Coating Thickness: How to Specify, Measure and Verify It

ASTM A123 and ISO 1461 can pass and fail the same galvanized beam, because they ask different questions of the same readings. What each standard requires, how to take the readings, which test settles a dispute, and the clause that ends the argument before the steel is dipped.

Same beam, two verdicts

Two inspectors can gauge the same galvanized beam, both read correctly, and reach opposite verdicts, because the contract and the inspector are working to different standards and the standards ask different questions.

A galvanizer working to ASTM A123 reads five spots on each third of a 12 mm plate girder, averages 78 µm against a Grade 75 requirement and passes the lot. A client inspector working to ISO 1461 takes three reference areas on the same girder, finds a mean of 78 µm against the 85 µm mean minimum and rejects it. Both applied their standard correctly. The dispute is a specification problem that surfaced as a measurement problem.

This article lays the two governing standards side by side, so the clause you write, the readings you take and the verdict you defend all answer the same question. What it settles:

• ASTM A123 tests averages. ISO 1461 puts a local floor on every reference area and adds a mean: per article above 2 m², across the control sample below it. Neither is stricter in general: on the same steel, which one demands more flips with material category and thickness.
• A123 assigns its grade by how the product was made, not by how it looks. A rolled beam and a plate girder of the same thickness can carry different grades.
• The gauge, the reference area and the referee test are part of the requirement. Write them into the order or expect to argue about them later.
• Neither standard sets a maximum thickness. A purchase order that adds one is writing a new requirement, usually one the galvanizer cannot control.
• A minimum is a floor, not a target. The average sitting above it is the largest controllable zinc cost in the plant, the argument of the Zinc Consumption KPIs article. Here it becomes a measurement routine.

Two standards, two different questions

ASTM A123 asks one question: does the average coating on each specimen reach the grade for its material category and steel thickness? A specimen average may sit one grade below the requirement, for example 85 µm where Grade 100 applies, provided the specimen averages, averaged together, meet the grade itself. For a large article that means its three specimens, and every test article in the sample has to pass. "One grade below" is read down the list of grades in Table 2, so 100 steps to 85, 75 to 65 and 45 to 35.

ISO 1461 asks two questions of every article. First, the average within each reference area must reach the local minimum, whatever the size of the article. Second, a mean must reach the mean minimum, and here size matters: on a large article, above 2 m² of significant surface, it is the mean of that article's own reference areas; on smaller articles it is the mean of all the reference areas in the control sample. For steel over 6 mm those figures are 70 µm and 85 µm. A coating can meet the mean and fail the local, which is the reason the standard states both.

Both standards average. A local value in ISO 1461 is itself the mean of at least five magnetic readings taken inside a reference area of at least 10 cm². Clause 6.2.3 says so directly: a single reading below the local minimum is not relevant, because only the average over the whole reference area is required. The difference is the size of the patch. An A123 specimen is the whole article, or a third of a large one, and its average may sit one grade below the requirement. An ISO reference area is a small patch, and every one is tested against the local floor, so a low area cannot hide behind a high one.

The consequence for a contract is direct. The same lot can pass one standard and fail the other, so the order must name one standard and its edition, and say which prevails if it names both.

The vocabulary that decides the argument

Most thickness disputes are disputes about one word. The two standards use different words for the same steps, and mixing them is how a five-reading average ends up compared against the wrong floor.

Two details change verdicts more often than the table suggests.

First, A123 never blends categories. Each material category and each steel thickness range on a fabrication is its own specimen, averaged on its own. A gusset plate and a chord member on one truss are two verdicts, not one. ISO 1461 works the same way: where an article includes several steel thicknesses, each thickness range is treated as a separate article.

Second, ISO 1461:2022 clause 6.2.3 now says that, unless agreed otherwise, reference areas are not taken from ancillary elements that are significantly smaller than the main elements of a larger article, because they are thinner or less reactive. The standard's examples are stiffeners, end plates and brackets. Measure the member that sets the coating, not the plate welded to it.

What each standard requires, and why the category matters

ISO 1461 sets thickness by steel thickness alone. These are the general-use figures for work that is not centrifuged; centrifuged work has its own table, and a product standard or contract can override both. The 2022 edition also adds separate requirements for ultra-low-reactivity steels in clause 6.5, covered in the new editions article.

ASTM A123 needs two tables to get to a number. Table 1 gives a coating grade from the material category and the measured steel thickness. Table 2 converts the grade into microns and mass, and the grade is the minimum average thickness in microns: Grade 100 is 100 µm, which Table 2 states as 705 g/m².

The category is decided by how the product is made, not by how it looks. Appendix X1 classes a rolled beam as a structural shape, a plate girder as plate, a pole from bent plate as plate rather than a structural shape, and bar grating as strip and bar. The AGA's note on the 2024 revision shows the choice moving the grade in both directions:

• A plate girder of 12.7 mm (½ in.) measured steel is plate, Grade 75, not a structural shape at Grade 100.
• A pole fabricated from plate of 5/8 in. (16 mm) or thicker is plate, Grade 100, not pipe and tubing at Grade 75.

Put the two standards on the same steel and the comparison has no single answer. The table below takes mid-band cases, because the band edges differ: ISO breaks at 1.5, 3 and 6 mm, while A123 breaks at 1.6, 3.2, 4.8 and 6.4 mm, with a further column at 16 mm.

The sign flips on steel of identical thickness. Which standard is stricter is a property of the category and the thickness, not of the standard. An order that cites both has, in practice, imposed the higher figure on every element of the fabrication.

One trap sits under both tables. A123 Table 1 uses the measured steel thickness, not the nominal, and on a closed section it cannot be measured after fabrication. Appendix X1.2 gives the fallbacks: measure the galvanized part and approximate, or use nominal values from the project drawings where the specimen is inaccessible. For tapered designs, and for shapes whose webs are thinner than their flanges, the thinnest section sets the row.

How many articles and how many readings

ISO 1461 sets the control sample from the lot size in one table.

A123 Section 7.3 uses a similar table, and a lot of 4 to 500 pieces also gives three test articles. The real divergence is the size at which an article is split. A123 sorts articles into single-specimen (up to 160 in², about 0.1 m²) and multi-specimen (larger), and applies different test rules to the two classes. ISO 1461 asks for three reference areas only when the significant surface is above 2 m²; between 100 cm² and 2 m² it asks for at least one. A 0.5 m² article is therefore three specimens under A123 and as few as one reference area under ISO 1461.

Take a lot of 120 structural beams of 12 mm steel, each about 3 m², one material category throughout.

Same forty-five readings, two different verdict rules. The arithmetic you do afterwards is what changes, which is why raw readings should be kept, not just the averages.

Placement matters as much as count. Unless otherwise agreed, ISO 1461 leaves the choice of reference areas to the galvanizer. On a long article they sit near the centre and about 100 mm in from each end, and no reading is taken within 10 mm of an edge, a flame-cut surface or a corner. A123 spreads readings for the widest dispersion across each specimen. If the purchaser cares where the gauge goes, agree the positions, or mark them, before the steel is dipped.

Measuring with a magnetic gauge

The magnetic gauge is the working instrument under both standards: ASTM E376 governs its use with A123, and ISO 2178 governs it with ISO 1461. The AGA lists three types. A pencil gauge depends on the skill of the inspector, so readings are repeated. A banana gauge works in any position without gravity interference. An electronic gauge is the most accurate and stores and averages readings.

The rules on where to place the probe are not identical.

• ISO 1461: at least five readings per reference area, a reference area of at least 10 cm², and no reading within 10 mm of an edge, a flame-cut surface or a corner.
• The AGA's summary of E376: stay away from edges and holes, keep at least 4 in. (about 100 mm) from the edge on large products, avoid curved surfaces where possible, read regular areas of the coating, disperse readings widely and take at least five.

The edge exclusions differ by a factor of ten: 10 mm under ISO 1461, about 100 mm in the AGA's summary of E376. When one inspection has to satisfy both standards, work to the larger distance.

Curvature deserves its own warning. The AGA notes that magnetic gauges are very accurate on flat surfaces and inaccurate on curved ones, which matters on poles, pipe and rebar. Choose a probe that fits the geometry, read the flattest regions and take more readings, not fewer.

Four habits protect the number:

1. Verify the gauge before the session and again after it, on reference foils over bare steel of the same grade and similar thickness as the work. If the check drifts, the session's readings are suspect. KTA-Tator makes the same point about verifying accuracy before and during use.
2. Read the coating as finished: clean, dry and cool. Avoid runs, drips, dross pimples and lumps, since the coating there is not regular.
3. Keep every raw reading, with the article ID, material category, measured steel thickness, specimen or reference area, and gauge ID. The same sheet can then be judged under A123 or ISO 1461, and a dispute becomes arithmetic.
4. Match the gauge tolerance to the margin you are trying to manage. If the plant aims to sit 10 µm above the minimum, a gauge whose tolerance is a large fraction of 10 µm cannot tell you whether it got there.

When the gauge is not enough: mass, microscopy and the referee

A gauge reading is an estimate of thickness. When it is disputed, or the article is too small or too curved for a probe, the standards fall back on other methods, and the two standards do not name the same referee.

• Weigh-galvanize-weigh. Non-destructive and suitable for single-specimen articles only. It sees the zinc added by the dip, and the AGA reports that it underestimates the total coating mass by up to 10%, since the coating also holds iron drawn from the steel.
• Weigh-strip-weigh (ASTM A90). Destructive and, again, single-specimen only. The article is weighed after cooling and again after the coating is stripped in acid, and the loss is divided by the steel surface area. The part is spoiled.
• Microscopy (ASTM B487). A destructive cross-section that gives thickness at one point. The AGA describes it as a method used to resolve measurement disputes, but A123 section 8.2.4.1 calls it appropriate for single-specimen articles and not practical for multi-specimen ones. A123's own referee, in section 8.2.5, is a new random sample with twice the test articles.
• Gravimetric test (ISO 1460). The referee under ISO 1461 clause 6.2.2, with a nominal density of 7.2 g/cm³ used to calculate thickness. The magnetic method is the everyday test. Where fewer than 10 articles are involved, the purchaser does not have to accept a gravimetric test that would destroy articles at unacceptable remedial cost.

The practical rule is to name the referee in the order, and to say who pays for the article sacrificed. A disagreement over gauge readings cannot be settled by more gauge readings, and a supplier who works to ISO 1461 and a client who expects microscopy will each believe they hold the last word.

Mass or thickness: three conversion factors in circulation

The standards do not convert microns to grams per square metre with the same factor. Dividing each published pair shows it.

The coating is layers of iron-zinc alloy under a layer of free zinc, not a slab of pure zinc, so no single factor is exact. The spread between the three is about 2%. The ISO 1461 mean minimum of 610 g/m² converts to 84.7 µm at 7.2, 85.4 µm at 7.14 and 86.3 µm at 7.067.

That is inside gauge tolerance, so it rarely changes a verdict. It still means a mass-specified order and a thickness-specified order are not the same contract at the margin. Choose one unit for the order, and if you must convert, state the factor.

For plant zinc planning, the 7.14 convention used in the Zinc Consumption KPIs article is fine. The error is about 1% of product zinc, well below the kettle level correction and the weighing errors that article warns about.

Minimum is a floor, not a target

Every micron above the minimum is zinc the customer did not specify. The Zinc Consumption KPIs article showed that average coating thickness is the largest controllable zinc cost. A 140 µm average where Grade 100 is specified is compliant and carries 40% more coating zinc than was specified, and bringing it to 115 µm cuts product zinc by about 18%.

The price of a micron follows from that article's own formulas. Surface area per tonne is about 255 divided by the average section thickness in mm, so 8 mm work carries about 31.9 m²/t.

On 8 mm work, each 10 µm of unintended excess is 2.3 kg of zinc per tonne, or 0.23 points of zinc consumption. On a plant galvanizing 500 t a month that is about 1,140 kg a month, or 13.7 t a year, for every 10 µm.

To manage it, keep a thickness gap log. For every inspected batch, record the gap: the average reading minus the specification minimum, in microns, by material category. Then:

1. Set a target gap per category from your own scatter. It should be wide enough to keep the lowest specimen average above the minimum on nearly every batch, say 19 in 20, and no wider.
2. Investigate any category whose median gap sits far above that target. The levers are withdrawal rate and drainage, which set how much free zinc leaves with the article, and immersion time beyond thermal equilibrium, which buys coating nobody pays for.
3. Keep reactive steel out of the statistics. Steel in the Sandelin range, silicon about 0.06 to 0.13%, will sit far above the minimum whatever the process does. Identify it from the mill certificate and price it separately.

The boundary condition from the KPI article applies here too. Thickness compliance stays at 100%. The target is the unintended excess only, and a narrow margin is only safe if the gauge is good enough to see it.

The maximum-thickness trap

Neither standard sets a maximum coating thickness. The AGA is explicit that the hot-dip galvanizing specifications set none, and ISO 1461 is written the same way: it states floors, and anyone who wants a heavier coating has to ask for one.

Purchase orders add a ceiling for understandable reasons: thread fit, mating surfaces, fear of flaking, appearance. The clause fails because the galvanizer controls only part of the result. Steel chemistry, section thickness, surface condition and immersion time all drive thickness, and a reactive heat can land far above any contractual maximum whatever the galvanizer does. A ceiling moves the risk of the steel's chemistry from the purchaser to the galvanizer, usually without a price attached.

There is a real limit behind the instinct. The AGA puts it at about 250 µm (10 mils). Above that, steel and the intermetallic layers contract at different rates on cooling, the stress at the layer interfaces rises, and thick members such as beams and poles can flake under handling, assembly or shipping. Sheet steel cools fast and rarely shows it. Coating that heavy also carries two to eight times the zinc that is needed.

What to write instead of a ceiling:

1. Control the cause. Specify steel chemistry at procurement and ask for the mill certificate, so reactive steel is identified before it reaches the kettle.
2. Specify the fit, not the thickness. Thread clearance, mating surfaces, and drips or runs that interfere with assembly must be dealt with. The AGA notes that the coating must suit the intended use, and that runs causing a fit problem are ground smooth.
3. Agree a flaking check for heavy members. Where a thick coating is expected, measure it and test for flaking before dispatch, as the AGA recommends, not after the steel is on site.
4. Price a thickness range as a special requirement. If a narrow range really matters, agree it with the steel, the price and the inspection method. Do not leave it as a rejection criterion the galvanizer discovers after the dip.

Writing the specification clause

A thickness clause that settles arguments in advance answers ten questions.

1. Standard and edition. ASTM A123/A123M-24 or ISO 1461:2022. Superseded editions still circulate in purchase orders. If both standards are cited, say which prevails on conflict.
2. Category and thickness basis. Material category assigned by method of manufacture, measured steel thickness as the basis, and drawings supplied where closed sections cannot be measured.
3. Anything heavier than the standard. Stated as a special requirement, with a grade or a minimum average in microns. "Heavy coating" is not a requirement.
4. Lot and sample. How the lot is defined, who draws the sample, and that it is drawn at random.
5. Test method. Magnetic gauge to ASTM E376 or ISO 2178, with calibration evidence, and the raw readings supplied with the report.
6. Positions. Where specimens or reference areas go, when it matters, agreed or marked before dipping. Say whether ancillary elements are excluded.
7. Referee. The method that settles a disputed reading: resampling under A123 section 8.2.5, the gravimetric test to ISO 1460 under ISO 1461, or microscopy where the articles are single-specimen. Say who bears the cost of any article sacrificed.
8. Failure. Resample first, with twice the articles under either standard, then repair or re-galvanize on agreed terms.
9. No maximum. Fit and flaking requirements stated separately.
10. Who inspects, where and when. At the plant before dispatch, third party or witness, and the declaration of compliance. ISO 1461:2022 uses that term and ties it to ISO 10474.

A model clause that covers the structure, to adapt to the contract:

Tools, related reading and sources

The Coating Thickness Finder, the Inspection Sampling Planner and the Zinc Coating Unit Converter on the free tools page work the same numbers interactively.

Related reading in this library:

• Zinc Consumption KPIs: How to Measure Plant Efficiency
• ASTM A123 and ISO 1461 New Editions: What Changed
• Corrosion Protection for Steel: Every Method Compared to Hot-Dip Galvanizing
• Vent and Drain Holes for Hot-Dip Galvanizing
• Standards Library and Defect Gallery

Standards cited, from the author's licensed copies:

• ASTM A123/A123M-24: sections 6.1, 7.2, 7.3 and 8.2, Tables 1 and 2, Appendix X1
• ISO 1461:2022: clauses 5, 6.2.2, 6.2.3 and 6.5, Tables 1 to 3

Further reading, pages opened on 2026-09-30:

• AGA, 2024 Revision of ASTM A123: Table 1, Appendix X1, "one coating grade below"
• AGA, Sampling
• AGA, Coating Thickness/Weight
• AGA, Maximum Coating Thickness
• KTA-Tator, Measurement of the Thickness of Hot Dipped Galvanizing
• V&S Galvanizing, Material Category Specification for Galvanized Structural Pipe and Hollow Structural Sections
• Sperrin Galvanisers, A Guide to EN ISO 1461

The figures apply to batch galvanizing of fabricated articles. Product standards, centrifuged work, fasteners and any contract that modifies the standard need their own check, and the edition named in the contract governs.

© Aladdin Mohammed · HDG Community. Requirements are quoted from ASTM A123/A123M-24 and ISO 1461:2022 for batch galvanizing of fabricated articles. They do not replace the governing standard, the project specification or the edition named in the contract.