Cohesion-tension sap ascent
A Socratic walk-through of cohesion-tension sap ascent — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can a tree draw water a hundred feet up a narrow pipe when the best pump on the ground cannot lift it past thirty?
A perfect suction pump at the bottom of a well fails at about ten metres however good it is, and that is arithmetic rather than engineering. The pump does not pull water — it removes pressure above the water so the atmosphere can push it up, and the atmosphere has only so much push. Divide atmospheric pressure by the weight of a water column, 101 kPa / (1000 kg m^-3 x 9.81 m s^-2), and you get 10.3 metres. Thirty-four feet, and there it stops.
A coast redwood is three hundred feet tall and moves water to its top leaves all day without a moving part or a calorie spent on lifting. So either the tree is doing something the pump cannot, or we have mis-stated what the pump's limit is a limit on. Which is it?
Reasoning it through
REASONING #Take the second possibility, because it is the productive one. The ten-metre ceiling limits pushing with atmospheric pressure. It says nothing about water genuinely pulled — pressure taken not merely toward zero but below it, into tension. Can a liquid be under negative pressure? Only if it can be stretched, which requires its molecules to hold on to one another, and water's hydrogen bonding makes it unusually good at that: water in a fine tube can take many megapascals of tension without separating.
So the mechanism must be a pull, not a push. Where does the pull come from, and what pays for it? Follow the water to its exit. In a leaf, water evaporates from the wet walls of the mesophyll cells into the air spaces inside the leaf. As it goes, the remaining water retreats into the pores of those cell walls, and its surface curves. A curved air-water interface carries a pressure difference across it — 2 x surface tension / radius. Put in water's surface tension of about 0.072 N/m and a wall pore of ten nanometres and you get 2 x 0.072 / 1e-8, near 14 megapascals of available tension. Now ask what the tree actually needs. Lifting water thirty metres costs 1000 x 9.81 x 30, about 0.3 MPa; a hundred metres costs roughly 1 MPa, ten atmospheres' worth. The evaporating surface has an order of magnitude more suction available than gravity demands.
Notice the energy budget: the tree spends nothing on the lift. Sunlight evaporates the water, the menisci left behind do the pulling, the column follows. The tree's contribution is not a pump but a plumbing specification — so what must that specification be?
Three things follow. The column must be continuous from soil to leaf, because tension travels only through unbroken liquid. The conduits must not implode, since a pipe with a strong vacuum inside is crushed from outside by the full atmosphere — which is why xylem conduits are dead, emptied cells with thick lignified walls reinforced by rings and spirals. And the system must contain its own failures, because a liquid under tension is metastable: not at equilibrium, merely holding on.
That last point is where the real limit lives. The column rarely breaks by pulling apart in mid-water. It breaks by air seeding — a bubble in an already-empty neighbouring conduit is drawn through the pores of the pit membrane in the shared wall, and the conduit fills with vapour and goes out of service. The threshold is the same capillary relation run in reverse: the widest pore in the pit membrane sets the tension that conduit can hold. Every conduit is therefore a compartment, and an embolism is a lost compartment rather than a lost tree.
The analogy
THE ANALOGY #A rope hanging down a shaft is not pushed up from below; it is held from above, and every strand carries the weight beneath it. It works to any depth so long as the rope is continuous and no strand parts. The failure is never gradual — it is a snap at the weakest point, and everything below that point falls.
a rope has tensile strength of its own, whereas a water column's strength is only its refusal to admit a gas phase, so it does not break where the load is highest but wherever the largest pore first lets air through — and unlike a rope, a broken column can sometimes be refilled and put back into service.
Clarifying the model
THE MODEL #The frequent misreading is that the tree "sucks" water up, which imports the pump's ceiling by the back door. Nothing in the plant lowers a pressure toward vacuum; evaporation at the top establishes tension, and the whole column from root to leaf sits at pressures below zero absolute — a state a mechanical pump cannot produce and could not sustain if it did.
The second, worth separating, is that this is the same story as a leaf closing its pores at midday. It is not. Stomatal regulation is about when to spend water at an hourly-changing price; this is about how a liquid can be lifted at all. The two meet only at the failure mode: a leaf economises because overspending puts the column past its air-seeding threshold.
Is it falsifiable? Squarely. It predicts that xylem sap in a transpiring tree is under negative pressure and that no living tissue is needed for transport. Kill a length of stem with heat or poison and water still ascends through it, which rules out a living pump; cut a transpiring stem and air is drawn violently inward, the wrong direction for anything pressurised. The refuting observation would be sap found at positive pressure in a transpiring trunk — and that is not hypothetical: pressure-probe measurements in the 1990s repeatedly failed to find the predicted tensions, a serious and public challenge. Evidence has settled back behind cohesion-tension, largely because the probe itself nucleates bubbles at high tension, but the theory was genuinely at risk rather than merely assumed.
A picture of it
THE PICTURE #How to readThis is a requirements diagram repurposed as a specification for a water column: the boxes are conditions that must hold at once, not steps in a sequence. Start at the top box and read downward — lifting sap a hundred metres derives two conditions, negative pressure and an unbroken column, and the unbroken column derives two constraints on the hardware. The two rounded elements are the physical realities attached to those conditions: the menisci that satisfy the tension requirement, and the embolism that occurs when the pit constraint is violated. Remove any one box and the whole thing fails.
What became clearer
WHAT CLEARED #The pump's thirty-foot ceiling was never a limit on moving water upward — only on pushing it with the atmosphere. A tree does not push. Evaporation at the leaf surface creates menisci whose capillary tension exceeds gravity's demand by roughly an order of magnitude, and the sun pays for it, so the tree's real problem is not generating the pull but building plumbing that survives being permanently in tension: continuous, collapse-proof, and compartmented so the inevitable breaks stay local.
Where to go next
ONWARD #- How some plants refill embolised conduits while neighbouring ones are still under tension, which remains contested.
- Why the tallest trees cluster in fog belts, and whether foliar uptake or reduced evaporative demand does more of the work.
Key terms
TERMS #| Term | What it means |
|---|---|
| Cohesion-tension theory | the account, from Dixon and Joly in the 1890s, in which sap ascends as a continuous column pulled by evaporation and held together by water's cohesion. |
| Water potential | the free energy of water per unit volume relative to pure free water, in megapascals and normally negative in a plant. |
| Air seeding | entry of air into a working conduit through a pore in a shared pit membrane, from a neighbour already embolised. |
| Embolism | a conduit rendered non-conducting by a gas bubble. |
Every term the collection defines is gathered in the glossary.