How Are Fake Gold and Silver Coins Caught? What Each Test Can and Cannot See

Tungsten is within 0.4% of gold’s density, so a gold-plated tungsten fake can match a real coin’s weight and size. We explain what the scale, calipers, magnet, ping, XRF, eddy-current and ultrasound tests can and cannot see, from the mints’ and the LBMA’s own documents, and why serial numbers and the seller matter as much as any instrument.

Published October 21, 2026 · 16 min read

Key Takeaways

• Tungsten is within 0.4% of gold’s density (NIST: 19.30 against 19.32 g/cm³; Los Alamos: 19.25 against 19.3), so a gold-plated tungsten fake can match a genuine piece’s weight and size. Lead, 41% lighter than gold, cannot: a lead Gold Maple Leaf of the right weight and diameter would be about 70% thicker.
• A magnet catches only a magnetic core, such as iron or steel; tungsten and lead are not drawn to one either. An XRF analyser reads less than 10 microns into gold, the LBMA says, so plating can fool it; eddy-current testers read below the surface to a limited depth, and ultrasound finds inserts but says nothing about fineness.
• Weigh a coin against its own mint’s figure: a 1 oz American Gold Eagle weighs 33.931 g and a Krugerrand 33.930 g, because both are 22-carat, against 31.11 g for a Gold Maple Leaf. None of the four mints’ pages gives a thickness.
• Reuters reported in August 2019, citing refinery and bank executives, that at least 1,000 kilobars worth at least $50 million carried forged Swiss refinery logos on real gold, two of three tested at 99.98% pure; word of them spread after JPMorgan found two bars with the same identification number.
• The LBMA says there does not appear to be a definitive non-destructive test, and that the best way to reduce the risk remains its Good Delivery system of refineries and chain of custody.

In the three years to August 2019, Switzerland’s four leading gold refiners identified kilobars worth at least $50 million stamped with Swiss refinery logos that no Swiss refinery had made, Reuters reported, citing executives at refineries and banks; four of them put the number at no fewer than 1,000 bars. Word of them had begun to spread in 2017, people familiar with the matter told Reuters, when JPMorgan found at least two kilobars in its vaults stamped with the same identification number. The gold inside was real: one refinery’s analysis of three found two at 99.98% pure and the third at 99.90%, short of the 99.99% that is the most common standard for such bars. A scale or a magnet would have passed them; what gave them away was a repeated number and, refiners said, a test of purity. The fakes the Perth Mint warns retail buyers about are the reverse, cheap metals such as tungsten or lead under a thin coat of gold or silver, and those are what the tests on a dealer’s bench are for.

This explainer takes the tests in turn: what a fake has to imitate, why tungsten is the filler for gold, what the magnet and the ping can and cannot tell you, and how deep an XRF analyser, an eddy-current tester and ultrasound each see. It then turns to the mints’ own security features, to assay cards and serial numbers, and to the seller. It describes how testing works; it does not recommend any product, dealer or device.

What Does a Fake Coin Have to Get Right?

A fake coin has to match the figures the mint publishes for that exact coin, its weight and its diameter, as well as its design and finish. Weight is the first check and the first trap. A 1 oz American Gold Eagle weighs 33.931 g and a Krugerrand 33.930 g, against 31.11 g for a Gold Maple Leaf, because both are 22-carat: the United States Mint gives the Eagle as 91.67% gold, 3% silver and the rest copper, and the South African Mint gives the Krugerrand as 916.67 parts gold to 83.33 parts copper. 91.67% of 33.931 g is 31.10 g, one troy ounce of gold, so the heavier coin is the correct one. A coin is weighed against its own mint’s figure, not against “an ounce”. The Academy’s guide to the world’s mints and their coins compares the coins themselves.

None of the four mints’ pages gives a thickness for these coins, and the Royal Mint’s pages give no weight in grams for the Britannia, only its troy ounce of fine metal. Weight and diameter are the published figures to check; for thickness the reference is a genuine coin of the same type, measured with the same calipers. The three measurements matter together because, at a fixed weight and diameter, a coin in a lighter metal has to be thicker and one in a heavier metal thinner. That is where density comes in.

Why Is Tungsten Used to Fake Gold?

Tungsten is used because its density is almost exactly gold’s, so a tungsten core plated in gold can match both the weight and the size of a genuine piece. NIST’s reference table gives gold 19.32 g/cm³ and tungsten 19.30; Los Alamos National Laboratory’s gives 19.3 and 19.25. On either, the two are within 0.4% of each other. Los Alamos makes the point in its own entry for tungsten: a brick of tungsten coated with a thin layer of gold will look and weigh almost the same as a real gold bar.

The other cheap metals miss by far more. Lead, at 11.35 g/cm³, is 41% lighter than gold, so a gold-plated lead coin of the Gold Maple Leaf’s weight and 30 mm diameter would have to be about 70% thicker than the real one. Silver is easier to imitate, because its own density, 10.50, sits between lead’s and copper’s (8.96). At a Silver Eagle’s weight and 40.60 mm diameter, a coin of lead would be about 7.5% thinner than the real coin and a coin of copper about 17% thicker: differences calipers find when there is a genuine coin to compare against.

Does the Magnet Test Work on Gold and Silver?

A magnet catches one kind of fake: a magnetic core, such as iron or ordinary steel. Gold is not drawn to a magnet, as the Royal Mint notes in its guide to telling real gold, and neither is silver. But, as the same guide adds, plenty of other metals are not drawn to one either, and tungsten and lead, which the Perth Mint gives as examples of the metals fakes are made of, are among them. A coin that ignores a strong magnet has ruled out a magnetic core and nothing more. A coin that is pulled towards one is not solid gold or silver, whatever is stamped on it.

What Does the Ping Test Tell You?

The ping test, balancing a coin on a fingertip, tapping it and listening to it ring, compares the coin with the sound of a genuine one. The tone depends on the coin’s metal, its mass and its shape, which a 2026 study of automated coin testing calls the coin’s acoustic fingerprint, so a coin of the wrong metal can ring differently from a genuine coin of the same type. By ear, it is a judgement made against a reference. The study’s authors, who analyse the sound with deep neural networks instead, write that the growing sophistication of counterfeits has made traditional inspection methods in general insufficient; their paper is a preprint on arXiv.

What Can an XRF Analyser, an Eddy-Current Tester and Ultrasound Actually See?

Each instrument sees to a different depth, and the London Bullion Market Association (LBMA), which sets the standards for bars traded in London, says that none of them settles the question alone: there does not appear to be a definitive non-destructive test that it can endorse. The only way to measure a bar’s fineness accurately, its guidelines say, is to take a representative sample and assay it, which destroys the bar. What the three common instruments do and do not see:

• XRF (X-ray fluorescence) reads the surface. The LBMA’s guidelines say X-rays penetrate gold generally less than 10 microns, so the method “can be deceived by simple plating”, and that a surface reading can verify fineness only if the piece is known to be the same all the way through. The Royal Mint calls XRF one of the most widely used methods: it reads what a surface is made of, and a plating thicker than its reach hides what lies beneath.
• An eddy-current or conductivity tester reads below the surface. Its probe induces electrical currents inside the metal and measures its conductivity, which the LBMA notes is changed by alloying metals, inclusions, porosity and foreign bodies. But the LBMA’s guidelines call it “not definitive”: it is only accurate to a certain depth, and it tests only the patch under the probe.
• Ultrasound looks through a bar. A probe on the top surface sends sound down and listens for echoes; the LBMA’s note on the method says any echo other than the one from the bottom surface probably means a foreign body or an internal pore, and because sound travels at very different speeds in gold and tungsten, the timing of the bottom echo shows whether the bar is gold at all. Its guidelines call ultrasound effective at finding inserts and air pockets, but it gives no information on fineness, and poor contact between probe and bar can produce false readings. Most London vaults installed it in the early 2000s, when tungsten inserts were the worry; the LBMA says such inserts have been found in small bars but never in the large bars held by central banks and in the London Good Delivery system.

Under 10 microns How far an XRF analyser’s X-rays generally reach into gold, in the LBMA’s guidelines. A hundredth of a millimetre: a reading of the surface, which a plating can fool. The same guidelines call eddy-current testing “not definitive” and say ultrasound gives no information on fineness.

Which Security Features Do the Mints Build Into Their Coins?

Three of the four mints in the table, and the Perth Mint, describe features built into their bullion to make it harder to copy. As each describes them:

• Royal Canadian Mint, Maple Leaf: a textured maple leaf micro-engraved by laser on a small area of the reverse, with the year at its centre (“26” on a 2026 coin) visible under magnification; radial lines in the field machined to within microns on the master tooling; and Bullion DNA, which the mint says facilitates the authentication of Gold Maple Leafs dated 2014 and later and Silver Maple Leafs dated 2015 and later. Every die used to strike them is laser micro-engraved with an anti-counterfeiting mark.
• The Royal Mint, Britannia: four features, which it lists as a latent image that changes from a padlock to Britannia’s trident, surface animation in the waves behind her, micro-text, and tincture lines on the Union flag on her shield.
• United States Mint, American Eagle: the redesigned one-ounce gold and silver Eagles of 2021 added anti-counterfeit features, the most visible of which, the mint says, is a variation in the reed pattern on the edge of the coin.
• The Perth Mint: a micro-laser-engraved letter on all its bullion coins, detectable only under a magnifying glass, and minted bars sealed in tamper-evident packaging that shows the word VOID around the edge of the blister if the card has been prised open.

Features raise the cost of a convincing copy; they do not make one impossible. “Any security feature can be duplicated that’s on the bar itself,” the LBMA’s chief executive, Ruth Crowell, told Reuters in 2019. The LBMA had also proposed, Reuters reported, a record kept off the bar: a global database of every kilobar produced, to cross-check the bars against. It has since built one: its journal, the Alchemist, describes a Gold Bar Integrity database that tracks the movement and provenance of gold bars, using data from London’s vaults and the refiners on its Good Delivery List.

Do Assay Cards and Serial Numbers Prove a Bar Is Real?

An assay card or a serial number is evidence of where a bar came from, and it is worth as much as the seal around it and the record behind it. The LBMA’s Good Delivery rules for large bars require the refiner’s stamp, an assay mark, the fineness, a serial number and, since January 2019, the month and year of manufacture. The kilobar specification endorsed by the LBMA and the Shanghai Gold Exchange requires a certificate of analysis to travel with each bar, carrying its serial number, weight, fineness and an approved signature, and it says serial numbers must be unique, never reused and referenced against inventory lists. That is the logic of the JPMorgan discovery: two bars with one number cannot both be what they claim.

For a retail buyer the same logic is smaller. A minted bar sealed in its card or blister carries the refiner’s word only while the seal is intact, and a serial number is most useful when the refiner or the dealer can check it against their records. Paperwork also varies by product: the Perth Mint says its collector coins come with a numbered certificate of authenticity and its bullion coins come with none.

Why Does the Seller Matter as Much as the Test?

The seller matters because no single test is final, so the professional market leans on the chain the metal has travelled. The LBMA’s guidelines draw that conclusion from their survey of testing methods: the best way to reduce the risk of sub-standard metal remains the Good Delivery system of refineries and chain of custody. The Perth Mint gives retail buyers the same advice in plainer words: buy from the mint or its approved distributors, avoid private or unverified online listings, and treat an unrealistically low price as the warning it usually is.

A price is low only against a reference. The spot price, which our explainer on who sets the price of gold traces to London, New York and Shanghai, is a wholesale price for large amounts, and a genuine coin costs more than it; our article on premiums and spreads measures by how much. A coin offered well under spot is being sold for less than its metal is worth on the wholesale market, which a seller of genuine metal has no reason to do. Our guide to buying physical gold and silver and the Academy’s sourcing guide cover choosing where to buy, and the Academy’s counterfeit detection section has a simulated test bench, the Counterfeit Lab, with six specimens to test and call.

The Bigger Picture

Every test on the bench answers one question: is this metal what it claims to be? The scale, the calipers and the magnet answer it cheaply for the crude fakes, and the instruments take it further, each to its own depth, though none of them settles it alone. None of them answers the other question, where the metal came from, and the forged Swiss kilobars show how much that second question can matter once the first has been passed. For a buyer, the two checks that cost least follow from that: weighing and measuring what arrives against the mint’s own figures, and knowing that the seller’s chain runs back to a mint or a refinery. The first will not catch a well-made tungsten fake, which is why the second matters as much. Neither makes a fake impossible.

Sources and Methodology

Coin weights, diameters, finenesses and security features are from the mints’ own product and programme pages, read on 14 October 2026; the Krugerrand’s weight and diameter are the South African Mint’s published figures for its 1 oz gold proof Krugerrand, and its Krugerrand brochure gives the same diameter; Rand Refinery’s product page for the bullion coin gives its composition but no weight or diameter. The US Mint’s edge feature is from its account of the 2021 redesign. Densities are NIST’s Table 1 values for the elements, cross-checked against Los Alamos National Laboratory’s periodic table; the percentage differences and the thickness comparisons are our own arithmetic from those densities and the mints’ weights and diameters, and treat a coin as a flat disc. What each instrument sees is from the LBMA’s Physical Market Guidelines and its notes on ultrasonic and eddy-current testing; the marking and certificate rules are from the LBMA’s Good Delivery technical specifications and the kilobar specification endorsed by the LBMA and the Shanghai Gold Exchange. The forged kilobars, their purity and the database proposal are from Reuters’ investigation of 28 August 2019. The ratings in the matrix are our editorial judgements from those sources, not measurements.

Conflicts of interest: the mints sell the coins they describe, and the Royal Mint’s and the Perth Mint’s guides are written by sellers of bullion. The LBMA accredits the refiners on its Good Delivery List, the system its guidelines recommend, and the Alchemist article that describes its bar database was written by an executive of AlpVision, which sells anti-counterfeiting technology. The authors of the acoustic study propose their own testing method. No dealer is named or recommended here; the mints are cited for their own specifications and guidance. GoldSilverPortfolio has no commercial relationship with any mint or maker of testing equipment.

Primary sources

• Royal Canadian Mint — “2026 Gold Maple Leaf Coin – 1 oz. 99.99% Pure (Bullion)” and “2026 Silver Maple Leaf Coin – 1 oz. 99.99% Pure (Bullion)”, product pages (read 14 October 2026)
• United States Mint — “Bullion Coin Programs” (read 14 October 2026), and “West Point Mint Strikes New American Eagle Gold and Silver Coins” (16 July 2021)
• The Royal Mint — “Britannia 2026 1 oz Gold Bullion Coin” and “Britannia 2026 1oz Silver Bullion Coin” (read 14 October 2026), and “How to Tell if Gold is Real”
• South African Mint — “2023 Collectable Coin Range” (1 oz gold proof Krugerrand) and the Krugerrand brochure
• The Perth Mint — “Tips for avoiding fake coins and bars” (read 14 October 2026)
• National Institute of Standards and Technology — “X-Ray Mass Attenuation Coefficients”, Table 1 (Hubbell and Seltzer)
• Los Alamos National Laboratory — Periodic Table of Elements, “Tungsten” and “Gold”
• LBMA — “Physical Market Guidelines”, non-destructive testing guidelines (read 14 October 2026)
• LBMA — “Ultrasonic Probe and Display” and “Sigmascope Eddy Current Tester”, Wonders of Gold
• LBMA — Good Delivery Rules, “Technical Specifications”, section 2.1.5 (Marks), and “Specifications for a 1kg 9999 Gold Bar”, endorsed by the LBMA and the Shanghai Gold Exchange
• Reuters — “Fake-branded bars slip dirty gold into world markets” (28 August 2019)
• LBMA — “The Silent Risk: Counterfeiting in Precious Metals”, Alchemist issue 117, with its sidebar on the Gold Bar Integrity ecosystem
• Siwek, Linh, Gryczka and Stodolski — “Hybrid Anomaly Detection for Bullion Coin Authentication Leveraging Acoustic Signature Analysis”, arXiv preprint 2604.27803 (30 April 2026)