Last time I promised a fluid that is only liquid while you keep hitting it, and that turns solid the instant you stop.
Two words in that sentence are wrong, and they are wrong in opposite directions.
Hitting is the wrong verb. Hit a bowl of cornstarch and water and it goes hard — that is the party trick, and it is the reverse of the material I meant. Instant is wrong more seriously. The material I had in mind is defined by not setting instantly. The delay is the entire point.
The last clause I’ll keep, because it was right: this is a material whose mechanical identity is a rate rather than a state.
The word is thixotropy.
Two rates, again
No. 118 left a reactor whose xenon level looked like a stock and turned out to be a race between two rates. This is the same shape in a jar.
Stirring tears a loose network of particles apart, at a rate set by how hard you stir. Brownian motion and short-range attraction put it back together, at a rate set by the material. The thickness you measure is wherever those two rates happen to stand at the moment you measure it. Howard Barnes, in the standard review, puts the scope of this bluntly: all liquids with microstructure can show it.
Ketchup. Non-drip paint. Some clays. Quicksand. And the one where it earns real money.
The word came out of an egg
Thixotropy sounds like an engineering term. It isn’t, quite.
It was coined in 1927 by Tibor Péterfi, a Hungarian biologist, in the Archiv für Entwicklungsmechanik der Organismen — an embryology journal — from the Greek thixis, a touching or shaking, and trepo, to turn. He was describing the inside of a cell: cytoplasm that behaved like a solid until it was disturbed and then flowed. Herbert Freundlich carried the word into colloid science, and it spent its working life in oil wells and paint factories.
It was never a name for a substance. It was a name for something a substance was caught doing.
The drillers write down three numbers
Drilling mud has two contradictory jobs. Pumped down a well it must flow freely. The moment the pumps stop it must stiffen enough to hold rock cuttings in suspension, instead of letting them settle onto the bit.
So: how thick is it?
The industry does not answer that. The standard API procedure measures gel strength after the mud has been left still for ten seconds, again at ten minutes, and again at thirty. Three numbers, not one, because there is no one.
That is the finding under the party trick. The thickness of a thixotropic fluid is not a value with a duration attached as a footnote. The duration is inside the value. Ask without saying when, and you have not asked anything.
The other fluid, properly
Cornstarch is not a slow version of ketchup. It is a different mechanism pointed the other way.
Push a dense suspension of hard grains hard enough and you force the particles into direct contact. Their rough surfaces cannot slide past one another, and they lock into transient force chains that carry the load like a solid — and under impact that locking spreads as a front, which is what holds a running adult. Shear builds the structure there. In ketchup, shear destroys it.
And oobleck barely has a clock. Its response tracks how fast you are pushing, not how long ago you stopped. Take the load off and it slumps. The two fluids get filed together as “weird non-Newtonian goo,” and they are close to opposites.
Whether it counts depends on your instrument
Barnes notes that thixotropy is expected wherever shear-thinning happens at all — microstructure cannot rebuild in zero time — and that it becomes significant only when its timescale is longer than the response time of the instrument watching. Faster rheometers have moved fluids across the line. He also warns that instrument inertia produces a lag routinely mistaken for the real thing.
So the honest sentence is not “this fluid is thixotropic.” It is “this fluid rebuilds slowly enough that my equipment can see it.”
Underneath sits an older, unsettled argument. In 1985 Barnes and Walters published “The Yield Stress Myth?”, arguing that with good enough instruments, materials that appear to have a stress below which they simply do not flow turn out to creep. The definition makes itself untestable: proving the flow rate is truly zero takes forever. Rheology now holds two views at once — a real solid-to-liquid transition, or two liquids with wildly different viscosities — and treats yield stress as an engineering reality regardless. Both camps are being reasonable.
What this is not
Wet sand (No. 80) also firms and loosens, but its strength comes from capillary bridges pulling grains together, and it rebuilds on no clock of its own — it waits for water. A sandpile (No. 62) holds an angle until a threshold is crossed, but nothing in it is quietly reassembling between avalanches. Thixotropy needs the rebuild to be ongoing and slow. Take away the clock and there is nothing left to name.
≈
One loop I’m watching
Next: an insect that counts to thirteen or seventeen underground, emerges by the billion for a few weeks, and is protected by nothing except how many of it there are. Next time.