Two words get used as though they meant the same thing, and on an examination they very much do not. A discontinuity is an interruption in the uniform structure of a weld or its base metal. A defect is a discontinuity that has been judged unacceptable against a stated requirement. Every defect is a discontinuity; most discontinuities never become defects.
The distinction is worth more than a vocabulary mark. It determines what an inspector is allowed to write down, what an inspector is allowed to conclude, and where the authority to reject actually sits. Getting it backwards in a report is how a finding turns into an argument that the inspector loses.
This guide covers what each term means, what does the deciding, the related words that orbit them, and how to phrase an observation so that it survives being challenged. It describes general inspection practice; the criteria that make any particular condition a defect come from the code or contract documents governing the work in front of you.
What a discontinuity actually is
A discontinuity is any interruption in the typical structure of a material - a break in the continuity of the physical, mechanical or metallurgical characteristics of the weld or the parts joined. Porosity is a discontinuity. Undercut is a discontinuity. A crack is a discontinuity. So is a change in grain structure across the heat-affected zone, and so is a legitimate design feature such as a deliberate change of section.
The definition is intentionally neutral. It says nothing about size, nothing about location, and nothing about whether anybody should care. It is a description of geometry and structure, and it belongs to the physical world rather than to anyone's judgement about that world. A discontinuity is simply a fact about a piece of metal.
That neutrality is exactly why it is the word an inspector reaches for first. Describing what is on the steel is always defensible. Describing what is on the steel and simultaneously declaring it unacceptable are two separate acts, and collapsing them into one sentence is where reports get into trouble.
- Discontinuity: an interruption in the uniform structure of the weld or base metal
- Defect: a discontinuity that fails a stated acceptance requirement
- Indication: the response a test method produces, which may or may not be a discontinuity
- Flaw: an imperfection requiring evaluation, used interchangeably with discontinuity in some documents
What turns a discontinuity into a defect
A discontinuity becomes a defect when it exceeds what the applicable acceptance criteria permit - in size, in length, in depth, in quantity, in spacing, in location, or simply in kind. Nothing about the discontinuity itself changes at that moment. What changes is the comparison between the measurement and a written requirement.
Because the requirement does the work, the same physical condition can be a defect on one job and perfectly acceptable on the next. A given depth of undercut may be routine on a statically loaded member and rejectable on a cyclically loaded one. A degree of porosity that passes on a structural fillet may fail on a pressure boundary. The steel has not changed; the document has.
This is also why some discontinuity types are treated as defects wherever they appear. A crack is the clearest example: codes typically prohibit cracks regardless of length, because a crack is a propagating feature rather than a bounded one. Even there, though, the prohibition lives in the document, not in the word.
Rejectable, defective, and the language of consequence
The word defect carries legal and commercial weight that the word discontinuity does not. Calling a weld defective asserts that it fails a requirement, which is a conclusion, and conclusions have to be traceable to the provision they came from. Calling a condition a discontinuity and reporting its dimensions asserts only what you measured.
Some organisations prefer rejectable or non-conforming over defective for exactly this reason, because those words point at the requirement rather than at the workmanship. Whatever term the paperwork uses, the sequence behind it should be the same: observe, measure, compare against the stated criterion, then state the disposition.
The habit that protects you is separating those steps on the page. A finding that reads as a description followed by a comparison followed by a disposition is auditable by someone who was not there. A finding that reads as a verdict is only as good as the reader's willingness to take your word for it.
Indication is a third word, and it is not a synonym
An indication is what a test method reports: a bleed-out on a penetrant examination, a particle build-up on a magnetic examination, a screen response on ultrasonic, a density change on a radiograph. An indication is evidence, and evidence has to be interpreted before it means anything at all. The interpretation is a separate act from the detection, performed by a different competency and recorded separately.
Indications resolve into three buckets. False indications come from the process itself - loose particles, poor cleaning, a scratch in the film. Non-relevant indications come from real geometry that is not a problem, such as a legitimate change of section. Relevant indications correspond to actual discontinuities and are the ones that get evaluated against acceptance criteria.
Keeping this chain straight matters because the words map onto stages of the inspection: an indication is found, it is interpreted as a discontinuity, and the discontinuity is evaluated as acceptable or as a defect. Skipping a stage in the vocabulary usually means a stage got skipped in the work as well.
- False indication: produced by the examination process, not by the part
- Non-relevant indication: caused by real geometry that is not a discontinuity of concern
- Relevant indication: corresponds to a genuine discontinuity requiring evaluation
- Evaluation, not detection, is what decides whether that discontinuity is a defect
Why examinations lean on this so hard
Question writers like this distinction because it is a clean test of whether a candidate understands the inspector's role. An inspector detects, measures, characterises and compares. An inspector does not decide what is acceptable - that decision was made when the code and the contract were written, and the inspector applies it.
So questions tend to be built around a trap: they describe a condition, sometimes with a dimension, and offer an answer that calls it a defect without any reference to a requirement. The correct answer is usually the one that names the condition accurately and leaves the acceptability question to the criteria, or the one that says more information is needed. The same trap appears in practical work, where the temptation is to answer from experience rather than from the document you were handed.
Writing a finding that holds up
A defensible finding contains four things: what the condition is, where it is, how big it is, and which requirement it was compared against. Drop any one and the finding gets weaker. Drop the requirement and the finding becomes an opinion, which is the form most easily argued away by anyone with a commercial reason to argue.
Location deserves more care than it usually gets. A depth measured somewhere along a two-metre weld is not actionable; a depth measured at a stated distance from a stated reference point is. The person who has to grind and reweld needs to find the spot, and the person who re-inspects afterwards needs to look at the same place you did. Photographs and sketches help, but they do not replace a measurement recorded against a datum.
- Name the condition using the standard term for it, not a colloquial one
- Fix the location against a reference the next person can find
- Measure the governing dimension, which is the worst one present
- Identify the acceptance provision you compared it against
- State the disposition separately from the observation
- Note anything you could not measure rather than estimating it silently
The judgement that is genuinely yours
None of this means an inspector is a measuring instrument with a clipboard. Deciding which conditions to examine closely, recognising that a cluster of small pores is telling you something about shielding, spotting that repeated undercut in one position points at technique rather than at luck - that is judgement, and it is the part of the job that experience actually improves.
What the discontinuity-and-defect distinction protects is the boundary between that judgement and the acceptance decision. You can and should report a pattern of acceptable discontinuities as a process observation, because a run of conditions that all pass is still evidence that something upstream is drifting.
Handled that way, the distinction stops being a vocabulary rule and becomes a working method. Describe generously, measure carefully, compare explicitly, and let the acceptance criteria carry the weight they were written to carry. That is also the version of the job that survives a challenge, which is the only test of an inspection record that ultimately counts.