Every drawer pulled out smoothly on 450 mm full-extension runners, but six of the twenty sampled pedestals carried a different slide model rated 15 kg where the order specified 25 kg, and they bound under 15 kg of contents.
Illustrative example This case is written from the patterns we see in furniture export orders, and the figures are representative rather than from a specific client file. It shows how a check works and what it finds. We will replace it with a real, anonymised case as soon as we can publish one with the buyer's permission.
The order, and the line that could not be checked
A European contract-furniture buyer ordered 120 office desks for a fit-out, each with a three-drawer mobile pedestal: one file drawer and two stationery drawers, 360 drawers in total. The unit price was $420 a desk, the order was $50,400, 30 per cent was paid on order, and the remaining $35,280 was held against a pre-shipment inspection report.
The purchase order had a hardware section, and the hardware section ran to nine lines. Desk 1,400 by 700 mm, melamine-faced particleboard top, 25 mm edge. Pedestal 400 mm wide, 500 mm deep, 620 mm high, on 50 mm castors. Drawer boxes 16 mm particleboard, 480 by 400 mm internal, melamine faced both sides. Drawer front gaps not more than 3 mm between adjacent drawers. Runners on System 32 fixing pitch. Soft close on the top drawer only. Then the runner line: 450 mm full extension, ball bearing, load rating not less than 25 kg, 50,000 cycles, to EN 15338.
That last line was the only line in the order that could not be checked as written, and it was the line that decided the case. It gave no manufacturer and no part code, so 450 mm full extension ball bearing describes a family of slides rather than a slide. It gave the load rating without saying whether the rating covers the drawer box alone or the box and its contents together. It gave 50,000 cycles without a test class, and under the current version of EN 15338 a cycle count without a class does not identify a test. And it gave a length where the buyer needed a travel, because a 450 mm slide can be a 450 mm slide and still move the drawer 340 mm.
Everything else in the hardware section was checkable on site in a few minutes. This one line held the entire order's worth of hardware risk and it was four sentences long with no number a person could measure except the length.
This is an illustrative example drawn from the pattern that runs through most hardware substitution complaints: the purchase order names a component type and its headline dimension, and the defect only exists in the relationship between that headline figure and the working conditions. The figures are representative of the category rather than taken from a named customer file.
The drawers all pulled out perfectly
The buyer's inspection request asked for a drawer alignment check on every pedestal in the sample: gaps between adjacent drawer fronts, the alignment of each front against the pedestal side, and that every drawer runs smoothly through its full travel. It is a reasonable request, it is the request a buyer makes after a bad experience somewhere else, and on this order it produced no finding at all.
All 20 sampled pedestals had drawer gaps inside the 3 mm the order allowed. All 60 drawers in those pedestals ran through their full travel smoothly, with no binding, no scraping and no measurable side play in the runner. The fronts were flush. Nothing in the sample looked wrong, and the factory was right that the goods matched the drawing.
That is worth stating plainly, because it is why the goods shipped and why nobody objected. A drawer slide is a very good-looking thing when it is empty. The friction that decides whether a slide is fit for the job only appears once the load is on it, and the load arrives after delivery, in a filing cabinet, in year two. An inspection that stops at the front of the drawer has confirmed that the drawer opens. It has not confirmed that the drawer works.
There was no visible difference between the six slides that failed and the fourteen that passed. Same height in the carcass, same silver plating, same nylon cage, same fixing holes, same 450 mm length. From the front of the desk they were the same slide.
Where to go next
The loaded test, and the six that bound
The loaded test was written down before the lot was opened, in the same form as the inspection request, so that the number in the report could be reproduced by somebody else. Fifteen kilograms of steel test weights, distributed across the base of the drawer rather than stacked in the middle, because a file drawer is loaded from the front and the weight ends up towards the rear of the box where the leverage is worst. Then the drawer pulled out to full travel and pushed back home, and the force it took to close was judged by hand and written down as free, sticky or binding.
Six of the twenty drawers bound. The drawer reached between 45 and 70 per cent of travel and stopped, and it could not be pushed home again with the load still in it, so the drawer front stood 4 mm proud of the pedestal line. With the drawer out and loaded, the rear of the drawer box sat 3 mm below the top edge of the runner. The other fourteen drawers took the same 15 kg through the full travel and closed without a complaint.
That is the whole finding: a lot of 120 desks, a sample of 20 pedestals, 60 drawers opened and closed under load, and 6 of those 60 that would not go home. Nobody had to open a single drawer with a screwdriver or strip a single rail off the carcass. The test was a set of weights and a pair of hands.
| Loaded drawer test on the 20 sampled pedestals | Result | Reading |
|---|---|---|
| Pulled to full travel and pushed home, 15 kg distributed across the drawer base | 14 of 20 pedestals | Closed flush, no binding, no measurable drop at the rear of the drawer box |
| Bound between 45% and 70% of travel, could not be pushed home under load | 6 of 20 pedestals | Drawer front stood 4 mm proud of the pedestal line; loaded and extended, the rear of the box sat 3 mm below the runner |
| Sample size and plan | 20 pedestals, 60 drawers | ISO 2859-1 at general inspection level II; a lot of 120 is code letter F, so the sample is 20 |
| Defect classification | Major, AQL 2.5 | A sample of 20 accepts one and rejects at two. Six is a rejection of the lot on this characteristic |
Where to go next
→ how to read an AQL sampling plan before the goods are opened
What was stamped on the rail
The slide is marked. On the outer rail of a ball-bearing runner, near the rear fixing, the manufacturer stamps the part code and the load class. The inspector photographed that marking on every drawer in the sample, and the six drawers that bound carried a different part code from the fourteen that ran, and the load class on the rail read 15 kg where the other rails read 25 kg.
The report did not rest on the photograph. A marking is a claim made by the factory, and the point of an inspection is to produce evidence the factory did not write. The functional result is the evidence: six drawers in a sample of 20 could not be opened and closed under a load the order's own 25 kg rating comfortably covers, and a slide rated 15 kg is a slide that cannot do that. The marking is what identifies which slides are in the wrong drawers and how many of them there are, which is the difference between a rejected lot and a rejected sample.
Then the arithmetic that decides the argument. The drawer box is 16 mm melamine-faced particleboard and weighs about 8 kg empty. The load rating on a slide is a system rating: the drawer box plus its contents, and the number that governs is the dynamic one, because at full extension the lever on the rear section and on the carcass screws is at its greatest. So the 15 kg of test weights was never a 15 kg test. It was 23 kg on the slide. A 25 kg runner carries that with margin. A 15 kg runner cannot, and the six drawers that bound were the ones that had it.
There is a second trap sitting in the same 15 kg figure, and it is worth naming because the buyer would have met it in the customer's office even if the slides had been correct. A slide rated 15 kg carrying an 8 kg drawer box has 7 kg of payload left, not 15. A buyer who reads 15 kg as 15 kg of folders fills a 15 kg slide with 15 kg of folders and puts 23 kg on it. Rating lines have to say which of those two numbers they are.
Where to go next
→ what a report records when a lot passes and a lot fails
→ our desk and casegoods verification scope, including hardware and drawer checks
Why one drawer can be right and the next one wrong
The factory had not supplied a bad slide. It had supplied a different slide from a range that includes the one that was ordered, and from the front of the desk the two are not distinguishable. Same 450 mm length. Same full extension. Same ball bearing and nylon cage. Same zinc plating. Same System 32 fixing holes. The difference is in the steel, in the ball size and in how the members are arranged, and none of that is visible once the slide is screwed into the carcass.
What the two had in common is what made the mistake possible, and what the buyer got wrong is that only one of those shared points was ever checked. Length was checked and it matched. Travel was checked and it matched. The load class was not checked at all, because the order gave a number without saying where the number came from, and the factory's incoming note gave a number without saying what was included in it.
Travel and load are independent properties, and each one has to be confirmed separately rather than inferred from the other. Length only tells you the travel once the extension type is also known, because the extension type is what sets the travel: a 450 mm single-extension runner moves the drawer about 340 mm and leaves roughly a quarter of the box inside the carcass, while a 450 mm full-extension runner moves it the full 450 mm and gives clear access to the back panel. And a load rating only means something once the test behind it is defined, because a catalogue rating is a rating in a laboratory test frame, and an assembly on a factory floor is not that frame.
EN 15338 is specific about this last point, and it is worth knowing the number because it is the kind of detail that gets left out of a hardware section. The standard requires the carcass used for the test to be particleboard of 0.65 g/cm3 density with an axial screw pull-out strength of 1,100 N, because slides work loose in a carcass that is too light. A slide that passes on a test rig in one factory can still fail on a pedestal built from a lighter board, and nothing on the purchase order said which board the test was supposed to be run on.
| What the order usually says | What that line actually determines | How it goes wrong |
|---|---|---|
| 450 mm | The nominal length of the slide, not the travel of the drawer | A 450 mm single-extension runner gives about 340 mm of travel and a 450 mm full-extension runner gives 450 mm. Same figure on the order, two different drawers |
| Full extension | Whether an intermediate member is fitted at all | On a partial-extension slide roughly a quarter of the box stays inside the carcass and the back of the drawer is reached by lifting items out at an angle |
| 25 kg | A class of slide, subject to whether the box is counted in or out | A 25 kg slide carrying an 8 kg drawer box has 17 kg of payload left, and a buyer who reads 25 kg as 25 kg of contents does not |
| 50,000 cycles | A number with no test class attached | EN 15338 now splits durability into three classes with different cycle counts and load factors, so the same 50,000 is not the same test |
Where to go next
→ our desk and casegoods verification scope, including hardware and drawer checks
→ how to read an AQL sampling plan before the goods are opened
The test the buyer could have named
The order already said to EN 15338, and that half of the line was the most useful half, because the current version of the standard gives a buyer a way to say what 50,000 cycles means. EN 15338:2024 splits the durability test into three classes: general household use at 20,000 cycles with a load factor of 1.25, office and commercial use at 50,000 cycles with a load factor of 2.5, and high intensity use at 80,000 cycles with a load factor of 2.5, all run at a controlled speed so the slide does not heat up. An order that says 50,000 cycles to EN 15338 Level 2 has removed the ambiguity. An order that says 50,000 cycles has left the whole test open.
The same standard supplies the static figures that an inspector can put in a report and a buyer can argue from. For a commercial grade slide: 300 N of vertical downward load on the open drawer, 150 N of horizontal lateral load, and 200 N of outward static load to prove the runner does not let go of the drawer when it is pulled. After the vertical load the guide rail must not deflect more than 1.5 mm, and the drawer front must not deform by more than 4 per cent of the stroke.
Two things the standard does not do, and both matter when a report is written against it. It tests the extension element and the parts used to fix it, not the cabinet and not the desk, so a report saying the slides comply with EN 15338 says nothing whatever about the pedestal they are screwed into. And the result is valid only for the extension element actually tested, which is another way of saying that a certificate for one slide is not a certificate for the slide in the drawer.
One more figure worth putting next to the case. 150 N of lateral load is about 15 kg pushed sideways. The failure in this order was found by pushing the drawer closed with the load still in it, which is the direction that load applies. Nobody needed a laboratory to find this one. They needed the weights, the method written down first, and the patience to open sixty drawers.
- Vertical downward load on the open drawer: 300 N for a commercial grade slide, with guide rail deflection after the load limited to 1.5 mm.
- Horizontal lateral load: 150 N, roughly 15 kg pushed sideways, applied with the drawer open.
- Outward static load: 200 N, to show the runner holds the drawer on the rail when it is pulled.
- Drawer front deformation: not more than 4 per cent of the stroke length after the durability test.
- Durability classes under EN 15338:2024: 20,000 cycles at K 1.25 for household general use, 50,000 at K 2.5 for office and commercial, 80,000 at K 2.5 for high intensity.
Where to go next
What the specification should have said
The whole case would have been caught by six lines in the hardware section of a purchase order. This is what a runner line needs to carry, and what each of those lines stops.
Five of the six lines are about the product. The sixth is about the test, and it is the one most hardware sections leave out, because the four that failed here were all compliant lines. Six lines about the slide and one line about the load test is the difference between a rejected lot and a conversation that takes a month.
None of this needs a laboratory. It needs a part code, a travel in millimetres, a load that says what it includes, a test class, and the words 15 kg of contents on the drawer.
| Line | What to write | What it stops |
|---|---|---|
| Model | Manufacturer and part code, with the marking on the rail photographed and reported for every drawer in the sample | A different model from the same range, dimensionally identical and in a lower load class |
| Travel | 450 mm nominal, full extension, minimum 420 mm of clear access at full travel | A partial-extension slide that is 450 mm long and gives 340 mm of travel |
| Load | Dynamic load rating not less than 25 kg with the drawer box and its contents combined, at full extension | A rating read as contents only, and the weight of the box itself discovered in the customer's office |
| Cycles | EN 15338:2024, Level 2, 50,000 cycles at load factor 2.5, test report on request | A cycle count with no class, which is not a requirement anyone can fail |
| Mounting | System 32 fixing pitch, 12.7 mm side clearance each side for a standard ball-bearing runner | A runner needing 17 mm or 12.5 mm clearance and a drawer box cut to the wrong width |
| Sampling and reporting | Drawer operation inspected as a major defect at AQL 2.5, with the rail marking reported for every drawer in the sample | A pass on drawer fronts that says nothing at all about what is behind them |
| Loaded test | Drawers to be opened and closed under 15 kg of contents distributed across the drawer base, at full travel | A lot that passes every visual check and binds in the customer's filing cabinet |
Where to go next
What to carry to the next order
Nothing on this order was mislabelled and nothing was fabricated. The factory built 120 desks to the order, fitted 360 drawers, and packed them correctly. The failure was one line in a purchase order that described a family of slides rather than a slide, and the factory's reading of it was not unreasonable: it said 450 mm, full extension, ball bearing, 25 kg, and the slides that came were 450 mm, full extension, ball bearing, and marked 15 kg.
The thing worth carrying away is that the drawer is the part of a desk that gets used, and the slide is the part of the drawer that gets specified by somebody else. A buyer who writes a part code, a travel in millimetres and a load that says the box is included will not have this conversation again. A buyer who writes 450 mm full extension ball bearing runner 25 kg will be told the slides look right, and they do.
And keep the two checks apart, because they do different work. The loaded test found six drawers in a sample of twenty. The marking told the buyer which slides were in those six drawers and what class they were. The lot could only be failed because the order carried a load rating at all, and the load could only be argued at 25 kg because the test was defined before the lot was opened and written into the report in those terms.
One set of weights, a written method and a camera on the rail would have caught this on any one of the 120 desks. That is the whole case, and none of it needed a laboratory.
Where to go next
Have an order you need verified? Send us your PI and factory address. Three levels, published prices, $149 per man-day from.