Purity, Weight, and How Fakes Actually Get Caught

Precious metals are sold using a measurement system that exists nowhere else in modern commerce, and the confusion is not accidental — it is where a surprising amount of retail overpaying happens. Two products can look equivalent on a shelf and contain different amounts of metal.

The troy ounce is not the ounce you know

A troy ounce is 31.1035 grams. The ordinary (avoirdupois) ounce used for groceries is 28.3495 grams. A troy ounce is about 10% heavier, which means anyone comparing a “one ounce” figure without checking which ounce is meant is off by a tenth before they start.

Useful conversions:

Unit	Troy ounces	Grams
1 troy ounce	1	31.1035
1 kilogram bar	32.1507	1,000
1 pennyweight (dwt)	0.05	1.5552
1 tola (South Asia)	0.3750	11.6638
1 tael (Hong Kong)	1.2034	37.4290

Pennyweights matter because scrap and jewellery buyers often quote in them, and a price per pennyweight looks reassuringly small next to a price per ounce for exactly that reason. There are 20 pennyweights to the troy ounce.

Fineness, karats, and why 22k can beat 24k

Fineness is the parts-per-thousand of pure metal: .9999 (“four nines”), .999, .9167. Karat is parts per twenty-four: 24k is pure, 22k is 22/24 = .9167, 18k is .750, 14k is .5833.

Here is where products get misread. The American Gold Eagle and the South African Krugerrand are 22-karat coins. The Canadian Maple Leaf is .9999. A buyer comparing “22k versus four nines” concludes the Maple Leaf has more gold.

It does not. All three contain one full troy ounce of gold. The 22-karat coins simply add copper and silver on top for hardness, so their gross weight is higher — 31.1035 ÷ 0.91667 = 33.93 grams for a Krugerrand versus 31.1035 grams for a Maple Leaf. The alloy is extra metal, not missing gold.

The practical difference is durability, not content: a 22-karat coin resists the handling marks that soft pure gold picks up easily. Neither is “better.” But a sales pitch built on the purity number alone is a pitch built on a misunderstanding.

On the silver side: bullion is normally .999 or finer, sterling is .925, and pre-1965 US dimes, quarters and half dollars are .900 fine — the “junk silver” of the trade, which contains roughly 0.715 troy ounces of silver per dollar of face value once ordinary circulation wear is accounted for.

Verification: what works and what does not

Counterfeiting bullion is a real, industrialised problem, and the popular home tests catch only the crude attempts.

Weight plus dimensions. Every genuine product has a published gross weight, diameter and thickness. A base-metal fake that gets the weight right will almost always get the dimensions wrong, because base metals are far less dense than gold. This is a genuinely useful screen and requires only calipers and a scale that reads to 0.01 g. It is also the test most people skip.

Specific gravity. Gold is 19.32 g/cm³, silver 10.49, lead 11.34. Weighing an item in air and again suspended in water gives its density. This defeats most base-metal substitutions — and it does not defeat the serious one, because tungsten is 19.25 g/cm³, close enough to gold that a tungsten core inside a gold shell passes a density check. This is the single most important limitation to internalise: density alone is not sufficient for gold.

The magnet test. Neither gold nor silver is ferromagnetic, so sticking to a magnet is disqualifying. More usefully, silver is the most electrically conductive metal there is, so a strong neodymium magnet slid down a silver bar descends visibly slowly, braked by induced eddy currents. Gold shows a weaker version of the same effect. This catches steel and nickel fakes and nothing sophisticated.

The ping test. Genuine bullion rings at a characteristic frequency; plated or cored pieces thud. Phone apps exist that analyse the tone. Suggestive, not conclusive.

XRF (X-ray fluorescence). Common at dealers, and frequently misunderstood: XRF reads the surface, a few microns deep. It confirms the outside is gold. A gold-plated tungsten core reads as pure gold all day long.

Ultrasonic thickness gauge. Measures the speed of sound through the whole piece, so an internal density discontinuity shows up. This is one of the few consumer-accessible tools that actually addresses the tungsten-core problem.

Electromagnetic resistivity meters. Purpose-built precious-metal verifiers read conductivity through the item rather than off its surface, and are widely used by dealers for exactly that reason.

The verification that costs nothing

All of the above is a fallback for a problem best avoided at the point of purchase:

• Buy recognised products. Major sovereign coins and major-refiner bars are the most studied, most familiar items in the market. A dealer sees thousands.
• Keep bars in their assay packaging. Sealed cards with matching serial numbers are a chain-of-custody claim. Breaking the seal to “look at it” permanently costs you resale value.
• Buy from established dealers, not marketplaces. The counterfeit risk concentrates almost entirely in anonymous peer-to-peer sales at suspiciously fair prices.
• Keep your invoices. Provenance is part of what you sell later.

A bar that never left a reputable chain of custody does not need to be tested. That, and not any gadget, is what most of the premium over spot buys.

Educational content only. Nothing here is financial advice.