Some characteristics on a made-to-print fastener or component are accepted by an attribute decision: a thread gauge passes or it does not, a visual standard is met or it is not, a presence check finds a feature or it does not. Others are accepted by a measured value compared with a tolerance. Those two kinds of evidence are not the same, and the inspection plan follows the characteristic and the method that have been agreed. A shop preference for one or the other does not override the print or the written plan.

What has to be visible in the plan

For an attribute check, the plan should name the gauge or the standard, the method of use, and what happens when a part fails. For a variable check, it should name the instrument, the reported unit, the location of the reading, and how many readings are taken if that matters. “Check the thread” is not complete if it does not say whether a gauge or a measured pitch diameter is required. Thread gauging as an attribute method is already described in the note on Thread standards, pitch and gauging.

If both an attribute sort and a recorded measurement are required on the same characteristic, that combination should be written down. A measured value taken on a sample does not automatically replace a 100% gauge check that the order still requires, and a 100% gauge check does not automatically generate the variable data needed for a capability study. Measurement-system agreement, when required, has to match the method actually used, as in the note on Measurement system and gauge agreement.

Why the choice matters for capability and sampling

Process capability, when the customer asks for it, is calculated from variable data obtained by an agreed method. It is not inferred from the fact that a lot passed a go / no-go sort. That point is already made in the note on Process capability on drawing characteristics. Sampling plans, when they are used, also have to match the kind of check: an attribute sampling plan and a variables sampling plan are not interchangeable without agreement, as discussed in the note on Sampling plans are agreed, not assumed.

Hardness, tensile, coating thickness and roughness checks are typically variable when a value is recorded, and may be attribute when they are reduced to a pass / fail against a limit. The method and the recorded evidence follow the specification. See the notes on Hardness and tensile checks when the print or the order requires them and on Measuring coating thickness.

Visual and functional checks stay in their place

A visual standard for a coating, a stamp, or an edge break is an attribute method when it is agreed, and it needs a defined standard, not an informal “it looks acceptable”. A “the spanner fits” or “the sample ring fits” check is useful only when the tool is a controlled gauge and the method is agreed; otherwise it is not a substitute for the dimension. Those limits have already appeared on specific features in earlier notes; the same rule applies generally here.

Optical or automated sorting, when it is specified, is usually an attribute sort against trained features. It does not replace a variable layout of a characteristic that the print still requires to be measured, as in the note on Optical or automated sorting when specified.

Changes of method, first article and conflicts

A change from attribute to variable inspection, or the reverse, after samples have been approved is a change to the inspection plan and is reviewed before serial use. First-article inspection, when required, should present the method that will be used in series, as in the note on First-article inspection on made-to-print parts. A laboratory measurement of a prototype does not by itself approve a later production gauge check if the decision rule will change.

A print that names a tight tolerance and only an attribute gauge that cannot resolve it, or a quality clause that asks for capability on a characteristic that is only gauged pass / fail, is flagged. Those conflicts are resolved on the revision or in a written plan, not by inventing numbers after a go / no-go sort.

Records, operators and mixed characteristics on one part

The inspection record should show the kind of evidence that was actually taken. Recording a pass / fail as if it were a measured value, or rounding a measurement into a tick without keeping the reading when the plan asks for a value, produces a file that cannot be audited. Operators have to be trained on the method that applies to each characteristic, including how to use a thread gauge or a micrometer, and how to react to a failure. A skilled operator’s opinion is not a third kind of official evidence unless the plan names a visual standard and the people qualified to apply it.

Many parts mix the two kinds of check on a single item: a heading may be gauged for a recess, measured for a shank, and visually compared for a stamp. The plan can be mixed. What it cannot do is let the easiest check stand in for the others. If a customer later asks why a capability figure was not supplied for a recess that was only gauged, the honest answer is that the agreed method did not produce that data — not that a number should be invented from the sort.

What to send with the RFQ

Send the current drawing revision, together with the material, finish and volume, and any already required inspection plan, gauge list or capability request. If particular characteristics must be recorded as values, or must be 100% gauged, say so. Unspecified points will be flagged before production rather than assumed.

This article is a capability and commercial note. It is not a certificate, a dimensional data book, or evidence of customer approval. Catalogue type codes and example grades do not replace the customer’s item number, material specification or drawing revision.