The PFAS-Free Claim

Act I — The Hanger

A single word on a hanger is supposed to do a great deal of work.

It is supposed to audit a rulebook that is plural. It is supposed to audit a method-set that is plural. It is supposed to audit a time-horizon that is plural. And the composition of that plurality — which rulebook, which method, which vintage — is the consumer's unpaid labour, performed on inputs the consumer is never given.

That is the finding. The rest of this report is how we arrived at it, and why it does not go away when the counter-readings are run.

The object under examination is Lululemon Athletica. On 2026-04-13, the Texas Attorney General served a civil investigative demand on the company, citing a 2022 third-party test, published by the consumer-testing site Mamavation, that reported fluorine above the investigator's detection threshold on several Lululemon activewear units. The demand arrived 1,547 days after the test was published.

Three days later, on 2026-04-16, Lululemon published a new corporate response document titled "Created Without PFAS — What to Know About Lululemon's Products."[¹⁸] What the company said about PFAS before the civil investigative demand, and what the company said about PFAS three days after the civil investigative demand, are not the same sentences. This report takes that fact as its first data point.

The pre-CID language — captured by this pipeline's analyst on 2026-04-18 from the then-still-resolving legacy chemicals-management URL[¹⁷] — read, verbatim:

"Since 2022, our apparel and accessories are free of intentionally added perfluorinated and polyfluorinated substances (PFAS). We are working to meet the EU's REACH legislation. Since 2022, our performance water- and stain-repellant coatings have been free of intentionally added PFAS."

Three sentences. Three architectures. Two modal verbs (are, are working). One method-status disclosure: zero. One temporal anchor: since 2022. One legal qualifier: intentionally added. One statutory reference: REACH.

The post-CID document, captured live on 2026-04-21 at 15:48 UTC from the response-document URL, carries a different pair of sentences. Verbatim:

"lululemon does not use PFAS in products today."

"Our ongoing focus is to help prevent the unintentional reintroduction of PFAS into our products through ongoing testing, monitoring, and collaboration with suppliers and third parties."[¹⁸]

The temporal anchor is gone. The intentionally added qualifier is gone. The REACH reference is gone. The free of outcome-claim is gone, replaced with a present-tense process-statement (does not use ... today). And in the closing sentence, a phrase appears that was not in the pre-CID language at all: unintentional reintroduction. A chemistry pathway the pre-CID sentence foreclosed by qualifier is, in the post-CID sentence, conceded by architecture.

This is not the paragraph a detective writes at the start of the investigation. This is the paragraph a detective writes when the suspect has already moved.

The accusation this report is preparing to bring is not against either pair of sentences. The accusation is against the architecture that wrote the sentences, and against the architecture that rewrote them — the architecture that permits a plural rulebook to be compressed into a singular claim, permits the claim to be reshaped under regulatory pressure, and permits the consumer to absorb the compression cost on both passes.

It is worth saying, before we go further, what this report does not argue. It does not argue that Lululemon's leggings are or are not PFAS-free. That is a chemistry question, and chemistry has a home in Report 071. It does not argue that the company is lying. Lying requires intent, and intent is not the object here. The object is the rulebook. The rulebook permits what it permits; the company writes what the rulebook permits it to write, then rewrites when the rulebook points at it. Architecture wrote the sentence. Architecture wrote the replacement sentence. The company wrote them both the way architecture permits.

What follows, then, is a detective story about a rulebook — five of them, actually — and about the shape of the space between them, read against a corporate claim that has now appeared in two registers eight days apart.

Act II — The Five Columns

Put five columns next to each other. This is the dyad, doubled. On one side, the two verbatim stacks at two timestamps: the pre-CID language captured 2026-04-18 and the post-CID response-document language captured 2026-04-21, three days after the civil investigative demand crossed the company's desk. On the other side, the rule each of those sentences would need to satisfy in each of the five jurisdictions a global retailer operates in.

Read the first row against the third row. California's trigger is 100 ppm total organic fluorine. The EU's trigger for PFHxA in textiles is 25 ppb. The two numbers are not arithmetically commensurable — California counts total organic fluorine as a screen, the EU counts one named substance as a restriction, and a chemist will tell you the two methods see different things on the same fibre. That incommensurability is the architectural problem, not a flaw in the comparison. Taken on the face of the statutes, nonetheless, a four-thousand-fold gap separates the "passes" line in Sacramento from the "fails" line in Brussels on the same fibre. And the method-version disclosure that would tell a consumer which measurement is running underneath the hanger word is absent from the hanger, absent from the tag, and absent from either version of the sustainability page.

That is the first column of the finding. There is no method-version wordmark attached to the PFAS-free claim that tells a consumer which trigger the garment has been tested against. The consumer does not know which column they are reading.

Read the fourth row. The United Kingdom — the jurisdiction in which a large share of YAN's readership buys this category — has no textile-specific PFAS rule in force in 2026. DEFRA's REACH review is ongoing. The HSE's 2024 risk management options analysis has not yet moved to restriction.[⁴] The Competition and Markets Authority, under the Digital Markets, Competition and Consumers Act 2024, has a greenwashing power of up to 10% of global turnover. On the question of textile PFAS, the CMA has not used it. This is the UK null, and it is a finding. The argument here is not that the CMA has failed to act. The argument is that its silence, in a power-arrangement of this size, is itself a data point.

Read the final row. OEKO-TEX STANDARD 100 is the certification most likely to be on a hangtag in a UK-facing store. Its 2026 specification logs a non-PFAS total-fluorine exception framework (effective 2026-06-01) — certain bound-fluorine configurations may remain compliant under the "absence of PFAS" heading if they meet the institute's bound-fluorine and migration tests. The architecture of the label permits a fluorinated input to survive the "absence" test. Not because the certifier is hiding anything; because the definition of "PFAS" in the chemistry literature, the California statute, the EU restriction, and the OEKO-TEX specification is not one definition. It is four.

Now put those four definitions back underneath the verbatim stacks at the top of this act. The pre-CID stack said "free of intentionally added PFAS." Which PFAS — the California one, the EU one, the OEKO-TEX one, or the one the peer-reviewed chemistry literature (van der Veen 2022, Schellenberger 2022) measures? It said "intentionally added." Which part of the production chain — the mill, the DWR application, the trim, the dye carrier, the supplier's upstream chemistry vendor? It said "since 2022." Against which permitted method — the total fluorine screen that returns fluorine in parts per million, or the targeted EPA Method 1633 that returns PFHxA in parts per trillion?

The post-CID stack does not answer those questions by specifying them. It answers them by removing the surface on which they were asked. "Does not use PFAS in products today" names no definition, names no method, and sets the vintage at today — the reader cannot locate the claim against any of the five columns because the claim has been rewritten to sit above all of them. "Unintentional reintroduction" then introduces a new conceptual frame: PFAS can re-enter the product without the intentionally added qualifier being engaged at all. The pre-CID sentence foreclosed that pathway by qualifier. The post-CID sentence concedes the pathway exists and redescribes the company's job as preventing it, ongoing.

Both stacks are grammatically complete. Both stacks are architecturally underspecified. The architecture is the finding.

Act III — The Lattice (Is This Category Unique?)

The counter that would deflate this report, if it worked, is simple: every regulatory category has plural jurisdictions, plural methods, and plural vintages, so the claim you are making about PFAS-free reduces to "regulation is hard." This is a real counter. It is the one that requires the most work.

To answer it, we built a three-domain test. If the pattern we are calling the Epistemic Lattice — definitional plurality in the statute, instrument plurality in the method, and temporal plurality in the vintage, all compressed by a single consumer-facing word — appears only in textile PFAS, then this is a textile-PFAS story. If it appears in three unrelated regulatory categories, then it is an architectural story, and the textile case is an instance.

We tested three domains. The test condition: does each domain show (a) a statutory definition that is plural across jurisdictions, (b) a method or certification step whose standard version is not disclosed on the consumer-facing label, and (c) a temporal gap of at least twelve months between third-party evidence of a problem and regulatory response?

Three-of-three. The domains were chosen for maximum structural distance — a surgical implant, a seed allergen, and a fund-classification letter — and the pattern reproduces in all three. The lattice is architectural. Textile PFAS is an instance.

The Q4 2022 SFDR reclassification is the domain that teaches this argument hardest. In a single quarter, 307 European funds marketed as the stricter Article 9 "dark green" category were down-shifted to Article 8 "light green," together representing €175 billion in managed assets. The funds did not change their holdings overnight; the architecture of what the label meant changed, under regulatory pressure, and the prior marketing survived on secondary shelves for months afterwards. Individual enforcement cases followed — the DWS Investment Management Americas consent with the SEC (Order IA-6432, 2023-09-25, $19 million civil penalty for the ESG misstatements, part of a combined $25 million in sanctions announced the same day, with an additional $6 million civil penalty under a separate administrative order addressing anti-money-laundering programme deficiencies) is the most cited — but the structural evidence is the mass reclassification itself.[⁸] The individual case comes later; the architectural move comes first.

What the cross-domain test earns us is the right to say the finding publicly. The Epistemic Lattice is not a textile-industry complaint. It is a shape that regulatory architectures repeat. The PFAS-free claim is what that shape looks like in a garment.

Act IV — The Unpaid Audit

Stand in a shop. Any of them. A rack of leggings, seventeen units of the same SKU, a hangtag on each, the words "PFAS-free" among the other words. You have sixty seconds before the child at your knee asks why. The five columns of the table you just read — California, New York, EU, UK, OEKO-TEX — are running behind your eyes. Four of them you cannot consult at the rack. The fifth you can see, because the OEKO-TEX mark is on the tag, but you cannot read which year's annex it was certified against. You make the call anyway. The architecture asks you to.

Who, then, is doing the work the hanger pretends to do?

In 2022, the consumer-testing site Mamavation commissioned Graham Peaslee's laboratory at the University of Notre Dame to screen a cohort of yoga pants by the particle-induced gamma-ray emission (PIGE) methodology the Peaslee group had developed for detecting total fluorine in consumer textiles. The Mamavation protocol reports a detection threshold above 10 ppm TOF and flagged several Lululemon units as returning a positive signal.[⁹] Van der Veen and colleagues (2022) subjected a C6 side-chain fluoropolymer on polyamide to ISO 6330:2012 domestic-wash cycles at 40 °C. Extractable 6:2 fluorotelomer alcohol rose from 87 μg/kg (aged fabric baseline, unwashed) to 430 μg/kg after five ISO 6330:2012 domestic-wash cycles on aged fabric at 40 °C — a 4.9-fold increase in the extractable fraction of a PFAS degradation product from the same fibre, measured across a realistic domestic wash protocol (10-cycle samples showed a further rise to 520 μg/kg).[¹⁰] Schellenberger and colleagues (2022) weathered a polyamide fabric treated with a commercial C8 side-chain fluoropolymer finish on a Sydney rooftop for six months; extractable PFOA rose from 12 ng/g to 513 ng/g — a 42.75-fold increase, measured by direct outdoor weathering on a C8 fluoropolymer substrate.[¹¹]

The state did not find the chemistry. The state was cited to the chemistry. Texas Attorney General Ken Paxton's civil investigative demand, served on Lululemon on 2026-04-13, cites Mamavation as the originating evidence.[¹²] The interval from publication to state action is 1,547 days. The audit that triggered the investigation was not performed by a public agency; it was performed by a consumer-testing site and a peer-reviewed lab paid by neither state, and the public reference to the sequence is the Paxton CID. That is the unpaid labour sentence in this report. The consumer, not the state, funds the first rail of evidence. The state is the follower.

This is the moment to cross-reference Report 046, The Verification Economy, which documented the same pattern in infant-formula labelling: the first rail of disclosure was absorbed by third-party verifiers operating on donation, subscription, or advocacy funding, not by the statutory body whose mandate implied it should have been the first rail.[¹³] The pattern is visible here because it is the same pattern.

What 072 adds to 046 is the method layer. In the infant-formula case, the question was whether the label said the right number. In the PFAS-free case, the prior question is which method produced the number. The three-method trichotomy — total fluorine (TF), extractable organic fluorine (EOF), and targeted analysis (EPA Method 1633 or equivalent) — sits inside the statute, not inside the chemistry. California trigger: total fluorine. EU trigger: targeted PFHxA. OEKO-TEX screen: extractable, total-fluorine weighted. Three methods, three floors. One hanger word.

The counter to this — the version that deserves a direct response — is that Lululemon has moved to the OEKO-TEX STANDARD 100 and to the Apparel Impact Institute's RSL 9.0, which is the 2025 revision of the industry's restricted substance list, and that RSL 9.0's fluorine threshold (50 ppm TOF in the fibre, with AFIRM alignment) is a higher-bar operational standard than the statutes.[¹⁴] This is correct, and it is worth naming. Does a voluntary industry threshold at 50 ppm TOF — tighter than California's 100 ppm, absent from UK statute — partially close the audit gap? On the measurement floor, the answer is yes: operational compliance at 50 ppm is a stricter number than any codified statute applying to a UK-shipped legging in 2026. On the audit-origin question, the answer is no. The rail that found the chemistry in 2022 was not RSL 9.0's conformance apparatus; it was Mamavation's commissioned test and van der Veen's peer-reviewed wash cycle. The rail that cited the state in 2026 was Mamavation's publication, not an industry conformance letter. RSL 9.0 tightens the floor after the fact. It does not fund the first audit of the floor, does not name the audit's author, and does not cite the audit's methodology in its public-facing marks. The architecture that underspecifies the hanger word is the architecture that leaves the first rail unfunded, whether the operational floor is 50 ppm or 500.

The OEKO-TEX certification matters on a related but distinct axis. In 2026, OEKO-TEX logged a non-PFAS total-fluorine exception framework, effective 2026-06-01, permitting certain bound-fluorine configurations to survive the "absence of PFAS" test if they pass the institute's leachability panel.[¹⁵] Under REACH Annex XVII Entry 79 as revised by Regulation 2024/2462, the EU's 25 ppb textile threshold for PFHxA does not apply to the bound-fluorine class the OEKO-TEX 2026 exception covers. The certification, the statute, and the consumer claim are in three different conversations. A single word on the label compresses all three.

Act V — The Leading Indicator

The cases above are stories in the past tense. The leading indicator is in the present tense.

ECHA's Committee for Risk Assessment issued an opinion on 2026-03-03 on the universal PFAS restriction proposal covering roughly 10,000 PFAS substances across fifteen sectors, including textiles, apparel and related accessories (the TULAC scope), recommending extension of Entry 79-type controls beyond PFHxA to further chain-length configurations in textile applications.[¹⁶] If the RAC opinion survives the SEAC socio-economic review and transposes into a regulation, the 25 ppb anchor will be joined by named-substance thresholds for additional perfluoroalkyl carboxylic acids — and the "intentionally added" modifier in a corporate sustainability sentence will start to fail category-specific chemistry that current total-fluorine screens would have cleared. The hanger word will not change. The rulebook will change under it.

That is the leading indicator. It is the reason this report is not a retrospective. A plural rulebook becomes more plural as the science moves, and the compression the consumer is paying to perform becomes costlier per purchase. The consumer's unpaid labour, in other words, is inflating.

There is one further jurisdiction the report must land before it closes: the UK. The CMA has a greenwashing power of up to 10% of global turnover, and on the question of textile PFAS, it has not used it. DEFRA's REACH review is ongoing. HSE's 2024 RMOA is on the desk. There is no current textile-PFAS restriction in force in the United Kingdom. This is not rhetorical. It is a finding. The UK is the jurisdiction in which a reader buying this category has the least codified protection, and it is the jurisdiction in which the certification most likely to appear on the hangtag (OEKO-TEX STANDARD 100) is most likely to have been certified to a total-fluorine screen whose 2026 exception the UK statute does not speak to. The silence in the UK column is the silence that the word on the hanger completes.

A last clarification on scope. This report is not a reading of the CMA's intent. The CMA has not opened a textile-PFAS investigation under the DMCCA; it may yet. The category sits inside the PFAS inquiry's silence, not outside it. The argument from silence is offered as an observation on statutory architecture, not as an inference about regulatory intent.

So what is left to say? This.

Two pairs of sentences on a corporate page, eight days apart across a civil investigative demand, read against five jurisdictional columns, three method floors, and three temporal lag periods — all compressed into a single word on a hanger — describe a shape we have now seen in medical devices, in seed allergens, and in investment-fund classifications. The shape is the architecture. The architecture wrote the sentence. The architecture wrote the replacement sentence. The rulebook moves under the hanger word, and the hanger word moves under the rulebook. The consumer is the one holding the hanger, performing the audit on both versions of the claim, and paying for the unpaid labour twice.

What Would Change This Analysis

The finding rests on three claims, each with a specification under which it would be revised.

Claim 1 — The Lattice is architectural, not categorical. The plural-rulebook-plus-method-silence-plus-temporal-gap pattern repeats across regulatory categories beyond textiles. Would change if the three-domain test (Essure, sesame, SFDR) fails on re-examination of any of the three tested domains with primary-source correction, or if two additional domains selected with equivalent structural distance from textiles fail to reproduce the pattern when tested to the same three-part condition.

Claim 2 — The temporal gap is an architectural permission, not a one-case anomaly. The 1,547-day interval between Mamavation's 2022-01-18 publication and the Texas Attorney General's 2026-04-13 CID represents a gap the architecture permits, not a case-specific delay. Would change if a comparable third-party-funded disclosure preceding state action by fewer than thirty days is found in the textile-PFAS record, or if a systematic survey of consumer-initiated regulatory evidence in the category returns a median lag under twelve months.

Claim 3 — The hanger word compresses irreconcilable architectures. The OEKO-TEX 2026 non-PFAS total-fluorine exception framework (effective 2026-06-01), the EU 2024/2462 Entry 79 25 ppb PFHxA threshold, and the California 100 ppm TOF trigger cannot be reconciled into a single "PFAS-free" hanger claim without method-version disclosure. Would change if a harmonised method-version disclosure framework is promulgated by the major textile certifications and adopted as a labelling requirement in the UK, EU or US by 2027, or if a single certification standard publishes an alignment annex that explicitly reconciles the three floors on the same consumer-facing claim.

The report will be revised when any of the three falsification conditions is met.

The question we leave open is the one a detective should leave open. If the first rail of textile-PFAS evidence in 2022 was Mamavation, what is the 2026 analogue, and who is paying for it now?