A 100% polyester flat weave passed 60,000 Martindale rubs to ISO 12947-2 and still pilled badly on the armrests of 320 sofas, because the abrasion coupon was flat, tension-free and cut from the middle of the roll, and the armrest is none of those three things.
Illustrative example This case is written from the patterns we see in furniture export orders, and the figures are representative rather than from a specific client file. It shows how a check works and what it finds. We will replace it with a real, anonymised case as soon as we can publish one with the buyer's permission.
The order, and the clause that was not in it
A European serviced-apartment operator ordered 240 three-seat sofas and 60 armchairs for a single building fit-out. The upholstery was a 100% polyester flat weave at 420 g/m², and the purchase order named one performance figure: abrasion not less than 60,000 rubs to ISO 12947-2. Fibre content, fabric weight, colour, and the full panel construction were all written down. There was no line about pilling.
The mill sent its documentation with the first production sample, and the documentation was accurate. Article 4478, 100% polyester, 420 g/m², Martindale 65,000 rubs to ISO 12947-2, specimen breakdown. A line further down the same sheet recorded pilling grade 4 to ISO 12945-2. Every number on that sheet was true of the fabric, and every one of them came from a flat specimen cut from the middle of a roll.
Nobody was going to find the missing clause by reading the paperwork, because nobody was looking for it. Pilling is not a defect that shows up in a container. It shows up four months later, on the arm of the sofa in unit 214, gone from the specified grey to a mottled speckled grey, rough under a hand.
This is an illustrative example built from the patterns behind furniture export disputes: the goods are representative of the category, the figures are representative of the trade, and no particular buyer, factory or operator is described. It shows what a check finds when the specification defines the strength of a fabric but not the way it gets worn.
Where to go next
→ upholstery checks on sofa orders
→ the three levels of check we sell, and what each one covers
What 65,000 rubs actually certifies
ISO 12947-2 is a specimen breakdown test. A 140 mm circular specimen is pressed against an abraded cloth and rubbed in a Lissajous path at 9 kPa until yarns rupture or a hole forms, and the count of rub cycles at that point is the number on the datasheet. It answers one question: how many rubs does this cloth survive before the structure fails.
It does not answer the other question a buyer cares about just as much, which is what the surface looks like after it has been rubbed for a while. That belongs to a different standard, ISO 12945-2 for the Martindale method, and it is run under a deliberately gentle load, 415 g for woven upholstery fabric, low enough that the fibres fuzz and tangle without the yarns breaking. The result is a visual grade from 1 to 5, 5 being no change and 1 being dense pilling across the whole surface, read at set intervals of 100, 1,000, 2,000 and 5,000 cycles for upholstery.
The two numbers on the datasheet therefore came from two different tests, with different loads, different endpoints and different units, and both were correct. The fabric was strong. It was also, on that specimen, not pilling. There is even a third part of the series that nobody quotes, ISO 12947-4, which assesses appearance change after abrasion, and that was not on the sheet either.
| Abrasion, ISO 12947-2 | Pilling, ISO 12945-2 | |
|---|---|---|
| What it measures | Structural failure of the weave | Surface appearance: fuzzing and balls of fibre |
| Endpoint | Yarn rupture, or a hole | Visual grade 1 to 5 against a photographic reference |
| Load | 9 kPa, 12 kPa for contract seating | 415 g specimen holder for woven upholstery |
| Evaluation | Cycle count to breakdown | Grade read at 100, 1,000, 2,000 and 5,000 cycles |
| A good number | 60,000 rubs or more for contract use | Grade 4 to 5 at 5,000 cycles |
| What it does not tell you | How the surface looks at 5,000 rubs | Whether the weave will survive 60,000 rubs |
Four months later
The first complaint was a photograph. A unit manager sent a picture of the arm of a sofa in a studio flat: the fabric along the top of the arm and around the front of the seat cushion had gone from the specified grey to a mottled, speckled grey, rough to the touch.
The operator's own team went back and looked. Pilling was on the arm, worst along the nose where forearms rest and slide, and on the front rail of the seat cushion. It was not on the back rest, not on the base, and not on the outside back panel that faced the wall. The unit manager had already worked out the pattern that matters, and it is the pattern reported across the trade: pilling follows contact, not cloth. The same fabric on the untouched back panel was clean.
The complaint was cosmetic, and the operator wanted a way forward rather than a laboratory argument. That is usually the point at which this kind of conversation turns into a dispute about whose test was right. Both tests were right. The question nobody had asked was whether a flat coupon was the right witness for the arm of a sofa.
Where to go next
Why a flat coupon cannot see an armrest
The specimen in a Martindale test is a flat disc, cut from a flat roll, mounted without tension, and rubbed evenly from several directions. It experiences no curvature, no pull across its width and no cut edge. The arm of a sofa has all three at once.
Curvature, because the cloth is stretched over the radius of the arm. Under tension the crimp in the yarn opens up and the fibre ends at the surface lift more easily, and that lifting is the first stage of pilling, where fibres migrate towards the surface under repeated low-pressure friction.
A cut edge, because the arm panel is cut and sewn, and cut yarns at a seam or a welt expose fibre ends straight to the first rubbing that arrives. A coupon from the middle of a roll has no cut edge in it at all.
Concentration, because a forearm on the nose of an arm loads a small curved area with the whole weight of the arm and then slides along it. The local pressure there is not the nominal pressure of a flat coupon, and the movement is directional rather than a Lissajous figure.
None of this is a fault in the fabric. It is the difference between a property of the cloth and a property of the cloth in service, and a test method that removes curvature and tension by design cannot report the second one.
How the difference was located
Nothing here needed a laboratory to establish, and that matters, because a sofa inspection can only report what it observed on the goods in front of it. On the delivered units the pilling zone was recorded against a fixed diagram, arm nose, arm front, arm top, seat front rail, seat front panel, back inside, back outside, and graded by eye against the photographic reference used for pilling assessment. Five units across the delivery, photographed, mapped and graded. The pattern was identical on all five: arm and seat front affected, everything else clean.
A comparative rub screen reproduced the complaint on the finished chair. A piece of standard wool cloth was drawn along the arm nose and then along the back of the same chair, 50 double strokes each, under the same hand pressure, and the lint was collected on white paper. Fuzz and small loose balls came off the arm. Nothing came off the back. The same comparison on a second unit gave the same answer, which is what makes it evidence rather than an anecdote.
Then the specimens went to an ISO/IEC 17025 laboratory, and this is where the case turned. Discs from the arm panel and from the back panel of the same delivered units, plus a coupon from the same mill lot, all read ISO 12945-2 grade 4 at 5,000 cycles. The arm fabric and the back fabric were the same fabric, and on flat discs they behaved identically, while on the chairs they plainly did not.
That result is not a disappointment, it is the finding. It proves that a flat laboratory specimen cannot settle this question, because the conditions that produce the pilling, curvature, tension and concentrated directional contact, are exactly what the standard removes. The honest limit of the report is therefore written into its final section: the pilling grade of the delivered lot in service was not established and cannot be established by any standard test, and the evidence available is a zone-by-zone differential on the delivered units plus a repeatable rub screen on the finished chair.
The mill was right, and the order was still wrong
This is the part that buyers tend to get wrong in their own favour. The mill tested what it was asked to test, using the standard method for that test, and reported accurately. A pilling figure taken on a flat coupon from the roll is the normal way pilling is measured, and the mill had no reason to do anything else.
The missing clause was the buyer's. The purchase order asked for a fabric that could survive 60,000 rubs and said nothing about what that fabric should look like after a year of forearms sliding along the arm of a sofa. Written that way there is no claim to make, because no requirement was breached.
There is a fair industry position worth stating plainly, because it is the mill's best argument and the buyer's real problem: pilling on upholstery is widely treated as normal wear rather than a manufacturing defect, and plenty of contracts say so explicitly. That position is defensible for a sofa that has been used for a year. It is much harder to defend when the first complaint comes four months after delivery and the fuzz appears on every unit in the building.
Which leaves one thing to do, and it is not to argue about the two numbers. It is to write down what the arm of the sofa has to look like after it has been in service, in a way that can be tested.
Where to go next
What the specification should have said
Six lines would have closed this. The order already had the fabric, the construction and the abrasion figure; what it lacked was a written statement of what the surface has to do under the conditions it will actually meet.
| Instead of | Write |
|---|---|
| no pilling line at all | Pilling to ISO 12945-2, or to ISO 12945-1 and ASTM D3512 for pile and loose constructions: grade 4 or better at 5,000 Martindale cycles, tested on the fabric as supplied to upholstery, after singeing and finishing |
| a grade with no cycle count | the cycle count and the evaluation point every time, so grade 4 or better at 5,000 cycles and not just pilling grade 4 |
| a mill datasheet accepted as evidence | a test report on the delivered lot from an ISO/IEC 17025 laboratory, with the article number and the finish state stated on the report |
| one fabric described once | the pilling requirement stated for each construction in the order, with the method that applies to each: ISO 12945-2 for flat wovens, ISO 12945-1 or ASTM D3512 for pile and loose constructions |
| no zone and no location | the acceptance zones named on the drawing: arm nose and top, front rail of the seat cushion, and back rest at head height, which are the places that take the friction |
| no mock-up | a first-off mock-up of the arm built from production fabric and rubbed to an agreed protocol, 100 double strokes with standard wool cloth along the arm nose, accepted with no visible fuzzing and no balls above 1 mm |
The last line is not a standard test, it is a protocol the two sides agree before the run, and it has to be agreed before the fabric is cut rather than after the first complaint.
What the factory can change, and what only the fabric can change
There are two levers, and they sit in different places, which is why conversations with a factory about pilling so often stall.
Construction is the factory's lever, and it is cheaper than changing the cloth. A larger radius on the arm nose spreads a forearm's load instead of concentrating it. Moving the seam out of the contact zone, or dropping the welt on the nose, removes the cut edge and the ridge from the place the complaint comes from. Cutting the arm panel from the roll with a different grain or a different curl changes how the cloth lies over the radius. A backing cloth under the panel limits how far the surface fibres can migrate. None of these change the fabric's test result, and all of them change the outcome.
The material is the mill's lever. Hairiness is the primary variable in pilling, and it is a property of the yarn rather than of the fibre. Comparative trials on cotton ring-spun cloth have graded 4.5 at a hairiness figure of 2.5 and 3.5 at 5.8 for the same construction; those are trial figures rather than standard limits, but the direction is the point. Yarn twist, fibre length and weave density all move it. A 50/50 polyester-cotton blend is the classic worst case, because the cotton supplies the loose fibre ends and the polyester anchors the balls so that they stay attached to the surface instead of shedding. Filament yarn, with no short ends to migrate, does not pill in the first place.
Finishing is the cheapest part of it and it is routinely missed. Singeing burns off the protruding surface fibres before they can tangle, and biopolishing, an enzyme treatment that breaks down the surface fibre ends on cotton-rich cloth, has been shown to lift pilling by one to one and a half grades, and unlike a softener it does not wash out. The catch is that a chemical finish wears off with cleaning, so if the buyer is relying on one, the requirement has to be a grade after a stated number of cleaning cycles, not before.
What it cost
The order was 320 pieces of upholstery on a serviced-apartment fit-out. The complaint was cosmetic, and every option was expensive in a different way, which is the usual shape of a complaint that arrives four months after delivery.
| Option | What it came to |
|---|---|
| Re-cover the arms on site, 320 units, material and labour | About USD 52,000, plus moving furniture out of 140 occupied units |
| Re-upholster in the factory before delivery | Not available at that point, the goods were already on site |
| One arm mock-up built from production fabric, plus one laboratory pilling run | About USD 380, and the defect would have been found before the container was sealed |
| Re-run the whole order on a fabric with a pilling grade | About USD 96,000 of fabric and a six-week delay, with no guarantee the new article performs better in service |
The third line is the one worth remembering. A single mock-up of the arm, built from the fabric that was going into production and rubbed the way an arm gets rubbed, is a fraction of one per cent of the order value. It is also the only item on that list that could have been done before anybody paid, and the only one that would have produced an answer.
Where to go next
→ how we work and what each check covers
→ the three levels of check we sell, and what each one covers
The short version
A fabric can pass the abrasion rating and still pill, because abrasion and pilling are different properties measured by different standards, and a rating only means something next to the method that produced it. A flat coupon cannot see an arm, because curvature, tension and a cut edge are the conditions that produce the pills, and a standard specimen removes all three. The position of the specimen is therefore not a detail of the test, it is the test.
The specification that follows from that is short: name the pilling standard, the cycle count and the grade, put the fabric through the test after finishing rather than before, name the acceptance zones on the drawing, and build one arm mock-up and rub it. A buyer who wants the arm of the sofa to still look right after a year has to say so in the order, because the mill tested what it was asked to test, and what it was asked to test was never the arm.
Where to go next
→ sofa inspection and upholstery checks in Foshan
→ the three levels of check we sell, and what each one covers
Have an order you need verified? Send us your PI and factory address. Three levels, published prices, $149 per man-day from.