The desk wobbled 4 mm at full height, and the 360 lock pins that would have stopped it were still in the hardware bag

The order, and the two lines that decided it

A European office furniture distributor ordered 180 electric sit-stand desks for a client fit-out: 1,600 mm by 800 mm melamine worktops on a two-stage steel column, 620 to 1,280 mm of travel, 80 kg rated, frames in RAL 9005 matt black. The order was $340 a desk, $61,200 in total, 30 per cent paid on order and $42,840 held against a pre-shipment inspection report.

The purchase order ran to eleven lines. Worktop 1,600 by 800 mm, 25 mm melamine on MDF. Two-stage lifting column, travel 620 to 1,280 mm, 80 kg dynamic load. Lifting speed 38 mm a second. Anti-collision function. Frame finish RAL 9005 matt. Column and frame in steel, 1.5 mm wall. Assembly by the buyer. Corner protection, cartons on pallets. Delivery to a single site, balance released against an inspection report.

Two lines decided the case, and both of them were about the sample rather than about the desk. Line four said the desk was to be supplied as per the approved sample submitted on 12 August 2026, sample number SS-07, including the hardware pack. Line eleven said that any change to the structure of the frame, the column or the worktop required written notice before production. Neither line said how far the desk was allowed to move, how the worktop joint was allowed to behave, or what the hardware pack had to contain beyond the bolts.

Everything the buyer actually wanted to know was therefore absent, and the one thing that was present pointed at the wrong document. A millimetre of play at full height is a number, and a number missing from a purchase order does not get supplied by an approved sample. But whether a bracket is bolted under a joint and whether a pin is pushed into a hole are counts, and a count is something an inspector can verify on a desk standing in front of them.

Why the factory's own check passed

The factory's final inspection record for the lot shows a play check on one desk, carried out at 750 mm, reading 0.3 mm, with a tick. Nothing on that record was false. The desk was stable at 750 mm. It was also 1,280 mm of travel with the inner column out of the outer one, and a two-stage column's clearance opens up as the inner stage leaves the outer one. The same desk is a cantilever at 1,280 mm and a flat panel on two legs at 750 mm. The same frame, the same welds and the same bolts are two different structures at the two heights.

The check was also a hand check. A palm against the front edge of a desk at sitting height says very little, and it says nothing about the descending behaviour that matters most at the top of the travel. An instrument reading at one height is a measurement. A hand on an edge at one height is an impression, and an impression does not survive a dispute.

The pins were in the box. They were supplied loose, 360 of them for 180 desks, two per column, with a hole-position chart showing which hole corresponds to which height. The assembly instruction in the carton had five steps and none of them was about the pins. The packing list called them part of the hardware pack, which they were. The instruction sheet, the only document the customer actually uses, did not mention them. This is the most common way a product is quietly wrong, and it is not a manufacturing defect at all: every desk in the lot was built correctly and then assembled without one step.

The buyer's own staff found the problem before the inspection did, which is unusual and worth naming. Six desks had been assembled in the client's fit-out, and four of them were reported as moving. The note from the fit-out team said the desks felt unstable at standing height. The distributor asked for an inspection before the balance was released, and asked for the specific thing to be checked: play at the top of the travel.

Two defects, two numbers

What came back from the site was two separate problems, and the report kept them apart because they have nothing to do with each other and because the fixes cost different money. The first is the 4 mm of movement in the desk at full height, and its cause is that the lock pin was never fitted, so the inner column was free to work inside the outer one under load. The second is the 1.5 mm gap that opened in the joint of the worktop, and its cause is that a support bracket under that joint was not fitted either.

The second is worth a moment, because it is the more interesting of the two and because fixing the first does nothing for it. A 1,600 mm worktop is built as two 800 mm panels butted together, and the desk's geometry decides what happens at the joint. The legs were set in at 1,120 mm on a 1,600 mm top, so the front beam is a 240 mm cantilever on each side of the leg and the joint at the centre of the top is not supported by anything underneath it. Under the 80 kg the desk is rated for, the unsupported centre of the joint opens. The bracket that closes it was quoted to the buyer on the accessories list and was not ordered, and the approved sample had one fitted.

After the pins were fitted on the trial desks, the play fell below 1 mm and the joint still measured 1.5 mm. That single pair of measurements is what separates the two cases. A desk can be perfectly rigid and still have a 1.5 mm gap in its top, and a desk can have a tight joint and still be loose at the top of its travel. Neither number substitutes for the other, and a corrective action on one does not correct the other.

That is also why the lot could not be written up as a single failure. The pins were a missing fitting against a sample, the bracket was a missing part against a sample, and the play and the joint gap were numbers against a purchase order that carried neither. Three findings, three different routes to a decision, and the buyer's commercial call depends on which is which.

How the play was measured

The method was written down before the lot was opened, because a play measurement without a stated method is an opinion with a decimal point in it. The desk was raised to the height to be measured with the motor off and unloaded, standing on the factory floor. A dial indicator with 0.01 mm resolution was clamped to the front edge of the worktop, 100 mm in from the edge, plunger horizontal, reading lateral movement. A spring scale applied 100 N laterally to the centre of the front edge over three seconds and held it for five. Both the peak reading and the reading after the load was released were written down.

The two readings mean different things, and reporting only one of them is how this kind of defect gets argued about instead of fixed. A peak of 4.0 mm that returns to zero is play: the parts are moving against each other inside a clearance and the frame is not bent. A peak of 4.0 mm that stays at 2 mm once the load is off is a distorted column or a joint that has moved, and no amount of re-tightening or re-assembly will fix it. The report recorded the residual, and the residual was zero on all 13 desks, which is what identified the cause as missing pins rather than a bent column.

The same check was repeated at three heights on every desk, because the clearance only opens at the top of the travel. Each desk was also pushed down at the front edge at maximum height with the motor off, 100 N downward at 50 mm from the edge, to see whether the worktop would descend.

A second round of measurements was taken on the same desks, at the same heights, after the lock pins had been fitted to the columns the way the sample had them, so the report could show the buyer what the defect was costing and what the fix was worth.

The joint that opened 1.5 mm

The worktop joint was measured separately, with a different method, because it is a different question. The desk was loaded with 80 kg distributed across the top, and a feeler gauge was read at three positions along the joint, 200 mm from each end and at the centre, with the reading taken at the front face of the joint where it opens. The joint measured between 1.3 mm and 1.5 mm on all 13 desks, and opened to 1.8 mm when the front-edge push from the play test was applied at the same time. On the approved sample SS-07, with its bracket fitted, the same measurement gave 0.3 mm.

The cause was visible once the worktop was turned over. There is no bracket under the joint on any desk in the lot. The sample has one, bolted to the frame and carrying the centre of the joint. The bracket was on the accessory list at $3.40 a desk, which is a price low enough to be read as a consumable rather than as a structural part, and it was never ordered because the purchase order described the desk as a two-piece worktop and said nothing about what held the pieces together.

This defect had a property that the wobble did not have: it was invisible until the desk was loaded. On an empty desk the joint closed, because the joint's own stiffness held it shut and nothing was pulling it apart. It opened under the 80 kg the desk was sold to take, and it opened at exactly the point where a person's forearms rest. An inspector who checked the top on an empty desk would have recorded nothing.

The two defects also had opposite profiles. The play appeared only at the top of the travel and vanished at 750 mm. The joint gap was there at every height and grew with load. One is a column problem and one is a worktop problem, and a single check on a desk at sitting height catches neither.

What the report could fail, and what it could not

Two of the findings were failable and two were not, and the difference came entirely from the purchase order. The missing lock pins and the missing support bracket were non-conformances against line four, which said the desk was to be supplied as per approved sample SS-07 including the hardware pack. A pin fitted and a bracket bolted are things that can be counted and photographed on a desk in the showroom, so the report recorded them as non-conformances and named the corrective action for each.

The 4.0 mm of play and the 1.5 mm joint gap could not be recorded as non-conformances, because there is no millimetre figure anywhere in the eleven lines against which to fail 4.0 mm or 1.5 mm. The report did what it could with that: it stated the measurement, stated the method, stated the conditions, and said plainly that the purchase order was silent and that the decision on the balance belonged to the buyer. Inventing a limit the buyer had not written would have put the whole document in doubt, and a report that fails goods against a standard it supplied itself is a report nobody can rely on.

There was also a piece of paper the buyer had been shown and had believed. The factory's datasheet said the desk was compliant with BIFMA G1-2013, and the buyer had taken that as evidence that the desk was stable. It is not. G1-2013 is a set of ergonomic design guidelines for furniture used with computers, and it contains recommended dimensions for users and clearances around the worktop. It carries no stability requirement, no strength requirement and no pass or fail criterion of any kind, so no document can say a desk complies with G1 and mean that the desk will not wobble. The product test standards for this desk are EN 527-2 for safety, strength and durability, with EN 527-1 for dimensions and EN 527-3 for worktop sizes, and ANSI/BIFMA X5.5 for desks and tables on the other side of the Atlantic. When the report asked for a laboratory report for the model, the factory's answer was that the desk was built to the same structure as a model that had been tested. That is not a test report either, and no batch of 180 desks is covered by a report for a different desk.

One piece of housekeeping decides what the report can cover at all. EN 527-1 assesses a desk inside a height band of 680 to 1,180 mm, and this desk travels from 620 to 1,280 mm, so the extreme positions of its travel sit outside the band the standard assesses and the dimensional clauses at those positions come out as not tested. The report said that plainly instead of implying a conformity the standard does not support. A dimension that was never tested is not a dimension that passed.

The rework, and the eight desks checked again

The factory's first reaction was to argue about the load. 80 kg was the figure on the datasheet, and a 1,600 by 800 mm melamine top does not weigh 80 kg. The inspector's position was that the rating is a rating: a desk sold as taking 80 kg has to be stable at 80 kg, and the shortfall had appeared with nothing on the desk at all. The argument ended when the factory ran the number itself.

The rework took four days. All 180 desks came back to the bench: the two lock pins were pushed into the columns at the delivery height, the hole-position chart was added to the assembly instruction as its first step, and the support brackets were fitted under the worktop joints with the panels re-fastened at 400 mm centres. Nothing was remade and no part was replaced. The brackets cost more than the pins.

The re-inspection took one man-day and covered 8 desks drawn from the reworked lot, all assembled, pinned and measured at 1,280 mm, with the joint gap read under the specified load. Play came out between 0.4 mm and 0.6 mm, the joint gap between 0.2 mm and 0.3 mm, and no desk descended under the 100 N downward push. The balance was released after that and not before it.

The useful number from this case is not the 4 mm. It is that the fix was 360 pins, 180 brackets and one line added to an instruction sheet, and the lot was still held for 12 days and re-inspected because the buyer asked for a report and a report was written. A desk that moves 4 mm at full height is a nuisance. A desk that moves 4 mm at full height and has no lock fitted is a desk that can come down under a hand on the front edge, which is about 10 kg of force, while somebody is standing at it.

How the specification should be written

None of the lines below cost the factory anything to comply with, and every one of them exists because a specific measurement was missing from the version that was signed. They are written as they would have appeared in the purchase order.

• Lock pin, one per column, part number stated, with a hole-position table giving the worktop height for every hole, and the pin fitted to the column at the height of delivery before the desk leaves the factory. The pin is to be refitted after any height change, and the assembly instruction is to say so as its first step.
• The lock is a safety function and is written as one. With the motor off and the desk at 1,280 mm, the worktop shall not descend more than 1 mm when 100 N is applied downward at 50 mm from the front edge.
• Maximum lateral play: 1.5 mm at the maximum specified height, measured with a dial indicator of 0.01 mm resolution clamped to the front edge 100 mm in from the edge, with 100 N applied laterally at the centre of the front edge over 3 seconds and held for 5 seconds. The same limit applies at the minimum height and at the seated height.
• Residual movement: after the load is released, the worktop shall return to within 0.5 mm of its original reading. A bent frame and a loose frame look identical under load and are not the same fault, and the residual is what tells them apart.
• Worktop joint: where the worktop is made in more than one panel, the joint gap shall not exceed 0.5 mm with 80 kg distributed across the worktop, read with a feeler gauge at three positions along the joint, and the joint shall not open under the downward push in the second line above.
• Support under the joint: a bracket is to be fitted under any worktop joint, and the panels are to be fastened to the frame at not more than 400 mm centres. The bracket is to appear on the assembly drawing and in the bill of materials, not on an accessory list.
• The heights at which the checks are made are part of the requirement. Play, descent and joint gap are to be measured at the minimum height, at the seated height and at the maximum height, because the clearance between the stages of a two-stage column only opens as the inner stage leaves the outer one.
• Test evidence: a laboratory report to EN 527-2 for safety, strength and durability, or to ANSI/BIFMA X5.5, covering the exact model in current production, supplied before shipment. A datasheet reference to BIFMA G1 is not acceptable, because G1 contains ergonomic design guidance and no stability or strength criterion.
• The height band: the desk travels from 620 to 1,280 mm, which is wider than the 680 to 1,180 mm band EN 527-1 assesses. The purchase order is to say which dimensions are to be verified by the buyer's own measurement at the two extreme positions, and the report is to state which clauses were therefore not assessed.
• Sampling is stated in assembled desks, not in cartons. A lot of 180 is code letter G and a sample of 13 under ISO 2859-1 at general inspection level II, and all 13 are assembled, pinned, loaded and measured at the three heights.
• The approved sample is used for countable features only, and the counts are written into the purchase order: bracket present, pin fitted, hole-position chart supplied, assembly instruction step present. A sample can confirm a number, but it cannot be the number, and a dimension that matters should never be left to a bag in the post.

360 pins, in the bag, the whole time

Nothing on this order was fabricated, mislabelled or concealed. The factory built 180 desks of a steel frame, a two-stage column and a melamine worktop exactly as instructed, and it supplied 360 lock pins in the correct hardware pack. The defect was that nobody wrote down the one step that turns those pins from parts into a locked column, and that nobody wrote down a millimetre.

The thing worth carrying to the next sit-stand order is the third line on the list and the eleventh. Play and joint gap are numbers, and a number has to exist before anyone can fail it. A count can be checked against a sample, but a count is not a specification. Where the requirement matters, write the number, write the height it applies at, and write how it is measured, and then the sample becomes confirmation of the number instead of a substitute for it.

And keep the two apart. This desk was loose because of a missing steel pin, and its worktop had a gap because of a missing bracket under a 1,600 mm panel. A buyer who had looked at one of the two and assumed the other was covered would have paid for a desk that was still wrong in one place, and would not have known which place it was.