The fabric was rated for 15,000 rubs

The situation

A European distributor ordered 320 upholstered sofas from a factory in Guangdong — two models, 190 three-seat and 130 two-seat. The purchase order named the upholstery in performance terms rather than by feel: 100% polyester, abrasion not less than 60,000 Martindale rubs to ISO 12947-3, the same fabric on every unit in the order. Unit price was $186, an order value of $59,520, and 70% of it was still unpaid when the factory asked for the balance.

Production ran over six weeks. The sofas came out of the sewing line looking right: tight seams, matched panels, an even hand across the batch. The factory submitted a third-party test report that had been in the file since the sampling stage, together with a cutting of the approved fabric, and the buyer's quality manager signed off the production photographs.

The buyer booked an inspection before the balance was released.

Why the composition told them nothing

The first check passed cleanly, and it passed in a way that should worry anyone who relies on it. The specification said 100% polyester. The delivered fabric was 100% polyester. A fibre composition check on site — microscopy on the panel and a burn test on the cuff — confirmed it.

**Abrasion rating is not a property of the fibre.** It is a property of the construction: yarn denier, twist, weave density, and the finishing applied after weaving. Two fabrics of identical 100% polyester construction can sit 40,000 rubs apart, and the fibre label will never show it. In this order the composition check confirmed the goods were made of the right material and said nothing whatsoever about whether that material was the right material.

Martindale is not a single method either. The same fabric tested to ISO 12947-3 and to ASTM D4966 does not return the same number, and the older Wyzenbeek method to ISO 12947-1 returns a lower figure again. **A rating without a method attached is not a specification.**

And the rating is a contract term, not an appearance item. A buyer who has not written the number into the purchase order has no claim at all, however obviously the fabric differs from the sample in the hand.

How it was found

Sampling was ISO 2859-1, general level II. At a lot size of 320 the sample is 32 units, and all 32 carried a sewn-in specification label naming 4471. One point of method is worth stating, because it is the one most reports get wrong: **article conformity is not a sampled characteristic.** It belongs to the whole roll, and one header card settles it for every unit cut from that roll. This was a lot-level fact checked at 100%, not a 32-unit count that happened to fail.

Where the honest limit sits

**What the inspection established:** the fabric in the sofas was Article 4471; that article's published rating is 15,000 Martindale to ISO 12947-3; the delivered fabric weighs 331 g/m² against a specified 380; the certified report held in the file does not cover the delivered article or the delivered lot; and the production-lot evidence required by the order does not exist in the factory's quality file.

**What the inspection did not establish:** the Martindale value of the delivered lot itself. A mill datasheet is a published figure, not a test on the lot standing in front of you. The inspector did not run an abrasion test and could not — the Martindale test is destructive, needs an accredited laboratory and about a week, and a number produced in a factory in Guangdong on a hand-cranked machine would be worth nothing.

**The only conclusive route** is a 300 mm × 300 mm cutting taken from the production lot, sealed, and tested at an ISO/IEC 17025 laboratory, combined with a mill report for the same lot. The report said exactly that, in those words, and gave the cost and lead time: one fabric, a few hundred dollars, five to seven working days. A finding that comes with a documented way to close it is worth more than one that comes with a number.

**Why that limit did not weaken the case.** The buyer's claim did not rest on a measurement the inspector had taken. It rested on the goods being a different article from the one ordered, and on the ordered article's requirement not being evidenced for the delivered one. A buyer does not need a laboratory result to know the requirement was not met, and the factory is free to produce one and close the question if it wants to.

How a 15,000-rub fabric gets under a 60,000-rub clause

Nobody buys a worse fabric on purpose. Three ordinary things happened here, and none of them required anyone to be dishonest.

**The quote was priced on the sample.** The buyer's cutting came from Article 4412, a 60,000-rub fabric, which the mill priced accordingly. The order was placed against that cutting and that price, and the approved sample in the buyer's office was the reference everyone continued to work from.

**The mill requoted.** Same mill, same construction family, same colour, four times cheaper per metre, 15,000 rubs. Across the 2,720 metres of fabric this order consumed, the difference was $1,900. Had that difference been passed on as a price concession, it would have been a perfectly ordinary commercial outcome rather than a problem.

**The paperwork did not reconcile.** The test report in the file covered the sample article, the datasheet for the delivered article sat in a sales brochure rather than in the quality file, and the spec label sewn into every cover named whatever article the sewing line was given. Nobody had been asked to make the three agree.

One related point the case makes visible: 60,000 rubs is a commercial-grade figure and 15,000 is light domestic. **This is not a tolerance failure at the edge of a band.** It is the difference between a sofa sold for an office reception and one sold for a hotel guest room, and it changes what the goods are worth.

What the specification should have said

The purpose of the specification is not to make an inspector run a test on site, because nobody can. It is to make the goods carry their own evidence, so that the only thing left to check on site is whether the evidence matches.

• **Name the method, not the trade word.** *Abrasion shall be not less than 60,000 Martindale rubs determined to ISO 12947-3. A rating determined to ASTM D4966 or to ISO 12947-1 (Wyzenbeek) shall not be offered as evidence, and no rating shall be accepted without the method stated.* The words *Martindale 60,000* on their own are satisfied by numbers measured three different ways, and the cheapest of the three is the one a mill will quote.
• **Name the article, and require the article on the goods.** *Upholstery shall be Article 4412 as supplied by the named mill through the named supplier. The article number shall appear on the header card of every roll delivered and on the sewn-in specification label of every cover, and goods cut from any other article shall not be stuffed.* This is the clause that would have stopped this order, and it costs nothing to write.
• **Require the report to be on the production lot.** *An independent test report from a laboratory accredited to ISO/IEC 17025 shall be supplied for the article and lot actually stuffed, stating article number, lot number, test method and test date, and dated no more than three months before stuffing.* The report in this file was 26 months old, was for the sample colour, and carried no lot reference — a valid document about a different piece of cloth.
• **Keep the sample, and give one to the buyer.** *A 300 mm × 300 mm cutting shall be taken from the first production lot, sealed, labelled with article and lot number, and retained for 24 months. One of each pair shall be released to the buyer at the pre-shipment inspection.* Without a cut taken from the production lot, a laboratory test on any other piece of fabric proves nothing about what is in the container, and this is the only route to a conclusive number.
• **Separate the properties so one certificate cannot cover four.** *Abrasion, pilling to ISO 12945-2, colour fastness to ISO 105-B02 and ISO 105-C06, and flame retardancy to EN 1021-1 shall each be evidenced by a separate report carrying its own report number, at the stated thresholds.* A Martindale figure says nothing about pilling, and pilling is what a buyer sees in the first year.
• **If a lower rating is accepted, accept it in writing against a price.** *Any upholstery rated below 60,000 rubs may be used only with the buyer's prior written approval, which shall state the agreed rating, the price adjustment, and which of the tests in the clause above are replaced.* A written 15,000 with a price adjustment is a decision. An undeclared 15,000 is a dispute.