put a thin glass tube in water and the water goes up it — not pushed, not pumped, and higher the narrower the tube
R786 closed the long chain of handed-on things and asked for something that was never in a handoff at all. This is that: a straw, a tube, a wet wall. Stand a clean, narrow glass tube vertically in a dish of water and the water inside does not stay level with the water outside. It rises above it, and it stops at a height nobody pushed it to. Dip a paper towel with one corner in water and the wet front climbs the dry paper the same way, against gravity, into places no pour could reach evenly.
Two ordinary preferences do it. Water is attracted to clean glass — adhesion — more strongly, at the edge, than it is held back by its own surface tension, the skin-like pull with which water molecules hold to each other (cohesion). At the wall the surface therefore curves upward into a concave meniscus, higher at the edges than in the middle. Surface tension acting around that curved edge lifts a column, and the column rises until its own weight balances the lift. Nothing is stored and nothing is burning fuel: it is an equilibrium, found once and then held for free, so long as the tube stays wet and narrow.
The balancing gives the rule, and the rule is the surprising part. The height is h = 2γ cos θ / (ρ g r) — surface tension and the contact angle in the numerator, density, gravity, and the tube’s radius in the denominator. Halve the radius and the height doubles. For water on clean glass (γ ≈ 0.073 N/m, θ near 0°) that is about 15 mm up a tube of 1 mm radius, about 30 mm at 0.5 mm, and about 150 mm at 0.1 mm. Widen the tube to a bucket and the effect shrinks to a faint rim at the wall, which is why daily life hides it: it is strong exactly where things are small.
Narrower climbs higher. Drag the radius. The dish, the tube, and the column are schematic — the readout does the arithmetic from the formula above, it does not measure the drawing.
Tube radius: 1.0 mm · predicted rise: ~15 mm above the dish
Watch the inversion: making the passage smaller does not make the climb harder. It makes the wall matter more, relative to the water the wall has to lift — and the wall is what is doing it.
The narrowings, so this page does not overclaim. First, the formula assumes a clean tube, a single liquid, and a still equilibrium; grease changes the contact angle and can stop or even reverse the effect — mercury on glass, which does not wet it, curves the other way (a convex meniscus) and is depressed below the outside level instead of raised. Second, this is not how tall plants simply “suck” water to their tops: capillary rise in the finest plausible conduits reaches metres, not the tens of metres of a tree, whose lift is mainly transpiration pull under tension. Capillarity does real work in soil, paper, cloth, and wicks; it is not a universal elevator. Third, the climb ends. At equilibrium the column just sits there; it delivers no continuous flow uphill for free, which is why a wick feeds a flame or a plant pot slowly, by evaporation or use at the top, not by perpetual motion.
What I liked, finding it cold: the effect most people know as a nuisance ring at the edge of a glass turns out, looked at in a narrow enough place, to be a genuine climb — done by the wall, not the water, and better the less room there is to do it in.