The Capstan Equation: Why Three Wraps of Rope Hold a 10-Tonne Ship

The governing equation and what each symbol means

Consider a flexible, inextensible rope wrapped around a fixed cylindrical post, in contact over an angular span β (the wrap angle, measured in radians). One end carries the large load tension T₂; the other carries the small hold tension T₁. At the verge of slipping — the moment the rope is about to slide in the direction of the larger tension — the two are related by the capstan equation:

• T₂ = T₁ · e^(μβ)

Here μ is the coefficient of friction between rope and post, β is the total angle of contact in radians (not degrees), and e ≈ 2.718 is Euler's number. The result is remarkable for what is absent: there is no radius R. A pencil-thin peg and a fat mooring post give exactly the same tension ratio for the same wrap angle. The load also holds against T₂ — the equation is an inequality in general, T₁·e^(−μβ) ≤ T₂ ≤ T₁·e^(μβ); slipping only begins when the tension ratio reaches the exponential bound. Between those limits the rope is static and can carry any load in that window.

Because β appears in an exponent, adding rope is astronomically more effective than pulling harder. With μ = 0.3, one wrap multiplies your force by e^(0.3·2π) ≈ 6.6, three wraps by ≈ 286, and four wraps by ≈ 1,880. That is why a deckhand snubs a hawser around a bollard rather than trying to out-muscle a ship.

Deriving it: a free-body diagram of an infinitesimal arc

The derivation is a small classic. Isolate a tiny element of rope subtending angle dθ. Its two ends carry tensions T and T + dT. The post pushes outward on the element with a normal force dN, and friction acts tangentially, opposing the impending slip, with magnitude up to μ·dN.

• Radial balance: The two tension vectors each tilt inward by dθ/2, so their inward components sum to T·sin(dθ/2) + (T+dT)·sin(dθ/2). For small angles sin(dθ/2) ≈ dθ/2, giving an inward pull ≈ T·dθ. This must be balanced by the normal force: dN = T·dθ (dropping the second-order term dT·dθ).
• Tangential balance: The difference in tension along the arc must be supplied by friction: dT = μ·dN at the slipping limit.
• Combine: substitute dN to get dT = μ·T·dθ, i.e. dT/T = μ·dθ.
• Integrate from T₁ (at θ = 0) to T₂ (at θ = β): ∫dT/T = ∫μ·dθ ⟹ ln(T₂/T₁) = μβ ⟹ T₂ = T₁·e^(μβ).

The physical heart of it is dN = T·dθ: the normal force at each point is itself proportional to the local tension. Because tension is higher on the load side, the post grips hardest exactly where the rope pulls hardest. Friction feeds on itself around the arc, compounding like interest — which is precisely why the answer is exponential rather than linear.

Why radius drops out, and what actually controls the number

Newcomers expect a bigger post to hold better. It doesn't — not for holding force. Doubling the radius doubles the arc length under any given angle, but it also halves the normal pressure per unit length (the same tension is spread over twice the contact area). The two effects cancel exactly, leaving only the dimensionless product μβ. This is a rare case where a mechanical result depends on a pure angle and a pure ratio, with no length scale at all.

So what does matter?

• Wrap angle β — the dominant knob, because it lives in the exponent. Going from two to three wraps (β from 4π to 6π) with μ = 0.3 jumps the ratio from ≈ 43 to ≈ 286.
• Coefficient of friction μ — manila rope on rough iron reaches μ ≈ 0.4–0.5; slick synthetic line on polished stainless can fall to μ ≈ 0.15. Because μ also sits in the exponent, a wetted, algae-slimed bollard can lose most of its grip.
• Direction of impending motion — friction reverses sign depending on which way the rope is about to slide, which is why the same wrap that holds a huge load can be paid out smoothly by easing the small end.

Radius, rope diameter, and even material stiffness enter only through second-order corrections (rope bending resistance, finite thickness), which real engineering standards fold into an empirical safety margin.

A worked mooring: 100 N holds 100 kN

Return to the sailor. Take manila hawser on a cast-iron bollard, μ ≈ 0.35, and three full turns, β = 3·2π = 6π ≈ 18.85 rad. The exponent is μβ = 0.35 × 18.85 ≈ 6.60, so:

• T₂/T₁ = e^(6.60) ≈ 735.

A hold force of just T₁ = 100 N therefore restrains up to T₂ ≈ 73,500 N — comfortably enough to snub a ship whose drifting momentum registers as tens of kilonewtons of line tension. Add a fourth wrap (β = 8π, μβ ≈ 8.8) and the multiplier leaps to e^(8.8) ≈ 6,600; now 100 N holds 660 kN, and the practical limit is no longer your hand but the breaking strength of the rope itself.

The same arithmetic explains a rock climber's belay device: a runner falling with, say, 5 kN of impact force is arrested by a belayer gripping the brake strand with well under 200 N. The device forces the rope through a sharp bend — a large effective β at high local μ — so the human end need only supply the tiny T₁. And it is why letting go completely is catastrophic: T₁ → 0 makes T₂ → 0, and the exponential grip evaporates instantly. The friction is a lever, not a latch; it needs some tension on the small end to work at all.

Where the exponential rules everyday machines

The capstan equation is the hidden operating principle of an enormous amount of hardware:

• Capstans and winches: a powered drum turns slowly while crew tail the rope with light hand force; the drum supplies β and μ, the operator supplies only T₁. This is literally the machine the equation is named for.
• Belt and pulley drives: the maximum torque a flat belt can transmit before slipping is set by the tension ratio e^(μβ) between the tight and slack sides. Serpentine belts in cars wrap accessories at large β precisely to raise this ratio; a V-belt cheats the geometry by wedging into a groove, replacing μ with an effective μ/sin(α) where α is the half-groove angle — a wedge of α = 18° multiplies grip by ~3×.
• Sailing and rigging: a sheet taken around a winch, or a line figure-eighted on a cleat, uses the same exponential. Sailors instinctively know 'three turns and a couple of hitches' will hold a straining sail.
• Knots and hitches: the prusik, the icicle hitch, and the constrictor all bootstrap their own T₁ from the wraps, which is why a well-dressed friction hitch grips a load far heavier than the cord's own weight.
• Band brakes and chain hoists: a steel band wrapped around a drum brakes with force amplified by e^(μβ) — the basis of the classic differential band brake.

Subtleties, limits, and common misconceptions

The clean exponential hides several real-world caveats worth stating precisely:

• It's the slipping limit, not the everyday state. The equation gives the maximum tension ratio at impending slip. In static equilibrium the rope holds at any ratio within e^(−μβ) ≤ T₂/T₁ ≤ e^(μβ). A moored rope usually sits comfortably inside that band.
• Radius does matter for real ropes. The ideal derivation assumes a perfectly flexible, thin line. A thick or stiff rope resists bending; on a small-radius peg this flexural rigidity adds tension the formula ignores, and can crush or over-stress the rope. Standards therefore mandate minimum bend radii (often ≥ 4× rope diameter for wire rope).
• μ is not a constant. Water, mud, ice, temperature, and even sliding speed shift the coefficient. A greasy winch drum or an iced-over bollard can silently drop e^(μβ) by an order of magnitude — a genuine safety hazard, not a rounding error.
• Heat and wear on moving belts. When a belt actually slips, the frictional power μ·N·v becomes heat; sustained operation near the e^(μβ) limit glazes and burns belts. Engineers design well below the slipping ratio.
• Degrees vs radians. β must be in radians. Using 180 instead of π inflates the exponent by 57×, a classic student error that turns a factor of 2 into a factor of 10²⁴.

Finally, the sign of friction flips with direction: the very wrap that lets 100 N hold 100 kN also lets you surge the line out smoothly by easing the hold end, because now the rope tends to slip the other way and friction resists that motion instead.

A short history: Euler, Eytelwein, and 260 years of rope

The result is often called the Euler–Eytelwein formula. Leonhard Euler analyzed friction on a curved surface in the 1760s, deriving the exponential relation as a piece of his broader work on flexible lines and mechanics. Johann Albert Eytelwein, a German engineer, presented and popularized it for practical machine design in his 1808 Handbuch der Statik fester Körper, which is why his name rides alongside Euler's in the belt-drive literature.

But the physics predates the mathematics by millennia. Egyptian, Phoenician, and Greek sailors were snubbing hawsers around wooden posts, and Roman engineers were driving capstans, long before anyone wrote e^(μβ). The formula simply codified what every rigger already knew in their hands: turns beat strength. Today the identical equation governs the tape path in a cassette or a 3D-printer's belt, the torque budget of an alternator drive, and the load rating of every climbing belay sold — a two-line derivation from a single free-body diagram, still doing heavy lifting after two and a half centuries.

Holding-force ratio T₂/T₁ = e^(μβ) for rope on a post, by wrap and friction coefficient

Wraps around post	Wrap angle β	μ = 0.2	μ = 0.3	μ = 0.5
Half turn (180°)	π rad	1.87×	2.57×	4.81×
1 full turn	2π rad	3.51×	6.59×	23.1×
2 full turns	4π rad	12.3×	43.4×	535×
3 full turns	6π rad	43.4×	286×	12,400×
4 full turns	8π rad	152×	1,880×	286,000×