Edge Curl, Wavy Selvedge, Bow and Skew: The Three Dimensional Defects That Decide Whether a Roller Blind Fabric Roll Is Accepted

A roll of 2.8-metre blackout coated fabric leaves Ningbo and arrives at Jebel Ali 26 days later. The buyer cuts the wrapping at the warehouse door, lays the fabric on the inspection table, and ten minutes later sends a photograph: the selvedge has rolled inward like a cigar, and there is a clear arch across the middle. The message is one line: “We cannot put this on the cutting table.” Nothing in that complaint concerns weight, shade variation or flame retardancy, and nothing concerns weaving precision either. It is distortion. And on wide-width roller blind fabric, distortion is the defect class that gets amplified most and written into contracts least.

This article separates three things that are constantly discussed as if they were one: edge curl, wavy selvedge, and bow with skew. Each has its own cause chain, its own measurement method, its own acceptance value and its own remedy. Separate them, and on the day of arrival you can tell whether you are looking at fabric memory that will recover in 48 hours, or permanent deformation that justifies a claim.

1. Why these three always erupt on the day of arrival

Three structural reasons matter more than any tolerance figure.

1. Wide width amplifies tension differentials. The wider the loom, the more a small warp tension difference between the two sides accumulates across the width. A deviation comfortably inside tolerance at 1.5 metres can cross the line at 2.8 or 3.0 metres. And roller blind fabric gravitates towards wide width precisely because the closer the finished width is to the greige width, the higher the cutting yield.
2. Coating and lamination break the fabric's self-balance. A plain woven grid is dimensionally stable on its own, but almost every roller blind fabric is coated on one or both sides — blackout, pearlescent, acrylic, foam. A single-side coating shrinks as it cures. Front and back shrink by different amounts, so the fabric must curl towards the side that shrinks more. That is physics, not operator error.
3. Winding hides the distortion, and unwinding releases it. Tension is unavoidable during winding, otherwise the roll is loose and collapses in transit. But that same tension flattens a 1% bow and straightens a slight skew. The buyer unwinds, lays it flat, and the stress releases. The defect returns. Which is why fabric that passed on the inspection machine can fail on the customer's cutting table — the two parties are looking at the same cloth in two different states.

2. Four forms, four mechanisms: do not mix them

The industry often lumps edge curl and bow into one conversation. Physically they are unrelated, and the corrective direction is opposite.

1. Edge curl. The selvedge rolls into a tube along the warp direction — one side or both, and possibly one side curling face-in while the other curls face-out. The mechanism is differential shrinkage through the thickness: fibres and coating across the edge cross-section contract unequally, forcing the edge towards the side that contracts more.
2. Wavy selvedge. The edge undulates in plane. The fabric itself does not roll, but the edge traces an S. The mechanism is a mismatch in warp elongation between the selvedge structure and the body, or unequal overfeed on the two sides at the stenter, so one edge is fed more length than the other.
3. Bow. The weft itself is curved, arching through the middle like a shallow U. The panel looks as though a piece has been pulled out of the centre.
4. Skew (also called bias). The weft is straight but not perpendicular to the selvedge; it runs at an angle. Stripes, checks and printed lines all lean together.

Why does the distinction matter? Because bow and skew are questions of angle and straightness, correctable at the stenter through weft straightening and overfeed adjustment. Edge curl and wavy selvedge are questions of edge structure, and must be solved back at the selvedge weave, the coating and the winding. Using a weft straightener to cure edge curl is the wrong medicine.

3. Six causes of edge curl, ranked by contribution

1. Single-side coating shrinkage (most common). The heavier the coating and the higher the curing temperature, the greater the front-to-back differential. Blackout fabric, especially three-layer constructions, is most sensitive here. The diagnostic is simple: lay a sample flat for 24 hours. If the curl consistently faces the more heavily coated side, this is your cause.
2. Mismatch between selvedge and body shrinkage. Selvedges normally use denser reed spacing and extra warp ends to prevent fraying. But if the edge warp density is too high, the edge shrinks more than the body during dyeing and finishing, and it pulls inward.
3. Low setting temperature or excessive overfeed. Insufficient setting temperature leaves internal stress unreleased: width becomes unstable, shrinkage rises, and curl follows. Excessive overfeed, conversely, leaves the edge slack and produces a wavy selvedge rather than curl. These two adjustments are usually a paired trade-off.
4. Winding tension and roll diameter. Wound too tight, the inner layers carry compressive stress for weeks. Wound on a thin core, the radius of curvature is small and the inner layers memorise the deformation. Horizontal stacking adds a flat spot at the bottom, and that one is irreversible.
5. Moisture regain. Polyester has low regain, but cotton, viscose and linen blends swell at the edge and move out of step with the body. Container condensation driven by day-night temperature swings aggravates this, especially on tropical routings.
6. Cut edge and thermal effects. Ultrasonic or hot-knife cutting creates a fused, hardened edge. A new stress interface forms between the hard edge and the soft body, and curl frequently starts along that line.

4. The cause chain behind bow and skew

Both almost always trace to one thing — uneven tension — but they arise at different stages and present differently:

1. Weaving. Unequal warp tension between the two sides, or a take-up mechanism that is out of sync, means the weft is inserted into a warp sheet that is not uniformly stretched, and it lands curved. Wider looms expose this more.
2. Dyeing, finishing and coating. If the two stenter chains are not in the same plane — a height difference of roughly 1 mm already shows — or if the left-right overfeed difference exceeds about 2%, the fabric is twisted as it passes and bow or skew appears. Cross-width temperature differential in the oven is a quieter contributor: where the difference exceeds about 3°C, the side that shrinks first drags the cloth off line, which is why common practice holds cross-width variation within 2°C.
3. Printing. Printing is the most sensitive stage, because the pattern line is the only visual reference. A 2% bow on a solid is invisible; the same 2% on a stripe or a check is obvious.
4. Relaxation and storage. Fabric keeps moving after it leaves the machine. Wool, viscose and high-shrinkage yarns shift again after wet processing and steaming. So cloth that measured within tolerance at the mill may have moved by the time it reaches port — which is exactly why acceptance must be based on measurement after relaxation.

5. How to measure: turning an argument into a repeatable number

Distortion disputes are hard because the two parties are not measuring the same thing. Fix the method and half the dispute disappears.

1. Standard. Bow and skew: ASTM D3882, Standard Test Method for Bow and Skew in Woven and Knitted Fabrics. The Chinese equivalent is GB/T 14801. Width measurement references ISO 22198. All three follow the same logic: lay flat, apply no tension, draw a reference line perpendicular to the selvedge, measure the maximum deviation.
2. Environment. Measure under standard atmosphere, 21 ± 1°C and 65 ± 2% relative humidity. If standard conditions cannot be met, record the actual temperature and humidity, otherwise two data sets are not comparable.
3. Sampling. Take three test areas per sampling unit, spaced as far apart as possible and covering at least 1 metre of length. For optical measurement, the specimen should be at least 400 mm wide. On wide-width fabric, sample at the head of the roll — that is where winding tension is highest and distortion is worst.
4. Relax first. This is the step most often skipped and the one that most affects the conclusion. After unwinding, lay the fabric flat or hang it without tension for at least 24 hours, ideally 48, so internal stress equalises and moisture content matches the room. Some mills already write “rest 24 to 48 hours after finishing before packing” into their standard procedure. The cost is warehouse space; the return is a clean arrival.
5. Calculate. Percent bow = maximum bow height ÷ measurement length × 100. Percent skew = weft offset ÷ measurement width × 100. Round to 0.1% or 1 mm. If a double bow (S-shape) is present, take the larger of the two.
6. Repeatability. ASTM D3882 reports interlaboratory repeatability standard deviations of about 0.289% for bow and 1.079% for skew. That number is genuinely useful: when the two parties' skew figures differ by around one percentage point, it is probably measurement variation and not enough to support a claim. Above two points, a technical discussion is warranted.

6. Setting an acceptance value: only meaningful in the contract

Standards define the method, never the pass mark. The pass mark is a commercial agreement and must appear in the purchase order — along with the measurement length, whether the fabric is relaxed, and where the sample was taken. Without those three, the two sides will measure to different rules forever.

1. Bow and skew. For solids and random prints, 3% is the common ceiling. For stripes, checks and jacquards requiring pattern matching, tighten to 1.5%, and to 1% on demanding work. On wide width (2.8 m and above), tighten a further half point.
2. Edge curl. Avoid a yes/no test. Specify instead: “after 24 hours flat under standard atmosphere, edge curl height shall not exceed X mm per metre and shall not form a tube.” The value of X depends on width, weight and coating structure — a 1.5 m lightweight solid and a 3 m three-layer blackout cannot share a number.
3. Wavy selvedge. Specify wave height and wavelength, or simply state that lateral deviation of the edge within 1 metre shall not exceed Y mm.
4. A relaxation clause. State that acceptance is judged on measurement after 24 hours of relaxed unwinding. This protects the supplier against instant rejection at the warehouse door, and protects the buyer against “just let it hang, it will settle” as an indefinite delay.
5. Sampling and decision rule. Agree the proportion of rolls inspected, the number of test areas per roll, and whether one out-of-tolerance roll condemns that roll or whether the batch is judged on average. Those two rules have very different commercial consequences.

7. How acceptance differs across five markets

The same cloth meets five different questions. Understanding the difference lets you build the right margin into the quotation.

1. Europe. Project and hospitality work often ties fabric performance to a finished-product standard. The fabric side looks at bow, skew and shade; the finished side looks at EN 13120 for the completed blind. Tender documents frequently cite the standard number without giving a value, so confirm the number during the Q&A window rather than after award.
2. North America. Closer to contract plus trade-association practice. Buyers arrive with their own inspection checklist and place a premium on “cuttable on arrival.” Tolerance for edge curl is lowest here, because North American roller shades are usually cut as a single full panel with few seams — one curled roll equals one scrapped roll.
3. The Gulf (Saudi Arabia, UAE, Qatar). Heat is an extra variable. Storage and transport around 50°C soften the coating and redistribute internal stress, so fabric that passed at the mill can develop fresh curl on arrival. Write a separate “post-high-temperature storage re-inspection” clause for these markets, and raise the thermal stability of the formulation.
4. India and South Asia. Monsoon humidity is the dominant variable. Moisture-driven edge swelling and wavy selvedge concentrate here, so state explicitly that judgement follows conditioning under standard atmosphere — otherwise monsoon-season arrivals produce disputes almost by default.
5. Africa (Nigeria, Kenya, South Africa). Long customs and inland transit cycles mean rolls experience repeated temperature swings between container and warehouse, reinforcing the distortion memory. Some markets also lack a standard inspection table, so the inspection conditions themselves are unstable. Put “inspection environment and equipment” into a contract annex, or supply a video work instruction for arrival re-inspection.

8. Six engineering remedies on the supplier side

Once the cause is located, work in the order structure → process → packing. Do not start by turning machine dials.

1. Reinforce the selvedge structure. Add edge warp ends or use a coarser edge yarn so edge and body shrinkage converge. This treats the root cause, at a small cost in cutting yield that belongs in the quotation.
2. Segment overfeed and tension. Reduce total overfeed by 2 to 3 percentage points, hold the left-right overfeed difference within 0.5%, and calibrate the tension roller so the pressure difference between the two ends is minimal — a large differential makes the cloth rotate as it enters the chain.
3. Geometric alignment of chains and oven. Scan along the chains with a laser level; where left-right height differs by more than 1 mm, adjust shims or replace worn links. Measure nine points across the oven and hold cross-width variation within 2°C.
4. Double-side or balanced coating. Single-side coating inevitably creates a shrinkage differential. Switch to thin double-side coating, or add a balancing layer on the reverse, to even out front-to-back contraction. Cost rises, but most curl complaints disappear — particularly worthwhile on wide width.
5. Roll engineering. Use a larger core, reduce winding tension, store vertically or on A-frames rather than stacking horizontally, and rest 24 to 48 hours before packing.
6. Cutting method. On wide width, prefer cold knife or laser over hot knife to avoid the fused hard edge and the stress interface it creates. Seal cut edges where appropriate.

9. Seven steps on arrival

Written for the buyer's floor team, and a sensible self-check for the supplier:

1. Before opening, record container internal temperature and any condensation on the walls, as background evidence for later discussion.
2. Inspect the contracted proportion of rolls; first check for crushed, collapsed or damp packaging.
3. After unwinding, do not judge immediately. Lay flat on the inspection table without tension and note the time.
4. Re-measure after 24 hours. Record both the “as-unwound” and “after relaxation” figures — that delta is the single most persuasive piece of communication you will have.
5. Measure bow and skew at three test areas per ASTM D3882 or GB/T 14801, recording direction, viewing face (front or back), width, temperature and humidity.
6. Measure maximum curl height with a rule, and note whether it is single-sided or double-sided, face-in or face-out.
7. Put photographs, data and the contractual acceptance value on one page as the basis for a claim or a release. With that page, negotiation time typically drops from three weeks to three days.

10. Three product types to watch in 2027

From a distortion-risk perspective, three areas of next year's range need an early conversation:

1. Wide-width three-layer blackout. Demand is rising, and it is the highest-risk construction for edge curl. Price the balanced double-side coating and the vertical storage into the quotation rather than absorbing them after a claim.
2. Cracked-ice (li bing) effects and high-shrinkage textured fabric. High-shrinkage yarn creates texture through shrinkage, and shrinkage is itself a source of distortion. Cracked-ice effects look forgiving visually, but once bow combines with the texture direction, the defect is magnified. These need their own acceptance values, not the solid-fabric numbers.
3. Jacquard and pattern-matched work. Skew tolerance must tighten to 1.5% or even 1%. Note also that jacquard floats are more easily creased during winding, and a crease combined with edge curl is very hard to resolve — interleaving paper at packing is advisable.

11. One sentence

Distortion defects are difficult not because they are hard to measure, but because they are hard to define. The method has a standard; the acceptance value often never reaches the contract; and both sides negotiate with the phrase “it looks crooked.” Write down six things — method, environment, relaxation time, sampling position, acceptance value and decision rule — and the cloth stops being a dispute and becomes a product again.

Shaoxing Wanbo supplies roller blind fabric, sunscreen fabric, blackout coated fabric, jacquard and high-shrinkage woven fabric, upholstery and carpet to buyers in the Gulf, India, Europe, North America and Africa, and treats distortion control as equal in importance to weight and shade. Wide-width products use balanced coating and larger-core winding; finished rolls rest before packing; and each shipment carries measured bow, skew and width data per ASTM D3882 and GB/T 14801, with before-and-after relaxation records available on request. We do not promise zero distortion — that would contradict physics. We do commit to stating each roll's actual condition in numbers before the vessel sails.

Note: test methods and repeatability data cited here follow the published content of ASTM D3882, GB/T 14801 and ISO 22198. Stenter, overfeed and oven temperature figures reflect common finishing practice and must be adjusted for specific equipment, fibre and construction. Freight rates and port conditions change daily; nothing here constitutes a same-day quotation.