Fibre composition
A composition declaration is a percentage of fibre mass. Buttons, zips, leather trims and interlinings are not in it. Four rules in a regulation already in force decide what the declaration may say, and two tolerances decide when a difference between the label and a laboratory is not an error at all.
On this page
What the field actually is
A fibre composition is a set of percentages over the fibre mass of the product. Two things follow from that sentence and both are routinely lost.
The denominator is fibre mass, not product mass. A metal zip, a horn button, a leather trim and a fusible interlining are all part of the garment and none of them is part of the composition. Leather sits in its own field rather than being folded in as a percentage.
And the percentages are not forced. Where a declaration does not add to 100, it is recorded as it stands rather than being renormalised into something tidier. A composition that adds to 97 is telling you something. A composition silently scaled to 100 has thrown that away.
Every figure downstream inherits that denominator, and we say so on the row.
Four rules already decide what it may say
Almost everything written about textile passports is written in the future tense. Fibre composition is the exception. Regulation (EU) No 1007/2011 on textile fibre names and labelling has been in force for years, and four of its rules settle what a composition declaration is allowed to be.
The framework that creates the passport is not the instrument doing that work. It sets no textile field list, and no textile delegated act has been adopted, so what binds a composition today comes from the labelling regulation rather than from the passport one.
| The rule | What it does to a declaration | Provision |
|---|---|---|
| A closed vocabulary | Only the fibre names on the list in the Annex may be used, and a name outside it may not be brought in as the root of a word or as an adjective attached to a permitted one. | Art. 5, Annex I |
| Descending order by weight | Percentages are by weight and are written from the largest to the smallest. An order chosen because it reads better is the wrong order. | Art. 9 |
| Accuracy sits with a named party | The manufacturer supplies the label and guarantees the accuracy of what is on it when placing the product on the market. Where the manufacturer is outside the Union, the importer stands in that place. | Art. 15 |
| Available before the purchase | The composition reaches the buyer before they buy, and that holds where the offer is made electronically rather than across a counter. | Art. 16(1) |
Read together they change what kind of thing a composition is. A description of a fabric can be written well or badly. A declaration under these four rules can be right or wrong, and the difference is checkable by anybody with the list and the product.
The practical consequence turns up in live product data constantly. A garment described as 70 per cent bamboo has a problem, because bamboo is not one of the permitted names. The fibre it almost always refers to is viscose, and the correct name is the one the law recognises. The root and adjective rule is why bamboo viscose does not rescue it either: attaching a permitted word to a name that is not on the list does not put it on the list.
One caution about the fourth rule. We have not read that article in full at the Official Journal, so it is stated here as we read it at the consolidated text and the source list below records that, rather than the caveat being quietly left out.
What the regulation permits between the label and the analysis
A declared composition and an analysed one do not have to be identical. The same regulation carries two tolerances. They measure two different gaps, and there is a narrow gate on adding them together. A reader who knows only the smaller figure is holding a third of the position.
| Provision | What it permits | The condition attached |
|---|---|---|
| Extraneous fibres Art. 7(2) | Fibres other than those named on the label, up to two per cent of the weight of the textile product, rising to five per cent of that same weight for a carded product. | Two conditions, and both have to hold. The extraneous fibres are technically unavoidable in good manufacturing practice, and they are not routinely added. |
| Manufacturing Art. 20(3) | A difference of three per cent between the composition as stated and the composition as it comes back from analysis by the prescribed method. | The comparison is against that method, on its stated mass basis and with its allowances. A number produced some other way is not the thing this tolerance is measured against. |
| Adding the two Art. 20(4) | The two allowances applied cumulatively rather than separately. | Permitted only where the extraneous fibres found are of the same chemical type as fibres already named on the label. Outside that circumstance the two stay apart. |
The carded figure is the one that goes missing from summaries. It is higher because carding is a process in which stray fibre is genuinely hard to keep out, and it is fenced by the double condition rather than being available on request. A product that is not carded does not get it, and a fibre somebody added on purpose does not get it either.
The gate on adding the two is where the comfortable error lives. Two per cent plus three per cent produces a cushion of five that feels defensible, and a carded product treated the same way produces one of eight. Neither is available unless the extraneous fibres found are of the same chemical type as fibres already on the label, which is a test about chemistry rather than about intention.
None of this is slack. A tolerance is a permitted difference between a composition somebody stated and the composition a prescribed method measures, and it protects a declarant who did the work. It does nothing at all for a value nobody evidenced, because there is no stated composition for it to be a difference from.
Among the attributes a passport carries, this is the one we have found with both a prescribed test method and a prescribed tolerance behind it. It is worth being exact about what is unusual in that, because it is not the shape. A prescribed method with a permitted deviation attached to it is a structure EU product law uses elsewhere.
What differs is who may lean on it. Here the allowance runs to the party making the declaration. In the comparable instrument we hold at second hand, the deviation belongs to the authorities carrying out official checks and is expressly denied to the manufacturer. We have not read that instrument at source, so it is named here by its shape rather than by its title.
How a declaration is read, and when it is refused
A composition can reach a passport from several places. A structured field the merchant already maintains. A variant option. A labelled block in a product description. And, sometimes, ordinary prose in a description that happens to contain the whole answer.
Prose is the interesting case, because it is where most systems either refuse everything or accept anything. The rule here is narrow. A description closes a composition only when it adds to exactly 100 across permitted fibre names with nothing in dispute. Anything short of that can propose a value for a person to confirm and cannot finish one.
Three refusals matter more than the acceptances.
Your own vocabulary is the authority
A term you list as a colour is not read as a fibre. Where a merchant has declared a structure, the declared structure beats a string that merely looks right, and a great many fibre names are also perfectly ordinary colour names.
A qualifier is not part of a name
Organic and recycled do not ride into a legal fibre name. They are separate claims with their own evidence, and the second of them turns on chain of custody records rather than on anything measurable in the garment, which is set out on recycled content.
A construction is not a fibre
A weave or a knit never implies a fibre content. Neither does a finish, a fabric trade name or a garment category. Where the composition is not stated, the honest output is the field left empty with a reason attached rather than a plausible guess.
One mapping is deliberately not made. Rayon is not treated as viscose, because in United States usage the word spans more than one of the legal fibre names and choosing one of them would be inventing information rather than translating it.
All of that happens before anybody tests anything. When a laboratory result arrives and differs from what a supplier declared, the question stops being how to read a value and becomes which of two values describes the goods. That has its own procedure, with the tolerances above as one gate among five, and it is set out on when a test contradicts the declaration. It is not answered here.
Two rows, side by side
Invented products, real value shapes. The point is the second one.
| Variant | What the store held | Passport value | State |
|---|---|---|---|
| Overshirt, size M | metafield: 100% cotton | Cotton 100% | Resolved |
| Knit jumper, size L | description: soft merino blend, colour: camel | empty | Not established |
The second row could have been filled. Merino is a permitted fibre name and camel is a permitted fibre name, so a system reading strings would have produced a composition with two fibres in it, no percentages and a fabricated denominator. Camel is the colour of the jumper. The merchant said so in their own colour option, and the merchant is right.
An empty box with a visible reason is worth more than a filled box that guessed.
What the reader gets instead is the field empty, the reason named and a question routed to the one person who can settle it. The vocabulary behind those two states is set out on the evidence page.
On the current 34 variant test catalogue this field resolves on 31 of 34 variants. That is a test catalogue rather than a customer, and it is the highest resolution rate of any field that depends on somebody outside the business.
Who is checking, and what they check with
A rule with nobody behind it is a style guide. This one has a regulator behind it and, less obviously, two commerce platforms.
The regulator's own measurement is worth putting on the record with the caveat that comes attached to it. A joint enforcement action published on 18 June 2026 examined 132 garments across 8 member states and recorded 49 of those 132 as non compliant on labelling.
In the same breath, here is what the publication does not give. It does not state the tolerance thresholds it applied. Without those a reader cannot tell whether a garment missed by thirty points or by a fraction over the three per cent manufacturing tolerance, so the figure is a count of recorded non compliance across 132 garments and not a rate of wrongness, and we do not republish it as one. Publishing that absence is more useful than publishing the number as though it settled something, and it is the same discipline we apply to our own figures.
The more interesting figure in the same publication is a gradient rather than a headline. Across three bands of declaration complexity, recorded non compliance rose as the declarations got more complex. That is a mechanism rather than a category: the more fibres a declaration carries, the more of it there is to get wrong, and the failures cluster where the arithmetic is hardest rather than in one kind of product.
Enforcement is not only public, and the second kind arrives faster. Two commerce platforms carry the labelling duty described above into their own published formats, which is where a catalogue actually meets it. One is a listing format that rejects a textile listing through its own error codes where the composition is missing or malformed, so the consequence is a listing that does not go live rather than a letter. The other asks for composition in layers rather than as one string, at product level and again below it, so a single free text sentence does not satisfy the format.
We describe both by class rather than by name. Our read of each format was made through platform documentation rather than at a first party console, both reads are dated to this page's source date on the list below, and neither supports any statement that a platform demands passport data. Whether a platform field named material can carry a composition at all is a separate question, and it is answered on what does not map.
Why this page exists, given what people search for
This vocabulary is barely searched. Our own keyword work puts the legal textile labelling terms in the lowest measured band, with no advertiser bidding at all on any of the terms that measured. Demand is not the reason to publish this page. The reason is above it: a rule with a live consequence is worth writing down even where almost nobody is looking for it.
What else on the passport depends on this
This is the field to fix first, and not because it is the most visible.
Several other attributes take fibre composition as an input. The modelled environmental figures need to know what the garment is made of before they can be produced at all. The microfibre supplement weights fibre categories, so a missing composition withholds the whole result rather than producing a smaller one, because a sum with a term dropped out of it gets quietly smaller and stays publishable and is wrong. A commodity code's last digits encode the dominant fibre, so a code cannot be checked against the garment until the composition is known.
We are not publishing an exact count of dependent attributes yet. Our own records give two different numbers for it and reconciling them against the governed source is work that has not been done. Publishing the wrong one of those on the strength of the record that was nearest to hand would be a small error in a page whose subject is not making small errors.
Where the modelling stops
The French regulatory calculator we use for the environmental figures does not recognise every legal fibre name. At the version we pin, 22 of 49 fibre names have no equivalent in it. A garment made of one of those cannot be modelled, and it is withheld rather than mapped to something adjacent.
What this means for a catalogue
Four things are worth doing and one is worth not doing.
Put the composition in a structured field rather than only in the description. A description that happens to parse today stops parsing the moment somebody edits it for tone.
Use the legal fibre names. Elastane rather than spandex, polyamide rather than nylon, viscose rather than bamboo. Merchants write the informal names constantly and every one of them is a value a machine has to guess at.
Write the percentages in descending order and keep the qualifier separate from the name. Recycled polyester is a polyester percentage plus a recycled content claim, and the second one needs its own evidence.
Where a laboratory result already sits in a folder somewhere, read it against the declaration now rather than when somebody else does. The tolerances decide whether there is anything to answer for, and they give the same answer whoever is asking.
The thing not to do is fill the field because it looks bad empty. A composition nobody has checked, sitting in a passport under a brand name, is a statement about a product that the brand is now making.
Where the values live and which of them a shop can already answer is covered in the Shopify guide.
You might want to read next
Sources
-
Arts. 5, 7(2), 9, 15, 16(1), 19, 20(3) and 20(4), Annexes I, VIII and IXCELEX 02011R1007-20180215In forceRelevant provisions reviewed
The spine of this page. The provisions read at the consolidated text are the closed fibre name list, the rule on order by weight, the extraneous fibre tolerance with its carded figure and both of its conditions, the manufacturing tolerance between stated and analysed composition, the provision governing when the two may be applied together, the analysis provisions with their method and allowance annexes, the provision placing accuracy on the party placing the product on the market and the provision making composition available before purchase. The article numbering carried on this page is our own transcription and is to be confirmed against the Official Journal, which is why every provision is also named by what it does. The provision on availability before purchase has not been read in full at the Official Journal, and the sentence stating it says so. Our transcription of Annex I is a transcription and has not had a legal review.
-
Commission joint enforcement campaign on textile labelling, 2026Regulator publicationReached through a secondary reproduction, primary text not read
Here for one figure, one gradient and one absence. The figures are published with their denominators. The absence is that the publication does not state the tolerance thresholds it applied, which is why its headline is not republished here as a rate of wrongness. One sub figure in the same campaign rests on five garments and is not published by us in any form.
-
Marketplace and platform seller policy documentationPlatform documentationRelevant provisions reviewed
The two platform formats described in the enforcement section. Read at platform documentation rather than at a first party console, which is why both are described by class rather than by name and why nothing here says any platform demands passport data. The read behind each format is dated to this page's source date rather than to a separate console session.
-
Arts. 7, 9, 11, Annex I and Annex IIICELEX 02024R1781-20240628In forceRelevant provisions reviewed
The framework that creates the passport. It does not set textile field requirements, and no textile delegated act has been adopted. It is on this page to mark the boundary between what binds this field today and what does not yet exist.
-
Version pinned at v7.0.0Government modelReviewed in full
The source of the fibre coverage figure on this page. The calculator's own answers are republished unaltered and attributed to it rather than to us.
Help someone else make sense of product passports.