Last time I promised you a tube with no moving parts, which is how every account opens; the standard review calls that absence its remarkable feature. True of the hardware: a pipe, a nozzle drilled into the side near one end, a small hole in the endcap beside it, a cone at the far end. Nothing turns.
Feed it compressed air and one end runs hot and the other runs cold, for as long as the air keeps coming.
Ranque’s, in 1933: twelve millimetres, seven atmospheres, thirty-eight degrees up at one end and thirty-two down at the other. Hilsch’s, in the 1947 paper everyone actually cites: 4.6 mm at eleven atmospheres, a hundred and forty up and fifty-three down.
The trick is that the air goes in sideways.
The nozzle is tangential, so the air doesn’t travel down the tube — it runs around the inside near the speed of sound, drifting along the axis only slowly. Thirty to fifty diameters along, a cone blocks most of the bore, leaving a gap at the wall. The outer sleeve squeezes past and leaves hot. What can’t get past has nowhere to go but back: it turns and runs the length again, inside the outgoing sleeve, still spinning the same way, and out beside the nozzle.
Two streams threaded through each other, going opposite ways. The one that comes back is the cold one.
Which sets up the detail that startles people meeting one on a bench. The control that decides how cold your cold air is sits at the hot end. The cone sets the share of incoming air that turns around instead of escaping hot: take a small share and it comes out very cold, take a large share and you get more of it, much warmer. Nothing adjusts at the cold end at all. You buy colder air by throwing more of it away.
Now the part worth holding onto: nothing is stored.
A refrigerator has a cold part — a box, a coil, some mass that stays chilled after you unplug it. This has none. The difference across the radius of that tube lives in air replaced thousands of times a second. Shut the supply and it doesn’t cool off or decay; it stops existing, and a second later there is nothing left but a slightly warm pipe. The difference isn’t a quantity the tube contains, it’s a shape the flow is holding — No. 21’s drain funnel with a thermometer in it.
Nor is it free. This is not a heat engine — the distinction owed to No. 17 — because no work comes out and no cold reservoir is drawn on. The second law is safe here because of the pressure: air arrives at several atmospheres and leaves at about one, and that drop isn’t converted into anything. It is spent. A compressor in another room did the work, and the tube buys a little cold with a lot of it at a dismal rate — which the literature says out loud, and I like it for that. Vortex tubes get used where “compactness, reliability, and low equipment costs are the major factors and the operating efficiency is of no consequence.” Anywhere compressed air is already piped and nobody is counting.
How it does it has been argued over since before the war, and the first person to reject an explanation of the vortex tube was Ranque. His 1932 patent had the inner layers expanding and cooling while they press on the outer ones and warm them; by 1933 he had abandoned that for friction between layers dragging energy outward. The argument opens with the inventor changing his mind.
The most-favoured account since descends from the second. Air enters as a free vortex, fastest at the middle — the way water speeds up nearing a drain — and shear converts it, down the tube, into a forced vortex turning like a solid body. That carries kinetic energy outward, while turbulent transport down a fierce radial pressure gradient takes the core lower still.
The rivals are live, each with an objection filed against it. That acoustic streaming from the tube’s own whistle drives the separation — except that simulations leaving the acoustics out still reproduce the measurements. That a secondary circulation acts as a refrigerant, a refrigeration cycle folded inside the device — except that other groups calculate the same circulation reduces the separation.
I cannot referee it. What I can say is why it hasn’t closed, and it’s less romantic than a hard theory problem: the measurements don’t agree. Investigators traversing probes across a running tube have reported the static temperature falling outward; others, rising. Ninety years in, the standard review’s verdict is that “it is nearly impossible to predict how a given tube will perform because the exact nature of flow inside the tube is in doubt.”
So “no moving parts” is quietly misleading. Nothing solid moves — but something in there is unmistakably doing mechanical work, dragging energy out of one region and depositing it in another against a temperature gradient, without pause. The machinery is real. It is made of air, and it is rebuilt from scratch, every millisecond, out of whatever arrives next.
≈
One loop I’m watching
Next: a glass tube packed with pond mud, sealed, stood on a windowsill and never opened again. Within weeks it sorts itself into coloured bands, and then it holds them — for months, sometimes years, with nothing going in but light. Each layer lives on what the layer beneath it excretes, and the waste travels back down to be remade, so the same atoms go round and round while the stripes stay exactly where they are. Next time.