How Science Works | Why Do Ripples Behave Differently at Different Sizes?

Raise a small patch of water above its surroundings and release it. Gravity and pressure differences encourage motion away from that elevated state. Fluid inertia means the surface does not simply return to flatness and stop immediately. It can overshoot, while neighbouring regions also move, allowing a wave to propagate.

At short wavelengths, the surface bends sharply over a small distance. Surface tension becomes especially important. It acts along the interface, and changes in the direction of that tension around a curved patch produce a net force. The resulting pressure difference is associated with curvature, not merely with the height of the patch.

Compare two small sinusoidal disturbances with the same amplitude but different wavelengths. The shorter wave bends more tightly. In the small-slope approximation, the magnitude of its curvature grows like amplitude divided by wavelength squared. Reducing wavelength therefore increases the surface-tension contribution relative to the gravity contribution.

Water still has inertia in both cases. “Gravity wave” and “capillary wave” name the dominant restoring influence, not two separate kinds of water. Around the crossover, both influences contribute substantially and neither should be discarded.

There is also a language trap: these surface gravity waves are not gravitational waves in spacetime. The shared word refers here to ordinary gravity acting on displaced fluid.

Contents · Next chapter

Wavelength λ is the distance between successive equivalent points, such as crest to crest. Frequency f is the number of oscillations per second at a fixed position. For a simple travelling sinusoidal wave, phase speed is c = fλ. It is the speed of a fixed phase, such as one crest.

Wavenumber k = 2π/λ measures spatial variation, while angular frequency ω = 2πf measures temporal variation. Then c = ω/k. These definitions keep the factors of 2π consistent when moving between a photograph, a time record and a mathematical equation.

A dispersion relation gives the connection between frequency and wavenumber. When phase speed depends on wavelength, the wave system is dispersive. A disturbance containing several wavelengths can change shape because its components acquire different phases as they travel.

For an invented measurement, crests spaced 0.020 metres apart pass a fixed point at 12 per second. Their inferred phase speed is 12 × 0.020 = 0.24 metres per second. That calculation does not identify the restoring force. To do that, compare the measurement with a model and the fluid’s properties, depth and surface conditions.

A camera can also mislead if frames are too widely separated. A crest seen in one frame may be mistaken for a neighbouring crest in the next. A scale reference, timing information and enough temporal resolution are part of a credible speed measurement, even when the wave pattern looks simple.

Contents · Previous chapter · Next chapter

For small-amplitude waves on an inviscid, incompressible liquid of uniform depth h, with no background current and negligible dynamical effect from the air, a standard linear relation is ω² = [gk + (γ/ρ)k³] tanh(kh). Here g is gravitational acceleration, γ is surface tension and ρ is liquid density.

The gravity contribution inside the brackets is gk. The surface-tension contribution is (γ/ρ)k³. The second grows more rapidly with wavenumber, which explains why short wavelengths favour capillary behaviour. The depth factor tanh(kh) records the influence of the bottom.

This relation assumes a clean, uniform interface and small enough slopes for linearisation to be appropriate. It omits viscous damping and many complications of contamination, currents, strong forcing and breaking waves. Its value is that the assumptions and competing terms are explicit.

MIT’s water-wave notes by C. C. Mei derive the relation from the fluid equations and surface boundary conditions. The equation is not an empirical label attached after seeing a ripple; it predicts which frequencies and wavelengths can travel together within its regime.

An informative diagram would place a long gentle wave beside a short tightly curved ripple. Label the same amplitude, different wavelengths, gravity acting downward and surface-tension forces along the surface. Then label the equation’s two contributions underneath. The drawing should not depict a solid elastic skin floating on the water: surface tension is a property of the interface.

Contents · Previous chapter · Next chapter

Compare the two restoring contributions by taking their ratio: [(γ/ρ)k³]/(gk) = γk²/(ρg). Equal contributions occur when k = square root of ρg/γ. The corresponding wavelength is 2π times the square root of γ/(ρg).

The square root of γ/(ρg) is the capillary length, commonly written ℓc. The crossover wavelength is 2πℓc. These are related lengths but they are not numerically identical. Forgetting the factor of 2π creates a substantial error when deciding which wavelength lies near the crossover.

Use illustrative water-like values γ = 0.072 newtons per metre, ρ = 1,000 kilograms per cubic metre and g = 9.81 metres per second squared. They give ℓc approximately 0.00271 metres, or 2.71 millimetres. The equal-contribution wavelength is approximately 0.0170 metres, or 17.0 millimetres.

Now test three wavelengths. At 2 millimetres, the capillary contribution is about 72.4 times the gravity contribution. At 20 millimetres, the ratio is about 0.724, so both matter. At 200 millimetres, it is about 0.00724, so gravity strongly dominates the restoring balance in this model.

These numbers are calculations from stated assumptions, not universal properties of every liquid surface. Temperature, composition and contamination can change surface tension. Different density and gravity values also change the crossover. There is no sudden physical switch at 17 millimetres: the relative contributions vary continuously.

Contents · Previous chapter · Next chapter

In deep water, kh is large enough that tanh(kh) is close to one. Dividing the dispersion relation by k² gives phase speed squared: c² = g/k + γk/ρ. One term falls as k increases; the other rises. Their sum therefore has a minimum.

In the long, deep-water gravity-wave limit, c is approximately the square root of g/k, or the square root of gλ/(2π). Longer wavelengths travel faster. In the short capillary-wave limit, c is approximately the square root of γk/ρ, or the square root of 2πγ/(ρλ). Shorter wavelengths travel faster.

Using the same illustrative parameters and the full deep-water expression, a 2-millimetre wave has phase speed about 0.479 metres per second. A 20-millimetre wave travels at about 0.232 metres per second. A 200-millimetre wave travels at about 0.561 metres per second. The middle wavelength is close to the slowest part of the curve; neither “shorter is always faster” nor “longer is always faster” describes all three.

The minimum deep-water phase speed is (4gγ/ρ) to the power one quarter, approximately 0.231 metres per second for our values. It occurs at the equal-contribution wavelength. This minimum belongs to the stated model, not to every possible disturbance on every water surface.

Raphaël and de Gennes used the dispersive properties of capillary–gravity waves to analyse wave resistance from a moving disturbance in a linear theory. The work connects the speed relation with energy carried away by generated waves. It does not imply that a slow object can never create any transient ripple when it starts, stops or changes its path. Original wave-resistance study.

Contents · Previous chapter · Next chapter

“Deep water” is relative to wavelength. A depth that is large for a millimetre ripple can be shallow for a much longer wave. The quantity kh, rather than the word “pond” or “ocean”, decides whether the deep-water approximation is appropriate for this linear model.

Take λ = 20 millimetres. At h = 10 millimetres, kh = π and tanh(kh) is about 0.996. The deep-water value is a close approximation. At h = 1 millimetre, kh is about 0.314 and tanh(kh) is only about 0.304. Ignoring the bottom would be a poor approximation.

Because the depth factor multiplies ω², the phase speed at the same wavelength is multiplied by its square root. Starting from our 0.232-metre-per-second deep-water value, the 1-millimetre-depth model gives about 0.128 metres per second. That is an ideal prediction; viscous effects can be particularly important in a very shallow real layer.

In the shallow-water gravity-wave limit, with kh much less than one and surface tension negligible, tanh(kh) is approximately kh. The relation then gives c approximately square root of gh. To leading order, phase speed no longer depends on wavelength. The familiar deep-water rule “longer waves are faster” has left its domain.

This is a useful model-reading habit: identify every dimensionless comparison before applying a limiting formula. A long wavelength and a shallow depth can change the physics even though the liquid and gravitational acceleration remain unchanged.

Contents · Previous chapter · Next chapter

A wave packet contains a narrow range of wavelengths combined into a slowly varying envelope. Individual crests can move through that envelope. Phase speed tracks a crest; group velocity describes the envelope’s leading-order propagation in the linear narrow-band setting and is associated with energy transport under the usual conservative-wave assumptions.

Mathematically, phase speed is ω/k while group velocity is dω/dk. They are different operations. Dividing a frequency by wavenumber need not give the same result as finding the slope of the dispersion curve.

For deep-water gravity waves, the group velocity is half the phase speed. For the pure deep-water capillary limit, it is one and a half times the phase speed. In the mixed regime the ratio changes continuously. These limiting values are not rules for arbitrary depth or strongly nonlinear packets.

For our 2-millimetre wave, retaining both restoring terms gives a group speed of about 0.712 metres per second, compared with phase speed 0.479. At 200 millimetres, the group speed is about 0.284 metres per second, compared with phase speed 0.561. The numbers approach the capillary and gravity limiting ratios respectively.

A useful animation description would show a small train of crests beneath a shaded envelope, with separate markers following one crest and the envelope centre. It should not show water particles travelling steadily at either speed. In a small-amplitude wave, fluid particles mainly oscillate, and any net drift requires a more careful account beyond the simplest first-order motion.

Contents · Previous chapter · Next chapter

Viscosity damps motion, so a real ripple can decrease in amplitude as it travels. Very short waves are particularly sensitive to small-scale gradients and dissipative effects. A formula predicting the frequency of an ideal wave does not automatically predict how far that wave remains visible.

Surfactants and contamination can also affect surface tension and interfacial stresses. A clean-interface assumption may fail even when the liquid looks clear. Changing the liquid or adding a substance may alter several properties at once, so a changed ripple pattern does not isolate surface tension by itself.

Boundary conditions can create an even more striking difference. Fauconnier and colleagues reported waves on a gas–water interface pinned to microstructures beneath water in 2025. That constrained interface is not the ordinary open water surface used in our calculation. Its measured propagation behaviour required attention to the trapped gas layer and geometry. The study is useful evidence that the interface and its constraints belong in the model. Original constrained-interface wave study.

Other experiments deliberately use wave fields to organise motion at a liquid surface. Francois and colleagues demonstrated rotating-wave patterns that guided floating microparticles into organised trajectories. The resulting particle motion reflects the driven wave system and its interactions; it is not equivalent to saying that every floating particle always travels at the linear group velocity. Original wave-based interface experiment.

These examples extend the basic picture without replacing it. First establish the simple restoring-force balance, then ask which additional property or boundary condition changes the prediction.

Contents · Previous chapter · Next chapter

A safe learning investigation can use supplied videos or numerical data rather than specialised apparatus. Choose one question: does the measured relation between wavelength and frequency agree with the mixed gravity–capillary model over a specified range?

For each record, identify wavelength from a calibrated spatial image and frequency from a time sequence at a fixed location. Document depth and the assumed surface tension and density. Compute c = fλ, then compare with the prediction using the appropriate depth factor. Do not fit a deep-water formula before checking kh.

Suppose a fictional record at λ = 20 millimetres gives f = 12 hertz. Its measured phase speed is 0.24 metres per second, while our deep-water prediction is about 0.232. The difference is about 3.4% of the prediction. That alone neither proves nor disproves the model: wavelength calibration, timing, depth and material-property uncertainty need to be compared with the discrepancy.

Now repeat at several wavelengths. A systematic trend in the disagreement may be more informative than one close match. If the shortest waves disappear quickly, record the visibility limit rather than silently omitting them. If reflections produce standing waves, a travelling-crest interpretation may no longer apply.

Avoid lasers, heated liquids, chemical additives and improvised high-power vibration equipment. Published experimental images and a paper graph are enough to test the reasoning. The important scientific step is to define what the observation measures and which assumptions connect it with the equation.

Contents · Previous chapter · Next chapter

Why does surface tension become important at short wavelengths? Shorter waves have greater curvature for the same small amplitude, making the capillary restoring contribution grow more rapidly with wavenumber than the gravity contribution.

Is the crossover equal to the capillary length? No. For the linear gravity–capillary relation, the equal-contribution wavelength is 2π times the capillary length. Our illustrative values gave 2.71 millimetres and 17.0 millimetres respectively.

Do all ripples travel faster when they get shorter? No. Short capillary-dominated waves and long deep-water gravity waves have opposite phase-speed trends. Depth can change the result again. The full relation explains why the deep-water speed curve has a minimum.

Does a crest show how fast wave energy travels? Not necessarily. Phase speed and group velocity differ in dispersive systems. Nor does either quantity automatically describe the net transport of floating matter.

What would make a comparison convincing? Calibrated wavelength and frequency measurements, documented depth and surface conditions, repeated observations and explicit uncertainty. A neat circular ripple is an invitation to ask these questions, not a complete answer.

The mechanism is a balance among restoring forces, fluid inertia and boundaries. Small changes of scale can change which term matters most, making a familiar water surface an accessible example of how scientific models connect appearance with quantitative prediction. Continue through the Fluid Dynamics guide and the How Science Works master guide.

Contents · Previous chapter · Continue to Fluid Dynamics