Why your chain and a pearl necklace wear out the same way

Neither one really stretches. Each joint gains a fraction of a millimetre, and dozens of joints add up to a strand you can measure with a ruler. Here is the shared geometry, the different physics underneath, and how to check both.

A long necklace of cultured baroque saltwater pearls laid out in loops
Photo: W.carter, CC BY 4.0, via Wikimedia Commons

A motorcycle chain and a pearl necklace wear out the same way geometrically: neither part really stretches, but every joint gains a tiny gap, and the gaps add up. A 520 chain is due for replacement when 20 links measure about 1.7% longer than new, which is only about 0.27 mm of extra play at each pin. A knotted pearl strand lengthens the same way, as the silk at each knot stretches and frays. The physics underneath differs (metal worn away by sliding in the chain, silk deformed and abraded in the necklace), so the fixes differ too: the chain is replaced, the pearls are restrung.

Key takeaways
  • 317.5 mm is the length of 20 new links of 520 chain; about 323 mm is a typical replacement limit.
  • That limit is only 0.275 mm of extra play at each pin and bush.
  • Chains start skipping teeth near 3% elongation, under 1 cm over 20 links.
  • A pearl strand lengthens the same way: about 0.18 mm per knot for 1 cm on a 45 cm strand.
  • Archard's law (V = KFs/H) drives chain wear; wet, abraded silk drives necklace wear.
  • The fix differs: a chain is replaced with its sprockets, a strand is just restrung.

This is a tier-4 "collision", pairing two unrelated things to prove the real link between them. It works through the maths of elongation, the wear law behind the chain, the material science behind silk, and a calculator that measures either one. The verdict says exactly how far the comparison holds.

317.5 mmlength of 20 new links of 520 chain
1.7%a typical 20-link service limit (323 mm)
≈3%elongation at which chains start skipping teeth
Once a yearGIA's restringing advice for pearls worn often

Does a chain actually stretch?

No. The steel plates do not get longer in any meaningful way; the joints get looser. A roller chain is a loop of links, and each joint is a hardened pin turning inside a bush.

Every time a joint wraps onto or off a sprocket, the pin rotates inside the bush under load, and both surfaces lose a little metal. The clearance grows, and each pitch, the distance between neighbouring pins, gets a few micrometres longer.

Shoji Noguchi and colleagues at Tokyo University of Science found that the pin and bush do not even touch along their full length: the pin bends under tension, and wear happens only at its tips.

The story

La Peregrina, a pear-shaped natural pearl first recorded at 223.8 grains (about 11.2 g), has outlasted the mounts made to hold it. While the Abercorn family owned it, it fell out of its setting twice: once into a sofa at Windsor Castle, and once during a ball at Buckingham Palace. Both times it was found.

In 1969 Richard Burton bought it at Sotheby's for $37,000 and gave it to Elizabeth Taylor for Valentine's Day. One day in the Burtons' suite at Caesars Palace it was gone. Taylor, thinking her puppy was chewing something, looked in its mouth and found the pearl: "It was—thank God—not scratched." She then had Cartier redesign the necklace, setting La Peregrina with pearls, diamonds and rubies. The pearl itself never wore. What failed, every time, was the joint that held it, and a chain and a strand fail the same way.

Wikipedia: La Peregrina pearl

Elongation is measured as the change in length over a known number of pitches:

\[ e = \frac{L_{\text{measured}} - L_{\text{new}}}{L_{\text{new}}}, \qquad L_{\text{new}} = n\,p \]
  1. New 520 chain. Pitch \( p = 5/8 \) inch = 15.875 mm, so 20 links measure \( 20 \times 15.875 = 317.5 \) mm.
  2. Service limit. Kawasaki's manual for a 520 chain gives 323 mm over 20 links: \( e = (323 - 317.5)/317.5 = 1.73\% \).
  3. Per joint. Spread over 20 pitches, \( 5.5/20 = 0.275 \) mm of extra length per pitch. That is the whole wear budget of each pin and bush combined.
  4. Whole chain. On a chain of, say, 118 links (an assumed length), 1.73% is about 32 mm, enough to move the rear axle several adjustment notches.
Evaluation of Wear between Pin and Bush in Roller ChainShoji Noguchi, Hideaki Yoshiba, Satoshi Nakayama, Tohru Kanada · Journal of Advanced Mechanical Design, Systems, and Manufacturing 3(4), 355–365 · 2009

Testing a whole chain takes too long, so the team built a wear tester that loads only two links in tension, measured the worn parts on a roundness tester, and used finite-element analysis to explain the pattern. Pins and bushes did not touch along the centre line; bending under load concentrated wear at the pin tips. The paper repeats the industry rule that a chain starts skipping teeth at about 3% elongation. The limit: a clean, lubricated lab rig, without the grit and water a road chain sees.

Read the paper (free)
A labelled rendering of a roller chain showing the outer plates, inner plates, pins, bushes and rollers
The parts of a roller chain. Wear happens where each pin turns inside its bush; the plates themselves barely change.Photo: Niabot, CC BY-SA 3.0, via Wikimedia Commons
New jointWorn jointpinpinbush (outer ring) and rollerclearance grows at the pin tipsEach joint adds about 0.275 mm at a 1.73% limit; 20 joints add 5.5 mmextra gap
The worn joint: the pin sits off-centre in an enlarged bush bore. That small offset, repeated at every pitch, is the whole of "chain stretch" (drawing not to scale).

What law decides how fast a chain wears?

Archard's wear law, published by J. F. Archard in 1953, is the standard model for sliding wear, and chain researchers use it to simulate pin-and-bush wear. It says the volume of material removed is proportional to the load and the sliding distance, and inversely proportional to the hardness of the softer surface:

\[ V = K\,\frac{F\, s}{H} \]

K is a dimensionless wear coefficient that can change by orders of magnitude between a clean, oiled joint and one packed with grit, which is why lubrication and sealing matter more than anything else.

The sliding distance s is where geometry comes in. Each joint rotates by 360°/N as it wraps onto a sprocket of N teeth, and again as it leaves. Taking a pin radius of about 2.5 mm, a 15-tooth front sprocket and a 41-tooth rear (both assumptions for a typical 200–400 cc bike):

  1. Sliding per articulation, front. \( r\theta = 2.5 \times (2\pi/15) = 1.05 \) mm.
  2. Sliding per articulation, rear. \( 2.5 \times (2\pi/41) = 0.38 \) mm.
  3. Per trip round the loop. Each joint articulates on and off both sprockets: \( 2 \times (1.05 + 0.38) = 2.86 \) mm.
  4. Laps per kilometre. The chain moves at the rear sprocket's pitch radius (\( 15.875/(2\sin(180^\circ/41)) = 103.7 \) mm) relative to a 140/70-17 tyre of about 314 mm radius, so 1 km of road is about 330 m of chain travel, or \( 330/1.873 \approx 176 \) laps of a 118-link chain.
  5. Sliding per joint. \( 176 \times 2.86 \approx 0.5 \) m per kilometre ridden, or about 10 km of loaded sliding at every pin over 20,000 km.

Two practical results fall out:

  • Small front sprockets wear chains faster. The articulation angle is larger, so on paper a very small countershaft sprocket costs chain life.
  • Load is wear. Because V is proportional to F, hard launches and wheelies, which multiply chain tension, eat chain life directly.

The sprocket gearing calculator shows the trade-off from the gearing side.

Wear Measurement on Chain Joint Components Using a Roundness InstrumentDaniel Sappok, Bernd Sauer · Periodica Polytechnica Mechanical Engineering 59(2), 51–59 · 2015

Working on engine timing chains, the authors developed a way to measure wear on individual pins and bushes with a roundness instrument, rather than only measuring total chain length, and compared pins with different surface treatments. They note a tolerable timing-chain elongation of 0.3–0.5%, which for an 8 mm pitch is about 40 µm, and use Archard's model, adapted for local contact pressure, to compute how the bush contour wears. Even new bushes showed about 14 µm of form deviation. The limit: timing chains in oil are a gentler world than an exposed drive chain.

Read the paper (free)

How does a pearl necklace "stretch"?

The same way, one joint at a time, but the joint is silk. A classic pearl strand is threaded on silk with a small knot between every pearl. The knots stop pearls rubbing each other and stop the whole strand scattering if the thread breaks.

Over time the silk at each knot loosens, stretches and frays where it passes through the drill holes, so a gap opens between each pearl and its knot. The Gemological Institute of America advises restringing pearls worn often about once a year, and warns that wet silk can stretch and attract dirt, so a strand should dry completely before being handled.

Chain: rigid plates, pin-and-bush jointsPearls: rigid pearls, silk-knot joints+0.275 mm per joint+0.18 mm per knotSame sum for both:total growth = joints × gap per joint20 chain pitches × 0.275 mm = 5.5 mm (1.73%)56 pearls × 0.18 mm = 10 mm (2.2%)
Both strands are rigid elements joined by a wearing link. Neither element grows; the joints do, and the gaps add up to a length you can measure with a ruler.

Count the joints and the same arithmetic appears. A 45 cm strand of 7 mm pearls with about 1 mm knots has about \( 450/8 \approx 56 \) pearls. If the strand has grown by 1 cm (an illustrative figure, not a standard), that is:

\[ \delta = \frac{\Delta L}{N} = \frac{10\ \text{mm}}{56} \approx 0.18\ \text{mm per joint}, \qquad e = \frac{10}{450} = 2.2\% \]

About 0.18 mm per joint and 2.2% overall, the same order as a chain at its 1.7% limit with 0.27 mm per joint. A jeweller's first check, that a strand hangs longer than it used to or that you can see thread between the pearls and the knots, is the necklace version of measuring 20 links.

Close view of a silver clasp joining a strand of baroque pearls
Every pearl is a joint. Like a chain's pins, each one contributes a fraction of a millimetre of slack as its silk stretches and wears.Photo: W.carter, CC BY 4.0, via Wikimedia Commons

Why does silk give way?

Because silk is elastic only up to a point and then deforms permanently, and humidity makes that easier. Silkworm silk (Bombyx mori) shows what materials scientists call elastic-plastic behaviour: it springs back under small loads, but beyond its yield point it keeps the extra length.

The tension in a hanging necklace is small, but the silk at each drill hole sees concentrated bending and rubbing every time the strand moves on the skin, and moisture from skin, perfume or a swim softens it.

Artificial and natural silk materials have high mechanical property variability regardless of sample sizeGabriele Greco, Hamideh Mirbaha, Benjamin Schmuck, Anna Rising, Nicola M. Pugno · Scientific Reports 12, 3507 · 2022

The authors tensile-tested 200 fibres each of Bombyx mori cocoon silk, degummed silk, native spider silk and an artificial spider silk, with carbon fibre for comparison. The relative standard deviation of strength and strain at break was 10–50% across fibre types, and testing more fibres did not shrink it; about 97% of the silkworm silk samples showed the typical elastic-plastic curve. They also note that humidity plasticises regenerated silkworm silk. The limit: single fibres in a lab, not twisted, knotted thread worn on skin, but the scatter explains why one knot fails long before its neighbours.

Read the paper (free)

Why does a worn chain ruin the sprockets, and a worn strand the pearls?

Because a longer pitch no longer fits the teeth. A sprocket is cut for one exact pitch, and its pitch circle diameter is:

\[ D = \frac{p}{\sin(180^\circ/N)} \]

For a 41-tooth sprocket and a 15.875 mm pitch, D = 207.4 mm. A chain worn 1.73% behaves as if it wants a pitch circle 1.73% larger, 211.0 mm, so its rollers climb toward the tips of the teeth, the load lands on fewer teeth, and the teeth wear into hooks.

A new chain on hooked sprockets then wears fast in turn, which is why chains and sprockets are normally replaced as a set.

The necklace version is gentler: once the silk loosens, pearls slide and touch, and nacre rubbing on nacre scratches. The knots are the pearl strand's sprocket teeth, keeping every element at its pitch.

Neither the plates nor the pearls get longer. Only the joints do, and the joints are where you measure.

Elongation checker: chain or necklace

Length of 20 links of 520 chain: new, at service limit, at skipping point
See the data as a table
20-link length (mm)
New317.5
Kawasaki service limit (1.7%)323
Skipping teeth (about 3%)327

Source: Pitch 15.875 mm; Kawasaki Z750 service manual limit; 3% figure from Noguchi et al. (2009) · The difference between a new chain and one that is skipping teeth is under 1 cm over 20 links.

How do you measure each one?

For a chain, measure 20 links under tension; for pearls, compare the strand with its original length and look at the knots. Step by step:

  1. Chain: take up the slack. On the stand, rotate the wheel so the lower run is slack and the top run is pulled taut, or hang a weight as your manual specifies.
  2. Chain: measure 20 links. Use a vernier caliper or a steel rule from the centre of one pin to the centre of the 21st pin. Compare with your manual; for a 520 chain, 317.5 mm is new and many limits sit around 323 mm.
  3. Chain: check at three places. Chains wear unevenly; take the worst reading. Also look for tight spots, kinked links and a chain you can pull away from the back of the rear sprocket.
  4. Necklace: lay it flat. Measure the strand without tension and compare with the length on the receipt or a previous measurement.
  5. Necklace: check the knots. If you can see thread between a pearl and its knot, or knots look fuzzy or dark, the silk is worn. GIA advises yearly restringing for pearls worn often.
Property520 motorcycle chainKnotted pearl strand
JointHardened pin in a bushSilk thread through a drilled pearl, knotted
Spacing per joint15.875 mmAbout 8 mm (7 mm pearl plus knot)
How it lengthensMetal removed by sliding wear (Archard)Silk yields, creeps and abrades
Typical limitAbout 1.7% over 20 links; skipping near 3%No standard; any visible gap at the knots
Per-joint growth at the limitAbout 0.27 mmAbout 0.18 mm in the 1 cm example
What it damages nextSprocket teeth hookPearls rub and scratch
RemedyReplace chain and sprockets as a setRestring with new silk and knots
Best preventionClean, lubricate, keep seals intactKeep dry, put on after perfume, restring on schedule
Silk threads being reeled from boiled silkworm cocoons
Silk reeled from cocoons. Strong for its weight, but elastic-plastic and sensitive to moisture, which is why a pearl strand slowly lengthens.Photo: Dave Proffer, CC BY 2.0, via Wikimedia Commons

How sensitive is the comparison?

  • Different chain limits. Makers set different limits; if yours is 2% instead of 1.73%, the per-joint budget rises from 0.275 to 0.318 mm. The geometry is identical; only the threshold moves.
  • Smaller pearls. A 45 cm strand of 5 mm pearls has about 75 joints, so the same 1 cm growth is only 0.13 mm per joint. More joints hide more wear, exactly as a longer chain does.
  • Dirt. In Archard's law, grit raises K, sometimes by orders of magnitude. An unsealed chain run through dust and monsoon water can reach its limit many times sooner than a clean, sealed one. The chain cleaning and lubrication guide covers how to keep K low.
  • Where the analogy fails. A chain's wear removes material and is irreversible; a necklace is restored completely by new silk, because the pearls themselves are not the wearing part. And a chain carries kilonewtons, a necklace well under a newton.

So do they really wear out the same way?

In geometry, exactly. Both are strings of rigid elements joined by a compliant link, both lengthen by the sum of tiny per-joint gains, both are judged by measuring a fixed number of joints, and both damage their neighbours once the pitch drifts. In mechanism, no: the chain's joints lose metal by sliding wear, while the necklace's silk deforms plastically and frays. The chain story in full is in what happens if you never change your chain, and checking it is part of any used bike inspection.

Verdict: same geometry, different physics

A 520 chain at a 1.73% limit has gained about 0.275 mm per joint; a 45 cm pearl strand that has grown 1 cm has gained about 0.18 mm per joint across 56 knots. The elongation maths, e = ΔL/(np), and the way to measure it are identical. The cause is not: metal pins and bushes wear by Archard's law, V = KFs/H, while silk yields, creeps and frays, especially when wet. So the link holds for measurement and diagnosis, and breaks at the remedy: a worn chain is scrap, a worn strand only needs new silk.

For more of the machines behind these posts, see Yantra.

What if next?

  • A cricket ball's swing and your helmet's buffeting are one phenomenon. One boundary layer, two very different results.
  • Your suspension is an idli. Springs that return and sponges that damp, explained with breakfast.
  • A gecko's foot and a racing slick stick for related reasons. Where grip comes from when surfaces meet.

Questions people ask

How do I know if my motorcycle chain is stretched?

Measure 20 links pin-centre to pin-centre under tension and compare with your manual. A new 520 chain measures 317.5 mm over 20 links; a typical limit is around 323 mm, about 1.7% longer. Tight spots and a chain you can pull off the back of the sprocket are other signs.

Can you fix a stretched chain by adjusting it?

Adjusting removes slack but does not restore the pitch. The worn chain still rides high on the sprocket teeth and keeps wearing them, so once it passes the limit it should be replaced, usually with the sprockets.

How often should pearls be restrung?

GIA advises about once a year for pearls worn often. Strands worn occasionally can go longer, but restring sooner if the strand has lengthened or the knots look loose or dirty.

Why are there knots between pearls?

Knots keep pearls from rubbing and scratching each other, hold each pearl at its spacing, and mean that if the thread breaks, you lose at most one pearl instead of the whole strand.

Sources

  1. Noguchi, S. et al.: Evaluation of Wear between Pin and Bush in Roller Chain, JAMDSM 3(4) (2009) (source for pin-tip wear and the 3% skipping rule)
  2. Sappok, D. and Sauer, B.: Wear Measurement on Chain Joint Components Using a Roundness Instrument, Periodica Polytechnica Mech. Eng. 59(2) (2015) (source for Archard's model in chain joints and the 0.3–0.5% timing-chain limit)
  3. Greco, G. et al.: Artificial and natural silk materials have high mechanical property variability, Scientific Reports 12 (2022) (source for silk's elastic-plastic behaviour, 10–50% scatter and humidity plasticisation)
  4. Kawasaki Z750 service manual, drive chain wear limit (via ManualsOnline) (source for the 323 mm 20-link limit)
  5. GIA: Tips for caring for your jewelry (source for yearly restringing and wet silk stretching)
  6. Archard, J. F.: Contact and rubbing of flat surfaces, Journal of Applied Physics 24 (1953) (source for the wear law, as cited by Sappok and Sauer)
  7. Wikipedia: La Peregrina pearl (source for the Abercorn losses, Burton's $37,000 purchase, the Caesars Palace loss and the Cartier reset)
  8. Lubricant performance in bicycle roller chains, Sports Engineering (2026)
  9. Identification of key parameter influence on efficiency of chain drive for track cycling, Mechanics & Industry (2025)
chain wearchain stretchpearl necklacerestringingArchard wear lawsilk thread520 chainelongation
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of RoadHook Machines.

Keep reading in RoadHook Machines

From the other blogs