The wood was 18% when the spec said 12%, and the kiln log said it was dry

The situation

A European garden-furniture importer ordered 420 solid teak folding chairs and 60 matching teak side tables from a factory in Foshan for a spring season. The chairs were 38% shipped flat, with the seat and back slats glued into a hardwood frame; the tables had 40 mm solid tops on four-leg bases. The order was €96,400, and 60% of it was unpaid when the goods were finished and loaded.

The purchase order carried a single line about moisture: kiln dried teak, moisture content 12% maximum. There was no mention of which species the teak actually was, no mention of a meter, no mention of where in the factory the reading had to be taken, and no mention of what the reading had to be at the point the container was sealed.

The factory had a kiln, a kiln log and a moisture meter, and all three had been used properly. The log recorded eleven kiln cycles for the chair frames, each with a target of 10% and a recorded average of 10.8%. The meter was a pin-type resistance meter, the kind with two probes 6 mm apart that you press into the end grain. The factory's final check read 11.4% to 11.8% on the frames, consistently, across the whole batch, for three days.

The buyer booked an inspection before the balance was released. The inspector's readings on the same chairs, with the same class of instrument, averaged 17.6%.

Two meters, two answers

Nothing had been falsified, and this is the important part of the case. Both numbers were real measurements, taken honestly, by people doing their job. They disagreed because resistance and capacitance meters do not measure the same thing.

A resistance meter pushes two steel pins into the wood and measures the electrical resistance between them. Water in the cell walls raises conductivity, so wetter wood reads wetter. It is accurate on a single species at a single temperature, and it is the instrument the trade has used for eighty years. What it cannot do is read through a coating, and what it needs is a species correction factor. Raw teak sits around a 0.8 to 0.9 correction against the instrument's baseline; with a coating on top, the pins read the wood, not the finish, so a lacquered slat has to be measured on an uncoated edge or an end-grain face.

A capacitance meter sweeps a small electric field through the surface of the wood and measures how the field is disturbed. It reads a larger volume of material, it needs no pins, and it can be used on a finished or installed surface. But the calibration belongs to the species and the machine, and it also carries a temperature correction of roughly 0.3% per degree Celsius, which means an uncorrected capacitance reading taken in a cold warehouse at 12°C overstates moisture by about 2 points on a cold morning.

Both instruments will give a different answer to the same board. The one this order turned on was simpler than that: the factory's meter was a resistance meter, it was being used on coated parts, and the readings that came back were being compared against a target with no correction factor and no reference calibration in the file. A resistance meter that has never been checked against a reference block, or against a capacitance meter, is a meter with a scale rather than a measurement.

• Ask which instrument, and write it into the order. Type of meter, model range, and whether readings are corrected for species and temperature. An order that says 12% moisture but does not say how 12% is arrived at has not specified a moisture requirement; it has specified a number.
• Insist on a reference check. Before production, the factory's meter shall be read against a certified reference block or a second, differently-principled meter, and the result recorded. A single comparison on one board settles whether the instrument is telling the truth.
• Decide where the reading is taken. End grain, uncoated edge, or a bored hole. If the part is coated, the order says so, because a coated part cannot be read with pins without exposing fresh wood.
• State the temperature at reading time, or state that readings are corrected to 20°C. Ten degrees of difference is three points of moisture, which on a 12% requirement is a quarter of the whole tolerance.

When a moisture reading means something

The second question is not what the meter says. It is when the number means anything at all, and the answer is uncomfortable for factories that dry wood and then pack it within a day.

Wood is hygroscopic. It gives up moisture to the air until the air and the wood reach the same equilibrium, and the level they settle at is the equilibrium moisture content for that species at that temperature and relative humidity. Teak in a factory at 28°C and 55% RH will stabilise around 10 to 12%. The same teak in a container at 35°C and 90% RH will stabilise near 16%. Both are correct measurements of the same board; they differ because the wood is still moving.

This is why a reading taken straight off the kiln line is not a result. The frame comes out of the kiln at 10.8%, and it has 40 mm of section, so it holds its interior moisture for days while its surface does the moving. Read at the surface while the surface is still adjusting, and the number describes the surface rather than the board. The mass needs to move through the whole section, not just through the outer few millimetres, and that takes time proportional to the square of the section size.

A factory that kiln-dries in the morning, assembles in the afternoon and packs the same afternoon measures a part that has not finished giving up water. If it then packs tightly with film and cardboard, the microclimate inside the pack is a different climate again, and the wood spends the sea voyage moving towards an equilibrium it was never measured at.

The rule that follows from this is the one the report used. Moisture content is not a property of a board at the moment it leaves the kiln. It is a property of a board, a species and a climate, and it only has meaning after the board has reached equilibrium with the air it will be measured in. So the order should not ask for a percentage at the factory. It should ask for a percentage in a stated condition, and it should say how long the parts stay in that condition before they are read.

What the report recorded, and what it refused to record

The inspector did three things on site. First, read 30 chair frames and 10 table tops with a capacitance meter, corrected to 20°C, taking five readings per part across different faces and recording the maximum rather than the average. Second, cut a small bore in the end grain of ten frames at a factory-authorised location and read the same parts with the factory's own resistance meter immediately afterwards, on the same board, at the same moment. Third, recorded the workshop temperature and relative humidity at the time of every set of readings.

The result was consistent across all three sets. Capacitance readings corrected for species and temperature averaged 17.6%, with individual parts between 16.9% and 18.4%. Resistance readings on freshly bores faces averaged 11.9%, with individual parts between 11.4% and 12.3%. The instrument difference alone accounted for about two points. The remaining four points were not an instrument effect at all: they were the difference between a dry outer fibre and a wetter core, which is exactly the gradient a part develops when it has been sealed up before it has finished drying.

The workshop was at 31°C and 62% RH on the day of the inspection. That single figure explains most of what the inspector found, because 31°C and 62% RH puts teak's equilibrium moisture content somewhere around 13%, and every board in the workshop was heading towards it. A batch measured at 12% in that room would reach 13% within days and keep climbing if it were packed.

The kiln log was entered into the report as the factory's account of the drying, not as a verified fact. It recorded ten cycles for the chair frames at 10.8% average, which is consistent with a kiln that works. It recorded nothing about the interval between the last cycle and the packing date, nothing about the workshop conditions in that interval, and nothing about whether the packed goods were held anywhere before loading. The report recorded that the log existed, that it was internally consistent, and that it could not be reconciled with a measurement taken on the parts as they were actually presented for shipment. That reconciliation is the buyer's problem to own, and the report did not do it for them.

Why the crack appeared after loading, not before

The chairs were still fine at the factory. That is the thing worth understanding about this case, and it is why a factory that only shows you the goods in the workshop is not hiding anything.

A packed container is a closed microclimate. Goods go in at the workshop's temperature and humidity, and inside a sealed steel box the heat of a sunlit yard drives the internal temperature up while the moisture released by the wood cannot escape. Over a three to four week voyage to a northern European port, internal conditions routinely cycle through 35 to 45°C at night-time ambient, with relative humidity climbing above 90% as the temperature falls. The wood in that box moves towards a much higher equilibrium than the one it was packed at, and it moves by shedding moisture from the outside in, which means the outside of the frame dries and shrinks while the core is still wet.

That gradient is the whole mechanism. An outer fibre in tension against a wetter core is a cup or a crook in a glued joint, and a glued and dowelled chair frame under that kind of stress does not bend visibly. It lets go along the weakest glue line, or it opens a joint at the rail-to-leg connection, and it does it weeks after the container was sealed and long after anyone could photograph the evidence.

The buyer's own records confirmed the timing: three shipments on the same mould, from the same factory, over two seasons. The first season's chairs had been packed from a workshop running at 24°C and 50% RH and had reached the destination without splitting. The second season's production had moved into a larger, hotter bay, and the third season's had moved again. Nobody changed the process on the furniture. The process changed around the furniture, and the order said nothing about where the reading had to be taken.

What the specification should have said

Seven lines. Each one is free to write, and together they turn a number on a factory form into a condition a buyer can defend at the container door.

• Name the species, not the trade name. The order specifies kiln-dried teak, species Tectona grandis, and where a species distinction matters, the supplier declares the actual material with supporting documentation. A moisture limit attached to a trade name rather than a species has no reference calibration, because meters are calibrated per species.
• Name the instrument and the correction. Moisture shall be determined with a capacitance meter, readings corrected for the declared species and reported at 20°C. Where a resistance meter is used, the species correction factor and the instrument's reference calibration record shall be stated. Two readings of the same board from different instruments shall not be treated as the same measurement.
• Name the condition, not just the percentage. Parts shall reach equilibrium with the workshop before measurement. Parts shall be held at 20 to 25°C and 45 to 55% RH for not less than 72 hours before the final moisture reading is taken, and the temperature and humidity at reading time shall be recorded with the reading.
• Set the limit as a maximum on every part, not an average. Moisture content shall not exceed 12% at every point measured on any part, with no averaging across a part and none across the lot. A batch average is a way of hiding the wettest board in it, and the wettest board is the one that splits.
• Make it a hold point before packing, not an inspection afterwards. The final moisture reading shall be taken and recorded on every part before packing, on the factory's own record, and that record shall be offered for verification before any carton is sealed. A reading taken after packing is a reading of the box.
• Put a condition on the shipment itself. Packed goods shall be held in a covered ventilated area at not more than 28°C and 60% RH from the final reading until container sealing, and the interval between the final reading and the sealing time shall be recorded. If the interval exceeds 48 hours, the moisture reading shall be repeated.
• Add the receiving end, because a one-sided requirement is unenforceable. On arrival, moisture shall be re-checked on not less than 5% of the shipment, selected across all positions in the container including the door end and the rear bulkhead, and the result compared against the value on the shipping record. Where the shipment was packed above the specified workshop condition, the buyer accepts a wider acceptance band at destination than at the factory, and the order says so in advance instead of after the container is open.