TIGERFASTENERS
A 10.9 on the Head Is a Claim, Not a Certificate
The number stamped on a bolt head feels like proof. It is a statement by whoever made or badged the part, and knowing exactly how much it asserts — and how little it verifies — changes what you ask for.
What the standard actually requires
ISO 898-1 does not treat the marking as decoration. Clause 10.1 permits the grade marking system to be used only when all the relevant requirements of the standard are met, and clause 10.2 requires the manufacturer's identification mark alongside the grade — with a distributor selling under its own mark treated as the manufacturer.
So the mark is a conformity claim carrying an accountable name. What it is not is independent verification: nothing in those clauses requires a third party to witness or certify anything. A counterfeiter can physically stamp the same characters, and that act is false marking rather than something the standard permits.
One thing worth stating carefully, because it is often mis-cited: ISO does not contain a sentence saying the mark is not proof of conformity. It says the marking may only be used when the requirements are met. The correct reading is that a mark is a claim and an identification, not that ISO disclaims it.
Most head styles are not required to be marked at all
Marking is mandatory for a narrower set of products than people assume:
| Product | Marking required |
|---|---|
| Hex and hexalobular bolts and screws | all classes, d ≥ 5 mm |
| Hex and hexalobular socket head cap screws | all classes, d ≥ 5 mm |
| Cup head square neck bolts | all classes, d ≥ 5 mm |
| Studs | only 5.6, 8.8, 9.8, 10.9, 12.9 and d ≥ 5 mm |
| Countersunk, raised countersunk, pan heads | usually not marked unless the purchaser requires it |
That last row matters more than it looks. An unmarked countersunk screw is not evidence of anything wrong — see what the markings actually mean. But it does mean the head carries no claim at all, so the claim has to live somewhere else.
It does: clause 10.5 requires the packaging of every size and type to carry the manufacturer or distributor identification, the property class, and the manufacturing lot number. For unmarked heads, the box is the record, which is why splitting a lot into an unlabelled bin destroys the only traceability you had.
The scale this reached once
In the late 1980s a US congressional investigation into counterfeit fasteners led to the Fastener Quality Act, signed in November 1990. The numbers from that period are worth knowing: the 1988 report described more than 30 million counterfeit fasteners found in Defense Industrial Supply Center inventory, with a further 2.6 million at Army depots.
Two cautions on those figures. They are counts found in inventory, not seizure rates or failure rates. And although the report said counterfeit fasteners may have been responsible for deaths and injuries, there is no reliable national total attributable to them — not every fastener failure is a counterfeiting incident.
The law was substantially amended in 1999. It is usually described as having been weakened, and that is half the story: the amendment narrowed the product scope considerably and dropped universal lot testing and certificate flow-down, but it also added explicit prohibitions on knowingly falsifying conformity records and manufacturer marks. It remains in force, and it covers far less than people assume — scope is defined by product characteristics, with exclusions including fasteners already in assemblies, small repair kits, and FAA-approved aviation fasteners.
What testing can and cannot settle
The instinct is to reach for a hardness tester, and it is a reasonable screen but not a verdict. ISO 898-1 makes tensile testing the referee where it can be performed; hardness is the referee only for bolts and screws that cannot be tensile tested.
The reason is concrete. A low-carbon boron steel can be heat treated to the same room-temperature hardness as a medium-carbon steel of a different grade, so an identical hardness reading does not identify the material or tell you how it behaves hot. Hardness sorts; it does not identify.
- Tensile, wedge and proof load — verifies strength and some ductility directly; the core where testing is possible
- Hardness — fast screening of heat treatment and strength band, not generally decisive
- Chemical analysis — confirms alloy and element limits, but not post-heat-treatment mechanical properties
- Metallography — confirms microstructure, decarburisation, carburisation, heat-treatment anomalies
- PMI, XRF, magnetic particle, penetrant, ultrasonic — useful screening, but no single one proves a whole property class
And the opposite overstatement is also wrong: visual inspection is not useless. It finds missing manufacturer marks, duplicated or hand-stamped characters, mixed lots and dimensional or surface anomalies. It just cannot confirm the grade. Treat it as the cheapest screen, not as a verdict.
The documentation chain, and where it does not help
The usual advice is to demand certificates, and it points the right way — but a common version of it is wrong. The EN 10204 3.1 versus 3.2 distinction is not the dividing line here. 3.1 is validated by the manufacturer's authorised inspection representative, independent of manufacturing; 3.2 adds countersignature by the purchaser's representative or a regulatory inspector.
That difference is who validates the document. It does not by itself determine whether the sampling was representative, whether the tests performed were the right ones, whether the lot was mixed, or whether the document is genuine. Asking for a 3.2 is not an unconditional guarantee.
Documents get falsified too, and the enforcement record is not hypothetical: a 2016 US case concerned a manufacturer certifying that required magnetic particle and penetrant inspections had been performed when they had not, settled at 2.7 million dollars, and a 2023 case involved contract testing on nuts and bolts that was not carried out.
So the sequence that actually works is: let the product or purchase specification decide what gets tested and how a lot is defined, then choose 3.1, 3.2 or independent laboratory testing according to risk, and keep the lot traceability the packaging was required to carry. What to check on the certificate itself is covered in reading a mill certificate.
One maintenance document goes further and says what an absent mark means: bolts with no markings are low strength, written as an identification rather than an unknown.
The same distinction appears far from steel. On a cable tie, the strength figure is defined as a reference characteristic declared by the maker, with a conformity class saying how much of it survives testing.
This page covers step 6, the documentation. The whole order is substrate, thread, head, drive, finish, documentation, and why doing it out of order is rework rather than a tweak is in specifying a screw.
Common questions
Can anyone legally stamp 10.9 on a bolt head?
No. ISO 898-1 permits the grade marking system to be used only where all the relevant requirements of the standard are met, and requires the manufacturer identification mark alongside the grade, with a distributor selling under its own mark treated as the manufacturer. A counterfeiter can physically apply the characters, but that is false marking rather than something the standard allows. What the mark is not is independent verification — nothing requires a third party to witness or certify it.
My countersunk screws have no grade marking. Is that a problem?
Not by itself. Countersunk, raised countersunk and pan head screws are usually not required to be marked at all unless the purchaser specifies it; mandatory marking applies mainly to hex and hexalobular bolts and screws, socket head cap screws and cup head square neck bolts at 5 mm and above, and to studs only in certain classes. What does apply to every size and type is the packaging marking: manufacturer or distributor identification, property class and manufacturing lot number.
Is a hardness test enough to check a suspect bolt?
Usually not. ISO 898-1 makes tensile testing the referee wherever it can be performed and leaves hardness as the referee only for bolts and screws that cannot be tensile tested. The concrete reason is that a low-carbon boron steel can be heat treated to the same room-temperature hardness as a medium-carbon steel of a different grade, so an identical reading does not identify the material or predict its behaviour at temperature. Hardness is a good screen and a poor verdict.
Does asking for an EN 10204 3.2 certificate solve this?
It helps but is not the dividing line people think. The difference between 3.1 and 3.2 is who validates the document — the manufacturer inspection representative independent of production, or additionally the purchaser representative or a regulatory inspector. It does not determine whether sampling was representative, whether the right tests were run, whether the lot was mixed, or whether the document is genuine. Certificates have been falsified in prosecuted cases, so let the specification define the tests and lot first, then choose the certificate type by risk.
References
- ISO 898-1 — Mechanical properties of fasteners: marking clauses 10.1–10.5, test clauses 8 and 9 (referee rule at 9.13.5)
- GAO-01-719 — Fastener Quality Act background, including the 1988 congressional report figures
- 15 U.S.C. §5402 — Fastener Quality Act definitions and scope exclusions
- NASA RP-1228 — Fastener Design Manual, including hardness equivalence between different grades
The 30 million and 2.6 million figures are quantities found in inventory as reported in 1988, not seizure or failure rates, and no reliable national casualty total is attributable to counterfeit fasteners. No verifiable case of a forged EN 10204 mill certificate for fasteners was found; the prosecuted cases cited concern inspection and test certifications.
Enquiries
If you are qualifying a new source, tell us what the part actually has to survive rather than only the class you want stamped on it. The class is what someone claims; the test plan and the lot definition are what you can check.