A rental operator ordered 360 bed frames whose slat base was described in a single line: thirteen solid slats. Every frame in the order had exactly thirteen slats, exactly as specified. The gaps between them were 85 mm, the mattresses sitting on those gaps were dipping 12 mm at every space, and the number that decided all of it appeared on nobody's document.

The order that said thirteen and nothing else

A rental-housing operator in Europe was fitting out 360 apartments and ordered bed frames from a factory in Guangdong. One model, a king-size timber bed frame with a solid slat base, at $186 a frame, which puts the order at $66,960. Thirty per cent, $20,088, was paid on order. The remaining $46,872 was due against a pre-shipment inspection report, which is the arrangement most buyers use and the one that puts the buyer in the strongest possible position when something turns out to be wrong.

The frame had been reordered from the same factory two years earlier for the operator's previous block of apartments. That first order had gone through without a single finding, and the operator's fit-out guide still carried the old frame's slat line, copied forward: 13 solid slats. That was the whole of the slat specification. No width, no pitch, no gap, and no mention of what the mattress maker's warranty required of the base the mattress would sit on.

The factory had redesigned the frame for this model and moved from 95 mm beech slats to 60 mm, which was a decision about the factory's own material and cost and had nothing to do with the buyer's mattress. The operator had also moved from its own in-house foam to a bought-in hybrid, which brought a warranty condition with it that the frame factory had never seen and that the frame order never mentioned.

The frame order was not wrong. Thirteen slats is what the operator's guide said, and the factory built thirteen slats on every frame in the order. The order was silent on the number that decided whether those frames could carry the mattresses that were going on them.

Where the eighty-five millimetres came from

The clear gap between two adjacent slats is not a free parameter and it is not a number anybody chose. It falls out of three others: how many slats there are, how wide each one is, and how much clear length there is to lay them across. Write S for the span available between the end rails, n for the count, w for the width of one slat and p for the pitch from centre to centre. The span the slats occupy is (n - 1) × p + w, so p = (S - w) / (n - 1), and the clear gap is p - w. Put those together and the gap is (S - n × w) / (n - 1).

For this frame the clear length inside the end rails is 1,980 mm and the margin left at each end is 90 mm, which leaves 1,800 mm of span to fill. Thirteen slats of 60 mm give a pitch of (1,800 - 60) / 12 = 145 mm and a clear gap of 145 - 60 = 85 mm. The order never said 60. The factory's new design used 60 mm slats and its own drawing said so, while the buyer's purchase order said thirteen and nothing else.

The number that made the old frame work was also never written down. Thirteen slats of 95 mm in the same 1,800 mm span gave a pitch of 142.1 mm and a clear gap of 47.1 mm. Same count. Same frame length. A gap inside the limit by accident, because of a dimension nobody had specified and the factory happened to choose well.

Work the equation the other way and the order's own count becomes the problem. For thirteen slats to hold a 50 mm gap in that span, each slat has to be at least (1,800 - 50 × 12) / 13 = 92.3 mm wide. The factory's new design was using 60. Read the table below as an argument about the specification rather than about the factory: the count never moved, what moved was width, and width was the one of the three inputs the order left open.

Recovering a number that nobody wrote down

Two of the three inputs to that gap were on paper somewhere, and the third was not on paper at all. The factory's drawing carried the pitch and the slat width, because a factory cannot build a deck without them. The buyer's purchase order carried the count, because a count is what a buyer thinks of as a slat specification. The clear gap, the one number the mattress maker's warranty is written against, existed only as a calculation that nobody had made.

So the gap has to be measured, not read. It is measured on the frame itself, unloaded, standing on a flat floor with the slats carrying nothing but their own weight, with a steel rule across the opening between two slat faces. Two rules apply and both are routinely skipped. First, measure at three positions, against the left end rail, at the centre and against the right end rail, and record the maximum, because a rail that is not straight opens the gap at one end while the drawing still reads the same everywhere. Second, measure the clear opening and not the pitch from centre to centre. A frame can meet its pitch and fail its gap, and on this order it did both at once.

Then measure it again with the mattress on. A 1,200 mm straightedge laid across the deck, with feeler gauges under it, gives the deviation over each gap, and that is the number a resident feels rather than the number a drawing lists. On the frames as built the surface between two slats dropped 12 mm under the mattress's own weight. The operator's fit-out guide allowed 6 mm of deviation over that straightedge, which is a tighter standard than most mattress makers apply to a claim and much tighter than the 20 mm at which a typical body-impression claim is even accepted.

One more measurement belongs on the same list: the margin at each end. The first slat sat 90 mm from the end rail, and the space beyond it is a gap too. A frame can have every slat correctly spaced and still leave a 90 mm opening at the head of the bed where the mattress is unsupported.

The dip that was found at handover

The first shipment was 60 frames. They went out, were installed, and the operator's handover check put a 1,200 mm straightedge on every mattress before the keys were handed over, because a rental operator that checks once at handover does not check again for two years. Six of the sixty beds failed, each with the same 12 mm hollow over the slat gaps. The other fifty-four passed, and nobody was satisfied with that, because fifty-four out of sixty is not a rate, it is a description of an installation.

The operator's own maintenance lead then did what the fit-out guide should have done months earlier and read the warranty on the mattress they had bought. It was the mattress maker's standard condition, and it sits at the strict end of what mattress brands publish: the slats must be no more than 50 mm apart. Published figures for this condition run from about 50 mm at the strictest to 120 mm at the most permissive, which is another way of saying that the number the buyer needed was sitting in a document belonging to a different supplier and had never reached the frame factory.

Here is what made this one hard to catch. The dip of 12 mm was below the 20 mm at which a typical body-impression claim is accepted, so the mattress itself would never have produced a complaint, and no one at the operator's office was going to read a warranty schedule on the off chance. Nothing in the product looked wrong. The condition was voided on all 360 mattresses by a base requirement of 50 mm, and the only way to find out was to go and read the page that nobody opens.

The mattress order was separate and larger in value than the frame order: 360 mattresses at $289, $104,040, with $31,212 already paid as a deposit. A warranty voided on every one of them was worth more than the balance still outstanding on the frames. The operator held the $39,060 balance on the remaining 300 frames and sent an inspector to the factory with two documents instead of one: the purchase order, and the page of the mattress warranty that sets the 50 mm condition.

What the inspector found on the three hundred

The remaining 300 frames were still at the factory and were packed, which meant the inspector could measure the real thing rather than a photograph of it. This was a Level 3 verification, two man-days at $299 a man-day, $598, because the job was not a conformity check against the buyer's own order. It was a check of the frames against a condition published by a third party, with a deflection test and a physical fit of the actual mattress on the actual deck.

The order was met. Every frame had thirteen slats, the slats were the specified material, the dimensions on the drawing were held. Under ISO 2859-1 at general inspection level II, a lot of 360 falls in code letter H and is checked on 50 units, and at AQL 2.5 that sample accepts three findings and rejects four. On the count of slats alone the sample would have passed with nothing to report.

What the sample found instead was that the spacing was 85 mm on every frame measured, at every position, with a maximum of 86 mm and a minimum of 84 mm across the run. The sample confirmed the count and the spacing at the same time, and the count was the wrong number to be confirming. The report recorded what was on the frames, what the frames were required to do, and the arithmetic connecting the two, and it said plainly that the frames were not defective. They were fully conforming to the purchase order. The defect was in the order, and the report had to be written so that the buyer could act on that without the factory accepting a fault it had not committed.

A bed frame inspection that starts and ends with the buyer's own purchase order would have passed all 300 frames on this order and reported nothing at all, because the order was satisfied in every respect it wrote. The specification was not the only document that mattered here, and the gap was not the only number on the order.

Why one frame was enough

There are two different failure patterns here and they deserve to be kept apart, because they call for opposite amounts of inspection. If the 85 mm gap had been on 12 frames out of 360, a mixed pallet, an offcut batch of slats, a jig left in one setting for an afternoon, then it is a random defect and AQL is exactly the right instrument. A level II sample of 50 from a lot of 360 finds a defect present in 12 units 83.9 per cent of the time and misses it 16.1 per cent of the time, and a 16 per cent chance of missing something is the honest price of not opening every carton on a 360-unit lot.

The gap was not on 12 frames. It was on 360 of 360, because every frame came off the same cutting list through the same jig with the same slats. A characteristic with no variation across the lot does not need a sample to be found, it needs one frame and a rule. The inspector took 8 frames across the run and found the same number on all of them, and the right conclusion from the ninth measurement was to stop measuring.

That distinction is the part worth carrying to the next order. The buyer's question was always whether the frames matched the order, and the answer was yes on every frame, and it would have been yes on every frame that factory ever made. The question that mattered was whether the frames do what the mattress warranty requires, and that question has an answer on the first frame you put a rule to.

Sampling protects you against defects that arrive by chance. It does not protect you against a number written into the product itself, and no sample size on earth will find a 50 mm condition in a frame that was never built to a 50 mm condition. The sample of 50 on the slat count proved the count was right and said nothing whatsoever about the gap.

Three ways to fix it, and what each one cost

The factory offered options once the report was read. All three were workable and one was wrong, and the difference between them was entirely in the arithmetic rather than in anybody's workmanship.

The 16-slat option is the one worth dwelling on. It is $1,044 cheaper than the 17-slat option, it uses the existing jigs, and it looks like the sensible compromise to anyone doing the arithmetic quickly. It leaves a 56 mm gap. Fifty millimetres is the limit, 56 is over it, the warranty is still void on all 360 mattresses and the rework has been paid for twice. A gap written as a target that has to be met, rather than a count that has to be hit, catches this at the drawing stage rather than at the bedside.

The plywood deck is the option that needs a second look, because it is the one that passes the warranty. The mattress maker's own wording accepts a solid platform between the mattress and the slats where the slats are more than 50 mm apart, so a deck satisfies the condition on paper. It also changes the product. Twelve millimetres of board takes the top of the bed from 380 mm to 392 mm, and the operator's fit-out had a fixed clearance at that height for a wardrobe run, so the deck would have moved the problem into the apartment. It also stops air moving under the mattress, which is the trade-off every mattress maker warns about when a solid deck replaces an open slat base.

The operator took the 17-slat option: $6,516 of rework across all 360 frames, which is 9.7 per cent of the frame order and 17 per cent of the balance that had been held, against $39,060 of balance and $104,040 of mattress warranty at risk. The first 60 frames had already been installed, so their four extra slats were fitted on site, and $696 of that $6,516 is material that went into sixty apartments rather than into the factory. The inspection that found all of this cost $598.

Four lines to replace one

The frame order needed one line replaced and one document attached. Nothing about the frame itself changed, which is the point. The product was always capable of carrying the mattress, and the order was what left the question open.

The third line is the one that would have prevented this order and the hardest to write. Buyers are used to specifying a count and treating the count as the requirement, because a count is easy to check and easy to argue about. The count is the wrong number. The gap is what the mattress feels, what the warranty names and what the factory can be held to, and once the gap is on the order the count and the width stop being independent choices and become arithmetic.

The fourth line is a habit rather than a dimension. The condition the mattress maker imposes on the base was in the operator's file the whole time, in a document that arrived with the mattresses and was never opened. Attaching it to the frame order costs nothing and moves the requirement to the supplier who can act on it.

Five lines to carry into the next order

None of this is unusual furniture practice and none of it needs a laboratory. It needs a tape measure, a straightedge and a decision to write down the number that a third party will later read your goods against.

• Specify the maximum clear gap between slats in millimetres. Not the pitch and not the count. It is the only one of the three that the mattress and the warranty both speak about.
• Say who the mattress is, and attach the condition its warranty puts on the base. A frame order is a base order, and the base requirement comes from the mattress maker rather than the frame maker.
• Measure the gap at three positions, unloaded, and record the maximum. Then measure it again with the mattress on, because 12 mm is what a resident sees and 85 mm is what a warranty reads.
• Treat a spacing attribute as systematic when the frames come off one jig. One frame settles it, and a 50-unit sample of slat counts settles nothing about spacing at all.
• Check the width the factory is using, not only the count it was given. On this order the count was never the variable. The width was, and it was never written down.

What the report could not settle

Two things stayed outside what the inspection could establish, and both of them belong in the buyer's own hands rather than in the report. The first is what the mattress maker would do about the 60 mattresses already installed. The frame was never the point of that warranty, the condition was, and the condition is now documented on the frames as built, so the maker's position is a commercial conversation the operator has to have, with the inspection report as the evidence of what the frames actually were.

The second is the operator's own fit-out guide, which carried the slat line from a two-year-old order and had never been checked against the mattress the operator was now buying. That line was the origin of the whole problem and it will be the origin of the next one unless somebody reads it against the mattress contract. No inspection report catches a guide that has not been checked.

What the report did settle is narrower and more useful. It recorded that 300 frames on a $66,960 order were built to a purchase order that did not contain the dimension that determined whether the $104,040 of mattresses sitting on them kept their warranty, and it put a number and a cost against the three ways to put that right.