The Sand Can Travel. Can the Song?

A dune seems like an implausibly large musical instrument. The more surprising possibility is that you can take some of its sand away and leave the landscape behind without leaving the voice behind.

In a laboratory experiment described by Stéphane Douady and colleagues, a motor drove a blade through sand brought from a singing dune near Tarfaya, Morocco. The apparatus was a circular channel, not a miniature desert. It made sustained tones. Changing how the blade pushed the sand changed the frequency. A whole dune was not necessary for that sand to sound. [1, Figs. 1–2]

That is a beautiful result. It is also a result with a precise boundary. Making a voice portable does not prove its usual surroundings have nothing to do with its song.

A heap of impacts is not yet a note

The phenomenon here is booming: the low, sustained sound that some sand avalanches produce. It is not the higher-pitched squeak of certain beach sands, nor every brief noise made by shaken grains. In Nathalie Vriend and colleagues' field study, a microphone on California's Dumont Dunes recorded a dominant frequency around 82 hertz with higher harmonics. The visible avalanche and the sustained tone were related, but an avalanche did not always produce that tone. [2, introduction and Fig. 2]

Individual collisions are not enough of an explanation. Grains moving without coordination can make noise without making a strong, sustained periodic sound. The interesting question is how enough motion becomes organized to push the air together.

Douady's group proposed that moving grains synchronize, allowing the flowing surface to act somewhat like a loudspeaker. They did not dispense with resonance. They moved the proposed resonant feedback into the sheared sand layer: waves within that layer would help coordinate the motion. In this account, grain motion sets the frequency, while resonance helps that motion become coherent. [1, pp. 2–3]

The blade experiment gives this idea something concrete to answer to. Neither blade speed nor pushed height alone organized the frequency measurements. The ratio of speed to height did—a measure of the mean imposed shear, not a direct measurement of the local shear in the active zone. The same sand could produce different frequencies when forced differently. Its grain size was not an immutable musical label.

For a free avalanche, the group connected grain diameter to a gravity-driven motion timescale. Their approximate relation predicts about 99 hertz for 160-micrometre grains, near their reported Tarfaya measurement of 105 ± 10 hertz. That arithmetic reconstructs the model; it is not another measurement or proof that grain diameter fixes every dune's note. [1, Table I]

The difficult part remained the proposed coordination. The authors used sound-onset thresholds to infer a remarkably low wave speed within the sheared layer. That speed came through their resonance model, rather than a direct travel-time measurement. The experiment showed controlled sound production more firmly than it established every detail of the proposed feedback.

The dune that would not sing

An independent Caltech-led team approached the instrument from underneath.

Vriend and colleagues compared two Dumont dunes in September 2006: one roughly 45 metres high, the other about 11 metres. Both had steep slip faces near their crests. People slid down them to start sand moving. The large dune boomed; the smaller one avalanched without producing sustained booming. When the team returned to the large dune in December, it would not sustain the boom either, although brief, faint squeaks remained. [2, methods and Fig. 2]

The explanation they investigated was not a hollow hill with air trapped inside. It was a waveguide: a layer through which vibrations could travel and reinforce one another because of the material above and below it.

The researchers struck a plate with a sledgehammer and recorded the resulting seismic waves using geophones along the dune. Their inferred September structure included a relatively slow-wave-speed upper sand layer beneath faster air and above faster material. The December profile lacked the same distinct layering. They proposed that the avalanche supplied energy while the layered dune selected the frequency. [2, Figs. 4–5 and discussion]

This model had a numerical test. For one August location, the reported layer depth and wave speeds give a calculated frequency of about 94 hertz, with an uncertainty of ±26; the measured frequency was 86 ±5 hertz. That is agreement worth noticing, not a prediction accurate to the nearest hertz. The paper reports closer correspondence in the upper region where booming was strongest, and less satisfactory correspondence farther downhill. [2, Table 1]

Their comparison also separated loudness from pitch. Two slides at different human sliding speeds produced overlapping frequency estimates, while the slower slide involved more surface area and about twice the acoustic amplitude. A stronger sound did not require a higher note. But human sliding speed is not the same measurement as the local shear within a flowing layer. This was not a direct repetition of the laboratory blade experiment. [2, paragraph 13]

What the sieve changed

A later report offers another useful handle: change the sand rather than the hill.

AGU's author-linked account of a 2012 study by Simon Dagois-Bohy and colleagues describes Omani sand producing a broad spread of frequencies. On a laboratory incline, a sieved fraction containing grains between 200 and 250 micrometres produced a clearer note near 90 hertz than the mixed-size sample. [3]

I could read that complete account, but not the full underlying paper, so this is a reported intervention rather than a methods-level assessment. It supports a role for grain-size distribution in tonal purity under those conditions. It does not say that any sand sorted to those dimensions will sing, or that every desert obeys one diameter-to-pitch rule. The account itself leaves the reason grains synchronize unresolved.

The sieve matters because a narrow note and a loud sound are different achievements. Narrowing the grain-size distribution changed the spread of frequencies in the reported experiment, without establishing how the grains synchronized or every condition needed for a landscape-scale boom.

More than one job inside an instrument

I started with the tempting picture of a dune as a giant resonator. The laboratory makes that picture insufficient. The field measurements give reasons not to assume that only the flowing grains matter in a natural dune.

These selected historical studies do not demonstrate that both proposed mechanisms form one settled theory. There is a real disagreement: the laboratory account locates frequency in grain motion, while the field account assigns it to the waveguide. They examine different forcing and different settings. Their results cannot be reconciled just by giving everything the name “resonance.”

What they do teach is more interesting than a slogan about singing sand. Supplying energy, selecting a frequency, coordinating motion and carrying sound into air are different jobs. A demonstration can isolate one without settling all four.

The next question I would ask is not whether another avalanche can be made louder. It is whether simultaneous measurements of local grain shear and subsurface waves, while independently changing one condition, could reveal which clock the note follows.

The sand can travel. Explaining what travels with it takes more than hearing the same word—song—in two places.

Sources and limits

Douady et al., “The song of the dunes as a self-synchronized instrument.” Complete author preprint, arXiv:nlin/0412047v3, version dated January 28, 2006; associated publication: Physical Review Letters 97, 018002. Main text, equations, table and captions read; the preprint was not verified word-for-word against the final journal version.
Vriend et al., “Solving the mystery of booming sand dunes.” Geophysical Research Letters 34, L16306 (2007), published paper and supplements at CaltechAUTHORS, DOI 10.1029/2007GL030276. Complete main text, methods, equations, table and captions, plus supplementary Table S1 read. Independent team, not a replication of the blade apparatus.
AGU, “Why ‘singing’ sand dunes hum certain notes,” October 24, 2012. Complete author-linked release, discussing Dagois-Bohy, Courrech du Pont and Douady, DOI 10.1029/2012GL052540. The full underlying paper was inaccessible in this study. This belongs to the related French research program, not an additional independent confirmation of it.

This is a comparison of selected historical experiments, not an exhaustive statement of present consensus. Calculations reconstruct reported models; no new experiment, figure-image inspection, supplementary movie viewing or audio playback is claimed.