Waves

The band that will not go sharp

Stretch a rubber band between two fingers and pluck it, then stretch it further and pluck it again. The note rises at first and then stops rising, though the band is longer and tighter each time. Stretching raises the tension and thins the band, so waves run along it faster; but they have further to run, and for rubber the two keep almost exactly in step. A band at three times its length is within two per cent of the highest note it will ever make, and that note belongs to the band's rest length and its rubber, not to how far it is pulled. A steel string is the opposite: a few hundredths of a per cent of stretch is a semitone.

Assumes: Only some notes fit, and that is where discreteness comes from · Entropy is a count, and the arrow of time is arithmetic

A child’s first musical instrument is often a shoebox with rubber bands stretched across its open top. It works, after a fashion: each band gives a note when plucked, and the bands give different notes. What does not work is tuning them in the obvious way. Stretch a band further and its note rises a little, then hardly at all; the band is longer and tighter, and visibly thinner, and still sounds much the same. A guitarist used to steel strings, where a quarter-turn of a tuning peg moves the note by a semitone, finds it disconcerting.

The behaviour is not a defect of rubber bands. It comes out of the formula every string obeys, applied honestly to a material that stretches by hundreds of per cent, and it singles out one special kind of spring whose note does not depend on its stretch at all.

Faster waves, and further to go

Only some notes fit derived a string’s notes from the requirement that a whole number of half-wavelengths fit between its fixed ends. The fundamental frequency is the speed of a transverse wave along the string divided by twice its length, and the speed, as the medium decides the speed put it for every wave, is set by a restoring stiffness against an inertia — here the square root of the tension over the mass per unit length:

f=12LTμ.f = \frac{1}{2L}\sqrt{\frac{T}{\mu}}.

Stretch a rubber strip of rest length L0L_0 to a length L=λL0L = \lambda L_0. Three things change. The tension rises. The mass per unit length falls as 1/λ1/\lambda, because the same rubber is spread over more length. And LL itself rises as λ. The first two make the waves faster; the third gives them further to go. Write the tension in terms of the mass mm and the product μL=m\mu L = m and the formula becomes f=12T/(mL)f = \tfrac12\sqrt{T/(mL)}: the note depends on how the tension grows compared with the length.

For an ideal rubber, the tension in a strip stretched λ times is GA0(λ−λ−2)GA_0(\lambda - \lambda^{-2}), where GG is the rubber’s shear modulus and A0A_0 its unstretched cross-section. At large stretches this is very nearly proportional to λ — to the length — and the ratio T/LT/L that sets the note settles to a constant. The fundamental becomes

f=c02L01−λ−3,c0=G/ρ.f = \frac{c_0}{2L_0}\sqrt{1 - \lambda^{-3}}, \qquad c_0 = \sqrt{G/\rho}.

Stretch a rubber band and its note climbs, then stops. The fundamental of a rubber strip clamped between two points, as a fraction of its limiting value c₀/2L₀, against how far it is stretched. Stretching raises the tension and lowers the mass per length, so the wave speed rises; but the length the wave must cross rises in step, and the two cancel. An ideal rubber network is at 0.935 of its limit at twice its rest length and 0.981 at three times: going from one and a half times to twice its length raises the note 189 cents, from twice to three times 83, and from three to six times only 29. A spring with a rest length approaches its own limit more slowly; a zero-length spring has one pitch at every length. Real rubber stiffens as its chains straighten, and the note rises again near breaking.
Fig. 1 The fundamental of a rubber strip clamped between two points, as a fraction of its limiting value c0/2L0c_0/2L_0, against how far it is stretched. An ideal rubber network is at 0.935 of its limit at twice its rest length and 0.981 at three times: going from one and a half times to twice its length raises the note 189 cents, from twice to three times 83, and from three to six times only 29. A spring with a rest length approaches its own limit more slowly; a zero-length spring has one pitch at every length. Real rubber stiffens as its chains straighten, and the note rises again near breaking.

The note climbs steeply from a slack band, which has no tension and no note, and then saturates: at twice the rest length the band is at 93.5 per cent of its limit, at three times 98.1 per cent. Going from one and a half times to twice the rest length raises the note by 189 cents, nearly two semitones; from three times to six, by 29 cents, about a quarter of a semitone, while the band doubles in length. The limit itself, c0/2L0c_0/2L_0, is the note a band of its rest length would have if waves crossed it at the speed c0c_0 — the speed of a shear wave in the unstretched rubber, about twenty metres a second for a typical band. The stretched band remembers how long it was before it was stretched.

Real rubber is not ideal at very large stretches. Its long-chain molecules, coiled at rest, are pulled nearly straight by about six or seven times their rest length, and the rubber then stiffens sharply before it breaks. The curve labelled with chains running out uses Alan Gent’s correction for this, and shows the note starting to rise again past four or five times the rest length — the last useful warning, to anyone listening, that the band is about to snap.

A spring whose note does not know its length

The comparison curves in that figure isolate what is special about rubber. A coil spring with a rest length, pulling with a force proportional to its extension, T=k(L−L0)T = k(L - L_0), also approaches a limit, but more slowly, as 1−1/λ\sqrt{1 - 1/\lambda}: its tension is not proportional to its length until it is stretched very far. And a spring whose tension is exactly proportional to its whole length, T=kLT = kL — a spring that would collapse to zero length if released, though in practice its coils stop it — has T/LT/L constant at every stretch and therefore one note at every length.

Such springs exist, and they are called zero-length springs. They are wound with a built-in twist that presses the coils together, so that the force they exert is proportional to their whole length over their working range. Lucien LaCoste invented them in the 1930s for long-period seismometers and then for gravimeters, where a zero-length spring holding a beam can be arranged to have an arbitrarily long period, which makes the instrument exquisitely sensitive to small changes in gravity. The property that makes a zero-length spring’s period independent of its geometry in a gravimeter is the property that makes its pitch independent of its stretch on a shoebox: tension proportional to length. Rubber at large stretch is close to a zero-length spring, and that is the whole of the shoebox’s mystery.

Every chain is a zero-length spring

Rubber behaves like a zero-length spring for a reason that goes down to its molecules. A rubber is a tangle of long polymer chains, cross-linked at intervals into a network. Each strand between cross-links is a random coil, and the number of ways it can be arranged falls as its two ends are pulled apart; the force it exerts is purely the tendency to recover those arrangements, and for a coil much shorter than its full contour length that force is exactly proportional to the distance between its ends. A strand of rubber is, by statistics alone, a spring of zero rest length — the same kind of entropic force that the attraction that is the small ones’ room found pushing large particles together because doing so gives the small ones more room.

Why then is a rubber band not exactly a zero-length spring? Because rubber does not change its volume. As the band is stretched it narrows, and the chains running across it are compressed while those running along it are pulled. The sideways squeeze is the −λ−2-\lambda^{-2} term in the tension, and it is what makes an unstretched band slack rather than pulling with a finite force. Once the band is stretched to two or three times its length that term is a few per cent, and what is left is a network of zero-length springs pulling along the band — whose note, accordingly, has almost stopped depending on the stretch.

A toy spring shows the same thing more directly. A slinky is wound with its coils pressed together, so that it pulls with a force nearly proportional to its whole length once it opens; the spring whose bottom does not know it has been dropped used one for a different purpose. Held stretched between two hands and flicked, a slinky’s sideways wave takes the same time to run from hand to hand however far apart the hands are, and its stretching wave keeps pace with it.

Two notes that close in on one

A stretched band carries a second kind of wave: compressions and rarefactions running along it, the band stretching and relaxing locally, as in the third of itself a spring carries where a spring’s own mass oscillates with whatever hangs from it. Its fundamental is set by how stiff the band is to further stretching, dT/dLdT/dL, rather than by the tension itself.

The band's two notes close in on one. The fundamental frequencies of a stretched rubber strip for waves that shake it sideways and waves that stretch and compress it along its length, both in units of c₀/2L₀, against the stretch. The sideways note starts from nothing — a slack band has no tension — and rises; the stretching note starts at √3 of the limit and falls, because the band's stiffness falls as it lengthens. At twice the rest length they are 1.118 and 0.935; far out they meet at c₀/2L₀. In a zero-length spring the two are equal at every stretch, which no ordinary string can arrange.
Fig. 2 The fundamental frequencies of a stretched rubber strip for waves that shake it sideways and waves that stretch and compress it along its length, both in units of c0/2L0c_0/2L_0, against the stretch. The sideways note starts from nothing — a slack band has no tension — and rises; the stretching note starts at 3\sqrt{3} of the limit and falls, because the band’s stiffness falls as it lengthens. At twice the rest length they are 1.118 and 0.935; far out they meet at c0/2L0c_0/2L_0.

For an unstretched band the stretching wave’s fundamental is 3\sqrt{3} times the limit: rubber is three times as stiff to a small extension as its shear modulus suggests, because stretching a rubber strip also squeezes its sides. As the band is stretched its stiffness to further stretching falls towards GA0/L0G A_0/L_0, and the stretching note falls. The sideways note rises from nothing. The two meet, at large stretch, at the same limit — and for an ideal zero-length spring they are equal at every stretch, because its tension and its stiffness are then the same thing per unit length. No steel string can arrange that: its stretching waves run at the speed of sound in steel, about five kilometres a second, its sideways waves at a few hundred metres a second, and the two families of notes are fifteen or twenty times apart. Piano tuners occasionally hear the stretching modes as a faint high squeal in the bass strings, far above the note being played.

The box of bands

A shoebox fixes the span, and the bands stretched across it differ in their rest lengths. The formula with LL fixed and L0L_0 varying shows which band sounds which note.

On a box of bands, the note belongs to the band and not to the box. The note of a rubber band stretched across a 10 cm box, against the band's unstretched length, for a rubber whose shear-wave speed is 20 m/s. Every band spans the same 10 cm, yet a band 5 cm long at rest sounds 187 Hz and one 8 cm long 87 Hz: the more a band is stretched the closer it comes to c₀/2L₀, the note of its own rest length, so short bands are high and long bands are low whatever the span. A steel string over the same box would be tuned by its tension at an almost unchanged length — the opposite way round.
Fig. 3 The note of a rubber band stretched across a 10 cm box, against the band’s unstretched length, for a rubber whose shear-wave speed is 20 m/s. Every band spans the same 10 cm, yet a band 5 cm long at rest sounds 187 Hz and one 8 cm long 87 Hz: the more a band is stretched the closer it comes to c0/2L0c_0/2L_0, the note of its own rest length, so short bands are high and long bands are low whatever the span.

A band five centimetres long at rest, stretched to ten, sounds 187 hertz; one eight centimetres long, barely stretched, 87 hertz. Short bands are high and long ones low, regardless of the span they cross, because each tends to the note of its own rest length. The tension along a band held between two fixed points is the same everywhere, so its modes stay the simple sines of an ideal string. That is not true of every string: the chain that swings in Bessel’s notes hung from one end carries more tension at the top than at the bottom, and its notes are no longer a harmonic series. A band hung vertically under its own weight would be in the same position, its upper part stretched more than its lower, though a rubber band is too light for its weight to matter.

Tuning a shoebox instrument is therefore done by choosing bands, or by changing the vibrating length with a bridge, which shortens LL at fixed tension and raises the note in proportion, just as a guitarist’s finger does on a fret. Pulling a band tighter is the one thing that hardly works.

The thickness of a band, which a child might expect to matter as it does for guitar strings, cancels out of the limit entirely: the limit c0/2L0c_0/2L_0 contains the rubber’s modulus and density and the band’s rest length and nothing about its cross-section. A thick band and a thin band of the same rubber and the same rest length, stretched equally, give the same note. A thick guitar string sounds lower than a thin one at the same tension because it is heavier per length; a thick rubber band is also stiffer in proportion, and the two cancel.

The rule makes the shoebox easy to lay out once it is understood. A scale spanning an octave needs bands whose rest lengths run over a factor of two, all stretched well past twice their rest length so that each sits near its own limit: for a twenty-centimetre span, bands from about five to ten centimetres long at rest. Each note is then almost immune to how firmly the band was hooked over the box and to the band settling as it creeps, which is exactly the stability a steel-strung instrument has to buy with tuning pegs. The rubber band’s indifference to stretch, which frustrates the player who tries to tune it by pulling, is what makes it hold its note once the right band has been chosen.

A steel string is tuned by stretch; a rubber band ignores it

The same formula governs a steel guitar string, but in a part of its range a rubber band never visits. A steel string is stretched by a fraction of a per cent — about 0.3 per cent for a typical tuned string — and its tension is proportional to that strain. Near zero strain the tension and so the note are exquisitely sensitive to length.

A steel string is tuned by stretch; a rubber band ignores it. How many cents the fundamental moves for each further 1% of length, against how far the string or band is already stretched, on logarithmic axes. A steel string's tension is proportional to its strain, so at a typical 0.3% strain a further 1% of length would move its pitch by about 2877 cents, more than two octaves, and a semitone is a change of length of 0.035%. A rubber band at three times its length moves by 1.0 cent per 1%. The two hang from the same formula; what differs is where on it each is used.
Fig. 4 How many cents the fundamental moves for each further 1% of length, against how far the string or band is already stretched, on logarithmic axes. A steel string at a typical 0.3% strain would move its pitch by about 2,877 cents for a further 1% of length, more than two octaves, and a semitone is a change of length of 0.035%. A rubber band at three times its length moves by 1.0 cent per 1%.

A further one per cent of length would move a steel string’s note by nearly three thousand cents, two and a half octaves — the string would break long before — and a semitone corresponds to a change in length of three and a half hundredths of a per cent, about a quarter of a millimetre of string wound onto the peg on a guitar’s scale. A rubber band at three times its length moves by one cent for a one per cent change. The two materials sit at opposite ends of one curve: the steel string so near the start that the note is all tension, the rubber band so far along that the note has forgotten the tension.

That is why a steel string’s tuning drifts with every change of temperature and humidity in the instrument and a rubber band’s does not drift with how hard it is pulled. The string whose end will not keep still found a string’s pitch shifted by a bridge that moves; for a steel string a bridge moving by a few micrometres is a cent or two, and instrument makers spend a great deal of effort keeping it from moving.

Warmth sharpens a band and flattens a string

Rubber’s tension has an origin that steel’s does not. Entropy is a count found a stretched rubber band pulling back because its long-chain molecules have vastly more coiled arrangements than straight ones, so that stretching it lowers its entropy, and what a system actually minimises put the consequence plainly: rubber’s tension at fixed length is proportional to the absolute temperature, because what it resists is a reduction in disorder, and disorder is worth more when it is hotter.

Warmth sharpens a band and flattens a string. The change in pitch of a rubber band held at three times its rest length, and of a steel string at 0.3% strain on a wooden box, as both are warmed with their ends fixed. Rubber's tension comes from the entropy of its chains and grows in proportion to absolute temperature, so the band goes sharp, +49 cents for 20 K, slightly less than the √T alone would give because the band also expands. Steel expands more than the wood that holds it, its strain falls, and the string goes flat by 47 cents. A metal string's tension is energy stored in bonds; a rubber band's is a count of arrangements.
Fig. 5 The change in pitch of a rubber band held at three times its rest length, and of a steel string at 0.3% strain on a wooden box, as both are warmed with their ends fixed. Rubber’s tension grows in proportion to absolute temperature, so the band goes sharp, +49 cents for 20 K, slightly less than the T\sqrt{T} alone would give because the band also expands. Steel expands more than the wood that holds it, its strain falls, and the string goes flat by 47 cents.

So a rubber band held at a fixed length goes sharp when warmed, by about half a semitone for twenty degrees. A steel string on a wooden instrument goes flat by about the same amount, for an unrelated reason: steel expands with heat faster than the wood holding its ends, the string’s strain falls, and with it the tension. Steel’s tension is energy stored in stretched bonds between atoms, which a little warmth does not change much; a rubber band’s is a count of arrangements, which warmth changes directly. The two warmings are about equal in size and opposite in sign, which is a coincidence of the numbers chosen. The physics behind them has nothing in common. Why heating a perfect spring changes nothing found that a perfectly harmonic solid would not expand at all, so the steel string’s flattening lives in the anharmonic part of its bonds, while the band’s sharpening needs no anharmonicity whatever.

What the ideal network leaves out

The ideal-rubber law describes natural rubber well up to two or three times its length and less well beyond; real bands contain fillers that stiffen them, show hysteresis so that the tension on stretching exceeds that on release, soften permanently the first time they are stretched far (the Mullins effect), and creep at constant length. The shear-wave speed of twenty metres a second is a round value; real bands vary by a factor of two. The steel-string figures use a strain of 0.3 per cent and expansion coefficients of 12 and 4 parts per million per kelvin, typical of a steel string and a spruce top along the grain. And every frequency here is the ideal string’s fundamental, ignoring the stiffness of a thick band, which raises its upper overtones, and the damping that makes a rubber band’s note dull and short.

Still open: what makes a rubber band sound like a rubber band

The pitch of a plucked band is easy; its tone is not. A rubber band’s note dies away in a few tenths of a second, far faster than a steel string’s, and the reason is internal friction in the rubber, which dissipates a large fraction of each cycle’s energy and does so more at higher frequencies — so the overtones die first and the sound is dull. How the damping depends on stretch, temperature and frequency, and how much of the sound reaches the air directly from a band so thin it barely moves any, is measured for engineering rubbers at small strains but hardly at the large strains a plucked band lives at, where the polymer chains are oriented and partly crystallised.

A string’s fundamental is the wave speed over twice the length, and a rubber band stretched further gains speed — tension up, mass per length down — almost exactly as fast as it gains length, so its note climbs to a limit c0/2L0c_0/2L_0 set by the rubber and its rest length: 0.935 of it at twice the length, 0.981 at three times, and only 29 cents more from three to six times. A zero-length spring, the gravimeter’s spring, has one note at every stretch; a steel string, at 0.3% strain, moves a semitone for 0.035% of length. And a band at fixed length goes sharp when warmed, because its tension is a count of molecular arrangements, where a steel string goes flat.

Part 13 of 13

This essay is one argument about Standing waves. The others:

The objects named here

The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.

Entropic forceHarmonicRubber elasticityTensionWave speedZero-length spring