.google.com/patent/US7399872B2/en
23. Methylene chloride. 21 CFR §173.255. Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-173/subpart-C/section-173.255
24. Advisory for Colorado's Craft has broad implications. Simplifya. https://www.simplifya.com/advisory-for-colorados-craft-has-broad-implications/
25. StayCon recalls 49 vape products over methylene chloride. The Marijuana Herald. September 2026. https://themarijuanaherald.com/2026/09/colorado-marijuana-company-recalls-49-vape-products-over-methylene-chloride-contamination-days-before-expanded-testing-begins/
26. Colorado marijuana vape recall: Craft and regulatory delay. ProPublica. September 25, 2026. https://www.propublica.org/article/marijuana-vape-recall-craft-colorado-regulatory-delay
27. New Zealand Institute of Chemistry. Turpentine production and processing. https://www.nzic.org.nz/unsecure_files/book/4F.pdf
28. Method for producing alpha-pinene from turpentine. CN patent 103497084B. https://patents.google.com/patent/CN103497084B/en
29. Tailoring turpentine for fine chemicals. Braz J Chem Eng. 2024. https://link.springer.com/article/10.1007/s43153-024-00520-5
30. PubChem. Dichloromethane (boiling point). https://pubchem.ncbi.nlm.nih.gov/compound/6344
31. PubChem. Alpha-pinene (boiling point). https://pubchem.ncbi.nlm.nih.gov/compound/6654
32. The Good Scents Company. Alpha-pinene, CAS 80-56-8. https://www.thegoodscentscompany.com/data/rw1006351.html
33. Marijuana Enforcement Division. Final administrative actions register (Ware Hause LLC and South Platte Distributors LLC stipulation, January 22, 2025). https://med.colorado.gov/final-administrative-actions
34. Colorado marijuana regulators consider changes in testing sample products. ProPublica and Denver Gazette. February 11, 2026. https://www.propublica.org/article/colorado-regulators-marijuana-lab-test
35. Colorado Natural Medicine Division. About us. https://nmd.colorado.gov/about-us
36. Kiselak TD, et al. Synthetic route sourcing of illicit at-home cannabidiol (CBD) isomerization. Forensic Sci Int. 2020. https://www.sciencedirect.com/science/article/abs/pii/S0379073820300359
37. Colorado Revised Statutes §44-10-202: Powers and duties of state licensing authority. https://colorado.public.law/statutes/crs_44-10-202
38. Marijuana Enforcement Division. Industry-Wide Bulletin 19-07: New health and safety rules specific to vaporizers. November 18, 2019. https://mjbizdaily.com/wp-content/uploads/2019/11/Colorado-Vape-Ban.pdf
39. Colorado Marijuana Rules, 1 CCR 212-3, Rule 3-335: Inactive ingredients in vaporizer delivery devices. Cornell Legal Information Institute. https://www.law.cornell.edu/regulations/colorado/1-CCR-212-3-3-335
40. Espino-Díaz M, Sepúlveda DR, González-Aguilar G, Olivas GI. Biochemistry of apple aroma: a review. Food Technol Biotechnol. 2016. https://pmc.ncbi.nlm.nih.gov/articles/PMC5253989
41. Dixon J, Hewett EW. Factors affecting apple aroma/flavour volatile concentration: a review. N Z J Crop Hortic Sci. 2000. https://www.tandfonline.com/doi/abs/10.1080/01140671.2000.9514136
42. El Hadi MAM, Zhang FJ, Wu FF, Zhou CH, Tao J. Advances in fruit aroma volatile research. Molecules. 2013. https://pmc.ncbi.nlm.nih.gov/articles/PMC6270112
43. Woronuk G, Demissie Z, Rheault M, Mahmoud S. Biosynthesis and therapeutic properties of Lavandula essential oil constituents. Planta Med. 2011;77(1):7-15. https://www.thieme-connect.com/products/ejournals/html/10.1055/s-0030-1250136
44. Colorado Marijuana Rules, 1 CCR 212-3. Adopted rule revisions of November 5, 2025, including the residual solvent table at Rule 4-215(J)(6)(a). https://drive.google.com/file/d/1WxClEnCvDiXNKLTAIx9Cg8xcLHLfCNXX/view
45. SC Labs. Updated residual solvent testing for Colorado cannabis compliance. https://www.sclabs.com/now-available-updated-rsa-testing-for-colorado-cannabis-compliance/
46. Synthetic flavoring substances and adjuvants. 21 CFR §172.515. https://www.law.cornell.edu/cfr/text/21/172.515
47. Marijuana Enforcement Division. Industry Bulletin 26-07: Delayed implementation of new testing requirements. June 29, 2026. https://drive.google.com/file/d/1F6oagX2q2dtzz0dA4D5RFaBH8UmxDIdm/view
48. Marijuana Enforcement Division. Industry Bulletin 25-16: Rulemaking adoption, new and revised testing rules. December 22, 2025. https://drive.google.com/file/d/19vyPO0L_Ezb7yculB0H3HzePKExb6wX1/view
49. Marijuana Enforcement Division. Industry bulletins index. https://med.colorado.gov/industry-bulletins
50. Massachusetts Cannabis Control Commission. Exhibit 7a: Concentration limits for residual solvents. https://masscannabiscontrol.com/document/exhibits-7a-and-7b-concentration-limits-for-residual-solvents-residual-levels-of-propane-n-butane-or-iso-butane-implemented-march-11-2021-and-revised-june-2-2021
51. California Code of Regulations, 4 CCR §15718: Residual solvents and processing chemicals testing. https://www.law.cornell.edu/regulations/california/4-CCR-15718
52. Washington Administrative Code, WAC 314-55-102: Quality assurance and quality control testing. https://www.law.cornell.edu/regulations/washington/WAC-314-55-102
53. Colorado Secretary of State. Basis and purpose attachment to 2025-00358. https://sos.state.co.us/CCR/Upload/AGORequest/BasisAndPurposeAttachment2025-00358.pdf
54. Colorado Secretary of State. eDocket tracking number 2025-00358: Adopted rules and attachments. https://sos.state.co.us/CCR/eDocketDetails.do?trackingNum=2025-00358
55. Marijuana Enforcement Division. 2025 rulemaking: Written public comments compilation. https://drive.google.com/file/d/1zqGTktll-hAvteaLtRnSwWfsxj70yDTP/view
56. Marijuana Enforcement Division. 2025 rulemaking announcement. July 11, 2025. https://drive.google.com/file/d/1Ru79nPZUznljCuuzT79WJDW2ySzeISYh/view
57. THC Pharm. Process for the preparation of dronabinol. DE patent 10106024B4. https://patents.google.com/patent/DE10106024B4/en
58. Bionorica. US patent 8,324,408 B2. https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/8324408
59. Noramco LLC. Process for the production of cannabidiol and delta-9-tetrahydrocannabinol. US patent 10,059,683 B2. https://patents.google.com/patent/US10059683B2/en
60. Socati. US patent 11,192,870 B2. https://patents.google.com/patent/US11192870
61. Canopy Growth. US patent 12,398,113 B2. https://patents.google.com/patent/US12398113
62. Canopy Growth. Improved methods for converting cannabidiol into delta9-tetrahydrocannabinol under neat or aprotic reaction conditions. US patent application 2022/0220090 A1. https://patents.justia.com/patent/20220220090
63. CCT Sciences. US patent 11,292,779 B1. https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/11292779
64. Novel delta-8-THC vaporizers contain unlabeled adulterants. Chem Res Toxicol. 2021. https://diverdi.colostate.edu/C442/references/concentrates%20and%20vaping/chem_res_toxicol_2021.pdf
65. Golombek P, et al. Conversion of cannabidiol into psychotropic cannabinoids. Toxics. 2020;8(2):41. https://www.mdpi.com/2305-6304/8/2/41
66. Ray CL, et al. Delta-8-THC product impurities. Molecules. 2022;27(20):6924. https://www.mdpi.com/1420-3049/27/20/6924/htm
67. Geci M, et al. The dark side of cannabidiol. Cannabis Cannabinoid Res. 2022. https://lcb.wa.gov/sites/default/files/publications/rules/2023-Proposed-Rules/CSWG3-DarkSide.pdf
68. European Monitoring Centre for Drugs and Drug Addiction. Technical report on HHC and related substances. 2023. https://www.euda.europa.eu/system/files/documents/2023-04/emcdda-technical-report-hhc-and-related-substances.pdf
69. Vape testing results. ProPublica and Denver Gazette. https://www.propublica.org/article/colorado-marijuana-hemp-thc-testing-results
70. Colorado Marijuana Rules, 1 CCR 212-3, Rule 4-215, version effective December 4, 2024. Cornell Legal Information Institute. https://www.law.cornell.edu/regulations/colorado/1-CCR-212-3-4-215
71. CDPHE. 6 CCR 1010-24: Regulated hemp facility manufacturing requirements, residual solvent list. https://www.sos.state.co.us/CCR/GenerateRulePdf.do?fileName=6CCR101024&ruleVersionId=11299
72. Joint FAO/WHO Expert Committee on Food Additives (JECFA). Evaluation of certain food additives: 63rd report (WHO Technical Report Series 928). 2004. http://apps.who.int/iris/bitstream/10665/43141/1/WHO_TRS_928.pdf
73. Flavor and Extract Manufacturers Association (FEMA). Safety and regulatory authority to use flavors: focus on vaping products. https://www.femaflavor.org/safety-regulatory-authority-use-flavors-focus-vaping-products
74. Sigma-Aldrich. Trans-2-hexenal safety data sheet, W256005. https://ehslegacy.unr.edu/msdsfiles/26630.pdf
75. FUJIFILM Wako. 2-Hexenal safety data sheet. https://labchem-wako.fujifilm.com/sds/W01W01080688JGHEEN.pdf
76. PubChem. 2-Hexenal: GHS classification. https://pubchem.ncbi.nlm.nih.gov/compound/5281168
77. US Food and Drug Administration. Using the Inactive Ingredient Database: Draft guidance. July 2019. https://www.fda.gov/media/128687/download
78. US Food and Drug Administration. Inactive Ingredients Database: Frequently asked questions. https://www.fda.gov/drugs/drug-approvals-and-databases/inactive-ingredients-approved-drug-products-search-frequently-asked-questions
79. THC vape recall hits 320 Colorado retailers. Westword. June 2026. https://www.westword.com/news/thc-vape-recall-hits-320-colorado-retailers-40894026/
80. Colorado DOR and CDPHE. Health and safety advisory: Workhorse Global LLC and WH Acoma LLC (PAX and Eureka Vapor). June 1, 2026. https://drive.google.com/file/d/1yFEI1eJSuhjnk8tqt75XekV5lOs2kJzb/view