Workmanship criteria in machined parts inspection: part 3
Dimensions feel objective until the gauge, the datum, or the sample selection goes wrong. This third installment of the workmanship series covers the measurement discipline buyers must demand: confirmed drawings, calibrated instruments, disciplined datums, random sampling, and reports that tie every verdict to evidence. It also explains the sampling logic that connects checked pieces to lot decisions and the trade-rule backdrop that makes written specifications worth their effort.
Key takeaways
- Confirm the drawing revision, gauge calibration, and datum setup before any measurement counts, because numbers without provenance are decoration.
- Require random carton selection so the sample represents the lot, a requirement sampling theory treats as foundational. [1]
- Read destination-market product rules as part of the order specification, since the trade framework on technical barriers explains how such measures function. [2]
- Record every finding with counts and photographs so rework scope and re-inspection have an auditable baseline.
- Act on the report the same day with a written ship, rework, or sort decision tied to the evidence.
The first two parts of this series built the workmanship sheet and its enforcement machinery: severity classes, boundary samples, and counting rules. Part 3 turns to the measurement side that buyers often take for granted. Dimensions are supposed to be the objective part of inspection, yet gauge practice, datum discipline, and sample handling decide whether a dimensional verdict deserves trust.
This matters because machining buyers mentally split inspection into hard numbers and soft appearance, then under-invest in the numbers. An inspector with an uncalibrated caliper, a drawing revision nobody confirmed, or a sample the factory pre-selected produces numbers that look precise while meaning little. Precision without discipline is decoration.
This installment covers measurement discipline step by step, explains the sampling mathematics that lets a sample speak for a lot, and places the whole exercise inside the trade-rule context every importer operates under.
Drawings, revisions, and datums first
Every dimensional check starts with a question most visits skip: which drawing revision governs this lot. Factories sometimes machine to a superseded revision while the buyer approved a new one by email, and the resulting argument cannot be settled with calipers. The purchase order should cite the revision explicitly, and the inspector should verify it before opening a carton.
Datum discipline comes next. A bore position measured from the wrong face, a flatness checked on an unqualified surface, or a thread judged without the specified gauge produces numbers that look authoritative and mean nothing. The inspection brief should state the datum scheme for critical traits so the inspector measures what the designer intended.
An independent machined parts quality check independent machined parts quality check earns its keep precisely here: the inspector has no stake in which revision was easier to machine and measures only the revision on the order. That neutrality is worth more than any single gauge reading.
Gauge practice buyers should demand
Sampling logic in plain language
A sample speaks for the lot only when three conditions hold: random selection, a sample size tied to the lot size, and acceptance numbers fixed before checking begins. Random selection prevents showcase cartons from standing in for production. The sized sample keeps the inspection proportionate. Fixed acceptance numbers keep the verdict honest.
The statistical reasoning behind these rules is developed in the literature on acceptance sampling, which shows how sample evidence supports lot decisions under defined risks for both sides. [1] Buyers do not need the mathematics on the shop floor, but they should insist on the three conditions, because dropping any one of them turns the visit into theater.
This logic applies to visual traits as readily as to dimensions. A scratch count from a random sample describes the lot; a scratch count from cartons the factory selected describes the factory's curation skills. The inspector's first job, before measuring anything, is therefore controlling the sample draw.
What random selection looks like
The inspector walks the stacked lot, counts the cartons, and pulls shipping units from spread positions including back rows and bottom layers. Factory staff carry cartons and open them, but they do not choose which ones. Any carton the factory refuses to make accessible is noted in the report, because an unsampled corner is an uninspected corner.
Mixed production dates deserve stratified attention. If the lot spans several machine runs, the sample should reach into each run rather than clustering in the most recent pallets. Date codes and heat or batch marks help the inspector spread the draw across the true production range.
Specification checklist for machined lots
| Brief element | What good looks like | Failure mode it prevents |
|---|---|---|
| Drawing and revision | Number and revision cited on the purchase order | Measuring against a superseded design |
| Workmanship sheet | Named traits with photos and severity classes | Floor debate over every flaw |
| Sampling plan | Level, sample size, and acceptance numbers pre-agreed | Moving goalposts after findings |
| Gauge expectations | Instrument types and calibration basis for critical traits | Authoritative-looking but baseless numbers |
| Logistics | Complete lot, accessible stacking, contact present | Half-lot visits and wasted travel |
Trade rules are part of the specification
Importers sometimes treat destination-market product rules as someone else's problem until a shipment is held. The safer view is that technical regulations belong in the specification from the first order: marking duties, restricted-substance limits, and documentation expectations are order terms, not surprises. The multilateral framework on technical barriers describes how such measures relate to legitimate regulatory objectives, which helps buyers explain to suppliers why a rule exists. [2]
Inspection then verifies the verifiable: presence and legibility of required marking, availability of stated documents, and conformity of the traits the specification actually named. No inspection can certify full regulatory compliance from a warehouse visit alone, and buyers should not ask it to. What it can do is confirm that the enforceable parts of the specification were met on the shipped lot.
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 belongs in the inspection brief for a machined lot?
At minimum the drawing revision, the workmanship sheet with boundary references, the sampling level and acceptance numbers, the factory contact, and confirmation the lot is complete. Anything the inspector must enforce but did not receive in writing effectively does not exist on the day.
How should buyers read dimensional findings in the report?
Ask for the gauge type, its calibration status, who performed the measurement, and whether the reading was repeated. Measurement is part of the verdict, not background detail, and an uncalibrated gauge can fail good parts or pass bad ones with equal confidence.
Do destination-market rules affect small machined-part orders?
Regulations of the destination market apply to every shipment size, and the trade framework distinguishes legitimate regulation from unnecessary obstacles without exempting small lots. Build compliance into the specification from the first order rather than discovering it at customs.
How should a buyer act on a workmanship report?
Compare the defect counts trait by trait against the agreed limits, read the photographs for borderline calls, and confirm which cartons were sampled. Then decide to ship, rework with re-inspection, or sort, and put the decision and its reasons in writing the same day.