Width tolerance is the specification that most often appears in a ribbon purchase order and most often does nothing. A number written as plus or minus 0.5 mm looks like a commitment, but it is not a commitment at all until the reader knows which edge is the reference, where along the roll the measurement is taken, how the ribbon is tensioned during measurement, and whether the number describes the ribbon as woven or the ribbon as slit and finished. The same ribbon, measured two ways by two competent inspectors, can differ by more than the entire tolerance band. Once that ambiguity is understood, the rest of the subject becomes surprisingly tractable: width drift is not random, it has identifiable causes, and each cause has a specific place in the process where it can be stopped.
A woven ribbon is not a uniform solid. It has two physically different edges, and they behave differently for the whole of its life.
The selvedge is the edge created by the loom itself, where the warp yarns turn around and return. It is compact, it is structurally integral to the cloth, and it is typically the most stable edge the ribbon has. The cut edge is created after weaving, by a slitter knife or a rotary knife passing through the fabric. It has no selvedge structure at all. It is a raw cut through the yarn system, and it will behave like a cut edge for the entire life of the product.
Here is the practical consequence that catches almost every brand buyer at least once: a ribbon 20 mm wide with a 19 mm selvedge on one side and a 1 mm trimmed margin on the other is not the same product as a ribbon 20 mm wide with selvedge on both sides, even though both measure 20 mm. The trimmed side will fray differently, will show a different print registration behaviour at the edge, will have a different tensile strength profile, and will look visibly different when the ribbon is folded or tied. If your tolerance says only "width 20 mm, tolerance plus or minus 0.5 mm", the inspector and the mill can each satisfy it and still deliver materially different goods.
Every usable width specification therefore has to name, at minimum:
Width variance in a finished ribbon is not one phenomenon. It is the sum of several distinct causes, and each has a different owner and a different fix.
Every loom has a nominal working width and a real working width that moves. Weaving tension on the warp, the take-up rate on the weft, the shrinkage of the warp in the reed, and the temperature and humidity of the weaving hall all shift the fabric width during the run. On a wide ribbon, this is absorbed by the selvedge margin and is invisible. On a narrow ribbon, the same instability is a meaningful percentage of the total width.
This is why narrow-width ribbon is disproportionately difficult and why the tolerance for a 6 mm ribbon should not simply be the tolerance for a 50 mm ribbon expressed as a percentage. A mill that produces a 50 mm ribbon to plus or minus 0.5 mm is working with a roughly 1 percent band on a width where a fraction of a millimetre of loom movement is easily absorbed. The same mill asked to hold 6 mm to plus or minus 0.5 mm is being asked for an 8 percent band on a width where the loom cannot deliver that precision without dedicated narrow-width weaving and a specific width-control strategy.
As a general procurement rule: ask the mill to state the tolerance band they can actually hold on that specific width, on that specific construction, and treat any request for a narrower band as a request for a different manufacturing route rather than a request for more care. Narrow-width ribbon is often better made on a wide loom and slit down, precisely because that is where width control is easiest.
When ribbon is woven wide and slit to width, the slitting operation introduces its own tolerance. Two mechanisms are common. A straight knife in a slitter produces a cut whose position depends on the tension running through the fabric at the moment of the cut; fabric that is not uniformly tensioned across its width will be cut with the knife at a slightly different distance from the reference edge at different points in the roll. A rotary knife produces a cleaner edge with less drag and less fraying, but still has a positional tolerance.
There is also a bow and skew component. If the fabric entering the slitter is skewed - the warp yarns running at an angle rather than square to the edge - then a slitter that cuts parallel to the machine direction produces an edge that drifts in and out across the width of the ribbon over the length of the roll. This is not a width error at all. The ribbon is exactly the right width everywhere; it is just not square. The result is a ribbon whose usable width varies, and whose two edges are not parallel.
This is why edge parallelism has to be measured and specified separately from width. A ribbon that is 20.0 mm wide and whose edges are not parallel is a worse problem than a ribbon that is 20.4 mm wide and whose edges are perfectly parallel. The first one cannot be cut reliably by the customer's converting process; the second one just needs a slightly wider allowance.
Wet processing relaxes and re-tensions the yarn system. Heat-setting stabilises it, but the width the ribbon has after finishing is not automatically the width it had off the loom. A ribbon that is 20.5 mm off the loom and finishes at 20.0 mm is perfectly healthy; the problem arises when the mill slits to 20.0 mm before finishing and then relaxes the ribbon during finishing, so the finished product is 19.4 mm. This is a sequencing error, and it is common, because slitting before finishing is faster and cheaper than slitting after.
The rule that follows from this is simple: the width in your purchase order is the width of the finished product. Any mill-side documentation that quotes a loom or greige width against your finished-goods tolerance is comparing two different things.
A ribbon wound onto a core under insufficient or excessive tension will measure differently at the roll surface and at the roll core. It will also develop a memory: a roll that has been held under tension for months will not lie flat when opened, and the ribbon will appear narrower or wider depending on how the inspector happens to be holding it. This is the cheapest of the four causes to eliminate and the one most often left in place, because it is invisible to everyone who is not measuring.
Given that the cut edge and the selvedge behave differently, the practical question for a brand buyer is which edge type the application actually requires.
| Edge type | Structure | Fray behaviour | Typical use |
|---|---|---|---|
| Selvedge (woven) | Compact, integral to the cloth, yarn turns around | Does not fray in normal use; may ravel slightly at a raw cut through it | Premium trims, visible edges on bows, narrow bands |
| Cut / slit edge | No edge structure; yarns severed | Will fray along the cut line, more so with low-density constructions and in high humidity | Interior trims, wrapped edges, anything later over-sewn or bound |
| Folded / self-hemmed | Edge turned and stitched or glued | None at the fold; adds width and thickness | Retail display trims, garment edging, visible cut edges |
If the application is a retail product where the ribbon edge is visible and handled by the end consumer - a hair bow, a gift bow, a favour ribbon on a package - a selvedge edge is worth specifying even though it costs more, because the failure mode of a fraying edge is a consumer complaint rather than a factory rework. If the edge is going to be bound, over-sewn or hidden inside a construction, a cut edge is entirely adequate and considerably more economical.
Most buyers think about width as a quality metric. Its real significance is that it propagates directly into the geometry of the finished bow, and bow geometry is what the end consumer sees.
A bow is made by folding ribbon at a defined point and forming the two legs and the tails. The relationship between ribbon width and bow proportion is close to linear: a wider ribbon at the same fold and same loop depth produces a wider, flatter bow with shorter-looking legs. Change the width by one millimetre on a 38 mm ribbon and you have changed the silhouette of the product. On a 20 mm ribbon, the same one millimetre is a proportionally larger shift, and on a 10 mm ribbon it dominates entirely.
Across a production run, width drift therefore shows up downstream as a spread in the finished bow: some bows in the same carton have visibly wider loops or longer tails than others. It is a subtle, cumulative inconsistency that is very hard to photograph and very easy for a customer to feel. Buyers who have chased "the bows look inconsistent" for months have frequently been chasing a width tolerance all along.
There is a second, more mechanical route by which width reaches bow geometry. A ribbon that is not edge parallel will not fold symmetrically. On a pre-tied bow, the fold sits on the diagonal, and a skewed ribbon will present an unequal fold angle on the two loops, so one loop stands up and the other lies flatter. This is visible in finished goods at a glance, and it is not a workmanship problem - it is a material problem that was detectable at the incoming inspection stage.
Pulling this together into something you can hand to a mill and an inspector and have both read the same thing:
| # | Element | How to specify it |
|---|---|---|
| 1 | Nominal width and tolerance | State nominal, state the band as plus/minus or min/max, and state the band per width band and per construction. Ask the mill to confirm the band it can hold on that width. |
| 2 | Edge type, both sides | Selvedge / trimmed / self-hemmed, stated per side, not just "raw edge" or "finished edge". |
| 3 | Measurement method | Instrument, gauge location, applied tension, whether relaxed or tensioned, and whether the reading is taken at the roll surface or the core. |
| 4 | Sampling rule | Number of points across the width and along the length, number of rolls per lot, and whether the acceptance is per-roll or per-lot. |
| 5 | Edge parallelism | Maximum permitted bow or skew as a millimetre figure per metre of length, measured against the selvedge or a declared reference edge. |
| 6 | Measurement state | Conditioning temperature and humidity and duration, and whether the value refers to loom, greige, post-dye, post-finish or slit state. |
Add two triggers: a re-test trigger clause naming the changes that invalidate an approved width - a change of loom, a change of yarn supplier or denier, a change of dyehouse, a change of finishing route, a change of width, a change of slitting route - and a bow consequence clause stating what the finished bow geometry must be when made from the approved ribbon width. The first clause prevents silent drift across a reorder programme; the second closes the loop between a mill-side number and the product the consumer actually receives.
A width inspection that only measures width will pass ribbons that will fail downstream. The minimum practical incoming check on a narrow-width programme is four measurements:
Cutting a bow from a ribbon whose edges are not parallel produces a visibly asymmetric product that no amount of care in the bow-making step will fix. That is worth remembering at the inspection table: some defects are created long before the product that shows them exists.
Most mills welcome a width and edge specification of this detail, because it removes ambiguity from the quotation. The mills that object are usually the ones whose process is not actually controlled tightly enough to meet it, which is useful information in itself.
Ask for the mill's own capability: not "can you hold plus or minus 0.5 mm" but "what tolerance do you routinely hold on this width and this construction, and how do you measure it". A mill with a real slitting and measurement discipline will answer with numbers and will tell you the measurement conditions. A mill that answers with "we always make it right" is telling you something else entirely.
Smith Ribbon has manufactured ribbon and bows since 2004, running in-house weaving, dyeing, finishing, slitting and bow assembly in Xiamen. Our engineering team will quote the width band we can actually hold for your specific width, construction and edge type, together with the measurement conditions we use, and we will tell you where we think a narrower band would require a different manufacturing route rather than more care. If you have a current width specification that is causing disputes, send it to us and we will tell you honestly whether the band is realistic for your construction.
It depends on the width and the construction, and the honest answer is that it is narrower than most buyers assume at narrow widths and wider than they assume at wide widths. For wide ribbon made on a wide loom, a band of plus or minus 0.5 mm is comfortably achievable. For narrow ribbon below about 10 mm, the same absolute band represents a much larger proportion of the width and is usually not achievable by weaving to width - the practical route is to weave wide and slit, or to accept a wider band. The correct approach is to ask the mill to state the band it routinely holds on your specific width and construction, with its measurement conditions, and to write that number into the purchase order instead of assuming a number.
A selvedge edge is created by the loom as the warp yarns turn around, so it is compact and integral to the structure. A cut edge is created by a slitting or rotary knife passing through the fabric, which severs the yarns and leaves no edge structure at all. The practical consequences are that a cut edge can fray along the cut line, particularly on low-density constructions, while a selvedge edge does not fray in normal use, and that the two edges have different print registration behaviour, different tensile profiles and different visual appearance when the ribbon is folded or tied.
Three things cause it. First, loom working width moves during a run as weaving tension, take-up rate and hall conditions shift, and the effect is proportionally larger on narrow ribbon. Second, slitting introduces its own positional tolerance and, if the incoming fabric is skewed, produces an edge that is cut square to the machine but not parallel across the ribbon. Third, yarn tension across the width is rarely perfectly even, so a ribbon will pull slightly narrower under tension in one region than another. This is why edge parallelism has to be specified and measured separately from width, rather than treated as a detail of the width measurement.
Yes, and this is where width stops being a quality number and becomes a product-design variable. The relationship between ribbon width and bow proportion is close to linear, because a wider ribbon at the same fold and loop depth produces a wider, flatter bow with legs that appear shorter. Change the width by one millimetre on a 38 mm ribbon and the silhouette changes visibly; on a 10 mm ribbon the same millimetre dominates the shape entirely. Separately, a ribbon whose edges are not parallel does not fold symmetrically, so one loop of a pre-tied bow stands up while the other lies flatter - a defect that is created at the slitting stage but only becomes visible at the bow-making stage.
The finished width, always. Wet processing relaxes and re-tensions the yarn system, and heat-setting then stabilises it, so the finished width is not automatically the greige width. The common error is for a mill to slit to target before finishing and then let the ribbon relax during finishing, delivering a product that is genuinely narrower than the purchase order. Any mill-side document quoting a loom or greige width against your finished-goods tolerance is comparing two different measurements, and that discrepancy should be corrected before the first order is placed.
Three causes account for most of it. Different rolls may come from different production runs with different loom conditions, so the mill should tell you the lot or run structure of your order. Different positions in the roll behave differently, because roll build tension varies from surface to core, so the same ribbon measures differently depending on where the sample is taken. And different storage history changes the reading, because ribbon that has been held under tension in storage develops a set. Measuring at a defined position in the roll, under a defined conditioning regime, with a defined tension, removes all three as sources of dispute.
Agree the measurement protocol before the order is placed, not after the first shipment arrives, and write it into the purchase order as six elements: nominal width with a stated band, edge type per side, measurement method including instrument and tension, sampling rule for points and rolls, edge parallelism limit, and the measurement state including conditioning. Then add a re-test trigger clause naming the changes that invalidate an approved width - loom, yarn supplier, dyehouse, finishing route, width, slitting route - so that re-qualification is prompted by the specification rather than by memory.
Yes, as a proportion, and this is where specifications most often go wrong. The absolute tolerance that a wide loom can hold on a 50 mm ribbon may be impossible to hold on a 6 mm ribbon, because loom movement that is absorbed by a wide selvedge margin becomes a large percentage of a narrow width. Rather than applying one absolute band across a range of widths, specify a band per width band and per construction, and ask the mill to confirm what it can actually hold. If the band you want on a narrow width is tighter than the mill can hold by weaving, the answer is often to change the manufacturing route - weaving wide and slitting - rather than to ask for more care.
Send us your current width specification, your construction and your edge requirement. We will quote the tolerance band we can reliably hold for that specific width and tell you plainly whether a tighter band is realistic or requires a different manufacturing route.
Talk to a compliance engineer: +86 13779951780 (WeChat) | xmmsd@126.com | www.smithribbon.com
© 2026 Xiamen Smith Ribbon & Bow Co., Ltd. | Professional Ribbon Manufacturer Since 2004
Smith Ribbon OEM since 2004 — woven ribbon, satin, grosgrain, velvet, organza, RPET, printed bows, decorative trims, and private-label accessories for global brand buyers.