Almost every activewear label carries a number nobody in the chain has actually tested. A mill writes 88% polyester, 12% elastane on a construction sheet; that figure goes into the tech pack, onto the care label, onto the product page and into the marketplace listing, and from there it is treated as a fact. In the EU it is a regulated claim, and the party who answers for it is the brand named on the label — not the mill that supplied the yarn. This guide is about how that number goes wrong, and what to do before somebody else tests it for you.
The short answer
- The fibre percentage on your label is a claim you are making. In the EU it sits under Regulation (EU) No 1007/2011, and the brand named on the label answers for it, not the mill.
- A 2026 EU market-surveillance campaign tested 132 clothing samples and found 37% non-compliant. Blended fabrics — which is essentially all activewear — failed at 64%.
- The number drifts in ordinary production: yarn substitution when a count is unavailable, finishing and dye-lot changes, a mill quoting the nominal recipe rather than a tested result, and occasionally a cheaper fibre swapped in.
- Fix it in the specification rather than after: declare composition per component, ask the partner facility for its own test report and check who it was issued to, and commission an independent quantitative test where the exposure justifies it.
The sequence that produces the failure
It starts with a construction sheet. The mill quotes a fabric as 88/12 polyester-elastane, and that pair of numbers is copied into the tech pack. From there it goes to the care label supplier, to whoever writes the product page, and into the marketplace listing where composition is a required field. Four systems now hold the same figure, all four inherited from one line on one document that nobody checked against the fabric.
Then somebody tests it. A market-surveillance authority buys a sample off a website. A retailer's QA team runs composition during onboarding. A marketplace verifies a listing after a complaint. The result comes back at 92/8, or with a fibre in it that was never declared, and the instruction that follows is addressed to the brand: relabel the stock or withdraw it. The mill is not the addressee. It sold a fabric; you made the claim.
Nothing dishonest has to happen anywhere in that chain for it to end this way. The mill quoted what it intended to knit. The factory sewed what it was sent. You printed what you were told. The failure is procedural rather than moral, which is why it is so common — and why trusting your supplier, however well founded that trust is, does not prevent it.
What the 2026 EU campaign actually found
The JACOP 2025 joint market-surveillance campaign published its results on 18 June 2026. Authorities in Cyprus, Denmark, Finland, Germany, Italy, Lithuania, Malta and Portugal tested 132 clothing samples — including 15 activewear items and 36 baby garments — against Regulation (EU) No 1007/2011. Forty-nine of the 132, or 37%, were non-compliant. The most common problems were incorrect fibre percentages and, in some cases, undeclared or substituted cheaper fibres. The consequence was withdrawal from the market or mandatory relabelling. We are working from an Intertek regulatory bulletin dated 19 July 2026 reporting the campaign.
The breakdown is where it gets pointed. Blended fabrics failed at 64%. Items bought online failed at 46%. By product type, scarves were worst at 80%, tops came in at 54%, and baby clothing — the segment under most scrutiny — was lowest at 25%. Read the first two figures together if you are a direct-to-consumer activewear brand: your product is a blend and you sell it online.
Be honest about what 132 samples can support. This is a surveillance sweep, not a census, and authorities pick samples they consider worth checking rather than at random. It does not show that 37% of everything on sale is mislabelled. It does establish that when trained inspectors go looking at blends specifically, they find problems in most of them — a narrower claim than a scare statistic, and a more useful one.
Why the number drifts between the sheet and the roll
Start with what a construction sheet usually is. In most cases it states the nominal recipe — the blend the mill intends to produce — rather than a laboratory result measured on the lot you are buying. Nominal and tested are two different kinds of statement, and a great many labels are built on the first while a regulator tests against the second. Ask a mill outright which of the two you have been given. The answer is often that nobody had asked before.
Then production intervenes. A yarn count runs out and a substitute goes in, sometimes at a different denier, sometimes with different elastane content in the covered yarn. Finishing shifts the balance: scouring, heat setting, brushing and any coating change how much mass each component contributes, and composition is declared by mass. Dye lots behave differently from one another. Each of these is small, routine and unremarkable on the day, and each moves the ratio.
This is also why blends drift and single-fibre fabrics largely do not. A 100% cotton jersey is either cotton or it is not; the claim is binary and hard to get wrong by accident. A blend is a ratio, so anything that changes the mass of one component changes the number describing all of them. Add a third fibre, or a bonded panel with its own composition, and the problem compounds rather than adds. Then there is the deliberate version — a cheaper fibre substituted, an expensive one quietly reduced. The campaign found that too.
What a composition test is, and what it does not tell you
A quantitative fibre-composition test does one narrow thing: a laboratory separates the fibres in a sample and reports each as a percentage by mass. Where the fibres can be picked apart physically — different yarns in different courses, for instance — the separation is mechanical. Where they cannot, it is chemical: a solvent dissolves one component and what remains is dried and weighed. A usable report names the laboratory, states the method, describes the sample it received, and gives the result by mass. Ask for the method to be named rather than assumed.
What the report does not do is describe your order. It describes the piece of fabric somebody put in an envelope. Which roll, which colourway, which lot and who selected it are part of the evidence, not administrative trivia. And a result close to your declared figure still has to be read against the tolerance rule for your market. Tolerances do exist. We are not quoting one here, because the number that matters is the one in the current rule for the market you sell into — worth confirming rather than recalling.
The cost question is real and we will not pretend otherwise. You are not going to test every colourway of every style, and a brand that tries will spend its budget on the least risky part of its range. Spend where the exposure concentrates: a blend construction you have not run before, a first order large enough that relabelling it would hurt, a retailer or marketplace onboarding where composition is checked as routine, and any product where composition is part of the selling story — a stated recycled percentage, a merino content, an elastane figure you market as the reason it holds shape.
What to ask a partner facility for
Ask for composition on the production fabric, not on the development sample. This matters more than it sounds. Sample yardage is frequently a different lot and occasionally from a different mill, and it exists to answer questions about hand feel, colour and fit. Stock samples with us run 3–4 days and a custom first sample takes around seven — a fast loop for those questions, and no evidence at all about the composition of a roll knitted months later for bulk.
Ask whether the mill holds a composition test report of its own, and if it does, ask two questions before relying on it: who was it issued to, and what sample does it describe. A report issued to a different buyer, on a different lot, is background rather than evidence about your goods. The same discipline applies as to any certificate — holder, article, date. A report with no sample description is not a document you can forward to a retailer.
Then ask what happens when a yarn is unavailable. The answer you want is that a substitution needs written approval before it goes into the fabric. The answer you will often get is that they will source something equivalent — a sound operational instinct and a labelling problem at once, because equivalent for knitting performance and equivalent for a declared percentage are not the same test. Settle this before the first bulk run, not during one.
Put the number in the specification, not in an email
Composition belongs in the written material specification, stated per component. A single garment often carries several: main body, mesh panel, waistband, lining, binding, gusset. A specification naming only the main fabric leaves the rest to whoever happens to be filling in the label. State which components are in scope, state the test method to be used if the figure is ever checked, and state which document governs if the mill's sheet and the label disagree. A number in a specification is contractual. The same number in a chat message is a recollection.
Add the re-declare triggers while you are there: a new mill, a new yarn supplier, a substituted count, a new finish, a new lot after a long gap. Each is a fresh question rather than a continuation of an old answer. And if you require evidence, tie it to release or payment in the purchase terms as you would any other condition — a test requested after the goods have shipped is a favour that can be politely declined.
One piece of adjacent housekeeping, since you are already editing labels. Under the US Textile Fiber Products Identification Act, the name required on the label is the name under which the business is done (16 CFR §303.19(a)). A registered identification number may be used in lieu of that name (§303.19(b) and §303.20) — meaning an RN is an option, not an obligation. Worth knowing, because suppliers and label vendors sometimes ask for one as though an order cannot ship without it.
Where we sit, and what none of this buys you
SEAMDANCE has been a buyer-side trading and supply-chain management company in Xiamen since 2018, coordinating independent specialist factories, mills and dye houses. On this problem, what we can do is bounded and worth stating exactly. We can put composition into the written material specification, per component, so the figure is contractual rather than conversational. We can ask the partner facility for its supporting data and tell you plainly when that data does not exist — an answer buyers need more often than they expect. And we can route an independent composition test to a third-party laboratory when a program warrants it, with the report issued in your name.
What we cannot do matters as much. We do not operate a laboratory, and we do not certify a label as correct — nobody in a supply chain can. Our inspection standard is AQL 2.5, and it will not help here: AQL sampling covers workmanship and measurement, and no visual inspection on any line detects a fibre percentage. A supplier who says their QC checks composition during production is describing something that does not exist.
Nor does testing make you compliant, and we are not going to tell you it does. A report describes what was in one sample on one day; compliance is a judgement about your market's rules, its tolerances and your whole shipment. What the work in this guide buys is narrower and still worth having: the number on your label came from a measurement rather than a construction sheet, and you can show where it came from when somebody asks. That is a defensible position, which is the part a brand can control. This is general information about labelling practice, not legal advice — check the current rules for each market you sell into, and take professional advice where the exposure justifies it.