The number of rolls to open is not a matter of opinion, and the answer stays unsettled until someone writes down the unit being counted. Four answers to the same question give three different sample sizes: 3, 3 to 5, and all 10. None of the four is careless. Each counts a different unit: 1. rolls, 2. metres of cloth, 3. defect points per 100 square yards. Fix the unit and the contradiction collapses, because ISO 2859-1 then gives one route from lot size and inspection level to a code letter, a sample size n, and an accept and reject pair. The clause naming the unit also decides whether one bad roll fails the delivery. According to ISO 2859-1, the plan is read from a lot size, an inspection level and an AQL, so a clause silent on the unit cannot be applied. What follows is the procedure for one afternoon, from the pass over all ten rolls to the clause for the next order.
Why Do Three Answers Give Three Sample Sizes for Ten Rolls?
A sample size is the number of units drawn from a lot, and the unit has to be fixed first because every number downstream depends on the choice. Three units are in play on a roll delivery. First, the roll: ten rolls is a small lot, so the plan opens a handful. Second, the metre: the same delivery becomes a long lot, and the sample is a length taken off a roll. Third, the point per 100 square yards, the unit the four-point system scores, which turns the sample into an area. Two checks decide the band: 1. the unit, 2. the inspection level. Each unit lands in a different lot size band, and ISO 2859-1 reads that band against an AQL to produce a code letter, then a sample size n and an accept and reject pair. Change the unit and the plan changes with it, which is how 3, 3 to 5 and 10 all stay legitimate.
Is This One Lot, or Two?
A lot is a homogeneous quantity: one product, one process, one material, one defined period. ISO 2859-1 draws the sample from that population, so homogeneity is a precondition rather than a formality. Ten rolls carrying two warp lots, or two dye lots, are not one lot. The check costs minutes and runs in three steps: 1. read every ticket against the purchase order, 2. sort the ten into piles by warp lot, 3. sort each pile by dye lot. Two piles means two plans, each built on the smaller sub-lot. The failure this prevents is expensive, because a delivery can pass a sample drawn across a mixed lot and still shade out at the cutting table. The average was acceptable and the rolls were not. Where a warp is split across two beams, the handover record proves the split, which is why a second warp lot on a shared beam belongs in the same file as the tickets.
What Does the External Pass on All Ten Rolls Cover?
An external pass is a check of every roll as delivered, before the wrap comes off, and the pass runs across all ten rolls whatever the sample size turns out to be. Five items are recorded per roll: 1. the ticket, meaning roll number, length, width, weight and shade reference against the order, 2. the wrap, intact, torn, water stained or crushed at one end, 3. the ends, flattened, dirty, rust marked from a strap or creased by standing on a hard floor, 4. moisture, since a damp wrap and a cold roll produce mildew inside a container, 5. contamination, meaning oil, dust or a foreign fibre under the film. The pass takes about a minute a roll, and two habits keep the record honest. Number the rolls before starting, then write every unopened roll as externally inspected only, never as passed, because an external pass says nothing about the cloth inside.
How Many Rolls Do You Open, and Which Ones?
An opened count is the sample size the plan produces, and the number is read out of the standard rather than chosen by feel. Four inputs set the number: 1. the unit, 2. the inspection level, 3. the AQL, 4. the lot size band, which returns a code letter. The code letter gives n, and the AQL column gives the acceptance number Ac and the rejection number Re. If the cell holds an arrow, the arrow overrides both, and the plan moves to another sample size with its own Ac and Re. At small lot sizes with a strict AQL the arrow can push the sample to 100%, which on ten rolls means every roll is opened. Two habits void the result: 1. drawing from the top of the pallet, 2. opening the first rolls off the machine. According to the AQL sampling plan guide, the sample has to be drawn at random from the lot, because pieces taken from the top of the pallet invalidate the decision.
How Much Do You Unroll, and Under What Light?
An inspection length is the metres opened on a sampled roll, and the length has to be long enough to expose a repeating fault, because a fault repeating across the width hides inside a short piece. Three measurements belong on every opened roll: 1. width at three places along the opened length, 2. ticket length against a tape, 3. roll weight against the shipping mark. The cloth then runs over a lit table at a speed an inspector can hold for the whole sample. Light and speed are the two variables buyers leave out of the clause and argue about later, so name both in writing: a light source at a stated distance, and a pass speed the table can sustain. On ten rolls with two people and one afternoon, the arithmetic decides the plan. Cost the minutes per metre honestly, and a plan needing four hours announces itself before the truck leaves.

Which Record Fields Survive a Dispute?
A usable inspection record is a document a third party can read without asking a question, so the record carries one row per roll and the same fields on every row. The fields are: 1. date, 2. purchase order and lot number, 3. roll number, 4. ticket length, measured width and roll weight, 5. packaging condition, 6. time opened and inspector initials, 7. defect description with position and size, 8. severity class and photograph reference, 9. disposition. Every roll gets a row. An unopened roll is filled in as external inspection only, which differs from passed and stops a three-roll sample from being described later as a fully inspected delivery. Add a header stating the plan: unit, sample size, Ac and Re, which rolls were drawn, and where on each roll the sample came from. Our own per roll four point inspection report follows that shape, because the argument at claim stage is rarely about the defect. The argument is about what the record can prove.
What Does Opening a Roll Cost You?
Opening a roll spends a resource no clause restores, and the cost is cloth. A roll unrolled and re rolled is never quite the roll it was, because the surface picks up handling marks, the end takes a crease, and a cut sample leaves a short length that cannot go back into a lay. Three habits keep the cost small. 1. Take the sample from the end of the roll rather than the middle, so the shortened piece is a piece a cutter would have written off anyway. 2. Re roll on the same tension and the same direction, then tape the end flat instead of folding. 3. Tag the roll with the metres removed and the date, so nobody resamples the same end twice. Record the metres consumed on the roll’s row. A sample that ruins a roll while proving a point about a roll is the one outcome an afternoon cannot recover from.
Which Failures Bypass the Defect Count?
A lot level failure is a condition that fails a delivery without any single roll being defective, and four conditions appear on roll deliveries often enough to check by name. 1. Tickets: a roll whose ticket does not match the order, a missing ticket, or two tickets for one roll. 2. Shade: two dye lots inside one delivery, which no defect count records because every individual roll sits inside its own tolerance. 3. Quantity: a total short against the invoice, measured by tape or by weight. 4. Grade: seconds or a different quality mixed into a first grade order. None of the four shows up in an AQL count, and each one stops a cutting room. Check the four during the external pass, write the finding as a lot level hold, and never let a passing defect count override the hold. A delivery can pass the sample and still be the wrong delivery.
What Is the Disposition Order After the Count?
A disposition is the decision recorded against each roll, and the order in which decisions are taken matters more than the wording. The sequence is fixed: 1. hold, 2. count, 3. release. Hold means no roll leaves the receiving area while the sample is still open. Then compare the cumulative defectives found in the sample against Re. Do not compare the worst single roll against your patience. The standard fails a lot on the accumulated count across the sample, not on one bad roll. If Re is reached, the lot fails and the delivery goes behind the lot-fail line. If Re is not reached, the rolls that failed on their own are segregated and rejected individually. The rest are accepted. Full inspection comes next when the arrow already pushed the plan there, or when a lot level condition appeared during the external pass. Sorting and a claim follow only after the count is written down.
How Do You Hold, Sort and Claim?
A hold is a physical control rather than an email, and the hold works only if the rolls are marked and moved to one place. Keep the ten rolls together until the sample is closed, tag each rejected roll with the defect and its frame number, then photograph the tagged roll beside the ticket. Sorting is a separate decision from rejection: a roll with one localised fault may be accepted with a short length deducted, and the deduction belongs in writing on the same row. A claim is the last step in the sequence, and the claim is won or lost on documents built in the first hour. Two pages carry most of the file. Inspection rights agreed before signing set what you were entitled to check, and the day one claim sequence sets what to send and in what order.
Which Sentences Go Into the Next PO?
The clause is the only part of this procedure surviving to the next order. Write the clause once, properly. Eight elements do the work: 1. the unit, 2. the standard, 3. sample size with Ac and Re, 4. the arrow rule, 5. sample length, 6. light and speed, 7. record fields, 8. disposition and claim window. According to ISO 2859-1, the code letter, the sample size and the Ac and Re pair all follow from the lot size band, the inspection level and the AQL. The numbers in the second column are yours to agree with the mill. A clause written in 2026 without an arrow rule fails on the first disputed delivery.
| Clause element | What it fixes | What it prevents |
|---|---|---|
| Sampling unit | Rolls, metres, or points per 100 sq yd | Three answers to one question |
| Standard and edition | ISO 2859-1, inspection level, AQL | A plan nobody can reproduce |
| Sample size, Ac and Re | n with the accept and reject pair | A verdict by opinion |
| Arrow rule | Follow the table, up to 100% inspection | An invalid small sample |
| Sample length and position | Metres opened, taken from the roll end | A ruined middle |
| Light and pass speed | Source, distance, metres per minute | An unfair defect call |
| Record fields | One row per roll, unopened rows included | A sample called full inspection |
| Disposition and claim window | Hold, sort, claim, days allowed | A claim that arrives late |