Hallmarks and Assay: How Anyone Proves Gold Is Gold

Gold has one commercial weakness, and everything in this article exists to address it: you cannot tell how much gold is in an object by looking at it. Colour can be alloyed. Weight can be faked with tungsten, which is inconveniently close to gold in density. Ring, feel and polish can all be reproduced. A buyer standing in front of a bracelet has, unaided, almost no way to distinguish 750 fineness from 375 with a good finish.

Societies that traded gold seriously worked this out early and built an institution rather than a gadget. The hallmark — a set of small punched symbols applied by a body with no stake in the sale — is one of the oldest continuously operating consumer protection systems in existence, predating trademark law, product labelling and every modern certification scheme by centuries.

What the marks say

A full traditional hallmark is a compact record. It typically carries a sponsor or maker's mark identifying who submitted the article; a fineness mark stating gold content in parts per thousand or karats; an assay office mark showing which office tested it; and, in some regimes, a date letter identifying the year of marking. Read together they answer four questions: who made it, how pure it is, who verified that, and when.

The date letter is the detail that reveals the design philosophy. A system that only certified purity would not need one. A system built to be auditable after the fact — to allow a piece to be traced to an office and a year decades later, when the maker is dead and the shop is gone — does. The hallmark is not a label. It is a record kept in the object itself.

Karat, fineness, and why 24ct is a fiction in practice

Karat divides gold content into twenty-fourths; fineness states it in parts per thousand. They are the same measurement in different clothing. Eighteen karat is 750 fineness, fourteen karat is 585, nine karat is 375. Investment bars are marked 995, 999 or 999.9 — the last of these, four nines, is the practical ceiling of industrial refining rather than a claim of literal chemical purity.

Alloying is not adulteration. Pure gold is too soft for most objects that get worn: a 24ct ring deforms, scratches and loses detail within months of daily use. Copper hardens and reddens; silver softens the colour toward green; palladium and nickel produce white golds with very different working properties and, in the case of nickel, a well-documented contact allergy problem that drove regulation across Europe. The alloy is an engineering decision, and the fineness mark is the honest disclosure of it.

Fire assay: destroying the sample to tell the truth

The referee method has barely changed in principle since the sixteenth century. A precisely weighed sample is wrapped with silver and lead and heated in a porous cupel. The lead oxidises and, along with base metals, is absorbed into the cupel wall, leaving a bead of gold and silver. The bead is treated with nitric acid to dissolve the silver — the step traditionally called parting — and the remaining gold is weighed. Fineness is the ratio of the final weight to the original.

Its authority comes from what it does not assume. It measures the entire sample, not a surface. It does not need calibration against reference standards of similar composition. It is slow, destructive, requires a skilled analyst, and consumes the material tested — and it is still what everything else is checked against.

XRF, and the honest limits of instant answers

X-ray fluorescence spectrometry is the workhorse of modern testing. Irradiate a sample, read the characteristic secondary X-rays each element emits, infer composition in seconds without damaging the piece. For a dealer processing a queue of walk-in sellers, it is transformative.

Its limitation is physical and unavoidable: XRF reads a shallow surface layer. A gold-plated base-metal item, a gold-filled item with a thick bonded layer, or a piece deliberately depletion-gilded so that its outer skin is richer than its core will all read high. Skilled operators know this and combine XRF with density checks, magnetism, ultrasonic thickness measurement, or a scratch through the surface before touch-testing with acids. An unskilled operator with the same instrument produces a fast, confident, wrong number.

• Touchstone and acid: a streak of the item on a basalt stone, compared with streaks from known needles under graded acids. Cheap, quick, surface-limited, still in daily use.
• Density by displacement: catches tungsten poorly and hollow fill well; useful as a cross-check, never conclusive alone.
• Ultrasonic thickness: detects internal voids and inserts in bars, which is why serious bullion desks own one.
• Fire assay: destructive, definitive, and the only method that ends an argument.

The erosion nobody advertises

Hallmarking's authority depends on a chain: compulsory submission, independent testing, and enforcement against unmarked or falsely marked goods. Each link is under pressure. Cross-border online sales route around jurisdictional requirements. Counterfeit punch sets circulate. Very light articles are exempt from marking in many regimes, and a great deal of contemporary jewellery is very light. Where enforcement funding falls, the mark degrades from a verified statement into a decorative convention.

That erosion is gradual and largely invisible to buyers, which is precisely the danger. A system like this works because it is boringly reliable; it stops working the moment the average buyer starts assuming a mark might be decorative. Nothing about the mark itself changes at that point. Only the belief does, and the belief was the product.

There is something clarifying about a technology that has survived five centuries of financial innovation without needing to be reinvented. The hallmark solved a specific, permanent problem — an object whose value cannot be observed — with an independent third party, a physical record and a standing threat of prosecution. Every digital provenance scheme now proposed for the gold supply chain is attempting the same trick with different tooling. The old system's advantage is that its record is punched into the metal, and the metal does not need a server.