Guide · 11-minute read
How to test gold at home: which checks mean something, and which will damage it
The short answer
Four checks at home (a magnet, the wear points, the marks, and weight against size) can sort "probably not gold" from "possibly gold". None of them gives a carat or proves a piece genuine. The toothpaste, vinegar, lemon, ice and float tests identify nothing, and the lighter, streak and acid tests mark the piece, so leave those alone.
What a home test can settle, and what it cannot
A home check has one job: sorting a drawer into the things that are probably not gold and the things that might be, so you can set the first group aside and ask about the second. It costs nothing and it is worth doing. It is also as far as a home check goes.
What it cannot do is give you a carat. The hallmarking guidance published by the Assay Offices of Great Britain states that it is not possible to detect by sight or by touch the platinum, gold, palladium or silver content of an item. That is why hallmarking exists at all. A piece can pass every check below and still be 9ct, 18ct, gold-filled or heavily plated brass, and those four are very different things.
The decision a home check is for
You are not deciding what a piece is worth, only whether it is worth photographing and asking about. A piece that a magnet lifts whole can usually go back in the drawer. A piece that passes everything has still not been identified.
The four checks worth running
None of these needs anything you would have to buy, apart from a cheap magnifier if you have none, and none of them puts a chemical on the piece.
- A magnetAny fridge magnet will do, though a stronger one is better. Hold it against the body of the piece rather than the fastening. Gold, silver, platinum and palladium are not attracted to a magnet, although a few white-gold alloys containing nickel and some platinum-cobalt alloys show a very faint pull. What matters is a strong pull on the whole piece. If a whole bangle, a whole ring or the length of a chain lifts towards it, that piece is almost certainly not solid precious metal. If nothing happens, you have learned that it is not steel, and nothing else.
- Wear points and colour break-throughLook at the tips of chain links, the underside of a clasp, the edge of a ring shank, the high points of any raised decoration and the back of a pendant where it has rubbed on clothing. Coatings wear from those places first. A pinkish, brassy or grey colour showing through usually means a different metal underneath. White gold is the exception: it is usually plated with rhodium, a bright white metal, and where that wears away the slightly yellow white gold underneath shows through. Seeing no break-through means only that the piece is not worn through yet.
- The marks, read under magnificationRest the piece on something plain and pale, bring the light in from the side rather than straight on, and look with the strongest magnifier you have. A phone camera at its closest focus works. You are looking for a number, for a small walking lion on silver, and for a shaped shield with initials in it. Marks are often worn, half-struck or filled with polish, and a mark you cannot read is not the same thing as no mark.
- Weight against sizePick two pieces up, one in each hand, and compare. High-carat gold is noticeably dense; 9ct is less so, but still heavier than brass of the same size. A chain that feels lighter than it looks may be hollow or plated. This is a comparison, not a measurement, and it works far better between two objects of similar size than on one object alone.
What each result allows you to conclude
| Check | If it passes | If it fails |
|---|---|---|
| Magnet | Not steel. Could be gold, silver, brass, bronze, pewter, aluminium or lead | Not solid gold or silver, unless only the clasp or a spring pulled, or the pull is very faint |
| Wear points | No coating has worn through where you looked. Says nothing about what is under an intact surface | Usually a coating over a different metal, unless it is white gold showing under worn rhodium |
| Marks | A full British hallmark means an assay office tested the piece before marking it. A number on its own is a claim by whoever stamped it | Nothing at all. Small items are exempt from marking and old items often lost theirs |
| Weight for size | Consistent with a dense metal. Gold, platinum, lead and tungsten are all dense | Hollow, plated, or a light alloy. Strong evidence, not proof |
Three of the four checks tell you far more when they fail than when they pass, which is true of nearly all cheap testing. The sorting guide takes the same checks through a whole box rather than a single piece.
Why a magnet passing proves nothing
The magnet is the most useful tool in the drawer and the most over-read. What it finds is iron, which in practice means steel, and to a lesser degree nickel and cobalt. It finds steel reliably. The Royal Mint explains the same effect in circulating coinage: the 5p and 10p changed from cupro-nickel to nickel-plated steel in 2011, and it is the iron in the steel core that makes them stick.
That is the limit of the information. Brass does not stick. Bronze does not stick. Pewter, aluminium, zinc alloys and most costume-jewellery castings do not stick. Neither does gold. A piece that sits quietly against a magnet has told you it is not made of steel. Plenty of base-metal jewellery is not made of steel either.
A clasp that pulls on a gold chain
Bolt rings, lobster clasps and spring-ring fastenings have a steel spring inside them, and that spring is steel on a 9ct chain exactly as it is on a plated one. If the magnet grabs the clasp and ignores the rest of the chain, that is the normal construction of a piece of gold jewellery and not a warning sign. Test the body, not the fastening.
The same goes for watch cases with a steel back on a gold front, spectacle frames with steel hinges, brooches with a steel pin and cufflinks with a steel swivel. Each of those is a mixed-material object where one small part is steel because steel is the right material for a spring. What the magnet is good for is the whole-piece pull, which rules out solid gold and silver.
Reading the marks, in British notation
A British hallmark is several small marks struck together, and the one people mean when they talk about carat is the fineness mark. It is a number in parts per thousand, not a letter and a k. The approved standards for gold under Schedule 2 of the Hallmarking Act 1973 are 375, 585, 750, 916.6, 990 and 999 parts per thousand. On jewellery, the first four are the ones you will meet.
British fineness marks for gold and what each one is
| Mark | Carat | Parts of gold per thousand |
|---|---|---|
| 375 | 9ct | 375 |
| 585 | 14ct | 585 |
| 750 | 18ct | 750 |
| 916 | 22ct | 916.6 under the Act, struck as 916 |
A piece marked 10k or 14k was made for a market that uses that notation, most often the United States, and those pieces reach British drawers constantly through inheritance and travel. They are not British hallmarks and the number is a manufacturer's claim rather than an assay office's finding. The hallmarks guide sets out what a full British mark contains and which foreign marks mean what.
A missing mark is what people most often misread. Articles below a certain weight are exempt from compulsory hallmarking, and for gold that weight is one gram, so a stud earring, a fine chain or a thin band may carry nothing and be perfectly good. The Hallmarking Act also exempts an item made before 1950 that meets the minimum fineness and has not been altered since, so it can be described and sold as precious metal without a hallmark where the seller can show both. A heavy unmarked piece, on the other hand, has not been identified yet, and only a test settles it.
The folk tests, taken one at a time
Search for this subject and you will be shown the toothpaste, vinegar, lemon, ice, lighter, ceramic streak and float tests, usually as a numbered list with no verdict attached. The verdict on each follows, with what it actually measures.
Seven household methods and what each is really measuring
| Method | What it actually measures | Verdict |
|---|---|---|
| Toothpaste rubbed on and wiped off | Abrasion. Toothpaste is a mild polish and nothing more | Measures nothing about the metal. On a plated piece it removes plating, which is the one result it can produce |
| Vinegar | Whether a dilute weak acid attacks the surface in a few minutes, which on most alloys it does not | Cannot separate gold from brass or bronze in the time anybody watches, and cannot separate 9ct from 22ct at all |
| Lemon juice | The same thing with citric acid instead of acetic | The same answer. It can also stain, and it attacks soft stones and old solder |
| The ice test | Thermal conductivity. Gold carries heat away quickly, so ice on gold melts faster than ice on plastic | The physics is real but the conclusion is not. Silver and copper conduct heat better than gold, and aluminium conducts it well enough to melt ice quickly too, so it separates metal from non-metal and nothing else |
| The lighter test | How the surface behaves when heated, which tells you nothing useful about the metal | Destructive. It discolours the surface, softens the metal, opens solder joints, ruins patina and can crack or burn a stone |
| The ceramic streak test | The colour of the trace left when metal is dragged across unglazed ceramic | Does separate gold from iron pyrite for a prospector. On jewellery it marks the piece, and a gilt surface leaves a gold streak too |
| The float test | Whether the object is denser than water | Brass, steel, lead, pewter and glass all sink. It excludes almost nothing |
The lighter test needs more than a line in a table. It does real harm, because heat does not stop at the surface. It softens the metal, takes the temper out of a spring and opens the solder at a link or a setting. It can craze or burn pearls, amber, coral, opal, turquoise and emerald without warning, and a piece worth keeping as jewellery can be spoiled in seconds.
The acid test is left out of the table because it is a long-established trade method, not a household one. The Goldsmiths' Company Assay Office describes touchstone testing as lightly rubbing the article onto a stone of basanite, a fine-grained black stone. Every other traditional method it lists needs a sample taken from the article by scraping or drawing. Testing acids are corrosive, and applied to the piece rather than to a streak they leave a permanent dull mark. No procedure for either is given here, and it is wise to be wary of any page that gives one.
Three methods that mark the piece
The lighter, the streak test and acid all change the piece. None of the marks can be undone, each lowers what an object is worth if it turns out to be worth more than its metal, and none of them would have told you a carat.
Carat is the number, and no kitchen test reaches it
Suppose every check goes your way. The magnet ignores the piece, the wear points are clean, there is a mark you cannot quite read, and it feels heavy for its size. You are holding something that might be gold, but you do not know whether it is 9ct or 22ct, and 22ct holds well over twice as much gold as 9ct in the same weight of metal.
That is why the home tests stop where they do. Establishing a fineness needs an instrument or a laboratory, and more than one method, because each has a gap. The Goldsmiths' Company Assay Office uses non-destructive X-ray fluorescence alongside touchstone work and laboratory methods that melt or dissolve a sample (cupellation, titration and, for platinum and palladium, ICP analysis). It notes that on occasion the surface layer has to be removed to find out what the alloy underneath is. The XRF guide explains where surface reading goes wrong. At Assaypost, every piece is weighed and checked for its metal before an offer is made; the methods used depend on the piece and are listed on the offer, and the testing page sets out their limits.
The most useful thing you can do at home is take a good photograph. A sharp picture of a mark, a worn link tip and the whole piece beside a ruler answers more than any of the seven methods above. From photographs we can give you a rough idea of what the marks appear to say before you post anything, though a photograph cannot prove what a piece is made of. That is what the photo route is for, and the sell gold page explains what happens after it.
Seller risks and exceptions
- Testing a piece that should not be tested. A signed brooch, a pocket watch or a piece of antique silver can be worth considerably more intact than as metal, and a scratch, a burn or an acid mark on it cannot be undone. Leave the testing to whoever buys or values it, and see what to keep back.
- Reading a pass as a result. Three of the four checks only tell you something when they fail. A piece that passes all four has not been identified, only not ruled out.
- Trusting a number stamped on a piece. A fineness mark on an unhallmarked item is a claim by whoever struck it. Gold-filled stock is stamped with a real carat and a fraction, and thickly gilded pieces are stamped with plating codes that look like fineness marks.
- Cleaning or polishing before you photograph. Polishing takes metal off the highest points of a piece, and that is often where the hallmark sits. See whether to clean before you sell.
- Assuming a magnet-free clasp means a gold chain. Plated brass does not stick to a magnet either, which is why the wear points and the weight matter alongside it.
- Forcing a watch case back to look for marks. Cases are snapped or screwed shut, and a slipped knife marks the case and the movement. Leave it closed, photograph the outside and ask.
When not to test it yourself at all
Put the piece down and ask first
Anything with a maker's name on it. Any watch by a known maker, running or not. Any coin, because a coin's surface is part of what it is worth and a test mark is permanent. Anything set with stones, because acids, heat and abrasives all attack the stones before they attack the metal. Anything old enough that the marks are worn shallow, because a rub or a scratch takes the last of them off.
For those pieces, photograph the marks and the piece and ask what it appears to be, or take it to a valuer or a saleroom for an opinion. The guide to valuation, appraisal or offer sets out which of those you actually need and which you do not, and what we buy says plainly which materials are taken here and which are not.
Questions sellers ask
Does the toothpaste test work on gold?
No. Toothpaste is a mild abrasive, so all it does is polish the surface. It cannot distinguish gold from brass and it cannot tell you a carat. On a plated piece the one thing it can achieve is taking some of the plating off.
My gold has no hallmark. Does that mean it is not gold?
Not on its own. Gold articles under one gram are exempt from compulsory hallmarking, so small studs, thin chains and light bands often carry nothing. An item made before 1950 and not altered since can also be sold as precious metal without a mark, if it meets the minimum fineness. A heavy unmarked piece is unidentified, and only a test settles it.
Why does my gold chain stick to a magnet at the clasp?
Because the clasp contains a steel spring, on real gold chains as much as on plated ones. Hold the magnet to the body of the chain instead. A pull along the links themselves means the chain is not solid gold; a pull only at the fastening tells you nothing about the chain.
Can I tell 9ct from 18ct at home?
No. Colour is not reliable, because alloys are mixed to hit a colour, and weight in the hand cannot separate them either. The fineness mark is the only thing at home that speaks to carat. In a full British hallmark it was struck by an assay office after testing; a number stamped on its own is the maker's claim until someone tests it.
Is the acid test safe to do myself?
It is not a home method. Touchstone testing works by rubbing metal off the article onto a stone, testing acids are corrosive, and used on the piece itself they leave a permanent mark. No procedure for it is published here.
What should I do instead of testing it?
Run the magnet over it, look at the wear points, photograph any marks with the light coming from the side, and send the pictures for a rough idea of what the pieces appear to be. It costs nothing, commits you to nothing and tells you more than any kitchen method, though a photograph cannot prove what a piece is made of.
Found a mark you cannot read?
Photograph the stamp close up, in good daylight, holding the piece still. A person reads it and writes back, including what the mark alone cannot tell you.
Nothing is sold until you say yes. Say no and everything comes back by tracked, signed-for post at our cost.
No WhatsApp? Call or text 07369 296063, 6am to 10pm, seven days a week.
When the mark is not the whole story
A stamp can be worn, faked or missing from perfectly good gold. So every piece posted is weighed and checked for its metal as well as read, and the offer lists the methods used.