How To Tell If Gold Is Real: 11 Measured Checks

How to tell if gold is real comes down to three numbers: a density of 11.5 to 19.3 g/cm3, a fineness stamp of 375 to 999, and zero pull toward a magnet.

Gold is the one luxury material with a legal definition. The FTC Jewelry Guides say what karat a piece may claim. The National Stamping Act says how far the metal may fall short of its stamp. The US Mint publishes the weight and diameter of every coin it strikes. Those documents are the checklist. A fake has to match a density, a stamp, an acid colour and a coin diameter at once, and most fakes match one. The cards on this page turn each of those into a number you can confirm with a kitchen scale, a cup of water, a tile and a loupe.

Run the four quick checks first, with the piece in your hand or on a video call with the seller. Then weigh and dunk. A piece that pulls to a magnet or reads 8.5 g/cm3 does not need the acid.

Gold Jewellery, Coins and Bars: What Changes and What Doesn’t

The same metal gets faked three different ways, and each way fails a different test. Jewellery fakes are brass or plated base metal, so density and acid catch them. Coin fakes are cast or struck in base metal to the right face, so weight and diameter catch them. Bar fakes are the expensive kind: real gold wrapped around a tungsten core, so only sound or resistivity catches them. Sort the item first, then pick the test.

Two densities in that table deserve a second look. Tungsten sits at 19.25 to 19.3 g/cm3 against gold’s 19.32, a gap the Royal Society of Chemistry’s own element pages round to nothing. A kitchen-scale water test cannot separate them, which is why a bar needs a different instrument. At the other end, yellow brass sits at 8.47 g/cm3 and red brass at 8.75 in the Engineering ToolBox density table. A “14K” chain made of either weighs about 40 percent less than a real one of the same size. The difference is large enough to feel in the palm before you reach for a scale.

Colour changes across karats and tells you little. 22K is a deeper orange-yellow than 14K because it carries less copper and silver. Plating can be made any colour, and rose, white and green golds are all alloys of the same metal. Treat colour as a way to notice mismatches, such as a pale edge on a yellow face, and never as a reading in itself.

Karat Marks and Hallmarks: What the Stamp Promises

A stamp is a claim about fineness, and the law says how true the claim must be. In the United States, the National Gold and Silver Stamping Act sets the limit. The assayed metal may fall short of the stamped fineness by no more than 3 parts per thousand on the test portion. Once solder is counted, the whole article may fall short by no more than 7 parts per thousand. A ring stamped 14K, which is 583 fine, must therefore assay at 576 or better with its solder joints included. The FTC Jewelry Guides add the marketing layer. The word gold on its own means 24 karat. Any alloy must carry its karat immediately before the word, in type at least as conspicuous.

Europe and the UK write the same information as a three-digit number. 375 is 9 karat, 585 is 14 karat, 750 is 18 karat, 916 is 22 karat and 999 is 24 karat. The UK goes further and makes the mark a legal event rather than a maker’s claim. The Hallmarking Act, described in the London Assay Office guidance, covers every gold article over 1.0 gram sold in the UK. Each must carry three compulsory marks: a sponsor’s mark, a fineness mark and the mark of one of the four assay offices. Describing an unmarked article as gold is an offence in itself. A British piece with a lone 750 and no sponsor or office mark is either imported and unhallmarked, or not what it says.

The stamp itself has a physical signature. Genuine marks are struck with a hardened punch or laser cut, so the numerals have sharp walls and a flat floor under a loupe. Copies are often cast with the piece, and cast marks have soft rounded edges, uneven depth and sometimes a faint seam through them. A US quality mark should also sit next to a maker’s trademark; a bare 14K with no trademark anywhere is legal to own and worth a second look before you pay.

None of that makes a stamp proof. A stamp is the cheapest part of a piece to fake. The “gas station gold” reported by Kansas police and a Junction City pawn shop was stamped 18K on magnetic, brassy metal. Read the stamp to learn what the piece claims, then test the claim.

Home Tests Ranked by What They Prove

Four tests need nothing more than a magnet, a tile, a scale and a cup. They do not prove the same amount, and this order runs from cheapest to most decisive.

The magnet test proves the least and takes the least time. Gold, silver, copper and brass are all non-magnetic, so a pull means a steel or nickel core and ends the check. No pull means the piece could be gold, brass, silver or tungsten. Use a small neodymium magnet rather than a fridge magnet, and slide it along a bar or chain rather than tapping it, because a slight drag shows up better than a click.

The ceramic streak comes second. Gold sits at 2.5 to 3 on the Mohs scale and leaves a golden yellow streak, per Geology.com’s gold entry. Pyrite, the classic fool’s gold, sits at 6 to 6.5 and streaks greenish black. Drag an edge of the piece across the unglazed back of a tile and read the line. A dark or grey streak is not gold. A gold streak is consistent with gold and with plating, so this test rules out and never rules in.

The density test proves the most and is the one worth doing carefully. Gold alloys have no neighbours at their densities except tungsten, and tungsten is not used in jewellery fakes. The steps take five minutes:

1. Zero a scale that reads to 0.01 g and weigh the piece dry. Write the number down as A.
2. Stand a cup of water on the scale and zero it again. Tie the piece to a thread and lower it fully under the water so it hangs free, touching neither the bottom nor the sides.
3. Read the scale. That number is the weight of water the piece displaces, which is its volume in cubic centimetres. Write it down as B.
4. Divide A by B. A 14K piece should give 13.0 to 14.3; 18K should give 15.0 to 16.0; 10K should give 11.5 to 13.0, using the bands from the Kassoy jewellery density table.
5. Reject the reading if the piece is hollow, has stones, or is under about 3 g, where a 0.01 g scale cannot resolve the volume well enough to matter.

The acid test proves fineness rather than metal, and it marks the piece, so it comes last. Commercial nitric acid is 68 to 70 percent by volume. Portland Gold Buyers describes what it does. 12K to 18K gold shows no visible reaction. 10K shows a faint smoke-black mark with no bubbles. Copper alloys turn the acid green and bubbly within seconds. The reaction only tells the truth at the depth you test. File a notch on a hidden spot before the drop, because a plated piece passes on the surface and fails in the notch. Aluminium, stainless steel and tungsten also stay silent under nitric acid, so a heavy item that neither reacts nor reads the right density has failed, not passed.

Plated, Filled and Vermeil: Where the Legal Lines Sit

Most “fake gold” in circulation is not counterfeit at all. It is plated or filled jewellery that was labelled honestly at the factory and dishonestly at resale. The FTC sets the lines, and the lines are thin. Under 16 CFR 23.3, the term gold electroplate needs at least 0.175 microns of fine gold, about seven millionths of an inch. Heavy gold electroplate needs 2.5 microns, about a hundred millionths of an inch. Gold filled needs a mechanically bonded layer weighing at least one twentieth of the whole article; rolled gold plate needs one fortieth. Vermeil, under section 23.4, is sterling silver under at least 2.5 microns of gold of 10 karat or finer. In each case the gold layer must itself be at least 10 karat.

Those numbers explain the wear pattern. A 0.175 micron plate is a few hundred atoms deep and wears through at a clasp or ring edge within months. That is why colour at the wear points is one of the four quick checks. A gold-filled layer is a solid sheet of karat gold bonded to brass, roughly 50 to 100 times thicker than electroplate, and can outlast its owner. Both are legitimate products, and both are worth a small fraction of a solid piece of the same size.

The economics are visible in the marks. A 20 g bracelet stamped 1/20 14K GF contains 1 g of 14K alloy, which is about 0.58 g of pure gold. The same bracelet in solid 14K contains about 11.7 g of pure gold. A seller who lists a GF piece at a solid-gold price has either misread the stamp or hopes you will. The Louis Vuitton hardware guide on this site covers the same trick on brass bag fittings, where a plated surface is meant to look like solid metal.

The detection problem is that a jeweller’s XRF gun cannot see through plate. Evident Scientific, which makes the analysers, puts the analysis depth at 15 to 20 microns and notes that a thick enough 18K plate over a steel core reads as solid 18K. Heavy gold electroplate at 2.5 microns is well inside that depth and reads as whatever karat the plating is. The XRF number tells you the karat of the surface. Only density, a filed notch or a cut tells you what is underneath.

Coins and Bars: Tungsten and the Tests That Catch It

Bullion fakes are built to pass the jewellery tests. Tungsten’s density of 19.25 to 19.3 g/cm3 sits within a few hundredths of gold, so a tungsten core inside a real gold shell weighs right, measures right and floats right. In September 2012 a Manhattan refiner named Ibrahim Fadl bought four 10 oz PAMP Suisse bars from a trade contact and drilled one after a warning from another dealer. Coin World reported that the bars were about 75 percent tungsten, carried genuine serial numbers, and had been peeled open, cored and refinished. At the gold price of the day the four bars should have held more than $70,000 in metal. The Secret Service and FBI took the case. The bars had passed XRF and a scale before the drill.

The cheaper bullion fake is the opposite kind and fails the simple tests. In October 2017 an Ottawa jeweller paid $1,680 at a Royal Bank of Canada branch for a 1 oz wafer in Royal Canadian Mint packaging. CBC News reported that a goldsmith could not roll it in a mill, that it snapped when bent, and that it failed both 18K and 14K acid. Gold is the most malleable metal there is, soft enough to cut with a knife by the Royal Society of Chemistry’s description, and a wafer that snaps is not gold. The Mint later confirmed it had not made, shipped or sold the piece.

Coins carry published specifications, so the first test is a ruler and a scale. The US Mint lists the 1 oz American Gold Eagle at 33.931 g and 32.70 mm, struck in 22 karat: 91.67 percent gold, 3 percent silver and the balance copper. The 24K American Buffalo weighs 31.103 g at the same 32.70 mm. The 1 oz Krugerrand weighs 33.93 g at 32.77 mm and 2.84 mm thick, in 91.67 percent gold with copper. A base-metal copy at 8 to 11 g/cm3 cannot reach 33.9 g inside those dimensions, so it comes in light, thick or both. Since the 2021 redesign, the 1 oz Eagle also carries a reeded-edge variation. Coin World locates it as a single notch in the reeding at 6 o’clock with the obverse up. The Mint has also added covert features it has not described.

Two instruments close the tungsten gap, and both live at coin shops rather than kitchens. An ultrasonic thickness gauge sends sound through the bar and times the echo. Sound travels through gold at about 3,240 m/s and through tungsten at about 5,170 to 5,180 m/s, per Testyourgold.com and Evident’s velocity table. A gauge set for gold therefore reads a tungsten-cored 10 oz PAMP bar, 8.3 mm thick by callipers, as about 5.2 mm. The second instrument is the Sigma Metalytics Precious Metal Verifier. Its bullion wand reads electrical resistivity through plating to a depth of about 2 mm and returns a result in about a second. A dealer who runs either test, and a scale, has done what XRF alone cannot. If you are buying gold watches from a grey-market dealer, the Jomashop review on this site explains what that paperwork warrants. The Rolex guide covers weight bands for gold-cased watches.

Where To Get Your Gold Authenticated

A local jeweller or pawn shop is the fastest route and usually free. Most run a handheld XRF, which gives a karat reading in about 10 seconds with an accuracy near plus or minus 0.4 percent under standard conditions, by Evident’s figures. Ask for the reading rather than a yes or no, and ask whether a density check or a filed acid test was done as well. XRF reads the surface; the combination reads the piece.

An assay office is the legal standard in the UK and issues the hallmark itself. The four UK offices, in London, Birmingham, Sheffield and Edinburgh, test articles by fire assay and XRF and strike the compulsory marks on articles that pass. Hallmarking is compulsory for gold over 1.0 gram sold in the UK, so a UK-bought piece with no hallmark and a price above scrap is already outside the law. In the US there is no equivalent office; a refiner’s fire assay is the destructive equivalent and is used when a piece is being sold for melt rather than kept.

Coin dealers and bullion dealers carry the instruments jewellers do not. A Sigma Metalytics verifier or an ultrasonic gauge reads inside the metal, and a dealer who buys bullion runs one on every bar that comes across the counter. For collectible coins, PCGS and NGC grade and seal the coin in a tamper-evident holder, which moves the authentication question from the coin to the holder. For gold bought at a live-stream auction, the Whatnot review on this site sets out what the platform’s buyer protection does and does not cover on precious metals.

Legitly sits in front of all of these. It reads your photos for the stamp format, colour at wear points, hallmark layout and coin edge detail, and flags what does not fit. A photo cannot weigh a chain or hear a bar, so Legitly pre-screens and does not certify. Use it to decide whether a listing deserves a deposit, then use the scale, the tile and the dealer before the balance changes hands. The gold hub and the wider jewellery category collect every related guide on this site. How to tell if silver is real runs the same density arithmetic against different target numbers. For a signed piece rather than raw metal, fake Van Cleef shows what a documented weight band adds to a hallmark, and the Cartier serial number check reads the code stamped beside it. Chains carry their own failure points, and how to tell if a gold chain is real reads the clasp stamp and the link joints.

Verdict: Two Numbers Off Means Walk

A genuine piece passes every card, so a single pass tells you little and a single fail tells you a lot. Fakes are built to pass the checks buyers run most: they carry a stamp, they take a magnet without moving, and the good ones survive a surface acid drop. What fakes cannot do is pass density and a filed notch at once, because a brass core is 40 percent too light and turns nitric acid green.

Rank the tests by cost to fake. A stamp costs nothing. A non-magnetic alloy costs nothing. A plating that survives a surface acid drop costs pennies per piece. A correct density costs the price of gold or the price of tungsten, and tungsten only makes sense in a bar. That is why the density test is the home test that matters, and why a bar needs a dealer’s ultrasound or resistivity reading before money moves.

Two failures end it. A 14K stamp with a density of 13.5 g/cm3 and a clean notch is gold. A 14K stamp with a density of 8.6 is not, and the acid is a formality. A coin that weighs right but measures 0.4 mm thick, or a bar that reads 5.2 mm on a gauge and 8.3 mm on callipers, has failed twice. Treat any two failed numbers as a fake and walk, whatever the stamp, the packaging or the story says.