Saramsha

No. 024  ·  10 October 2026

Why It Doesn't Stop Halfway

A siphon moves water uphill over a hump with no pump at all — it only needs a tube with no air in it, and an outlet lower than the source's surface. Raise the outlet platform and watch the flow slow to nothing, or let one bubble into the crest and watch it stop at once.

About this piece

Not suction — a column that cannot pull apart

The usual story people tell about a siphon is that it "sucks" liquid over the hump, the way a straw does. That is backwards. A straw works because your lungs lower the air pressure inside it, and the ordinary air pressure outside pushes liquid up to fill the gap — the moment you stop sucking, it stops. A siphon, once it is running, needs nobody to keep sucking at all. What it actually needs is far stranger: a tube that is completely full of liquid, start to end, with no gap of air anywhere inside it. As long as that is true, the tube behaves like a single connected object, and the only thing that matters is the height of the liquid's surface at each end — not the height of the hump in between, and not the length of either leg.

Here is why the surfaces are what matters. Picture the full tube as one long, bent column of liquid. Gravity pulls down on every part of it, but the column is connected, so what counts is the imbalance: the weight of liquid in the downhill leg, past the crest, is pulling that side down, and atmospheric pressure is pushing up on both open surfaces to stop a vacuum from forming behind it. Work through the pressures at the crest from each side and they only balance when the two surfaces are the same height; whenever one surface sits higher than the other, the pressure at the crest from that leg is lower, and the column accelerates toward the side with the taller surface pushing it. Raise the outlet platform above here and you are raising that surface — shrink the gap and the flow slows exactly as the surfaces approach the same level, which is also where it stops on its own, with the hump no different in height than when the flow started. (For a few decades, textbooks disagreed about whether the liquid also needs some tensile strength to hold itself together over the hump, since the pressure there briefly drops below atmospheric; demonstrations in the 2010s run in a vacuum, where atmospheric pressure cannot be doing the pushing at all, showed ordinary water siphons keep flowing anyway — meaning the liquid's own cohesion can carry at least part of the job. Either way, the two surface heights are still what the flow answers to.)

Why one bubble is fatal and a litre of water is not

Letting air into the crest does something a drop in liquid level never does: it cuts the column in two. Each half is now an ordinary open tube reaching down to its own reservoir, with ordinary air pressure pushing on both of its ends — nothing left to create an imbalance, so nothing left to move. It makes no difference that both containers still hold plenty of liquid, or that the two legs of the tube are each still mostly full; the argument above only works for one continuous column, and a single bubble is enough to end that. This is also why priming a siphon — filling the tube completely before starting it, classically by sucking on the downhill end, though a syringe or a quick dunk works just as well — is the one indispensable step, and why topping the tube back up after a break restores the flow immediately if a height difference still exists.

The same two facts — driven by surface height, killed by one bubble — show up constantly outside the lab. A toilet cistern's trap is a small siphon that, once the rising water tips past the top of its bend, empties the bowl completely rather than stopping halfway, for exactly the reason in the title. A fuel thief with a hose is relying on the same geometry, which is why pinching the hose at the top kills it instantly. Aquarium keepers run a "python" gravel cleaner the same way to drain a tank without lifting it, and the Romans crossed valleys with masonry pipes shaped as giant inverted siphons rather than build an arched bridge the whole way. In every case, somebody who understands the mechanism can also defeat it — which is the other half of why it is worth knowing how it actually works.

The simulation treats both containers as having a straight-sided, equal cross-section, so liquid height is directly proportional to volume; the platform under the right-hand container is the only height you control, the tube is always drawn full enough to look primed when it is, and the flow rate is modelled as proportional to the square root of the surface-height difference, in the same family as Torricelli's law for liquid draining from a hole.