The desks passed. The load they were tested at was two thirds of the number in the contract.

The situation

A European contract furniture supplier ordered 350 height-adjustable office desks for a fit-out programme. The purchase order carried a performance clause in one line: *static load capacity 1,500 N, tested to EN 527-2.*

The factory accepted the order without comment and ran its standard pre-shipment load test on every batch. The test sheets came back with a signature, a date, a measured deflection figure and the word PASS.

Nobody had read 1,500 N next to the number on the test sheet. The desk was strong. The record did not say it was strong to the number that mattered.

The number on the test sheet was not wrong

Read carefully, the factory's test record was entirely accurate. It said the load applied was 1,000 N. It said the hold duration was ten minutes. Both figures were correct measurements of what the machine actually did.

The problem is that neither figure was derived from the contract. They came from the factory's own internal test procedure, which had been written years earlier for a different desk model and a different customer, and had never been re-checked against a purchase order.

Unit and duration mismatches are the ordinary failure

None of these is unusual, and none is anybody's fault in the ordinary sense. They happen because a load figure is a number in a document rather than a test anyone has to perform, so it never gets checked against anything.

• **Newtons against kilograms.** 1,500 N is roughly 153 kgf. A factory working in kilograms who reads 1,500 as 1,500 kg is testing almost ten times the required load — which passes trivially. The reverse error, reading 1,500 N as 150 kgf and testing 150 kg, produces exactly the 1,000 N figure this batch was tested at once rounding is applied. **Both parties can be sincere and the numbers can still be off by an order of magnitude.**
• **Peak load against sustained load.** A contract may require a brief peak load to check for catastrophic failure and a separate sustained hold to check for creep. A test that applies 1,000 N for ten minutes is testing neither. Creep in a particle board top or a die-cast aluminium leg shows up over hours, not minutes — the desk can pass a ten-minute hold and still sag under a full working day.
• **Test standard version.** EN 527-2 has been revised, and the deflection measurement and load application differ between editions. A test sheet citing a standard without an edition or year cannot be compared to a purchase order that cites one.

When the weights do not add up, what the report says

That wording is deliberately unflattering to the inspection. It records a gap in what was checked rather than a problem in what was found, and it means the buyer cannot treat the result as a clearance. **A report that says it could not verify something is more useful than one that says pass**, because the first produces a decision and the second produces a false one.

• **Wrong: record a pass at the reduced load.** A test sheet that says *tested to 1,000 N for 10 min — PASS* is being read as a pass at 1,500 N by everyone who receives it. The number is on the sheet and nobody re-reads it.
• **Wrong: record a fail.** Nothing in the data shows the desks cannot take 1,500 N. A desk tested to 1,000 N and passing tells you it survives 1,000 N. It does not tell you it fails at 1,500 N, and reporting it as a failure claims knowledge the inspection did not obtain.
• **Right: report the test as performed, and the requirement as unverified, as two separate statements.** The measured result and the contract requirement both appear, with the gap between them stated in the same table. The verdict is on the specification: *the load requirement could not be verified on site with the equipment available.*

Written acceptance, and why it has to be written

The factory accepted the deviation without argument, and it did not dispute the test record — because the record was accurate and the report had not accused anyone of falsifying it. That is the practical return on writing these findings as observations: **the finding closes before the container ships instead of during the demurrage window.**

• **The requirement quoted from the contract** — *1,500 N, one hour, per EN 527-2* — not a paraphrase.
• **The test actually performed** — *1,000 N, ten minutes, 3 units*, with the rig identified and its calibration status noted.
• **The gap, quantified** — *33% below the specified load, 83% below the specified hold duration.*
• **The risk being accepted, in words.** The buyer stated that a sustained 1,500 N condition was not expected in service, because the desks were specified for general office use rather than for a sit-stand station loaded with reference documents. That reasoning belongs in the document. Without it, a signature on a deviation is a blank cheque.
• **What happens if it fails in service.** The buyer reserved the right to claim against the factory for consequential loss, so the acceptance did not extinguish the warranty — it deferred the point.
• **Signature, date, and both party names.** A verbal acceptance given at the loading dock is worth nothing at the point where it is needed.

What to put in the specification

A load capacity is normally one line in a purchase order, and one line is not enough to make it checkable. Six additions:

• **The unit, twice.** *1,500 N (newtons)* — and, if the desk will also be described in kilograms anywhere in the documentation, state the equivalence: *1,500 N ≈ 153 kgf.* This one clause prevents the most common error in the category.
• **Load type and duration together.** *Static load 1,500 N applied to the centre of the top for one hour, deflection measured at 10 minutes and again at 60 minutes.* Duration is not a detail; a ten-minute hold is not a weaker version of the same test, it is a different test.
• **The standard with its edition.** *EN 527-2:2003* rather than *EN 527-2*, because the measurement method changes between editions and an unversioned citation cannot be compared.
• **The deflection limit and where it is read.** *Maximum deflection 10 mm at the top centre after one hour.* A limit without a point is a limit nobody can apply consistently.
• **The sample size.** *Five units drawn from the front, middle and rear of the batch, tested to EN 527-5 sampling.* Three units chosen by whoever is holding the timer is a demonstration, not a result.
• **What happens when the requirement cannot be tested on site.** *Where the specified load cannot be applied with equipment available at the inspection location, record the test as performed and the requirement as unverified. A pass at a reduced load shall not be recorded as a pass against the specified load.* That sentence is the entire subject of this case, and it costs one line to write.