Every inspection report rests on a sample, and the sample size is not a guess. Standard sampling tables convert your lot size and chosen inspection level into a precise number of pieces to examine. Buyers who understand these two inputs can read any report critically and can argue for larger samples when the stakes justify it. This guide explains sampling levels for glassware and ceramics without the statistics fog.
Key takeaways
- The lot size and the inspection level together fix the sample size through a standard code letter. [1]
- General Inspection Level II is the default for most tableware orders, with Level I for low-risk bulk goods and Level III for high-risk shipments.
- Special levels exist for checks that damage the product or take a long time, such as physical stress tests.
- Acceptance numbers translate defect counts into pass or fail, separately for critical, major, and minor classes.
- Trade rules in each destination market may demand particular evidence, which is why sampling choices sit inside a wider compliance picture. [2]
Nobody inspects ten thousand mugs one by one. Instead, the inspector draws a few hundred pieces according to a published plan and judges the lot from what the sample reveals. This method works because the mathematics of random sampling is well understood and has been codified into tables used across industries.
The tables take two inputs. The first is the lot size, meaning how many pieces the shipment contains. The second is the inspection level, which expresses how discriminating you want the examination to be. Together they point to a code letter, and the code letter points to a sample size.
For buyers of plates, bowls, cups, and vases, the practical questions are which level to choose, how to read the resulting numbers in a report, and when to pay for a tighter examination. The sections below answer each in turn.
From lot size to code letter to sample size
Sampling standards organise lot sizes into ranges, and each range maps to a code letter under the chosen inspection level. The code letter then fixes the number of pieces to draw. A small parcel lot might require a few dozen pieces, while a full container of tumblers can require several hundred. The relationship is deliberately non-linear: doubling the lot does not double the sample, because large random samples gain precision more slowly.
Acceptance sampling theory explains why this works: a properly randomised sample carries information about the whole lot, and larger samples narrow the uncertainty around the true defect proportion. [1] Reports should state the lot size, level, code letter, and sample size together. When any of these fields is missing, ask for it, since the conclusion cannot be evaluated without knowing how much evidence sits behind it.
Which level fits which order
| Level | Typical use in tableware | Sample character | When to choose it |
|---|---|---|---|
| General Level I | Plain bulk tumblers, low unit value | Smallest general sample | Margins thin, risk tolerance higher |
| General Level II | Standard mugs, plates, bowls, vases | Balanced default | Most pre-shipment inspections |
| General Level III | High-value sets, strict retailers | Largest general sample | Chargebacks or safety concerns loom |
| Special levels | Breakage and capacity tests | Very small samples | Each test destroys or delays goods |
| Tightened examination | Repeat failures at one factory | Stricter acceptance numbers | Supplier is on probation |
Level II is the default for good reason: it balances cost against discrimination for ordinary consumer goods. Dropping to Level I saves inspector time but noticeably weakens the detection of scattered defects such as pinholes. Moving to Level III makes sense when a retail customer imposes penalties or when earlier batches failed, since the extra sample sharpens both finding defects and defending good lots.
Acceptance numbers and defect classes
Finding the sample size is only half the job. The tables also give acceptance and rejection numbers for the agreed quality limits, worked out separately for critical, major, and minor defects. If the count of major defects in the sample stays at or below the acceptance number, that class passes; reaching the rejection number fails it. Critical defects normally allow no tolerance at all.
A scheduled glassware ceramic inspection applies these numbers mechanically and records the counts that produced the verdict. Buyers sometimes ask inspectors to soften a borderline result, but the numbers exist precisely to remove that temptation. The right response to a borderline failure is a documented decision: rework, re-sort, or accept with eyes open.
Understanding acceptance numbers also helps buyers set smarter limits. Tightening the major-defect limit without raising the inspection level produces more borderline disputes, while raising the level without tightening limits spends money without changing decisions. Adjust the two together.
When the standard plan needs adjusting
Standard plans assume a reasonably uniform lot. Mixed shipments containing several designs, colours, or sizes break that assumption, and each distinct subgroup deserves its own sample. State in the booking how the lot divides so the inspector can allocate sample pieces across the variants instead of clustering on one.
Destination-market requirements can also shape the plan. Because members maintain their own technical regulations under shared transparency disciplines, the evidence a buyer needs varies by market. [2] Where a market demands specific test evidence, align the inspection scope with the laboratory plan so samples serve both purposes.
For zero-tolerance attributes such as food-contact safety claims, laboratory testing of representative samples complements visual sampling rather than replacing it.
Practical booking tips
Tell the inspection company the exact lot breakdown, the levels you want, and any subgroups needing separate verdicts. Confirm that finished and packed quantities will be complete on the inspection day, since sampling from incomplete lots skews every number in the report.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- WTO: Agreement on Technical Barriers to Trade —— WTO portal for the TBT Agreement covering technical regulations, standards and conformity assessment.
Frequently asked questions
What is General Inspection Level II?
It is the middle, default discrimination setting in widely used sampling tables. For most tableware orders it produces a sample that balances cost and defect detection. Levels I and III sit below and above it for low-risk and high-risk situations.
Can I ask for a bigger sample than the table says?
Yes. You can raise the inspection level or specify tightened examination. Expect higher inspection cost and a longer site visit. The gain is sharper detection of scattered defects and stronger evidence when disputing results with a supplier.
Does one sample cover mixed designs in one container?
Not well. Each distinct design or colour behaves like its own lot with its own defect pattern. Ask for the sample to be split across the variants, or book separate verdicts per subgroup so a failing design cannot hide inside a passing average.
Who decides the acceptable defect limits?
You and the supplier do, before production, in the specification or contract. The sampling tables convert those limits into acceptance numbers. Inspectors apply the numbers; they do not invent the limits on site.