Tolerance: why a garment can measure 1 cm off the chart
Tolerance is how far a finished garment may measure from its spec and still pass. Fabric, cutting, sewing and the tape itself all move the numbers a little, so a chart value is a target, not a promise to the millimeter. Keep the tolerance under half the step between sizes.
A garment that measures 1 cm off its size chart is usually not a mistake. Every point of a spec sheet (chest, length, sleeve) comes with a tolerance: how far a finished piece may land from the target, plus or minus, and still be accepted. Two pieces of the same size, from the same batch, can measure a little differently, and both be right.
The number that matters is not the tolerance alone. It is the tolerance against the step between sizes. If a piece can drift further than half a step, an M and an L can measure the same, and the chart stops telling them apart.
Where the centimeter goes
A pattern is exact. The garment made from it isn't, for reasons nobody can fully remove:
- Fabric. Knits relax after cutting and wovens can shrink at the first wash. The same pattern in two fabrics can come out at two sizes.
- Cutting. Fabric is cut in stacks, and the layers at the top and bottom of a stack don't come out quite alike.
- Sewing. A seam a few millimeters wider, repeated on both sides of a body, takes width off the finished piece.
- Finishing. Washing, pressing and steaming set the final size, and move it.
- Measuring. Where the tape starts, how flat the piece lies, how hard it's pulled. On a piece laid flat, any error doubles when you go around: 1 cm across the chest is 2 cm of circumference.
None of these is a fault on its own. Together, they are why a spec sheet gives a range around each target rather than a single number.
Tolerance against the size step
Take the garment measurements of Ruban Size's Men's T-shirts template, and suppose your spec allows 1 cm either way:
| Size | Half chest, laid flat (cm) | Accepted range | Length (cm) | Accepted range |
|---|---|---|---|---|
| S | 50 | 49-51 | 70 | 69-71 |
| M | 53 | 52-54 | 72 | 71-73 |
| L | 56 | 55-57 | 74 | 73-75 |
The half chest steps 3 cm a size, so a 1 cm tolerance leaves a clear centimeter between neighbors: no S can measure like an M. The length steps only 2 cm, so the same tolerance lets an M at its longest (73 cm) match an L at its shortest. A shopper choosing by length could get the same garment in two sizes.
The rule that falls out of it: keep each tolerance under half the step of that measurement. Here, under 1.5 cm on the half chest and under 1 cm on the length. Lengths and small points (collars, sleeves) step the least, so they need the tightest tolerance. How size grading works has the steps of the other templates.
Try the chart, in its body and garment tabs:
The shopper's tape has a tolerance too
The other side of the chart is a person with a tape measure, and they make errors of their own. Ruban Size's Find my size assumes a typical error of about 1.7 cm on a body circumference, taken from a study of self-measured waists (Barrios et al. 2016) and the spread of waists in ANSUR II.
That error is why where a shopper falls in a range matters. On the women's tops chart, the M covers chests of 90 to 95 cm, or 89.5 to 95.5 cm once the gaps between sizes are split halfway. A 92.5 cm chest sits in the middle: typical measuring error keeps it inside the M about nine times in ten. A 95 cm chest is half a centimeter from the L, and the same error keeps it in the M only about six times in ten.
Find my size's confidence starts from that chance: high from three chances in four, medium from one in two, then lowered when other measurements point to another size. It shows on every answer, so a shopper near a boundary sees it, along with the neighbor size when it is a real alternative.
Body charts also carry a tolerance of their own, built in: each size is a range (80 to 84 cm), the spread of bodies it is made for. Where ranges overlap, as the cycling template's heights do (166 to 167 cm is in both XS and S), the answer says the shopper is between two sizes.
What to publish
- Measure production pieces, not only the spec. Measure two or three pieces of each size if you can, and publish what they measure. If production runs a little small, a chart copied from the spec promises more room than arrives.
- Round to what you can measure. Half centimeters, or quarter inches. A chart that claims a 52.37 cm half chest promises a precision no tape gives.
- Say it in a line. "Measured flat. Each piece can vary slightly." If you work to a set tolerance, give your own figure.
- Label garment measurements as such, and laid flat values as laid flat (garment measurements explains the difference). Half a chest read as a whole one is a bigger error than any tolerance.
- Watch for drift. A new batch that comes out small shows up in returns before anywhere else. Reading your returns by size shows how to spot it.
How to measure a garment flat covers the points and the method, so every piece is measured the same way.
How Ruban Size handles it
Values as you measure them. The size chart editor reads decimals with commas (52,5), fractions (20 3/4) and ranges (52-54) as numbers, so they convert to inches. A header like "Half chest, laid flat" tells Find my size to double the value and add the room each kind of garment needs.
Answers that admit the margin. Every Find my size answer says where the shopper falls ("Chest 95 cm: M, top of the range"), how sure it is, and the neighbor size when it's close.
Drift, caught in returns. On the Fit plan, returns by size reads the return reasons of your last 60 days in your browser and flags a size from 3 size returns when at least 70% point the same way. Fix the chart's fit scale in one click, or measure that batch again. The analytics page shows what it reads.
Every remeasure kept. Each save keeps the version it replaced, so a chart updated after a new batch can be compared with the old one, or restored.
Questions
Why doesn't my garment match the size chart exactly?
Fabric, cutting, sewing, washing and the way it's measured each move the numbers a little. A difference of about a centimeter can be within the tolerance the maker allows, especially on a measurement taken laid flat.
What is tolerance in garment measurements?
The amount a finished garment may differ from its spec, plus or minus, at each point of measure, and still be accepted. It is set per point: measurements that step little between sizes, like collars and lengths, need a tighter one.
How much tolerance should a size chart allow?
Less than half the step between two sizes for each measurement. If your sizes step 2 cm in length, a tolerance of 1 cm either way already lets two sizes measure the same.
Should I publish the spec or the measured garment?
The measured garment, when you can. Shoppers compare your chart with pieces they own, so numbers taken from production pieces match what arrives.
Keep reading
- Sizing explainedHow size grading works: from one sample to a size rangeGrading turns one approved sample into a full range of sizes. Every point of the pattern moves by a set amount per size, the grade rule. Girths grow by several centimeters a size, lengths by one or two, and the steps often widen toward the larger sizes.6 min
- How to measureHow to measure a garment flat for your size chartLay the garment flat and measure straight across from armpit to armpit, from the top of the shoulder to the hem, and along the sleeve. Then write "laid flat" in the header, so nobody takes half a chest for a whole one.6 min
- ReturnsReading your returns by sizeCount each size's returns as too small against too big. When one size keeps coming back the same way, the chart or the cut is off for that size, so fix the numbers or the fit scale. Ruban Size does the counting from Shopify's return reasons.6 min
- Sizing explainedWhat "true to size" means, and how to show itTrue to size means an item fits the way its size chart says, so shoppers take the size the chart gives them. A store shows it best on a fixed scale from runs small to runs large, with one line of fit note.6 min