In short
No single home test proves gold is real; each one rules something out. A magnet rules out steel and nothing else. Baking soda, vinegar and water rule out nothing, whatever the videos say. The test that works at home is density: weigh the piece, weigh it again submerged, divide, and compare with the table below; brass and silver are a third to half lighter than gold. An acid kit then reads the karat, and a jeweler's XRF gun settles anything the others cannot.
Contents
What each test proves
| Test | Catches | Misses | Cost |
|---|---|---|---|
| Read the stamp | unstamped costume, GF/GP/925 marks | fake stamps; real unstamped gold | free |
| Strong magnet | steel and iron cores | brass, copper, silver, tungsten, plated anything | a neodymium magnet, a few dollars |
| Baking soda, vinegar, water | nothing reliably | everything | free, and worth exactly that |
| Ceramic streak | plating worn to black streak | plated pieces with intact surface pass | an unglazed tile |
| Density (water displacement) | brass, silver, hollow and plated pieces by weight | tungsten cores, stone-set pieces | a 0.01 g scale, ~$20 |
| Acid kit on a touchstone | plated over base metal; reads karat | surface only: gold filled and thick plating pass unless notched | $20–40 |
| XRF at a jeweler or refiner | alloy composition to 20–50 µm depth | very thick plating over a core, unless filed | usually free with a sale |
The pattern in the table is the thing to remember: plated pieces have real gold on the surface, so every surface test passes them. A stamped, non-magnetic, acid-passing "14k" chain can still be a brass chain with a good plating job. The tests that look inside are weight and XRF, and weight is free.
Stamp decoderWhat 417, 585, 750, GF or GP meanOpen calculatorTests to skip, and why
Baking soda, vinegar, water
The recipe going around is to dissolve baking soda in water, or mix it with vinegar, drop the piece in and watch for a reaction or a color change. Nothing in it can work. Baking soda is a weak base and household vinegar a weak acid; neither dissolves gold, and neither does anything visible to brass, copper, silver or steel in the minutes people wait. Vinegar will slowly darken copper alloys over hours to days, which tells you the piece is not pure gold, but so is every ring below 24k. A piece that "passes" has passed a test with no fail condition.
The bite test
Pure gold is soft enough to dent under teeth, which is where the idea comes from; 14k is not, and lead is softer than either. A test that fails real jewelry and passes lead is worse than no test, and it chips teeth.
The float test
Gold sinks; so does brass, steel, silver and every metal fake. It catches only hollow plastic.
The lighter test
Heating a piece and watching for darkening finds copper, which is in every karat alloy below 24k, so it fails real 14k. It also melts solder and ruins stones.
Skin discoloration
Real 10k and 14k can mark skin; plated brass can go a week without. Not diagnostic in either direction.
The density test, step by step
You need a scale that reads to 0.01 g and a glass of water.
- Weigh the piece dry.
- Hang it from a thread so it is fully under water without touching the glass, and read the scale with the glass on it and the piece suspended: the increase over the glass alone is the weight of the displaced water, which in grams equals the volume in cubic centimeters.
- Dry weight divided by that number is the density.
A 14k ring should land near 13; brass lands near 8.5, sterling near 10.4. Stones, hollow links and porous castings throw the number off, so use this on plain metal pieces.
| Material | Density, g/cm³ |
|---|---|
| 24k gold | 19.3 |
| Tungsten | 19.3 |
| 22k gold | 17.7–18.0 |
| 18k gold | 15.2–15.6 |
| 14k gold | 12.9–13.6 |
| 10k gold | 11.3–11.6 |
| Sterling silver | 10.3–10.4 |
| Brass | 8.4–8.7 |
| Stainless steel | 7.9–8.0 |
Tungsten is the reason density is not the last word: it has the density of pure gold and is used for bar and coin fakes, less often for jewelry because it cannot be cast or soldered like gold. A tungsten-cored bar passes a magnet, a stamp check and a density test, and fails XRF only if the gold shell is thin; refiners drill or cut suspicious bars. For jewelry, tungsten fakes are rare and density is a strong test.
Acid kits and electronic testers
An acid kit is a set of nitric acid solutions calibrated for 10k, 14k, 18k and 22k and a black touchstone. Rub the piece on the stone, apply the acid for the karat you expect: a streak that survives is at least that karat, one that dissolves is below it or not gold. It is what pawn shops use and it works, with two limits. It tests the surface, so file a notch in a hidden spot on anything that could be gold filled or heavily plated before testing; and it is acid, so gloves, ventilation and no gemstones nearby. Consumer electronic testers under about $150 measure surface conductivity and are fooled by thick plating; the professional ones and XRF analyzers at jewelers and refiners read composition to a few tens of microns and are usually run free of charge if you are selling, which is the cheapest way to get a definitive answer before you sell.
Scrap payoutA mixed lot: melt value and what buyers actually payOpen calculatorCombining the tests by situation
Everyday jewelry with a stamp
Stamp, then magnet on the links away from the clasp, then a density check if the piece is plain. Three passes and the piece is almost certainly what it says; an acid kit adds the karat if you are selling.
Unmarked pieces
Density first, because it is the only free test that sees inside. If it lands on a gold value, an acid kit or a jeweler's XRF names the karat; if it lands near brass or silver, you are done.
Coins and bars
Weigh and measure against the mint's published specifications, which are exact to a hundredth of a gram and a tenth of a millimeter; a coin that is the right diameter and thickness but the wrong weight is the classic tungsten or lead fake. For bars above an ounce, an ultrasonic or XRF check at a dealer is worth the trip.
Inherited or antique pieces
Do not scratch, file or acid-test anything that might have value as an object; a density check and a jeweler's XRF are non-destructive. Old American pieces are often unmarked and often gold, and a signed piece is worth more than its metal regardless of what the tests say.
How professionals test
XRF
An X-ray fluorescence gun excites the atoms in the surface and reads the emitted spectrum, giving the percentage of each element in seconds without a mark. It reads 20 to 50 microns deep, enough to see through plating, not enough to see a thick shell over a core; refiners file a spot on anything doubtful.
Fire assay
The reference method: a sample is melted with lead, the gold is separated and weighed. Destructive, accurate to a part in ten thousand, used by refiners to settle payment on large lots, not for your ring.
Ultrasonic and electrical conductivity
Used on bars and coins to find cores; sound travels differently through tungsten than gold, and gold's conductivity is distinctive. Dealers use both on anything over an ounce.
When to stop testing
When the stamp, the magnet and the density agree, and the piece is ordinary jewelry, you know what you have; more testing adds risk of damage and no information. When they disagree, or the piece is unmarked and heavy, or it is a coin or bar worth real money, stop guessing and take it to someone with an XRF, which most jewelers, pawn shops and refiners will run for free in the hope of a sale. The one thing not to do is act on a single test in either direction.
Questions people ask
Does real gold stick to a magnet?
No: gold, silver, copper and brass are all non-magnetic, so a piece that pulls toward a strong magnet has steel in it. Not pulling proves little, though, because plated brass does not pull either. Test the links, not the clasp: spring-ring and many lobster clasps hide a steel spring that twitches on a perfectly real chain. Use a neodymium magnet; a fridge magnet is too weak to tell you anything.
How do you test gold with baking soda?
You cannot, and the millions of searches for it every year are chasing a test with no fail condition. Baking soda is a mild base and vinegar a weak acid; neither dissolves gold, and neither does anything visible to brass or plated steel in the minutes people wait. A $20 pocket scale for a density check, or a $25 acid kit, can actually fail a fake.
Is 14k gold real gold?
Yes: 58.5% gold alloyed with copper, silver and sometimes zinc or nickel, the standard alloy for American jewelry. Real and pure are different questions. 14k is real and hard-wearing; 24k is pure and bends under a fingernail, which is why almost no wearable jewelry is made of it.
How can I tell if gold is real without acid?
Density. Weigh the piece dry, then suspended in water, and divide: 14k lands near 13 g/cm³, 18k near 15.5, brass near 8.5, sterling near 10.4. On a plain piece with no stones or hollow sections that one measurement separates solid gold from every common fake except a tungsten core, which is a bullion problem, not a jewelry one.
Is gold fake if it sticks to a magnet?
The part that sticks is not gold. Often that part is the steel spring inside a clasp on an otherwise real chain, so test a link away from the clasp. If the links themselves pull, the chain is steel with a gold color, and no other test is needed.
Are electronic gold testers accurate?
The professional XRF analyzers at jewelers and refiners are, reading composition 20 to 50 microns deep in seconds, and most will test for free if you might sell. Pocket testers under about $150 measure surface conductivity, are fooled by thick plating and cannot separate gold filled from solid; they answer is-this-plated-brass and not much else.
Does 925 mean real gold?
No, 925 is sterling silver. A yellow piece stamped 925 is gold-plated silver, called vermeil when the gold is at least 2.5 microns of 10k or better. Its value is the silver, a few dollars for a typical pendant; the gold on it is measured in cents.