Sixty clinic waiting-room chairs had armrests 40 mm below the order, and the purchase order and the factory drawing both said 220 mm. The difference was what the number was measured from.

The order, and the number both sides wrote down

A clinic operator in Europe ordered sixty waiting-room chairs for a public outpatient waiting area: steel frame, foam and PU upholstery, linked in twos and fours, armrests on both sides. The purchase order gave the seat height as 460 mm, the seat depth as 500 mm, and the armrest height as 220 mm above the seat surface. The value was not unusual. It sat inside the range that accessibility guidance gives for armrests on seating that people have to push themselves out of, and comfortably above the 180 mm that appears on a great many mass-market waiting chairs.

Production ran and the chairs were finished. At the pre-shipment visit the inspector worked through the purchase order dimension by dimension with a steel rule, the way he had on every other line on that order. Seat height was good. Seat depth was good. Armrest height measured 180 mm above the cushion on every chair he sampled, in all three seat widths, across the whole batch.

The factory was not being difficult about it. The factory's own production drawing specified the armrest at 220 mm, and the factory had built the chairs to that drawing. The purchase order specified 220 mm, and the factory had built the chairs to that drawing. Two documents, one number, both true, and a chair that did not have the armrest anyone had asked for.

What the inspector measured, and from what

The first job was not to decide whether the chairs were wrong. It was to decide whether the factory and the buyer were measuring the same thing, because at that point each side had a number and a drawing and no shared reference point. So every dimension was measured twice and the datum was written down next to each reading.

Seat height was measured from the finish floor to the top of the cushion, once with nobody in the chair and once under a 75 kg load plate. The armrest was measured from the seat frame deck, from the uncompressed cushion top, and from the loaded seat surface. The floor-to-armrest-top height was measured directly. Eight chairs were sampled, spread across the two-seat and four-seat units and across the batch.

The cushion was 40 mm of moulded foam over the frame deck. That single number turned out to explain most of what followed.

Why 220 mm on both documents was still a 40 mm miss

The factory's drawing and the buyer's purchase order were describing the same dimension from the same point in space: the horizontal plane of the seat frame, the deck the cushion is stapled to. From there to the top of the armrest was 220 mm on every chair. That is why the factory could show the inspector a drawing with 220 mm on it, and why the inspector could measure 220 mm with the same rule in his hand. Both readings were correct. Neither of them was the height of the armrest above the surface a patient actually sits on.

Above the uncompressed cushion it was 180 mm. Above the surface the seat presents under a 75 kg adult it was 168 mm, because 40 mm of moulded foam does not stay 40 mm thick under someone sitting on it. The report recorded all three readings and noted which one the purchase order had intended.

This is worth being precise about, because it decides who owes what. The factory built to the drawing it was given, and built it consistently: the armrest height was 220 mm on every chair in the batch, with no variation to speak of. This was not a workmanship defect, not a tolerance problem, and not something a tighter incoming inspection would have caught. A batch of chairs carrying a 40 mm systematic offset is a design and specification outcome. Writing it up as a factory quality failure would have been wrong, and it would have wasted a great deal of money arguing about it.

168 mm also sits below the 178 mm lower bound of the general elderly-seating band, though it is inside the band that waiting-area and street-furniture guidance works to, where armrests run from about 220 mm to 300 mm above the seat. The number the order named, 220 mm, was in that band. The order was not asking for something unusual. It was asking for something the drawing failed to describe.

The number that decides whether a wheelchair can use the chair

A 40 mm difference on an armrest looks like nothing, and to a fit and finish inspector it is close to invisible. It becomes real only when you ask what the armrest is for. For most people in a clinic waiting room, the armrest is a push-off point: plant a hand, lean forward, stand up. For a wheelchair user, the armrest is the landing point for a lateral transfer, and it is judged against a completely different reference.

The relevant reference is the user's own armrest. Anthropometric study of wheeled mobility devices puts the mean armrest height at 716 mm above the floor for all device users and 704 mm for manual chair users, with a 5th percentile of 650 mm. The chair being delivered put its armrest top at 620 mm above the floor. For the median manual chair user that is roughly 85 mm below the surface their hand is already resting on. They have to reach down and across to find it, while their trunk is unsupported between two seats, and then bear part of their weight through a hand that is not at elbow height.

Then there is the clearance that a height measurement never touches. A wheelchair user transferring sideways moves their hips across the gap between the chair and their wheelchair. A fixed armrest on the transfer side cuts that gap by the width of the arm and the structure carrying it, and the 2010 ADA Standards already require folding or retractable armrests on at least 5% of aisle seats in assembly areas for precisely this reason. Nothing in the purchase order asked for a removable armrest, so nothing in the purchase order asked for the one thing that would have made the seat usable from the side.

This is where the defect stops being dimensional. A 40 mm armrest error on a chair used only by ambulant people is a specification dispute. The same 40 mm on a chair that has to work for a wheelchair user is a barrier to use, and barriers to use are what accessibility regulation is about.

What the report had to say, and what it had to leave out

The report had to make one distinction sharply, because the buyer's next question was always going to be the same: is this a safety problem or an annoyance? It is not a safety problem. The chairs are stable, the frames are sound, the armrests carry the loads that armrests carry. What these chairs do not do is support a lateral transfer for a wheelchair user. That is a usability failure with an accessibility consequence, and it is a failure of the specification rather than of the factory's workmanship.

The report recorded the defect as a non-conformance against the purchase specification: armrest height 168 mm measured above the loaded seat surface, against 220 mm required above the seat surface, with the datum stated in full, the loading condition stated, the instrument stated and the sample count stated. It recorded all three armrest readings, not only the one that failed, because a report showing just the failing number hides the mechanism and invites an argument about which measurement was the fair one.

It also recorded what had not been done. No type approval testing under EN 16139 had been carried out on these chairs. The steel frame, the arm brackets, the seat pan and the finish had been checked on site against the drawings and the purchase order. Nothing in the report meant that the chairs met any product standard, and the report said so in those words, because a document that prints 168 mm in a table of dimensions will otherwise be read by a buyer as a compliance document.

• Armrest height at three datums: seat frame deck, uncompressed cushion top, and loaded seat surface, with the loading condition and the instrument named
• Seat height and seat depth at both loading conditions, so the relationship between the three could be read from one table
• Sample count and spread: eight chairs, three seat widths, one continuous batch, and no variation in armrest height across the batch
• The root cause, stated as a datum difference between the factory drawing and the purchase order, rather than as a production fault
• That no EN 16139 or ANSI/BIFMA X5.4 type testing had been performed, and that the report was a specification check and not a compliance certificate
• That the repair route had not been costed, because it depends on whether the arm brackets are welded or bolted, and that was a question for the factory to answer in writing

How the specification should be rewritten

The last section of the report was a rewritten clause, because a buyer who only learns what went wrong will buy the same thing again. The clauses were written so that the next order cannot be built to the wrong datum by accident, and so that an inspector can turn each one into a pass or a fail without interpretation.

The first-article clause, which is the one that would have caught it

Every clause above is worth having, and only one of them would have caught this before the batch was finished. That one is the first-article check, and it fails on bare frames: a first article that is a welded frame with no upholstery measures 220 mm from the deck and passes. The datum only separates when there is a cushion on the frame, a cover over the cushion, and a person or a load on the seat. That is also why a waiting chair inspection that measures a bare frame proves nothing about armrest height, however carefully the frame was checked.

The clause itself is short. One fully upholstered chair, in the production cover and the production foam, delivered to the buyer's site before the batch is cut, measured under load by both parties, with the signed measurement sheet becoming part of the inspection instructions so that the same readings are taken again at the pre-shipment visit.

• One fully upholstered chair in the production cover and production foam, not a sample frame and not a showroom unit
• Load the seat to 75 kg and measure armrest height above the loaded surface, not above the deck
• Measure seat height unloaded and loaded, and check that the two readings together sit inside the specified band
• Bring the chair into the actual waiting area and attempt a lateral transfer from a wheelchair, measuring the gap the user has to move across and the armrest height against the user's own armrest
• Sign the first article off before production, and write the signed sheet into the inspection instructions so the same readings are taken at the pre-shipment visit

What to carry to the next order

Nothing on this order was mislabelled, nothing was substituted, and the factory did not cut a corner. It built a good chair to a drawing that was internally consistent and that did not say what the buyer meant. That is a different failure from a substituted drawer slide or a lighter foam, and it is worth naming separately because the remedy is different too.

Two habits would have caught it, and both are cheap. First, when a purchase order gives a height above a surface, write the load condition into the same sentence: the armrest height is 220 mm above the seat surface under a 75 kg load, not above the seat surface, which is a different measurement on a different day in a different factory. Second, order the first article upholstered, and put it in the room before the batch is cut, not in the container after.

There is a general lesson about specifications for public seating that goes well past armrests. The dimensions that decide whether a chair can be used by the person who needs it most are the ones nobody measures, because they are invisible and they do not fail visibly. A chair can pass every other line on the specification sheet and still be a chair that some people cannot get into.