Two hundred and forty hotel mattresses were invoiced as eighty percent natural latex. The band carrying the eighty percent label turned out to be polyurethane foam, and the paperwork was right about the ratio.

Two hundred and forty hotel mattresses, invoiced as eighty percent natural latex

A hotel group refurbishing three properties ordered two hundred and forty king-size mattresses for a single shipment. The purchase order specified the core by volume: a two hundred millimetre natural latex comfort core over a fifty millimetre high-density polyurethane support base. That is eighty percent latex, twenty percent polyurethane, and it is the layout most hotel mattress specifications at this level describe.

The supplier's quotation and the packing list on every carton repeated the number in the same form. 80% natural latex core. The invoice carried it. The sample had been approved at the factory and, as far as anyone had recorded, it was a genuine latex mattress.

The only part of the order that made the material claim measurable was a minimum density of 60 kg/m³ for the comfort core, together with a prohibition on polyurethane and memory foam inside it. Nothing else in the specification said how to find out what the core was made of.

That gap is the whole case. Eighty percent in a purchase order can describe a layer schedule without describing a material. When a factory builds to the schedule and fills it with something cheaper, the paperwork stays correct in every place the buyer wrote a number and wrong in the only place the buyer asked a question.

What a corner cut actually shows

Cutting a corner off a mattress is the cheapest material check in furniture inspection, and it happens during almost every mattress inspection because the corner has to come off anyway before the unit is compressed and bagged for shipment. What the section gives you is the layer build: how many bands there are, how thick each one is, and whether each band runs the full width of the panel or stops short.

It also gives you specimens. A block cut square from the middle of each band, of known dimensions, carries the mass of that material in that volume. That is the raw material for the one measurement that decides this case.

What the section does not give you is a material identification. Colour, cell size and the feel of the cut face are consistent with several foams at once. Natural latex and polyurethane at comparable densities look more alike than anyone selling either will admit, and an inspector who has decided what the answer is before the saw is out will find it every time.

So the section is the step that makes measurement possible, not the step that settles the question. Two independent measurements settle it.

Density is the measurement that decides it

Apparent density is the cheapest decisive test in foam, and it is what the purchase order had already specified. ISO 845 and ASTM D3574 Test A both determine it from the mass and volume of a prepared specimen, and a squared corner block gives you both without a laboratory. The numbers on the cut sample came out like this.

One correct band next to one wrong band

The support base is what makes the table readable. A factory that had misunderstood the order would not have got the base density right and the comfort density wrong. One correct band sitting next to one wrong band is a substitution in a specific place, not a different factory.

This is also where the weight argument comes from, and it is the version that needs no saw at all. A cubic metre of the specified material weighs at least sixty kilograms. The foam in the comfort core weighs 28.4. Same volume, less than half the mass.

Commercial mattress inspection rarely starts with a saw. It often starts with a pallet on a platform scale and the purchase order next to it, because a substitution of this size changes the weight of the shipment by tonnes and nobody weighs a shipment unless something has already looked wrong.

Why polyurethane can be sold as latex

Latex is not one material, and in mattress language it usually stands for one of three things. Natural rubber latex is made from the sap of the rubber tree. Synthetic latex is made from styrene butadiene rubber and is petroleum-derived. A blend mixes the two. Natural rubber latex leaves the tree at around thirty percent rubber content and is concentrated above sixty-one and a half percent solids, of which sixty percent is rubber and the balance are compounds found nowhere else, including the proteins that cause latex allergy in a small number of people. Synthetic latex contains none of them.

Polyurethane belongs to a different family altogether, built from polyols and isocyanates with a gas that expands as the reaction runs. It is the material in almost every mainstream foam mattress sold in the world. It is also sold, widely and with little resistance, as latex foam, latex-like foam and latex feel, because in trade the word describes a feel as much as a chemistry and nobody has stopped it.

The certification route is where the absence becomes obvious. GOLS, the Global Organic Latex Standard, requires at least ninety-five percent certified organic raw material, which means a product described as eighty percent natural latex cannot carry it at all. A blend cannot carry it either. CertiPUR-US certifies polyurethane foam and cannot cover latex. Eco-Institut measures emissions from the finished product and says nothing about what the core is made of. OEKO-TEX and GOTS cover textiles.

So on an eighty percent natural claim the honest finding is that no certificate in the ecosystem could have verified it, because the number sits below the threshold of the only scheme that measures the thing. The buyer's own specification set a target that no recognised certificate reaches.

The second measurement, and the limit of the third

Density alone is strong but it is one number, and one number can be argued. The second measurement is resilience: ball rebound, ISO 8307 and ASTM D3574 Test H, in which a steel ball is dropped from five hundred millimetres onto the sample and the rebound height is read as a percentage of the drop.

The ranges are well separated. Natural latex at bedding density sits around sixty to seventy-five percent. Conventional polyurethane runs thirty to fifty. High resilience polyurethane runs fifty to sixty. Memory foam is often below ten. The sample came back at thirty-four, consistent with polyurethane and well outside the latex range, landing on the same conclusion the density reached by a completely different route.

What infrared spectroscopy cannot do

There is a third measurement that closes the question and one that does not. Infrared spectroscopy identifies the polymer, and polyurethane is unmistakable in a spectrum through the carbamate carbonyl at roughly seventeen hundred to seventeen hundred and thirty wave numbers and the N-H stretch near thirty-three hundred. The test runs on a small piece of the cut sample and reports as a laboratory result.

What infrared spectroscopy cannot do is separate natural rubber latex from synthetic latex, because both are polyisoprene and their backbones look the same in the spectrum. If the dispute is latex against polyurethane, the spectrum settles it. If the dispute is natural against synthetic, only a certificate with a defined scope, or a chemical analysis aimed at rubber content, will do it. A report that says infrared confirms natural latex is simply wrong.

The part that turns into a regulatory problem

For a hotel order sold in the United States, this stops being a material dispute. Mattress sets are subject to two mandatory federal flammability standards: 16 CFR Part 1632 for cigarette ignition and 16 CFR Part 1633 for open flame. Part 1633 measures heat release under a propane burner and requires that the peak rate of heat release not exceed two hundred kilowatts at any point in the thirty-minute test, and that total heat release not exceed fifteen megajoules in the first ten minutes.

Both standards are written around the prototype. A prototype is a specific design of mattress set serving as the model for production units, and it is the same with respect to materials, components, design and methods of assembly. A mattress sold with a foundation and the same mattress sold alone are two separate prototypes. The permanent label must carry the prototype identification number, and the recordkeeping rules require records establishing that any substituted core or interior materials do not influence the ignition resistance of the prototype.

So a change of core material is not a detail inside an approved product. It is a different prototype, and the test report sitting in the file describes something that is no longer being built.

That is the part of this case that would have cost the buyer the most, and it is invisible to anyone checking only the mattress.

How the report was written

The report was written in four separations, because the buyer had three different questions and one decision to make.

First, what was measured, with the method and the specimen identified: density for each band, ball rebound on the comfort core, the section described band by band, and the pallet weights.

Second, what was specified, quoted from the purchase order rather than paraphrased, because a paraphrase is where a fifty millimetre band quietly becomes a sixty millimetre band.

Third, what could not be verified and why. The eighty percent composition claim on the invoice was not tested, because no published method defines a composition percentage in a finished mattress, and no certificate capable of confirming natural rubber content could exist for a product at that percentage. That sentence matters more than the failure it sits next to, because it tells the reader which part of the dispute is evidence and which part is interpretation.

Fourth, the classification. The delivered core does not meet the specified material and density in the comfort band, and this is a major non-conformance against the purchase order. It was not written as a safety finding, because both the specified build and the delivered build are complete products and nothing observed made either one unsafe. It was not written as fraud either, because a section and two measurements establish what the core is, not what anyone intended.

The report also stated its inference plainly. The conclusion applies to the whole lot on the basis that the run used one foam block, which is normal production practice but is an assumption and not a measurement. Every mattress in the shipment was weighed, and every carton was checked against the sectioned sample.

The specification, rewritten

The same order, with the same mattress, written so that it can be verified on a factory floor by someone who was not in the negotiation. Five lines, each one replacing a sentence that could not be checked.

None of these lines make the factory work harder

None of these lines make the buyer more demanding of the factory. They make the buyer specific, and specificity is what turns a dispute into a measurement. The factory that built to the specification as written would have passed every one of these lines without changing a thing about how it works.

The on-site procedure that would have caught it

The substitution was visible on the first day and needed no laboratory. Here is the order of the checks and what each one costs.

• Weigh a pallet against the order. The specified core was at least 49.5 kg per mattress and the delivered core was 26.7 kg, so the shipment should have been about five and a half tonnes lighter than the declared weight across two hundred and forty units. A platform scale and the purchase order are enough.
• Cut the corner, read the band thicknesses and check that each band runs the full width. A band that stops short is a different product.
• Measure apparent density per band to ISO 845 on a squared block. This is the decisive test, and it is the one the purchase order had already specified.
• Run ball rebound to ISO 8307 on the same block, so the material conclusion rests on two measurements that do not depend on each other.
• Send a small piece to a laboratory for infrared spectroscopy when the finding has to survive a dispute, remembering that it settles polyurethane against latex and not natural against synthetic.
• Check the flammability prototype. If the core changes, the prototype changes, and the test report and the label have to change with it.

The commercial outcome

Two hundred and forty mattresses were already made, compressed and loaded before the section was cut. That is the normal sequence and it is why these checks are priced by the day rather than by the mattress. A core substitution discovered after loading is a negotiation; the same substitution discovered during assembly is a line change.

The buyer had three realistic options. Accept the delivered product and take a price credit for the material that was not supplied. Reject the shipment and re-order, which means a new prototype, new flammability testing in the United States and a second production slot that is rarely available on the same delivery date. Or accept the material and re-label the product as a polyurethane core mattress, which changes what the hotel can claim about it and moves the flammability position again.

What the credit was worth came down to arithmetic once density was known. A cubic metre of the specified material at sixty kilograms against the delivered twenty-eight point four, over the volume of the comfort core, on two hundred and forty units. No test was needed for that number, only the corner block and a scale.

The part of the loss that never appears in that calculation is the certification and flammability work that has to be redone, because the product that was tested is not the product that was built. A buyer who had written the prototype requirement into the purchase order would have known that before the container was sealed. A buyer who had not would have found out in a customs query or a hotel opening.