How Fake Gold Is Detected

The popular image of gold fraud is a plated brick — something that looks right and fails the first serious test. Real counterfeiting is more disciplined than that, because the counterfeiter is not trying to fool a person. They are trying to fool a specific test at a specific point in the chain, and they know which test that is.

That reframing explains everything about how authentication has developed. Each defence closed a route; each closure pushed forgers towards a harder and more expensive one. The current state of play is an equilibrium in which sophisticated fakes are technically possible and economically marginal.

Why tungsten, and why it broke the old test

For most of history, density was sufficient. Gold is extraordinarily dense at 19.32 grams per cubic centimetre, and nearly everything cheap enough to substitute — lead at 11.3, copper at 8.96, brass lower still — is dramatically lighter. Weigh the object, measure its displacement, divide. A lead-cored bar of the right weight is visibly too large; one of the right size is far too light.

Tungsten ends that comfort. At 19.25 g/cm³ it sits within a third of one percent of gold, comfortably inside the error band of any field density measurement. It is cheap, industrially available, and — this is the part that matters — machinable into a core that can be encased in genuine gold. The resulting object has the correct mass, the correct dimensions, the correct density and a completely genuine surface.

The layered approach

No dealer runs one test. Authentication is a sequence, ordered by cost and speed, in which each stage is designed to catch a different class of forgery. A bar that reaches the end of the sequence has been examined by methods with independent physical bases, which is the actual source of confidence — not any single instrument.

• Visual and provenance check: refiner marks, serial number, casting texture, and whether the bar has a documented custody record. Most fakes fail here, on details a specialist notices instantly.
• Dimensional gauging: calipers to a hundredth of a millimetre against published mint or refiner specifications.
• Precision weighing: to 0.01 g, cross-checked against the stated fineness and nominal mass.
• XRF spectrometry: surface elemental composition in seconds, non-destructive.
• Ultrasonic velocity: transit time through the object, the primary defence against core filling.
• Conductivity or magnetic-response testing: eddy-current and related methods that probe deeper than XRF.
• Fire assay: destructive, definitive, and reserved for settlement disputes and refinery intake.

XRF: fast, non-destructive, and shallow

X-ray fluorescence works by exciting inner-shell electrons in the sample and reading the characteristic X-rays emitted as the atoms relax. The spectrum identifies elements present and quantifies them well. It takes seconds, requires no preparation, and damages nothing, which is why every serious dealing room owns one.

Its limitation is not a calibration problem to be solved but a physical one. The exciting radiation penetrates only tens of microns into a dense metal, and the fluoresced X-rays that escape come from the same shallow layer. A tungsten core under two millimetres of genuine gold produces a perfect 9999 reading, because the instrument is genuinely looking at 9999 gold — all of the gold it can see.

Ultrasound: listening to the inside

The technique that actually addresses core filling borrows from industrial non-destructive testing. A transducer couples to the bar's surface and emits a pulse; the instrument times the echo from the far face. Because thickness is known from the calipers, transit time yields the speed of sound through the material.

Gold transmits longitudinal sound at roughly 3,240 metres per second. Tungsten manages over 5,100. That is not a subtle difference requiring statistical judgement — it is a failed reading. A layered bar also produces an internal reflection at the interface between shell and core, which shows up as an anomalous intermediate echo. The test takes under a minute and costs a fraction of an assay.

Coins are a different problem

Bullion coins are, counter-intuitively, harder to counterfeit convincingly than bars, and the reason is manufacturing tolerance. A sovereign mint strikes to specifications published to a hundredth of a millimetre and a hundredth of a gram, from dies whose relief and lettering are executed at a standard that is expensive to reproduce.

A forger must therefore match diameter, thickness, mass and design simultaneously, in an object too thin to hide a useful core. Slide gauges — machined blocks that accept a genuine coin and reject anything outside tolerance — catch the overwhelming majority of fakes in seconds. What remains is die-quality forgery, which requires real tooling investment and tends to appear in numismatic rather than bullion markets, where a premium above metal content makes the effort worthwhile.

Fire assay, and why the old method survives

Everything above is a proxy. The reference method is cupellation, essentially unchanged in principle since antiquity. A drilled or cut sample is wrapped with silver and lead and fused in a bone-ash cupel; the lead oxidises and carries base metals into the porous cupel wall, leaving a precious-metal bead. The bead is parted in nitric acid to dissolve the silver, and the remaining gold is annealed and weighed.

It destroys the sample, takes hours, and is the only method that measures the entire sample rather than a surface, an echo or an electromagnetic response. Refineries assay on intake for exactly this reason, and every non-destructive instrument in the chain is ultimately calibrated against results obtained this way.

The practical conclusion

Sophisticated fakes are real, and they are also rare in the wholesale market, because the chain of integrity means that most institutional metal never leaves accredited custody and therefore never needs re-verification. Fraud concentrates where the chain is broken: private sales, informal dealers, and marketplaces where the buyer's only test is a photograph.

The defence that costs nothing is structural rather than technical. Buy within the accredited chain, keep the documentation, and treat any deep discount to spot as what it almost always is — a price that reflects a risk the seller understands better than the buyer does.