How to Tell If Silver Is Real — Hallmarks, Magnet Test, Weight

By Dmitry Strugovshchikov, founder of STRUGA

A stamp is the weakest evidence on a piece of silver, and almost every guide puts it first. The order below is the one we use on pieces customers hand us across the bench: two minutes of physics before anyone reads a mark. What settles the question is density, and a kitchen scale and a glass of water measure it.

Real silver is confirmed by density, not by a mark: 925 sits at 10.36 g/cm³ against 8.53 for brass and 7.93 for steel.

A magnet rules out a steel core and nothing else. Ice, tarnish and the egg test all pass on silver-plated brass.

A missing hallmark proves nothing: unmarked silver is lawful in Russia, Germany, Sweden and the United States.

Updated August 28, 2026

What does «real silver» actually mean?

The word carries two separate questions, and mixing them is why home tests seem to disagree. The first is whether the metal below the surface is the same alloy as the surface. The second is how much silver that alloy contains. A plated brass ring fails the first; an antique at 800 passes the first and fails a claim of 925. Home tests answer the first question well and the second badly.

Pure silver is a single element, Ag, atomic number 47. At 99.9% it is too soft to hold a stone or a ring profile, so it is alloyed for wear. Sterling means at least 925 parts of silver per thousand: the UK Hallmarking Act 1973 and the US FTC Jewelry Guides fix the silver content and say nothing about the remainder. Copper is traditional rather than required: germanium alloys such as Argentium are lawfully hallmarked 925.

Which test should you run first, and what does each one prove?

Run them cheapest first, and read the last column before you trust a result. Guides list eight or ten tests and imply that passing them adds up to proof. It does not: five pass identically on silver-plated brass, because the surface being tested really is silver.

Test	Cost	What it proves	What it cannot show
Magnet	$5	no ferrous core	brass, copper, nickel silver, 304 steel all pass
Ice	free	the bulk conducts like silver	copper is faster; useless on thin pieces
Tarnish, cloth, egg	free	silver on the surface	nothing about the core
Loupe	$10	wear-through, cloned stamps	unworn plating looks clean
Weight alone	free	little without known volume	our rings at one size run 3.3–17.0 g
Density by displacement	free	solid silver against every substitute	925 from 875 — a 0.9% gap
Acid	$18–40	fineness at the point tested	anything, unless you file through
XRF	$20–50	surface to about 40 microns	plating over 20–50 microns reads as 925
Laboratory assay	varies	the answer	it destroys a sample

Does a magnet prove anything about silver?

It proves there is no iron inside, and that is the whole of it. Silver is diamagnetic, so a neodymium magnet will not lift it — nor will it lift brass, copper, aluminium, nickel silver or 304 stainless. As a test for silver the magnet is nearly empty; for a steel core it is instant. Test the body, never the clasp: lobster clasps carry steel springs.

The slide test, which is the useful version

Tilt a flat face to roughly 45 degrees and let a strong neodymium magnet slide down it. On silver the magnet crawls: the moving field induces eddy currents, and by Lenz's law those currents push back. Silver is the best electrical conductor of any metal, so it brakes hardest.

Two limits. Copper runs 100% IACS against silver's 106%, so a plated copper core brakes almost as well. And the effect needs mass and a flat face several centimetres across, so on a ring you will see nothing.

How much should a real silver ring weigh?

Less than the round numbers in circulation, and the figure alone settles nothing. A flat sterling band in US 9, 6 mm wide and 2 mm thick, holds 0.79 cm³ and weighs 8.2 g; the same band in brass, 6.7 g — a gap of 18%, not the quarter most guides imply. Our own catalogue shows why the absolute number is useless:

Ring, size 19	Weight	Across sizes 16–22
Signature Links V.2	3.3 g	2.1–5.5 g
Thorn	8.0 g	4.7–12.7 g
Brutalism V.1	17.0 g	9.6–24.0 g

A fivefold spread at one size, all of it 925. Weight is evidence only when the geometry is known exactly, and between 925 and 875 the density gap is 0.9%.

How to measure density with a kitchen scale

1. Weigh the piece dry and note the number.
2. Hang it on a thread, fully submerged in a cup of water, touching neither the sides nor the bottom, and weigh it again.
3. Divide the dry mass by the difference between the two readings.
4. Read the result: sterling lands near 10.36, brass 8.53, nickel silver 8.73, 304 steel 7.93.
5. Allow for casting porosity, which reads one or two per cent low: treat anything above 10.1 as solid silver.

This is the only home test that separates solid metal from plating without marking the piece.

Does ice really melt faster on real silver?

It does, and the usual explanation for why is wrong. Silver conducts heat better than any metal at 429 W/(m·K), against 401 for copper and about 16 for stainless steel. The common claim is that plating fails because the layer is too thin to hide the base metal. The physics runs the other way: a 2.5-micron layer carries thousands of times less thermal resistance than the two millimetres beneath it, and in one second heat travels 13 mm through silver. The coating takes no part in the test.

What the ice reads is the bulk. Silver-plated brass melts ice about 1.7 times slower than solid silver, stainless steel four times slower. Two limits: it cannot separate 925 from 999, and copper melts ice faster than silver. Mass matters too, so a thin pendant gives no reading.

Why does real silver turn black, and is that a good sign?

Darkening is the strongest free evidence that there is silver on the surface, and no evidence about what lies under it. Silver does not rust and does not oxidise in air at room temperature. It reacts with hydrogen sulphide in the atmosphere to form silver sulphide, and fractions of a part per billion are enough. A piece that never darkens is the suspicious one. In 925 the copper works too, reacting with sulphur and, unlike silver, with oxygen — which is why sterling darkens faster than fine silver.

Does the cloth test work?

Rub the piece with a soft white cloth for half a minute. A grey-black smear is silver sulphide lifting off. No smear proves nothing: freshly polished or rhodium-plated silver leaves none.

Does real silver turn skin green?

The copper does, not the silver. Copper marks skin green when it meets sweat, so a green finger points to an alloy containing copper, which sterling is.

What can you see with a 10x loupe?

Wear, and wear is where plating gives itself away. Plating breaks through first where a piece rubs most, and the base metal shows its own colour at the break:

• the underside of a ring shank, the ridges of a bracelet, the links beside a clasp, the bail of a pendant;
• brass reads yellow, copper orange-pink, steel grey-blue;
• on solid metal the surface, the stamp area and the inside of every opening share one colour;
• tool marks from hand finishing are expected and say nothing about authenticity.

The stamp deserves the same lens. A punch displaces metal, so a genuine mark carries a small raised lip; a stamp that sits too crisp on an untouched surface was applied to a finished, plated piece.

What do the numbers on a silver stamp mean?

The number is a fineness: parts of silver per thousand, and nothing about craft or origin. The table gives the jurisdiction each number belongs to, which is where most guides, our own earlier version included, get it wrong.

Mark	Silver	Where it is a legal standard
800	80%	the floor in Germany, Sweden, Denmark, Norway; lawful in the UK, France, Russia
830	83%	Norway and Russia; the Scandinavian fineness
835	83.5%	a Convention standard and German trade custom, not German law
875	87.5%	Russia, from Soviet practice
900	90%	American coin silver under the FTC guides, not French
925	92.5%	sterling; what ordinary jewellery is made to
958	95.84%	Britannia — not 950, which in a UK mark means platinum
999	99.9%	fine silver; lawful everywhere, too soft for wear

STERLING and Ster. are marks the FTC recognises for at least 925. SS is not a recognised silver mark anywhere and proves nothing either way.

Does real silver always carry a hallmark?

No, and this is the assumption that sends people to the wrong conclusion. Unmarked genuine silver is normal rather than exceptional. Marking is voluntary outright in Germany and Sweden, not required in the United States, and voluntary at any weight for domestic silver in Russia. Where it is compulsory there is usually a weight floor.

Britain exempts anything under 7.78 g, France under 30 g and all exports, Denmark under 3 g; Norway alone marks at any weight. A hallmark on a heavy British piece is therefore strong evidence, because selling it unmarked is an offence. A missing mark on a light or non-British piece is no evidence at all, and a mark that is present is worth what the physical test agreeing with it is worth.

Which «silver» names contain no silver at all?

Several, and they carry most of the fakes. They fall into three groups:

• Alloys with no silver in them. Nickel silver, German silver and alpacca are one copper-nickel-zinc alloy under three names, about 8.73 g/cm³. Britannia metal is a tin alloy, unrelated to Britannia silver at 958.
• Regional labels. Tibetan, Mexican and Bali silver are marketing terms, not standards: genuine Bali silver is 925, but the same label sits on pieces containing little or none.
• Wording on the tag. Silver-tone, silver-coloured and silver finish all mean plated base metal, and the vagueness is deliberate.
• Rhodium-plated 925 is the honest inverse: real silver underneath, but the plating hides the metal until it wears through, usually within 12 to 18 months, and the tarnish that follows is patchy.

When should you stop testing and pay for a laboratory assay?

When the piece is worth more than the test, and when the home tests disagree. Two paid methods sit between the kitchen and the laboratory, and each has a limit that is rarely stated.

• The acid test uses Schwerter's solution: on 925 the drop turns dark red, on 800 brown, on brass dark brown, on nickel blue. On an unbroken surface it reads the plating and confirms the fake — it tests the piece only if you file a notch through to fresh metal first.
• XRF answers what the surface is made of, not what the piece is. It sees through plating of one to two and a half microns, but its depth runs to roughly 40 microns, so heavier plating reads as solid 925. It over-reads fineness on cast pieces whose surface is copper-depleted.
• Laboratory assay is the only final answer, and it is destructive. For silver the reference method is ISO 11427, titration with potassium bromide; cupellation, named most often, is the reference for gold.

What STRUGA puts on every piece, and what to do if yours fails

Every object we make is 925, and its exact weight in grams is printed on its own page — which is what made the table above possible to write. Nothing is rhodium-plated, so each piece darkens rather than hiding under a coating. Seventeen objects out of 134 are blackened at the bench before they ship: thirteen earrings, two necklaces, one ear cuff, one ring.

If a piece of ours ever fails the magnet, density or hallmark check, write to us and we replace it. If a piece from elsewhere fails, act in this order:

• Photograph the piece and the failed result before anything else.
• Return to the seller while the window is open; after it closes the options narrow fast.
• Get a jeweller's second opinion before pushing a claim.
• Correct the declared value if the piece was insured as silver.

The rings above sit in the ring collection, in the Thorn and Brutalism families; the pre-blackened objects are in blackened silver, mostly earrings. For the metal explained rather than tested: the 925 chapter of the Atlas, the sterling silver guide, gold against silver, how an object is made.

Objects from this article

• Signature Links V.2 Ring | STRUGA — 2.1–5.5 g
• Thorn Ring | STRUGA — 4.7–12.7 g
• Brutalism V.1 Ring | STRUGA — 9.6–24.0 g
• Brutalism Ear Cuff V.1 (M) | STRUGA — 1.6 g