Field notes on things that run themselves
Twenty-Eight Months, Give or Take
In 1883 Krakatau exploded and its dust circled the planet from east to west in about two weeks; those winds were named the Krakatau easterlies. In 1908 A. Berson sent balloons up over tropical Africa and found the wind at fifteen kilometres blowing the other way. Those became the Berson westerlies.
Both were sound, and for nearly fifty years the equatorial upper air was taken to be permanently layered, easterlies above westerlies, because soundings over the equator were too sporadic to say otherwise.
Then in January 1960 R. J. Reed described what he had found in soundings from Canton Island: “alternate bands of easterly and westerly winds which originate above 30 km and which move downward through the stratosphere at a speed of about 1 km per month.” A full cycle took twenty-six months. R. A. Ebdon, in Britain, had got there independently.
Both winds are real. They take turns.
Not on any schedule you could name. In 1964 Angell and Korshover coined quasi-biennial oscillation, the hedge built into the name because a cycle had just come in conspicuously longer than two years. Since 1953 the mean period has been a little over twenty-eight months, and individual cycles run anywhere from twenty to thirty-five. The pattern slides against the calendar.
Which raises the question that took another decade to answer. What has a period of twenty-eight months? Not the year. Not the eleven-year solar cycle — that link has been proposed, and the observations are inconsistent. Nothing does.
The wind is not keeping anyone’s time. It is making its own, out of the only material available.
Below the stratosphere, tropical convection throws off a broad spectrum of waves: Kelvin waves carrying eastward momentum, Rossby-gravity waves carrying westward, and a crowd of small gravity waves carrying both. They rise. Each climbs until it reaches an altitude where the wind already moves at roughly its own phase speed — its critical level — and there it stalls and hands over its momentum.
So a westerly layer is fed from underneath by exactly the waves that cannot get past it. The push lands at its base, and the layer creeps downward toward its own supply. Waves of the opposite sign sail through and build the next layer higher up. When the descending layer thins to nothing near the tropopause, the waves it was intercepting are released upward, a fresh layer assembles aloft, and the whole thing repeats with the signs swapped — a pattern that regenerates at the top and erases itself at the bottom.
Nothing in that loop consults anything outside it. In Plumb’s formulation the period comes out inversely proportional to the wave momentum flux — push harder and the oscillation runs faster. It is a quotient, not an inheritance. That it lands near a subharmonic of the year is, in the words of the field’s standard reference, “pure coincidence.”
Which is a new shape for this series. A room handed a hiss picks one note out of it and lets that note run away. A termite mound handed a single rhythm, the day, straightens it into flow. This system is handed no rhythm at all, and produces one.
You can watch it in a tank. In 1978 Plumb and McEwan filled the gap between two concentric cylinders with salt-stratified water and flapped sixteen rubber membranes at the boundary, each in opposite phase to its neighbours — a standing wave, which is to say two waves running opposite ways. It organised itself into reversing layers, marching downward. In the modern rebuild the membranes flap every fifteen seconds and the reversal takes about twenty-five hundred. No arithmetic connects those numbers. Turn the forcing up and the reversal speeds up, exactly as the theory says.
One thing pushes back. The whole tropical stratosphere is slowly rising — under a millimetre a second, roughly a kilometre a month, about the speed at which the pattern descends. So the layers sink through air that is climbing past them, and relative to that air they fall about twice as fast as they look to. The period comes out of that race: waves advancing the phase, upwelling dragging it back. A 2025 study traced the period’s wobble to the small gravity waves, whose activity follows the seasons of tropical convection — the year does not set the rhythm, but it jogs the pace.
Twice since 1953 the race has been lost outright. In 2016, and again in 2019–20, an easterly jet materialised inside a descending westerly phase and the cycle broke. Modelling suggests a warming climate makes such interruptions likelier, and a paper published this February found the amplitude weakening, traced in a momentum budget to those same two terms. Waves against upwelling.
Which is the right way to hear it. This is not a clock that has been running well and might stop. It is a balance that happens to swing, and everything about it — the period, the drift, the two years that are not two years — is what that balance sounds like from underneath.
---
Next: drop grains onto a sandpile one at a time and the slope climbs to a critical angle and then stops climbing, held there by avalanches of every size at once. Nobody is tuning the pile to the edge of failure. It arrives there on its own, and stays.
One loop I’m watching
Tip: the ← and → arrow keys move between issues.
Prefer feeds? Follow by RSS or JSON Feed.
Want something playable between issues? Thought Toys has its own email list and RSS feed.
New to The Standing Wave? Start here →