Tolerance in Bag Making: What Is a Realistic ±2cm?
On a soft bag, plus or minus 20 mm is realistic on an overall dimension above 400 mm and tighter below it. On a cut panel it is 2 mm, on a weld 1.5 mm and on a print position 3 mm. A single tolerance for the whole bag is the specification error that causes most of the rest.
A realistic tolerance on a soft bag is about plus or minus 10 mm on a finished dimension up to 400 mm and plus or minus 20 mm above it. Cut panels hold 2 mm, sewn panels 3 mm, a welded seam width 1.5 mm, print placement 3 mm and embroidery placement 5 mm. Those figures are programme conventions rather than values printed in a standard, and they must be confirmed against the cited test method where they will be defended.
Tolerance is the field buyers leave out and the one that produces the most arguments, because an unstated tolerance is read as zero. A dimension written as 300 mm with no band means any departure is a defect, and no soft bag made from cloth holds to that across a run of 500 pieces.
The correction is not to widen everything. It is to set the band where it belongs per feature, which is what the table below does: tight on anything that is cut or placed, looser on anything that is assembled, and loosest on the overall dimension of a soft body.
Why Is a Soft Bag Looser Than a Cut Panel?
Because every assembly step adds variance. A panel cut to a pattern holds the pattern's accuracy, which is machine accuracy. Once that panel is sewn, turned, closed and topstitched, it has been handled four times, and each handling adds a little. The seam allowance, the feed of the machine and the stretch of the cloth under the foot all move it.
The cloth itself is the other source. A woven polyester moves less than a knit, a coated shell moves less than an uncoated one, and a laminated foam panel moves less than either. A tolerance that is realistic on 600D polyester is optimistic on a stretch shell.
What Tolerance Band Belongs on Which Feature?
Set by what the feature controls. A dimension that has to mate with something — a divider slot, a zip length, a hardware aperture — gets a tight band because it has to fit. A dimension that only has to look right — the overall height of a soft body — gets a loose one.
That principle is more useful than any table, because it generalises. Where a buyer is unsure what band to set, the question to ask is what this dimension has to fit against. If the answer is nothing, the band can be wide; if the answer is a specific part, the band is set by that part.
| Feature | Realistic tolerance | Why that figure | What breaks if it is tighter |
|---|---|---|---|
| Cut panel, any axis | plus or minus 2 mm | Machine accuracy on a pattern cut | Panels rejected that would have assembled correctly |
| Sewn panel after assembly | plus or minus 3 mm | Handling, feed and seam allowance variance | A high reject rate with no functional gain |
| Finished soft bag, up to 400 mm | plus or minus 10 mm | Accumulated assembly variance on a small body | Rejects on a dimension nobody can see |
| Finished soft bag, above 400 mm | plus or minus 20 mm | Variance scales with the dimension | Rejects on a dimension nobody can see |
| Welded seam width | plus or minus 1.5 mm | The weld fixture holds it, and strength depends on it | Nothing, but it does not improve strength |
| Foam thickness, 5 mm | plus or minus 1 mm | Sheet tolerance and compression in assembly | Sheet yield loss with no performance gain |
| Foam thickness, 10 mm and above | plus or minus 2 mm | Compression variance scales with thickness | Sheet yield loss |
| Strap length, laid flat | plus or minus 10 mm | Webbing stretch and cut-length variance | A visible difference in drop between units |
| Print placement from a seam | plus or minus 3 mm | Placement is fixtured and visible | Rejects on a logo nobody can see 3 mm off |
| Embroidery placement from a seam | plus or minus 5 mm | Digitising and hooping variance | Rejects on a logo that reads correctly |
| Hardware position | plus or minus 3 mm | It has to sit where the strap expects it | A strap that pulls the panel out of line |
| Carton dimension | plus or minus 10 mm | Board tolerance and packing variance | Nothing, unless the pallet pattern depends on it |
Is Tolerance Additive Across a Stack of Features?
Not in the way buyers assume. Two features each holding plus or minus 3 mm do not produce a combined error of plus or minus 6 mm in the ordinary case, because the variances are independent and partly cancel. The statistical sum is smaller than the arithmetic sum.
What is additive is bias. If the pattern is cut 2 mm long and the seam allowance is consistently taken 2 mm wide, the two biases add and the finished bag is 4 mm out in the same direction every time. Random variance partly cancels; systematic bias does not, and it is the thing worth looking for when a whole run measures out.
Four rules for setting tolerance on a bag pack
- Set the band by what the dimension has to fit against, not by what looks reasonable.
- Tighten on anything fixtured — cut panels, welds, print placement — and loosen on assembled soft dimensions.
- State the band on the same row as the dimension, so it is not read from another section.
- Treat a run that measures consistently out as a bias to be found, not as variance to be tolerated.
How Does Material Choice Move the Band?
By more than most packs assume. A stable woven polyester with a coating holds the figures above comfortably. A stretch shell, a knit-backed laminate or an uncoated light fabric does not, and on those materials the realistic band on a finished dimension can be half again what the table gives.
Neoprene is the case buyers notice, because it compresses. A panel measured under a rule compresses under the rule, and two people measuring the same panel get two numbers. On compressible materials the pack should state the measuring force or, more practically, state that the dimension is taken with the material relaxed and unsupported.
How Is Tolerance Checked at Inspection?
On the sample size drawn from ISO 2859-1 at General Inspection Level II with an AQL of 2.5, measured with a steel rule on a flat table against the method and the band stated on the pack. Where the pack gives no band, the inspector applies a convention of their own, which is how a correct bag ends up failing.
The failure mode worth naming is the dimension that is out of band and immaterial. A bag 12 mm over on an overall height at the 400 mm band is within tolerance; a bag 12 mm over at the 300 mm band is not, even though neither difference is visible or functional. Stating the band per dimension band rather than per bag is what prevents that.
| Measurement | Band on the pack | Result at inspection | Correct outcome |
|---|---|---|---|
| 300 mm overall, reads 310 mm | plus or minus 10 mm | At the limit | Accept |
| 300 mm overall, reads 312 mm | plus or minus 10 mm | Outside the band | A major defect, counted against AQL 2.5 |
| 520 mm overall, reads 538 mm | plus or minus 20 mm | Inside the band | Accept |
| 520 mm overall, reads 545 mm | plus or minus 20 mm | Outside the band | A major defect |
| Print at 48 mm from a seam, spec 45 plus or minus 3 | plus or minus 3 mm | At the limit | Accept |
| Print at 50 mm from a seam, spec 45 plus or minus 3 | plus or minus 3 mm | Outside the band | A major defect, because placement is visible |
| Whole run 8 mm long on every unit | plus or minus 10 mm | Inside the band | Accept, and investigate the bias |
| Whole run 8 mm long on every unit | plus or minus 3 mm | Outside the band | Reject, and find the systematic cause |
What Happens When the Buyer Asks for a Tighter Band Than Is Realistic?
One of three things, none of them good. The run is rejected on a dimension that does not matter, which costs the programme and damages the relationship. Or the band is quietly ignored on the floor, which means the pack has stopped governing production. Or the price rises, because holding a tight band on a soft body costs inspection time and yield.
The third is the honest answer and the one worth asking for. Where a dimension genuinely has to hold to plus or minus 3 mm on a soft body, that is achievable with fixturing, slower handling and more inspection, and all three cost money. It is a legitimate request, and it should be priced rather than assumed.
How Should Tolerance Be Written on the Pack?
On the same row as the dimension, in the same units, with the method. "300 mm flat, empty, plus or minus 10 mm" is complete. A tolerance table in an appendix, referenced by a note, gets misapplied — and the misapplication only appears when a container is being inspected.
Where a general note is unavoidable, it should give the default band per dimension band and then list the exceptions explicitly. That is shorter than writing a band on every row and it survives being read by somebody in a hurry.
How Does Tolerance Apply to Hardware and Components?
Through the component's own tolerance rather than the bag's. A 25 mm webbing is sold to a width tolerance of its own, and a buckle made for 25 mm webbing has an aperture tolerance of its own. Where the two tolerances run in opposite directions, the webbing will not feed — and that is a specification problem rather than a quality one.
The practical fix is to specify the mating pair rather than each part separately: state the webbing width with its tolerance and the buckle aperture as the range of webbing it must accept. That is how the two parts are actually bought, and it removes a class of failure that no inspection catches.
One more source of variation is worth naming because it surprises buyers: the difference between the first unit off the line and the five hundredth. Tooling wears, operators change, and a run measured at the start is not always the run measured at the end. A tolerance band applies across the run, not to the first unit alone.
How Does Tolerance Fit the Rest of the Programme?
With the measurement method, which is the other half of the same field. A band on a taped dimension and the same band on a flat dimension are different specifications, and the pack has to say which it means before the band means anything.
Then it holds through sampling and bulk. Sampling runs 6-10 working days from a signed pack, bulk runs 35-50 days from written sample approval, and inspection applies the band at AQL 2.5 under ISO 2859-1. Where the band is right, the inspection result is a measurement; where it is wrong, the result is an argument.
How Does Conditioning Change a Measured Dimension?
More than buyers expect, because textiles move with moisture. A shell measured straight off a machine in a humid week and the same shell measured conditioned in a standard atmosphere give different numbers, and the difference is not an error — it is the material behaving as it does.
Where a dimension will be defended, state the conditioning. The standard practice for conditioning textiles before testing sets a temperature and relative humidity and a time the material must sit in them, and a measurement taken against that condition is repeatable in a way that a measurement taken on the bench is not.
| Source of variation | Rough size of the effect | Whether it is controllable | What to do about it |
|---|---|---|---|
| Pattern and cut accuracy | Up to 2 mm | Yes, by machine and die | Hold the tight band here |
| Sewing and handling | A few millimetres | Partly, by fixturing | Loosen the band on assembled dimensions |
| Material stretch under the foot | A few millimetres | Partly, by operator and machine setting | Test on the specified material |
| Moisture conditioning | Several millimetres on some fibres | Yes, by conditioning before measurement | State the conditioning where a figure is defended |
| Fill state | Up to 20 mm on a soft panel | Yes, by stating it | State empty or filled on every taped figure |
| Operator and rule | A millimetre or two | Yes, by method | State the measuring tool and the method |
What Is the Honest Answer on Holding 2 cm?
Yes, on an overall dimension above 400 mm on a soft bag, comfortably. Not on a cut panel, where the answer is 2 mm, and not on a print placement, where the answer is 3 mm. The question is always which feature, and a supplier who answers without asking that has not understood the question.
A buyer who states the band per feature up front gets a more accurate price as well, because the inspection effort and the yield are part of the cost. It is one of the few fields on a pack where being more precise makes the bag both better specified and better priced.
Get the tolerance band set per feature
Send the drawing and the material list. You get a band on every dimension that reflects what the material and the construction hold.
Frequently Asked Questions
What is a realistic tolerance on a soft bag?
About plus or minus 10 mm on a finished dimension up to 400 mm and plus or minus 20 mm above it, with 2 mm on a cut panel, 1.5 mm on a weld width and 3 mm on print placement.
Are those figures in a standard?
No. They are programme conventions. Where a tolerance will be defended, confirm it against the cited test method rather than against a table.
Why is a soft bag looser than a cut panel?
Because every assembly step adds variance: handling, machine feed, seam allowance and the stretch of the cloth under the foot.
Is tolerance additive across features?
Random variance partly cancels, so the statistical sum is smaller than the arithmetic sum. Systematic bias does add, and that is what to look for when a whole run measures out.
How does material choice move the band?
A stable coated woven holds the figures comfortably; a stretch shell or a knit-backed laminate does not. Compressible materials need a stated measuring condition.
How is tolerance checked at inspection?
On the sample size from ISO 2859-1 at General Inspection Level II with an AQL of 2.5, measured flat against the method and band on the pack.
What if the pack gives no tolerance?
The inspector applies a convention of their own, which is how a correct bag ends up failing.
Should every dimension get the same band?
No. Set it by what the dimension has to fit against. Anything that mates with a part gets a tight band; an overall soft dimension does not.
What happens if I ask for a tighter band than is realistic?
One of three things: rejects on a dimension that does not matter, the band being quietly ignored on the floor, or a price rise to cover fixturing, handling and inspection.
Can a tight tolerance be bought?
Yes, with fixturing, slower handling and more inspection. It is a legitimate request and it should be priced rather than assumed.
Where should the tolerance be written?
On the same row as the dimension, in the same units, with the method. A table in an appendix gets misapplied.
Does the measurement method change the band?
Yes. A band on a taped dimension and the same band on a flat dimension are different specifications.
What is the tolerance on print placement?
Plus or minus 3 mm from a named seam. Embroidery is plus or minus 5 mm, because digitising and hooping add variance.
What are the programme windows?
Sampling 6-10 working days from a signed pack, bulk 35-50 days from written sample approval, inspection at AQL 2.5 under ISO 2859-1.
People Also Ask
What is a realistic tolerance for bag manufacturing?
Plus or minus 10 mm on a finished dimension up to 400 mm and plus or minus 20 mm above, tighter on cut panels, welds and print placement.
Why does a bag come out a different size from the spec?
Because every assembly step adds variance, and because a tolerance that was never stated gets read as zero.
How tight can bag tolerances be?
Two millimetres on a cut panel, 1.5 mm on a weld width and 3 mm on print placement. Tighter than that on a soft body has to be bought.
Is tolerance the same across a whole bag?
No. It is set per feature by what the dimension has to fit against.
How do you check tolerance during inspection?
Measure the sample size from ISO 2859-1 at Level II against the band and method stated on the pack, with an AQL of 2.5.
Sources and further reading
Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.
- ISO 2859-1:1999 — Sampling procedures for inspection by attributesInternational Organization for Standardization
- ASTM D751 — Standard Test Methods for Coated FabricsASTM International
- ASTM D6193 — Standard Practice for Stitches and SeamsASTM International