Define the construction exactly before you set tolerance
Use one construction assumption throughout the PO and the inspection guide. Here the article refers to a 100% polyester brushed fleece blanket, nominal 320gsm finished fabric basis weight, finished size 130 x 170 cm, face and back brushed, knife-cut perimeter, then decorative blanket stitch on all four sides. It is not a woven plaid throw, not a jacquard woven blanket, and not a printed imitation check unless the supplier states that clearly. Those categories hold repeat geometry differently and should not share one tolerance table.
For mainstream supply, most polyester check fleece blankets are either printed on brushed knit fleece or made from piece-dyed solid fleece cut into blanket form. A true yarn-dyed checked knit fleece is a niche construction and should be treated as a confirmed exception, not a default assumption. If a supplier claims yarn-dyed check in warp-knit or circular-knit fleece, ask for the actual formation route, greige photo, and pre-brush fabric approval because many market offers use printed plaid visuals instead of engineered checked knitting.
For this guide, the construction assumption is: brushed knitted polyester fleece with an engineered or printed check effect, cut and sewn into blanket form. Buyers should state which route is approved because the tolerance logic changes. Printed check usually needs tighter print-repeat and shade review but looser geometry acceptance after brushing. Engineered checked knit can hold cleaner repeat continuity but may still move during heat-setting, brushing, relaxation, cutting, and sewing.
A practical PO note could read: nominal check repeat 80 x 80 mm; dominant charcoal bar 20 mm nominal; secondary cream bar 8 mm nominal; decorative blanket stitch thread polyester Tex 40-60; stitch pitch 10-14 mm; stitch inset from finished cut edge 8-12 mm. If the program uses a different edge finish, see 230gsm-polar-fleece-stadium-blankets-with-whipped-stitch-edges-specify.
Separate required PO limits from starting values
Blanket sourcing goes wrong when buyers mix hard acceptance limits with illustrative mill targets. Write the PO in two layers: required acceptance limits that control shipment release, and starting process values used by the mill to stay safely inside those limits. The required limits belong in the contract or tech pack. The process values belong in internal QA sheets or approved supplier control plans.
For this 320gsm fleece blanket, a workable required acceptance set is: finished size 130 x 170 cm, tolerance ±2.0 cm each direction; finished stripe-to-edge parallelism max 6 mm on any side; fold-face alignment offset max 8 mm; decorative stitch inset 8-12 mm; dominant-bar corner cut-through max 5 mm where the dominant bar is 20 mm nominal; dimensional change after agreed laundering max ±3% length and width; commercial lot-average fabric mass 312-328 gsm. Those are commercially plausible numbers for mid-market retail, not universal law.
A workable starting process set for a vertical mill is tighter: fabric bow/skew target not over 1.5% before cutting, review at 2.0%; cut-panel stripe-to-cut-edge deviation within 4 mm over 100 cm span; sewing waviness below 3 mm in any 200 mm segment; internal stitch inset target 9-10 mm. A trader buying open-market fleece and subcontracting sewing will often struggle to hold these process values consistently because fabric distortion, spread relaxation, and cut accuracy are not fully under one roof.
The commercial reason for each number should be stated. Stripe drift and uneven fold trigger retail refusal and markdown risk. Excessive corner cut-through creates a first-impression defect on folded presentation. Underweight fabric drives consumer return risk for hand-feel mismatch. Loose dimensional stability claims create post-wash complaint exposure. Buyers who can explain the cost of each defect generally get cleaner supplier alignment. For adjacent sourcing context, see low-moq-startup-blanket-sourcing and blanket-quality-control-inspection.
GSM and piece-weight acceptance must use a sampling plan
Do not write a single GSM number without saying how many specimens are tested and how acceptance is calculated. For fabric mass per unit area, cite ASTM D3776 or ISO 3801 on conditioned specimens taken from the finished blanket body, excluding edge sewing, labels, and obvious distortions. Condition under standard textile atmosphere before testing. If the buyer and supplier do not align on conditioning, brushed fleece can show enough spread to create avoidable disputes.
A practical PO clause is: take 5 specimens per tested blanket, distributed across the blanket body; calculate the blanket average; then calculate the lot average from 5 blankets sampled from the inspection lot. Under this approach, the required lot-average acceptance can be 312-328 gsm, while a single tested blanket average below 304 gsm or above 336 gsm triggers investigation and expanded sampling rather than immediate automatic lot rejection. That makes the numbers enforceable instead of decorative.
Keep fabric GSM separate from finished blanket net weight. A blanket can hit GSM and still feel light if usable area runs short, brushing is aggressive, or moisture content at pack-out differs from the approval sample. Many buyers therefore add a finished net weight line, typically agreed from the sealed sample with a tolerance band that reflects size and moisture variation. For 130 x 170 cm at 320gsm plus sewing thread, a finished piece weight often lands roughly in the 700-760 g area, but this is only a starting reference and should be confirmed by construction and finish route.
Use tighter controls for premium retail packs and looser ones for donation, promotional, or compressed freight-sensitive programs. If the blanket is ribbon rolled or shelf folded, weight perception matters more because the customer handles it before purchase. Related planning context appears in fleece-weight-throw-blanket-program and custom-blanket-lead-times-shipping.
Standards language: fabric-stage distortion is not the same as finished appearance
Write the standard next to each measurable claim, but do not stretch a test method beyond its scope. ASTM D3882 is a fabric-stage method for measuring bow and skew in woven or knitted fabrics. It does not directly define a finished blanket acceptance rule for stripe-to-edge appearance, fold presentation, or corner cut-through. Use it to control the fabric before cutting, where the mill can still adjust stentering, overfeed, relaxation, and brushing balance.
The finished article needs its own visual and dimensional rules. For this blanket, stripe-to-edge parallelism, fold-face alignment, corner bar cut-through, and sewing waviness are separate finished-product checks. That split matters because a fabric with acceptable bow/skew at fabric stage can still be cut off-grain or sewn with draw-in, while a fabric with borderline bow/skew may still look acceptable after careful spreading and cutting segregation.
For dimensional change, cite ISO 6330 for laundering and ISO 5077 for calculation. If the product carries a wash claim, state the program, temperature, detergent type if relevant, and drying route. A clean PO line is: maximum dimensional change after one agreed home-laundering cycle: ±3% length and width. If the blanket is for value retail with low wash-performance sensitivity, some buyers may relax this. For premium or catalogue channels, buyers often tighten it.
For edge durability, do not cite ASTM D5034 as if it were a seam test. ASTM D5034 is a grab tensile fabric test. If the buyer needs a sewn-edge durability requirement, use a seam-appropriate method agreed by both parties, such as a seam-strength evaluation based on the stitched construction, or define a practical in-house pull test against the approved sample. If no formal edge-strength claim will be tested, leave the standard out and control edge durability through workmanship inspection and wear trials. For adjacent test-method reading, see astm-d5034-seam-strength-targets-for-300gsm-fleece-stadium-blankets-wi and anti-pilling-test-requirements-for-240gsm-polar-fleece-blankets-iso-12.
Convert bow and skew percentages into finished-blanket risk
Percent bow or skew on fabric and millimetres of stripe drift on the finished blanket are related, but they are not interchangeable unless you state the geometry. A simple working conversion for sourcing discussions is: allowable deviation in mm ≈ measured span in mm × distortion percent. On a 1300 mm blanket width, 1.5% distortion is roughly 19.5 mm; 2.0% is roughly 26 mm. Those numbers are too large for shelf appearance on a checked blanket, which is why fabric-stage control has to be tighter than final visual acceptance.
The process linkage is this: the mill controls bow/skew on fabric before cutting; the cutting room selects the cut line against the repeat; sewing then either preserves or worsens the geometry. A fabric running at 1.5% distortion may still produce an acceptable finished blanket if panels are spread properly and cut with reference to the check line. A fabric running near 2.0% leaves less margin and usually increases panel loss, re-cut risk, and mixed appearance within the same lot.
That is why a buyer should not write only one blanket-level mm tolerance and ignore fabric-stage distortion. A vertical mill can manage the chain from knitting through finishing, cutting, and sewing. A trader sourcing open-market fleece often cannot see the true pre-cut distortion history, so promising tight finished parallelism across the whole lot is less realistic unless inspection sorting and cut loss are built into the price.
As a starting PO logic, use fabric-stage bow/skew target not over 1.5% under ASTM D3882, with 2.0% as an internal review point, then hold finished stripe-to-edge parallelism to max 6 mm. The first number protects process capability; the second protects shelf appearance. Related sourcing context appears in choosing-picnic-beach-camping-mat and promotional-stadium-throw-sourcing.
How to measure stripe-to-edge parallelism without ambiguity
The inspection rule should define the datum, reference stripe, conditioning, and sample count. Condition finished blankets for at least 4 hours in a dry inspection area close to standard room conditions, then lay each blanket flat without stretching on a hard inspection table. Use the finished cut perimeter line as the edge datum, ignoring small oscillation of the decorative stitch thread. If corners are visibly drawn, do not measure through the curved or distorted corner zone.
Select one continuous dominant stripe visible along the side being checked. On each long side, measure the perpendicular distance from the finished edge datum to the same edge of that stripe at three points: about 200 mm from one end, midpoint, and 200 mm from the opposite end. The difference between the maximum and minimum readings is the side parallelism deviation. Repeat on the second long side as a minimum. Check the short sides as well if the retail presentation exposes them or the PO requires four-side compliance.
For lot inspection, a practical rule is to measure this on all sampled blankets selected for visual AQL inspection, not just one showcase piece. If the lot sample under ISO 2859-1 is 32 pieces, measure all 32 pieces for stripe parallelism because the measurement is quick and appearance-sensitive. Record the worst side on each blanket. If any blanket exceeds the agreed major limit, classify it under the AQL defect table instead of averaging away a visible defect.
A workable acceptance line is: finished stripe-to-edge parallelism max 6 mm on any side; local sewing-induced waviness over 3 mm within any 200 mm segment counted separately as an edge appearance defect. That separation prevents arguments where a supplier blames sewing waviness on print or fabric geometry. For related edge-finish detail, see 280gsm-polyester-fleece-throws-with-lockstitch-hemmed-edges-spi-hem-de.
Define corner cut-through, off-square, and uneven fold in inspector language
A tolerance is only usable if two inspectors will call the same defect the same way. For corner dominant-bar cut-through, inspect each corner individually on the retail face side. Measure from the notional corner bisector into the dominant stripe area and record how much of the 20 mm nominal dominant bar is lost by the finished corner shape. The maximum allowed loss is 5 mm, equivalent to 25% of a 20 mm dominant bar. If the blanket is reversible and both faces are meant to present equally, inspect both faces; otherwise grade by the sale face.
Define off-square as a visible mismatch between blanket geometry and check geometry, usually seen when the dominant bars are not perceived as parallel to the finished sides. Do not use diagonal size difference alone as the only criterion because brushed fleece can relax asymmetrically while still looking acceptable. In this article, off-square is captured primarily by the stripe-to-edge parallelism rule and secondarily by size tolerance.
Define uneven fold as a retail presentation defect where the approved fold exposes visibly different check-bar widths from left to right, or the centre fold line causes more than 8 mm mismatch between corresponding dominant bars on the front face. This matters because many retailer refusals happen on folded appearance, not on open-blanket lab data. If the product is sold banded, strapped, or ribbon rolled, include the folded-face approval photo in the sealed sample set.
Define edge waviness or draw-in as local undulation of the finished edge datum exceeding 3 mm in amplitude within any 200 mm segment, usually caused by sewing tension imbalance or feed mismatch. This should be counted separately from stripe parallelism because the commercial corrective action is different: sewing reset, not fabric re-approval. For presentation-sensitive pack formats, see 150gsm-polyester-fleece-blankets-with-satin-ribbon-rolls-presentation- and 280gsm-polyester-flannel-throws-with-white-label-neck-ribbon-packs-fol.
AQL framework: classify what is major, minor, and shipment-relevant
Tolerance tables are incomplete without a sampling and grading framework. A practical finished-goods inspection basis is ISO 2859-1, General Inspection Level II, with acceptance by the buyer’s agreed single-sampling plan. Many blanket buyers use AQL 2.5 for major defects and AQL 4.0 for minor defects as a starting point, then tighten for premium retail or loosen for donation and promotional channels. These are starting PO values, not universal defaults.
For this article, workable major defects include: stripe-to-edge parallelism over 6 mm on any side; dominant-bar corner cut-through over 5 mm on any corner; finished size outside ±2.0 cm; wrong colourway; obvious sewing damage, hole, oil stain, or face-side contamination; severe fold-face misalignment over 8 mm on the retail front; open seam or skipped decorative stitching over 30 mm cumulative on one edge. Any one of these can trigger retailer refusal, rework, or consumer return.
Workable minor defects include: local edge waviness over 3 mm but not affecting overall shelf appearance; slight stitch pitch variation outside target but within workmanship acceptability; small brush marks removable by hand grooming; minor shade variation not visible under normal inspection distance; thread end not exceeding agreed trim limit. These tend to generate factory touch-up cost rather than outright refusal if kept within AQL.
State the lot and sample logic in the PO or inspection SOP. Example: inspect per ISO 2859-1, Level II, normal inspection; classify geometric appearance defects as major when over listed limits; rework allowed only if re-folding or thread trimming corrects the issue without changing fabric geometry. For an adjacent checklist model, see aql-2-5-inspection-checklist-for-200gsm-coral-fleece-promotional-blank and aql-inspection-for-280gsm-jacquard-flannel-throw-blankets-with-satin-b.
Vertical mill versus trader: what tolerances are realistic
A vertical mill with knitting, brushing, heat-setting, cutting, sewing, and final inspection under one system can usually control tighter appearance tolerances because the fabric-stage distortion data is visible before cutting. That supplier can segregate rolls, adjust overfeed, tune spreading, and align the cut line to the check repeat. The result is better odds of holding a 6 mm finished stripe-parallelism limit consistently.
A trader buying open-market fleece and subcontracting cut-and-sew may still deliver acceptable blankets, but the process window is narrower and lot-to-lot variation is usually harder to control. If the check effect is printed, not engineered, brushing and heat history can vary by fabric source. In that supply model, a buyer should rely more heavily on sealed sample approval, pre-production fold review, and mid-line inspection, and may need to allow a slightly looser appearance tolerance unless the trader prices in higher sorting and rejection loss.
For premium retail, catalogue, and club-store programs, ask whether the supplier is controlling knit formation, finishing, cutting, and sewing in-house or assembling from outside fabric. For donation, event, or highly price-driven tenders, tolerance can be loosened slightly if the commercial risk of shelf presentation is low. The key is matching the tolerance to the channel instead of copying a premium spec into a budget program.
If the supplier cannot confirm the construction route, pre-cut distortion controls, and measurement method in writing, the tolerance table is not yet ready for PO issue. That is usually the point to simplify the design, widen appearance limits modestly, or shift to a construction with lower geometric sensitivity. Related sourcing context appears in 230gsm-solution-dyed-polyester-fleece-blankets-uv-color-retention-lot- and solution-dyed-230gsm-polyester-polar-fleece-blankets-moq-color-range-l.
PO clause set buyers can paste into a tech pack
A compact clause set is often more useful than a long article. A buyer-ready version for this blanket could read: Construction: 100% polyester brushed knit fleece blanket with approved checked appearance, nominal 320gsm, finished size 130 x 170 cm. Edge finish: decorative blanket stitch, polyester thread Tex 40-60, stitch inset 8-12 mm, pitch 10-14 mm. Fabric mass testing: ASTM D3776 or ISO 3801, conditioned specimens from blanket body; 5 specimens per blanket, 5 blankets per lot sample; lot-average acceptance 312-328 gsm; blanket average outside 304-336 gsm triggers review and expanded sampling. Dimensional change: ISO 6330 plus ISO 5077, max ±3% length and width after agreed cycle. Fabric distortion before cutting: ASTM D3882 target not over 1.5%, review at 2.0%. Finished appearance: stripe-to-edge parallelism max 6 mm on any side; fold-face alignment max 8 mm; dominant-bar corner cut-through max 5 mm per corner; local edge waviness over 3 mm in 200 mm counted as defect. Inspection: ISO 2859-1, Level II, AQL 2.5 major / 4.0 minor unless otherwise agreed.
If the blanket is printed check rather than engineered checked knit, revise the clause to say so. Then add artwork approval, print registration expectations, and wash-fastness or rubbing-fastness limits as needed. If the blanket is sold compressed, vacuum packed, or under a strict freight target, add a finished net weight line and packaging compression limit so the supplier does not solve cost pressure by brushing lighter or cutting short.
If the blanket is for a tighter channel such as department-store retail, it is reasonable to tighten fold-face alignment and corner presentation after pilot production proves capability. If it is for donation or event use, buyers may keep the same safety and gross workmanship thresholds but relax shelf-appearance limits modestly. The right spec is the one the channel will actually enforce, not the most severe number a buyer can write.
For adjacent PO drafting help, see custom-blanket-decoration-methods, travel-airline-blanket-weight-packing, and textile-certifications-explained-buyers.
Frequently asked
What is the most common construction for a 320gsm polyester check blanket? Most market supply is brushed polyester knit fleece with a printed check appearance, or solid fleece printed later. A true yarn-dyed checked knit fleece exists but is less common and should be confirmed with the supplier before you write tight geometry tolerances.
Can ASTM D3882 be used as the finished blanket acceptance method? No. ASTM D3882 is for measuring bow and skew at fabric stage. Use it to control finished fabric before cutting. Finished blanket appearance still needs its own rules for stripe-to-edge parallelism, fold presentation, and corner cut-through.
How should buyers write GSM acceptance so it can actually be enforced? State the test method, specimen count, and acceptance basis. A practical setup is ASTM D3776 or ISO 3801, 5 specimens per blanket, 5 blankets from the lot sample, accepted on lot average. For a nominal 320gsm fleece, 312-328 gsm is a workable lot-average starting value, while 304-336 gsm can be used as a review trigger band for individual tested blanket averages.
What is a realistic stripe-to-edge tolerance for a retail fleece check blanket? For a mid-market retail program using a checked brushed polyester fleece blanket with decorative blanket stitch edges, max 6 mm on any side is a realistic starting PO value if the supplier controls fabric finishing, cutting, and sewing. Printed checks or trader-sourced fleece may need slightly looser limits unless the supplier prices in more sorting and panel loss.
How do I measure corner dominant-bar cut-through consistently? Inspect each corner individually on the sale face. Measure how much of the nominal dominant stripe width is lost at the corner. If the dominant bar is 20 mm nominal and your maximum loss is 25%, the reject point is over 5 mm on any corner. If both faces matter commercially, inspect both faces.
What AQL should be used for these appearance defects? A common starting point is ISO 2859-1, General Inspection Level II, with AQL 2.5 for major defects and AQL 4.0 for minor defects. Use tighter levels for premium retail and looser levels for donation or promotional channels if the commercial risk of shelf appearance is lower.
Have a project in mind? Send us your spec — we'll reply within one business day with indicative pricing and a sample plan.