Every cushion in a 480-piece patio order passed the check. Underneath the cover, the foam took on nearly 22 times its own weight in water and did not give it back.

The order, and the two words that decided it

A European outdoor furniture importer ordered 480 patio dining chairs with upholstered seat cushions, for a hotel terrace and a restaurant courtyard. The chairs were aluminium frame with a powder coat, the cushions were 500 by 500 by 80 mm with a colourfast outdoor cover. The order was split into two shipments of 240, made eleven weeks apart, and the first 240 went out first.

The cushion line in the purchase order ran to two sentences. Cushion foam: 38 kg/m3, outdoor grade, water-resistant, quick-dry, must not absorb water. Every one of those words describes a performance that nobody had agreed how to measure. There was no class of foam named, no number, no test method and no limit. What the line did say was density, and the goods that arrived had the density exactly right.

The consequences of that gap were not visible in the factory, on the day of loading, or on the delivery documents. They became visible on a terrace facing the sea, six weeks after installation, in the first real storm of the season.

What passed, and what nobody was looking for

The pre-shipment check on the second shipment of 240 cushions followed the standard routine. Thirty-two cushions were drawn from the lot and examined for appearance against the approved sample, seam quality, zip function, cover fabric, colour, and foam density at three points on the core. Density came back at 38.1, 38.0 and 38.2 kg/m3 against a specified 38 plus or minus 3. Nothing was wrong with any of it. The lot passed at AQL 2.5, where two defects in a sample of 32 are acceptable and three would have rejected the lot.

There was nothing wrong with any of it, because none of the checks were pointed at the cell structure of the foam. A cushion can have perfect density, perfect stitching and a spotless cover and still be built from a foam that drinks. Cell structure has no colour, no texture and no weight difference you can feel in the hand. The only way to see it is to put a piece of the foam in water and watch what happens.

It is worth being precise about the sampling here, because this is the part people get wrong. The defect was in every cushion cut from that foam batch, one hundred per cent, not in a scatter. If the wrong foam had been in a tenth of the lot, a sample of 32 would still have found it ninety-seven per cent of the time. If it had been in one in twenty, the sample would have found it eighty-three per cent of the time. Sampling was never the problem. The sample was being asked the wrong question, and it answered that question correctly.

How a terrace in the rain told the buyer what the factory could not

The first 240 cushions had been installed six weeks earlier at the buyer's pilot site, on a terrace facing the sea. Two weeks after installation, in the first real storm of the season, the site manager sent one photograph and one sentence: the cushions were wet through the bottom and heavy enough that the chairs were hard to move.

Three cushions were taken off the chairs and weighed dry, then weighed again after the storm, then again as they dried. The same cushion was used throughout, so the figures below are one cushion's history rather than an average of three.

The test that finally separated the two foams

With the balance payment on the second shipment held, the buyer asked for one thing: a measurement of the foam itself, on site, in the factory, before anything else was loaded. The method was written down first, which is the point of doing it that way. A 50 by 50 by 25 mm specimen was cut from the core of each cushion sampled, away from the edges and away from any visible seam or tape line. Each specimen was weighed dry to 0.01 g, immersed in distilled water at 23 degrees C for 24 hours, lifted out, blotted on both faces with a dry cloth, and weighed again.

The specimen from the delivered cushions weighed 2.38 g dry and 54.6 g wet. It had taken up 2,194 per cent of its dry weight in water, nearly twenty-two times its own mass. A specimen cut from the closed-cell block the factory had on its own shelf, same density, same dimensions, weighed 2.38 g dry and 2.44 g wet, an increase of 2.5 per cent.

Why the number on the purchase order was never going to be measured the same way twice

Which is why a purchase order that says water absorption no more than five per cent, with no method attached to it, is not a specification. It is a wish. The factory is not obliged to read it the way the buyer intended, and the buyer's own inspector is not obliged to test it the way the factory did. Both readings are defensible, and they will not agree.

The way out is not to remove the number. It is to write the number and the method together, so that any factory and any inspector working from the same document produce the same figure, and so that the figure can be compared with the same figure next season, and with the same figure from another supplier.

Closed cell, reticulated, and the foam that failed

The number that decides which one you have is the closed-cell content, the proportion of the foam's volume that is sealed rather than connected. A closed-cell foam sits above 90 per cent by volume, with the remainder as solid cell walls. An open-cell foam sits at essentially zero, because all of its cells are connected, and that is exactly why it drinks.

The physics behind the immersion result is worth spelling out once, because it is simple and it explains the number on the report. Foam at this density is about 97 per cent void. The 62.5 cubic centimetre specimen held roughly 60 cubic centimetres of connected pore space, which is 60 grams of water if the pores filled completely. The specimen weighed 2.38 grams dry and 54.6 grams wet, so it had taken up 52 grams, about 86 per cent of what the structure could hold. The same foam reported by volume instead of by mass gives 84 per cent. Read as a volume figure it looks unremarkable. Read as a mass figure it is one of the most absorbent materials in the building trade. Both numbers were measured on the same piece of foam on the same day.

Reticulated foam is the one to understand properly before writing a specification. It is still an open-cell foam, and on this test it would score badly, and yet it is the standard choice for exposed outdoor seating because the water moves through it and drains instead of sitting in it. A closed-cell foam would score almost perfectly on absorption, and it is stiffer and less comfortable, and it also will not drain: water that gets past the cover sits against it rather than running away. Which one a buyer wants depends on whether the cushion lives in the rain or under a roof, and on how it is covered. That is a decision, and it belongs in the specification rather than in a factory's interpretation of the phrase outdoor grade.

What the report said, and what it could not say

One practical instruction went in with it: the factory was to cut and seal a 300 by 300 millimetre piece of the actual production block from each batch, label it with the batch number, and hold it for twelve months. A retained sample is the only thing that makes a laboratory test possible three months later, and in this case it was the difference between a claim the buyer could prove and a claim the buyer could only assert.

• Closed-cell content was not determined on site. The immersion test establishes beyond argument that the material is not closed-cell, but the actual percentage of sealed volume needs a gas pycnometer, which means an accredited laboratory. A three-specimen test on the retained block sample was recommended at the buyer's cost.
• The mould on the first 240 cushions was not attributed to this material. Mould in a foam cushion has several possible sources, including the fibre wrap and the storage conditions before installation. The observation was recorded as an observation, not as a conclusion.
• The drying performance of a compliant cushion could not be demonstrated on site. The reference block was weighed and immersed, but a two-hour drying curve needs a controlled environment and a day of measurements, so it was listed as something the buyer could require from the factory's own test record rather than something we had verified.

The purchase specification, put back in a form that can be tested

The cushion line went back to the buyer as a replacement, written to be used rather than admired. The first version was six lines long. The version below is longer, and that is the point: every clause is either a number somebody can measure or an instruction somebody can be held to.

• Foam class, named explicitly. Closed-cell content at or above 90 per cent by volume, determined by gas pycnometer to ISO 845, stated on the batch datasheet. If reticulated open-cell foam is acceptable instead, say so in the same sentence, and require the cell membranes to be removed, evidenced by a drainage test, not by the word reticulated on a datasheet.
• Water absorption, with the method attached. Specimen 50 by 50 by 25 mm cut from the cushion core away from seams and edge tape, weighed dry to 0.01 g, immersed in distilled water at 23 plus or minus 2 degrees C for 24 hours, blotted on both faces, re-weighed. Limit 5 per cent by mass, reported to the nearest 1 per cent, and a statement that the figure is valid only by this method.
• Drying time, as a functional requirement. One litre of water poured over the seat face of the finished cushion, blotted after 5 minutes, then the cushion on edge in moving air at 25 degrees C and 60 per cent relative humidity. The cushion shall be back within 5 per cent of its dry weight in 2 hours, and shall not exceed the dry weight plus 1 per cent at 24 hours.
• Test the finished cushion, not only the core. The cover fabric, the fibre wrap and the seam tape are all part of the water path. A core specimen that passes does not clear the cushion.
• Retention of the production material. A 300 by 300 millimetre piece of the actual block, sealed, labelled with the batch number, held by the factory for 12 months and released to the buyer's laboratory on request.
• Outdoor cover and drainage. Cover fabric declared as solution-dyed acrylic or PVC-backed, with the colour fastness rating and the backing stated. Bottom seam and any vent specified, so that the bottom of the cushion is not a sealed pocket.

What it cost, and what it would have cost

Set against that, the cost of getting it right first time is easy to read. The closed-cell upgrade across the whole order was 2,016 dollars, twenty per cent of the cushion value, and it would have made the goods compliant with a specification line that takes two sentences to write. The targeted investigation that found the problem in two days cost 398 dollars. Replacing the 240 cushions already sitting on a terrace and starting to grow grey cost 5,040 dollars, and that was before anyone had to talk to the hotel about the chairs.

There is also a version of this order where the second shipment was loaded and shipped, and the cushions were only found to be wrong when the hotel's guests sat on them. Nobody would have been able to prove what the foam was, because the block would have been gone and the specification would not have had a method attached to it. A patio furniture inspection china that stops at the carton count, the model number and the cover colour passes that order every time. The gap between the two orders is one immersed piece of foam and 398 dollars.

None of this is a claim that every order needs an immersion test. It is a claim about which order needs one: the one where the cushion is going to be rained on.

The two numbers to put in the next purchase order

Most outdoor cushion problems are not mould problems. They are specification problems that arrive as mould problems months later, in a different country, with the foam block already gone. The cushion in this order failed on the terrace, and the number that decided it was never written anywhere except in a report, on a piece of foam, on one day.

If you buy outdoor cushions, write two things down. Write the foam class with a number attached: closed-cell content at or above 90 per cent by volume, or a named reticulated grade, measured by a method you also write down. And write the drying clause: one litre of water on the finished cushion, back within five per cent of dry weight in two hours, and never more than one per cent above dry weight at 24 hours. Everything else in the specification can be negotiated later. Those two lines decide whether the cushion is still dry in October.