ISO 2859-1 Sampling for Knit Bag Buyers: General Inspection Levels, Switching Rules and the Lot Definition That Decides Everything

What ISO 2859-1 Governs and What It Leaves Out

ISO 2859-1 covers acceptance sampling by attributes for lot-by-lot inspection, indexed by acceptance quality limit. An attribute is a judgement that a piece conforms or does not: a hole is present or it is not, a handle is misaligned beyond tolerance or it is not. The standard governs how many pieces to look at and how many nonconforming pieces to allow before rejecting the lot.

What it deliberately leaves out is every measurement. Abrasion life, tensile force, colour fastness and dimensional change are variable properties measured on separate specimens against their own standards, and no amount of attribute sampling substitutes for them. This is why a bag can pass an ISO 2859-1 inspection and still fail in month two, and why the two halves belong in the same quality plan.

The standard is maintained by ISO/TC 69/SC 5 under ICS 03.120.30, the classification for applications of statistical methods. Knowing the committee matters practically, because it tells a buyer that the document is a statistical one rather than a textile one, and that nothing inside it knows or cares what a knitted loop is.

The 2026 Edition Versus the 1999 Edition

Most bag specifications written before 2026 cite ISO 2859-1:1999, which was withdrawn when the third edition was published in January 2026. The replacement runs to 82 pages and introduces skip-lot sampling procedures into Part 1 along with updated guidance on applying sampling strategies, while retaining the AQL-indexed single, double and multiple schemes that buyers actually use day to day.

For a buyer the practical consequence is small but real. The sample size and accept number tables a buyer has been using remain the shape they were, so an inspection plan built on the 1999 edition is not wrong, but it is citing a withdrawn document. The fix costs one line in the order acknowledgement: name the standard and the edition, and require the inspection report to do the same.

The larger change is philosophical. Bringing skip-lot procedures into Part 1 formalises something good programmes were already doing informally: reducing inspection on a proven, stable supply and concentrating it where results have moved. A buyer running repeat seasonal orders can now point at a clause rather than at a habit.

Inspection Levels I, II and III Compared

The general inspection level scales the sample size to the lot size, and choosing it is the single highest-leverage decision a buyer makes inside ISO 2859-1. Level II is the normal default and the one most bag programmes should use. Level I roughly halves the sample, which halves inspection cost and roughly halves the sensitivity of the check.

Level III roughly doubles the sample and is the right choice where a buyer has reason to be cautious: a first lot from a new programme, a lot following a change of yarn, or a style with a history of shade complaints. The AQL does not change between the three; only the sample and therefore the discrimination of the plan does.

There are also four special inspection levels, S-1 to S-4, used where the test is destructive or expensive and a small sample is unavoidable. They appear in bag programmes when a buyer wants a handful of pieces cut open for joinery inspection without sacrificing twenty units, and they should be named explicitly because they produce very small samples that surprise people.

Single, Double and Multiple Sampling Compared

Single sampling takes one sample and decides: on a lot of 3,201 to 10,000 pieces at level II and AQL 2.5, that is 200 pieces with 10 accepted and 11 rejected. It is easy to administer, easy to explain to a colleague, and it is what most bag inspections use. Its weakness is that a borderline result produces a full re-inspection rather than a second look.

Double sampling takes a first smaller sample and only takes a second if the first is inconclusive. The average number of pieces inspected is lower than single sampling for the same protection, which is the argument for it, at the cost of a procedure that has to be administered carefully and explained patiently. It suits programmes where inspection capacity is the constraint.

Multiple sampling extends the same idea to up to seven samples and is rarely worth it for bag programmes. The administrative overhead exceeds the saving unless inspection is extremely expensive per piece, which it is not for a knitted bag. Where a buyer is tempted by multiple sampling, double sampling almost always delivers the practical benefit.

Switching States: Normal, Tightened and Reduced Compared

ISO 2859-1 is a system of switching rules rather than a single table, and the state a programme sits in changes the plan. Normal inspection is the starting point. Tightened inspection follows a defined run of poor results and lowers the accept number without enlarging the sample, so the bar rises at no extra inspection cost.

Reduced inspection becomes available after a sustained run of good results and cuts the sample size, which genuinely reduces inspection cost. It also reduces sensitivity, which is why a buyer should decide explicitly whether reduced inspection is permitted. In most bag programmes it is applied by default and nobody notices the sample has halved.

Discontinuation is the end state the system is really protecting against. Where tightened inspection keeps failing, the standard's logic is that acceptance sampling should stop and the process should be fixed rather than inspected harder. For a buyer that is a useful sentence to have available during a dispute about a persistently failing programme.

ISO 2859-1 Against Its Sibling Parts

The ISO 2859 series has parts beyond the first, and knowing which one applies avoids a common error. Part 1 is for a continuing series of lots. Part 2 is indexed by limiting quality and covers isolated lots, which is the right reference when a buyer purchases a single one-off lot and has no history to switch against.

Part 3 covers skip-lot procedures in its own right, Part 4 covers assessment of declared quality levels where the question is whether a supplier's claim is supported rather than whether a lot can be released, and Part 5 covers sequential sampling plans indexed by AQL. Part 0 is the introduction and is genuinely worth reading, because it carries the worked examples the other parts omit.

For a knit bag buyer the practical split is simple. Repeat seasonal programmes sit in Part 1. A single promotional order bought once sits in Part 2. Everything else is a specialist case, and a buyer who finds themselves citing Part 4 or Part 5 should first check whether a simpler question was being asked.

Attribute Sampling Against Variable Sampling

ISO 2859-1 counts; ISO 3951 measures. Where a characteristic can be measured on a continuous scale, such as a handle breaking force in newtons or a dimensional measurement in millimetres, variable sampling under the ISO 3951 series extracts more information from a smaller sample, because it uses the distribution of results rather than a pass or fail count.

The catch is that variable sampling assumes a known and roughly normal distribution and needs one measurement per piece rather than one judgement. On a knitted bag, where the interesting properties are often structural and the inspector is looking for dropped stitches rather than measuring anything, attribute sampling is the honest choice despite using more pieces.

A blended approach is usually best and is what a well-written quality plan does: attribute sampling under ISO 2859-1 for the visible defects at goods-in, and variable measurement under ISO 13934-1 or ASTM D5034 for strength, ISO 12947 for abrasion life, and the ISO 105 series for colour. Neither half replaces the other.

Choosing a Plan: Three Scenarios, Three Answers

A first order of 100 pcs per style and colourway with no history: Part 1, level II, single sampling, AQL 2.5 major and zero critical, with tightened inspection available and reduced inspection not permitted. The sample will be small at that lot size, which is precisely why the pre-bulk laboratory tests matter more than the inspection on a first order.

A repeat 12,000 piece seasonal programme in three colourways with two clean seasons behind it: Part 1, level II, single sampling, lots defined per colourway, and skip-lot or reduced inspection considered on the strength of the history. Here the inspection is a confirmation rather than the main control, and the main control is the pre-bulk testing.

A one-off promotional buy of 5,000 pieces from a partner the buyer has not used: Part 1 with level III for the first lot, or Part 2 if the buyer wants the isolated-lot framing. Level III costs more inspection and buys real discrimination on a lot where nothing is known about the process behind it.

How to Cite the Standard So the Report Is Checkable

A citation needs four elements to be checkable: the standard number and edition, the inspection level, the plan type and switching state, and the AQL per defect class. Add the lot definition and a fifth, and the inspection can be audited by someone who was not there, which is the entire point of writing it down.

Ask for those five elements to appear on the report itself rather than only in the order. A report that shows the standard, the level, the sample size and the accept and reject numbers lets a buyer verify the outcome in about thirty seconds; a report that shows only a pass result requires a phone call.

Frequently Asked Questions

People Also Ask