Measurement tolerance: when a fabric difference is only rounding

Measure the same roll of fabric in three ways and you can get three figures: the counter on the measuring table shows 148.6 m, the label says 150 m, and the value calculated from the scale comes out at 151 m. Part of this difference is not a loss but the nature of measuring. Without a tolerance written in advance you can make two mistakes: treat small measuring differences as losses and spend hours investigating them, or wave a real shortage through as "just rounding". Here is how to set a tolerance and what it must never hide.

Why the figures differ

• Meter counter: the tension in the fabric and the counter's calibration. Pull the fabric while measuring and it stretches; let it go slack and it shortens.
• Scale: the step size of the scale, the moisture in the fabric, the weight of the cardboard core and packaging.
• Label: the metres written by the manufacturer or supplier are often the measure at the machine; a trimmed end or an opened roll changes it later.
• Conversion: a kg-to-metre calculation based on weight per m² and width magnifies a small deviation in the weight.

None is "wrong" on its own, but when they meet in one record, small differences arise.

How to write the tolerance

The tolerance is written at two levels: per roll and per lot (line) total. For example, 1 per cent per roll and 0.5 per cent on the line total. The rules might be:

1. The decision is written before the count, together with the measuring method (counter or scale).
2. The same method is used on every roll.
3. Results fall into three bands: green (within tolerance), where no difference is written and the recorded figure stands; yellow (one to two times the tolerance), which is re-measured; and red (more than twice), which goes into the variance table as a line and gets a cause.

Setting the tolerance after the count invites adjusting it to make a difference look acceptable. It is a company decision and may vary by product: a stretchy knit measures less steadily than a woven fabric.

Example: within tolerance, but always in one direction

An illustrative case: a lot of 40 rolls labelled 150 m each, with a tolerance of 1 per cent per roll, which is 1.5 m. The team measures 10 rolls. The differences between measured and labelled (metres) are: −0.8; −1.2; +0.4; −1.0; −0.6; −1.4; −0.9; −1.1; −0.7; −1.3.

None exceeds 1.5 m, so taken one by one every roll is within tolerance. But the total is −8.6 m, an average of −0.86 m (about 0.57 per cent), and nine of the ten rolls fall short. Applied to 40 rolls, the same average comes to 40 × 0.86 = 34.4 m. This is not rounding; it is a systematic deviation with a direction. Possible causes: the supplier's label overstating the metres, the counter drifting one way, or the fabric being squeezed while measured. To find out, measure the same rolls by a second method or on another counter; if the counter is not the cause, talk to the supplier.

So put a direction rule next to the tolerance: if most of the measured rolls (say eight or nine out of ten) deviate the same way, the line is flagged even though each roll is within tolerance.

Scales and resolution

If you calculate from the scale, its step size is part of the tolerance. On a scale with 0.1 kg steps, the reading of a 24.8 kg roll is uncertain by half a step, 0.05 kg: 0.05 ÷ 24.8 = about 0.2 per cent. Converting to metres adds the deviation in the fabric weight. Metres from the scale are therefore a last resort when there is no label or counter result, and should be marked "calculated" in the report.

Misusing the tolerance

• Adjusting it after the count to make the difference acceptable.
• Writing "within tolerance" for every difference and never looking for a cause.
• Looking only at the percentage; on a long roll the same percentage is more metres, so the table needs a metre and a value column too.
• Applying one tolerance unchanged to every type of fabric.

In short

A tolerance stops you counting small measuring differences as losses; a direction rule keeps a real deviation from hiding behind rounding. StockSystem measures rolls without a label, or opened or cut ones, and writes the result in the unit on the item card; the report shows the recorded unit and the quantity found on site side by side.