To untangle a necklace, you unclasp it, lay it flat on a hard pale surface, and open the knot from the outside with two straight pins — you never pull the two ends apart. That is the whole method, and it works on almost every open-link chain. The more useful question is the one nobody answers: how did a chain sitting still in a closed drawer tie itself in the first place? Two physicists measured that, and the answer changes how you store the thing.
Key Takeaway
A knot needs a loose end to thread itself through. Work the knot open from the outside with pins instead of pulling, and store the chain with its clasp fastened — a closed loop has no end to do the weaving.
Get the Knot Out First: Six Steps
Give yourself ten minutes and good light. Rushing this is how a chain that needed patience turns into a chain that needs a bench jeweler. The method works on every open-link weave among the solid silver chains we make — cable, curb, Figaro, Byzantine and the rest.
Open the clasp
A closed loop has nowhere to unthread to. Releasing the clasp gives the knot two free ends to travel back out through, which is the reverse of how it got in.
Lay it flat on something hard and pale
A white plate, a sheet of paper, a kitchen worktop. Fabric swallows links and hides which loop sits on top. Flat also stops the chain's own weight from pulling the knot tighter while you work.
Spread the knot before you touch it
Push the slack chain toward the knot from both sides. A knot with slack in it is a loose tangle of loops. A knot with the chain pulled straight either side is a solid lump you cannot read.
Two pins, from opposite sides
Straight sewing pins or a pair of fine needles. Slide a tip under an outer loop and lift, then do the same from the other side. You are widening the knot, not driving anything through it. Never lever a pin against a link — the pin is harder than the silver.
Free one loop at a time
Find the outermost loop, walk it back over the bundle, then stop and look again. Most knots come apart in three or four of these moves once the bundle has room in it.
Check the links before you wear it
Run the chain through your fingers. Any link that now sits open, oval where its neighbours are round, or flattened, is a break waiting to happen on a night out. That is a five-minute job for a jeweler and a lost pendant if you ignore it.
Why a Chain Ties Itself Inside a Closed Drawer
In 2007, two physicists at UC San Diego, Dorian Raymer and Douglas Smith, published an experiment in the Proceedings of the National Academy of Sciences that answers this properly. They dropped a length of string into a clear box, rotated the box, tipped the string out, and checked whether it had knotted. Then they did it again. And again — 3,415 times.
Their standard run used a 3.2 mm braided string in a 0.30 m cube, spun at one revolution per second for ten seconds. Ten seconds. In that time they collected 1,127 knots, and 1,007 of them were identifiable prime knots with three to eleven crossings. They logged around 120 distinct knot types, including every prime knot up to seven crossings. Complex knots form in seconds, in the dark, with nobody touching anything.
Three of their results matter if you own chains, whatever weave yours is built from.
Short strands hardly knot at all
Below 0.46 m — about eighteen inches — they got no knots whatsoever. The confinement and tumbling simply did not bend the string enough. From 0.46 m up to 1.5 m the probability climbed steeply, and from 1.5 m to 6 m it flattened out around 50%. Not 100%. Even a six-metre string, tumbled, knotted itself only about half the time.
Shaking it harder made things better, not worse
This is the one nobody expects. In their 0.15 m box, spinning at one revolution per second knotted the string 50% of the time. Spinning three times faster dropped it to 17%. At high speed the string got pinned against the walls and stopped tumbling properly. Meanwhile, tripling the time from ten seconds to thirty seconds took the same setup from 50% to 74%. Knots are not built by violence. They are built by long, gentle, repeated agitation — which is a fair description of a chain riding in a jacket pocket all day.
The knot is made by the end
Their explanation is the part worth keeping. Confined in a box, a stiff string coils. That coiling parks several roughly parallel runs of string right beside the loose end, and as the box turns, that end weaves over and under its neighbours — what the authors describe as random braid moves. The knot is not made by the middle of the string. It is made by the end, travelling.
⚠️ Read this before you quote the numbers: Raymer and Smith tested string, not jewelry. No metal chain, no clasp, no pendant, no comparison of link styles. The eighteen-inch figure is only the point at which that particular string stopped knotting in that particular box. It is not a safe necklace length. Anyone who tells you an 18-inch chain has a 50% chance of knotting is inventing it. What transfers is the mechanism, not the percentages.
The One Move That Tightens Every Knot
Grabbing both ends and pulling. Everyone does it, because it feels like the direct approach, and it is the one action guaranteed to make the problem worse. A knot converts tension into grip: every gram of pull you apply gets concentrated at the handful of links inside the bundle, and those links are the softest thing in the equation.
Sterling is .925 — 92.5% silver, with the balance mostly copper. That copper buys strength over fine silver, but a thin chain is still a chain of small soldered rings — a 2mm cable chain is about 5 grams of metal spread over twenty inches — and it gives way at a link bend or a solder seam long before anything else. A knot you pulled tight for thirty seconds can take an hour to open. A link you pulled open takes a jeweler.
Oil, Powder, Freezing: What's Worth Trying?
Search this problem and you will be told to freeze the chain, tap it with a spoon, hang a weight off it, dust it with talc, or soak it in window cleaner. We went looking for who actually stands behind those. Freezing, tapping and hanging a weight have no jewelers' association or conservation body behind them that we could find — and two of the three apply the exact tension the step above tells you to avoid. Skip all three.
A drop of baby oil or mineral oil is different: it is genuinely common practice among working jewelers, and on a plain solid silver chain it is a reasonable last resort once the pins have stalled. Put it on the knot, not the whole chain, and wash it off with warm water and a little mild soap afterwards, then dry it completely. Keep it well away from anything strung on silk, anything glued, anything with a porous or treated stone, and any chain with a deliberately darkened finish. Ammonia-based window cleaner earns you nothing that mild soap does not, and it belongs nowhere near pearls or soft stones. Talcum powder gets recommended a lot; we have never found a source behind it stronger than repetition, and powder packed into a dense weave has to come back out again.
Does the Weave Change How Much It Tangles?
Here is the honest answer, and it is not the one most jewelry sites give: nobody has published a tangle-rate ranking of chain weaves. Not the Jewelers of America, not the bench-supply houses, not anyone with a lab. So when a page tells you that links above a certain length-to-width ratio tangle 4.7 times more often, ask it for the study. We went looking for that one, and the journal it cites does not appear to exist.
What the trade does document is kinking and repairability — two things that decide whether a bad tangle costs you an evening or costs you the chain. A kink is a bend the metal keeps. A knot is a topology problem you can undo. Confusing the two is why people panic at the wrong chains. Here is what the sources actually say — and what they conspicuously do not.
| Weave | How it is built | Kink and repair reality |
|---|---|---|
| Cable | Single links joined end to end, each one perpendicular to the last | Open links, so a jeweler can work on it. No published tangle data. |
| Curb | A cable chain whose links are twisted so they lie flat | Jewelers of America calls curb among the most durable, and relatively simple for an experienced jeweler to repair. |
| Figaro | One longer link alternating with three to five shorter ones, all twisted flat | Because the links interlock like a curb, JA rates it a simple repair job. The long links carry no documented tangle penalty. |
| Byzantine | Repeating pairs of interlinked jump rings — substantial, but very flexible | Dense enough that a knot struggles to seat in it. That is an observation from handling them, not a measurement. |
| Wheat / Spiga | A rope variation — figure-eight links braided into an almost square profile | The braided section has little room to fold sharply on itself. No published repair or tangle rating. |
| Rope | Twisted links each connected to two others, giving a continuous spiral | JA: can be difficult to repair, especially hollow ones — the spiral has to be rebuilt to match. |
| Franco | An intricate square-profile weave — curb from two faces, wheat from the other two | Rio Grande's bench guide says outright that Franco resists kinking. That is a kink claim, not a tangle claim. |
| Herringbone | Rows of flat V-shaped links facing opposite ways — low, liquid, mirror-like | JA: thin, with a structure that gives it a tendency to twist and kink, and challenging to repair. The riskiest chain on this list. |
| Snake | Curved plates or bands packed tightly into a smooth, flexible tube | JA describes it as thin and delicate. The links cannot be opened and rewoven the way an open-link chain can. |
Handling this stuff daily, the pattern we notice is about slack rather than link shape. A fine 2mm cable chain is light, floppy and full of slack — close to the condition the tumbled string was in. Our 3mm Byzantine puts 33 grams over the same length and barely has a loose joint in it. The 18mm Cuban runs 315 grams and 7mm thick; it does not tangle for roughly the reason a scaffolding pole does not tangle.
None of that is a measurement and we are not going to dress it up as one. But if tangling is the thing that drives you mad, mass and density are the honest levers, not link shape. A 3.5mm wheat chain at 19 grams on the 20-inch length sits about halfway along that spectrum, which is why it is the one we hand people who want a daily chain and never want to think about this again.
The Weaves We Do Not Stock, and What That Tells You
We sell thirteen plain silver chain necklaces: cable, curb, Figaro, Cuban, rolo, Byzantine, wheat, ball. Every one is an open-link weave, and there is not a herringbone, a snake or a Franco among them. That is not a verdict on those chains — a herringbone catches light in a way nothing on our list can. It is a decision about the bad day. Every chain we sell can be opened, rebuilt and closed again at a bench, right down to the 5mm rolo with its individually soldered oval links. A herringbone with a crease in it stays creased, and Jewelers of America's own guidance flags it as challenging to repair.
That matters more here than at most shops, because none of our pendants ships with a chain — the chain is always a separate decision. If you are picking one to carry a heavy cast pendant, the chain that survives being knotted is worth more than the chain that photographs best.
Storing Chains So This Stops Happening
Everything above is repair work. This part is the fix, and it follows straight from the mechanism: the knot is tied by a travelling end, so take the end away and give the chain nothing to travel through.
💡 The one habit worth building: fasten the clasp before you put the chain down. A clasped chain is a closed loop, and a closed loop has no free end to weave over and under its neighbours. It takes half a second and it is the single highest-value thing in this article.
Be clear about what that buys you, though. Closing the clasp removes the route the physicists described. It does not make a chain tidy. A closed loop still coils, still folds, still hooks its own clasp through a link, and still wraps itself around the other three chains in the dish. Fastening the clasp is necessary. It is not sufficient.
So the second rule is separation. One chain per hook, one chain per bag, one chain per compartment. Two chains sharing a pocket do not need a free end to become one object — they simply thread through each other, and untangling two chains is meaningfully harder than untangling one. This is also the point where anti-tangle storage and anti-tarnish storage turn out to be the same job: an individual sealed bag keeps a chain away from its neighbours and away from the sulfur in the air at the same time. Our guide to keeping silver from going dark in the drawer covers that side of it in detail.
Hanging or flat is mostly a matter of what you own. An ordinary open-link chain is happy either way, as long as it hangs on its own hook rather than sharing one. Anything flat and fluid — herringbone, snake, omega — should be laid straight rather than hung, folded or balled, because those are the constructions that keep a bend once they take one. And for travel, a roll only counts as separation if each chain gets its own secured slot. Four chains in one pocket of a travel roll is a dish with a zip on it.
One more thing the experiment implies that nobody says out loud: the worst place for a chain is not the drawer. It is a coat pocket, a gym bag, or a glovebox — hours of gentle, low-speed agitation, which is precisely the condition that produced the highest knot rates. If you take a chain off at work, clasp it and put it somewhere still. Do not carry it loose all afternoon.
When to Stop and Hand It Over
Twenty minutes of pin work with no progress is a message. Stop if a link has already opened or gone flat, if the knot has swallowed the clasp or a pendant bail, if the chain is hollow or very fine, or if the pins are marking the metal instead of moving it. Stop immediately on a herringbone or snake chain that has started to crease — every extra minute of handling is what turns a recoverable tangle into a permanent one. A jeweler has bench tools, magnification, and no emotional attachment to your chain, which is a genuine advantage at eleven at night.
Frequently Asked Questions
Can a necklace really tie itself into a knot?
Yes, and fast. In a 2007 experiment published in PNAS, physicists tumbling a string in a box for ten seconds at a time produced 1,127 knots across 3,415 trials, including every prime knot up to seven crossings. The mechanism is a loose end weaving through nearby coils. No handling required.
Does fastening the clasp actually stop a chain tangling?
It removes the main route. A knot forms when a free end threads back through the chain's own coils, and a clasped necklace is a closed loop with no free end available. It will still coil, kink and wrap around other chains, so pair the habit with one chain per bag.
Which sterling silver chain tangles the least?
Nobody has published a tangle ranking of chain weaves, so treat any numbered claim with suspicion. In practice the denser chains have less slack to fold: a 3mm Byzantine carries 33 grams over the same length where a 2mm cable chain carries 5. Mass is the honest lever here.
Should I put baby oil on a knotted chain?
Only as a last resort, and only on a plain solid silver chain. Put one drop on the knot itself, keep working with pins, then wash it off with warm water and mild soap and dry the chain fully. Keep oil away from strung pearls, glued settings and porous stones.
The chain that never tangles is the one you clasp before you set it down. Everything else — the pins, the flat plate, the patience — is just cleaning up after the ten seconds you skipped. If you are replacing a chain that lost this fight, the heavier weaves in our men's silver necklace range are the ones we see come back for repair the least.