Fakes are a real problem at the eBay end of the bullion market. Sovereigns, Krugerrands and 1oz silver coins are by far the most common counterfeits in the UK. Five quick tests will catch almost every fake before money changes hands.

1. Buy from a reputable source

The single best defence is sourcing. UK dealers worth using:

  • The Royal Mint direct (highest premium, lowest fake risk).
  • BullionByPost, Atkinsons, Bairds, Chards, Hatton Garden Metals — all members of recognised trade bodies.
  • Coin shops with a London Bullion Market Association (LBMA) or British Numismatic Trade Association (BNTA) affiliation.

Avoid:

  • eBay listings priced suspiciously close to spot.
  • “Estate sale” Facebook Marketplace coins.
  • Cash-only sellers at car-boot sales pricing coins below dealer buyback rates.

2. The magnet test

Pure gold and silver are non-magnetic. Hold a strong neodymium magnet near the coin. If it sticks or tugs noticeably, the coin contains iron or a magnetic alloy — fake. (Tungsten-cored fakes will pass this test, so it’s a starting filter, not a final answer.)

3. The dimension test

This is the most powerful single test. Every legitimate bullion coin has published dimensions accurate to a fraction of a millimetre — diameter, thickness, weight. Counterfeiters can match one or two of these but rarely all three at once.

For example, a Gold Sovereign:

  • Diameter: 22.05 mm
  • Thickness: 1.52 mm
  • Mass: 7.988 g
  • Composition: 22ct gold-copper, density ~17.55 g/cm³

If your “Sovereign” weighs 8.10g, or is 1.65mm thick, it’s not a Sovereign. Even a 0.1mm thickness deviation is a red flag.

Compare your coin against the specs on this site’s coin pages — every entry lists diameter, thickness and gross weight.

4. The ping test

Drop the coin gently onto a hard non-metallic surface (a wooden table) from a height of an inch. Silver and gold produce a long, clear ringing tone that lasts a second or two. Tungsten and lead-cored fakes produce a dull thud with no ring.

The smartphone app CoinTrust records the ping and compares it to a reference. It’s not infallible but it catches the common Chinese fake Sovereigns.

5. The Sigma / ultrasound / XRF test

For higher-value purchases (anything over £500), a dealer should let you watch them use a Sigma Metalytics Precious Metal Verifier or similar device. It passes a current through the coin and reads its electrical conductivity — a strict test that catches tungsten-cored fakes that pass weight and dimension tests.

Better still: many UK bullion dealers operate XRF analysers that read surface composition. If a dealer won’t XRF a coin you’re buying at material value, walk away.

Common UK fakes — what to watch for

  • Chinese-fake Sovereigns: dimensions slightly off, weight 0.1–0.3g light, ring is dull. Frequently aged with chemicals to look “old”. Often sold in batches on eBay.
  • Tungsten-cored 1oz gold bars: tungsten has almost exactly the same density as gold, so weight + dimensions can match. Sigma test is the only quick catch.
  • Silver-plated copper 1oz “Eagles”: weight is ~7% light. Magnet doesn’t stick, but the ping is wrong and dimensions are slightly off.

What to do if you’ve been sold a fake

  1. Don’t try to resell it — that’s a separate offence.
  2. Photograph the listing, packaging and the coin itself.
  3. Open a dispute with the platform (eBay’s MoneyBack Guarantee covers counterfeits).
  4. Report to Action Fraud and the relevant police force.
  5. If you bought from a UK dealer, report to the BNTA / LBMA.

The fake-resistant alternative

If counterfeiting worries you, lean on modern tamper-evident assay cards:

  • PAMP Suisse Veriscan — barcode on every assay card that links to PAMP’s verification app.
  • Royal Mint DNA Certified — micro-engraved verification.
  • All sealed cards from LBMA-accredited refiners (Metalor, Heraeus, Argor-Heraeus, Umicore, Valcambi).

A sealed 1oz PAMP Fortuna bar with intact Veriscan is essentially impossible to fake in the field — the card and barcode would have to be perfectly replicated, and the bar inside has to match the published dimensions exactly.