AQL and Textile Inspection: Building a Sampling Plan for Towel Orders

Start with the decision: accept or reject a defined lot

Acceptance sampling sits between no inspection and inspecting every unit. A random sample is checked, and its result is used to decide whether the submitted lot should be accepted or rejected under an agreed plan. The NIST Engineering Statistics Handbook stresses that its main purpose is to decide the disposition of a lot, not to estimate that lot’s quality.

That distinction matters. If 2 towels in a sample are classified as nonconforming, it is not valid simply to announce that the same proportion exists in the full shipment. Nor does an accepted result prove that the lot contains no defects. Sampling carries producer and consumer risks; a sound commercial brief defines those risks through a recognised plan instead of treating a sample as certainty.

Use the current standard and the right type of lot

ISO 2859-1:2026 is the current third edition for AQL-indexed inspection by attributes. It provides single, double and multiple sampling schemes for lot-by-lot inspection of a continuing series of lots from one process or supplier. Its system includes switching among normal, tightened and reduced inspection, plus skip-lot provisions. The 2026 edition replaces the 1999 edition and its amendments, so purchase documents should name the edition rather than say only “ISO 2859-1”.

ISO 2859-1 is not automatically the right framework for every purchase. ISO 2859-2:2020 addresses isolated lots or short series where the switching rules of Part 1 cannot operate and indexes plans by limiting quality rather than AQL. ISO 28590:2017, confirmed current in 2024, introduces the ISO 2859 family and helps users choose an inspection system. Buyers should have a qualified quality professional select the applicable system instead of borrowing a familiar table from another order.

AQL is an index—not a shipment’s permitted defect percentage

In ISO 2859-1, the selected AQL works with lot size, inspection level, sampling scheme and current inspection severity to lead to a sample size and acceptance/rejection numbers. It is associated with an acceptable process average across a continuing series under the scheme. It does not mean that each submitted lot may knowingly contain that percentage of defective towels, and it is not a contractual promise that an accepted lot will contain no more than that percentage.

AQL alone therefore cannot generate an inspection result. “Major defects: AQL 2.5”, for example, still lacks the standard edition, lot definition, inspection level, scheme, severity, defect catalogue and counting basis. Do not select numbers from an online calculator unless the inputs and the licensed standard behind them have been verified. This article intentionally does not reproduce ISO’s proprietary sampling tables.

Define the lot and sampling unit before choosing a plan

A lot is the specific quantity submitted for one acceptance decision. State the purchase order, style, colour, size, branding and packaging combinations included, the total quantity, and how production is divided by batch or shipment. Combining unlike products can hide a concentrated problem; splitting one homogeneous production run after seeing results can distort the decision. Agree the lot structure before selection begins.

Also define the sampling unit. It may be one finished towel, one retail set or another stated unit. Explain whether inspectors count nonconforming units—units with one or more nonconformities—or count nonconformities. One towel can have a wrong label, an open hem and a stain, so the two approaches do not produce the same tally. Use the terminology and method required by the purchased standard and inspection brief.

Build a towel-specific defect catalogue

“Critical, major and minor” are useful operational classes only when buyer and supplier agree what each one means for this product. There is no universal list that assigns every towel fault to the same class. Classification should consider safety or legal impact, intended use, function, saleability and the approved appearance standard.

• Critical: reserve for a nonconformity presenting an agreed safety or serious regulatory risk, where applicable. Do not label an ordinary cosmetic fault critical merely to make the plan stricter.
• Major: could include an open seam, hole, missing branding, incorrect product or legally required label, dimensions outside the agreed tolerance, or colour outside an agreed standard when the issue materially affects use or saleability.
• Minor: could include a limited cosmetic deviation that does not materially affect intended use, such as a small removable loose thread, if its location and extent fall within the written boundary.

Add photographs or sealed samples, measurement method, viewing conditions, tolerances and edge cases. The towel sample approval checklist helps record what the physical approval establishes, while the towel specification-sheet guide turns dimensions, construction, colour, branding, packaging and testing into inspectable requirements.

Complete every sampling-plan input

An inspection level affects the relative amount of inspection; it is not a synonym for defect severity. Once all inputs are fixed, an authorised user derives the code letter, sample size and accept/reject numbers from the licensed current standard. Record those outputs on the inspection instruction so the inspector is not forced to infer them on site.

Select randomly, inspect consistently and apply the decision once

Random selection protects the plan from convenient samples. Draw units across cartons, pallet positions and the beginning, middle and end of the available lot; do not accept only top-carton pieces prepared for inspection. If the lot contains agreed subgroups such as colours or sizes, state how the random sample is allocated so smaller groups are not invisible.

Check every selected unit against the same current specification, approved sample and defect catalogue. Record each observation by unit and class, then compare the final counts with the specified acceptance and rejection numbers. The NIST handbook explains that single, double and multiple schemes have different decision paths; an extra sample is legitimate only when the chosen scheme calls for it. Repeatedly drawing fresh samples until one passes invalidates the agreed decision rule.

Define what rejection triggers—and what sampling cannot replace

Rejection is a lot-disposition decision, not an automatic instruction to destroy goods. The contract should identify who may release or hold the lot and whether the next step is investigation, 100% sorting, authorised rework, corrective action or formal resubmission. Any reinspection must follow the agreed standard and resubmission rules, not an improvised second chance.

Visual attribute sampling also does not replace in-process quality control, accurate specifications or laboratory tests. Absorbency, colourfastness, fibre composition and dimensional change may need separate methods, specimens and acceptance criteria. A plan should state which checks occur on every sampled towel, which measurements use a subset and which tests are performed by a qualified laboratory.

Questions buyers often ask

Is AQL the percentage of defects a supplier is allowed to ship?

No. AQL indexes a statistical sampling scheme. It is not a fixed defect allowance for an individual lot, and a supplier should not knowingly build defects up to that number.

Does an accepted lot mean every towel conforms?

No. Acceptance is a probabilistic lot decision based on a random sample. It does not certify every uninspected unit or estimate the lot’s exact defect percentage.

Should critical, major and minor defects use the same AQL?

Not automatically. Define each class and its consequence, then agree the applicable index or zero-tolerance rule with qualified quality input. There is no universal towel value.

Can the inspector take another sample after a rejection?

Only when the agreed sampling scheme or formal resubmission procedure permits it. Taking repeated independent samples until a pass is obtained defeats the plan.

Is an AQL inspection the same as laboratory testing?

No. Attribute inspection classifies defined observations. Performance and composition tests may require separate standards, equipment, specimens and decision rules.