The air a wing does not carry

The no-slip condition is one of the few facts in this subject that everybody meets early and nobody doubts: the fluid immediately at a solid surface moves with that surface. It is the reason the blades of a fan stay dusty and the reason a boundary layer exists at all.

The usual next sentence is that a moving wing therefore drags a layer of air along with it, like a blanket, and that skin friction is what it costs to haul that blanket. The sentence is memorable, it follows from something true, and almost every quantity in it is wrong.

How much air is actually being carried #

The question has an exact answer, and the quantity that answers it has been in this collection since the layer’s three thicknesses were measured.

In the ground’s frame, the air in the layer is moving forward at $U - u$: at the wall that is the full flight speed, and at the outer edge it is nothing. The total mass moving with the wing, per unit span, is therefore

$$\rho\int_0^\infty (U - u)\,dy \;=\; \rho U \delta^{*},$$

the displacement thickness — the same quantity that measures how far the outer flow is pushed aside, appearing here in its other role.

At a Reynolds number of a million, over a metre of chord, $\delta^{*}$ is 1.72 millimetres. The layer itself is 4.91 millimetres thick. So even at the moment of counting, the air being carried amounts to a slab a third the depth of the layer, and a fifteen-hundredth of the wing’s chord.

It is worth putting that number beside something familiar. A 1.72-millimetre slab of air at sea-level density, over a wing of 200 square metres, weighs about 0.4 kilograms. An airliner’s wing carries a few hundred grams of air with it, and it does so while producing a hundred tonnes of lift and several tonnes of drag. Whatever skin friction is the price of, it is not the freight.

It is not the same air #

The blanket picture requires the same air to stay with the wing. It does not, and the profile says so without any further calculation.

Only the fluid exactly at the wall has the wing’s velocity: a set of zero thickness. Move a tenth of a millimetre out and the air is moving at a substantial fraction of the free stream relative to the wing, which is to say it is going backwards past it. Sixty-nine per cent of the layer’s depth carries air at more than half the free-stream speed relative to the wing, and that air is streaming past at tens of metres a second.

Nor does the layer keep what it has. It entrains: it grows by taking in fluid from outside, and the fluid it takes in was in the free stream a moment earlier. A layer at the trailing edge of a wing contains almost none of the air that was in it at the leading edge — it contains air that joined somewhere along the way, was slowed a little, and will be released into the wake.

How fast the turnover is has a number, and it comes out of quantities already on this page. At a Reynolds number of a million over a metre of chord the free stream is 15 metres a second, so the mass flowing inside the layer past the trailing edge is $\rho U(\delta - \delta^{*})$, which is 59 grams a second for each metre of span. The mass actually resident in the layer is $\rho$ times the area under $\delta(x)$ along the whole chord, which for a layer growing as $\sqrt{x}$ is two-thirds of its trailing-edge value: 4.0 grams per metre of span. Divide the second by the first and the layer’s entire contents are replaced in 68 milliseconds — about fifteen times a second, at a flight speed slow enough to cycle at.

A blanket that is thrown away and rewoven fifteen times a second is not being carried. It is being continuously made, at the leading edge, out of whatever air happens to be arriving, and the wing has no more relationship with the particular molecules in it than a candle flame has with any particular wax.

What travels with the wing is a deficit, not a parcel. The blanket is a bookkeeping entry, and the air filling it is different air every instant.

And the drag goes the wrong way #

The third claim in the story is the one that fails hardest, because it fails in sign.

Skin friction is $\tau_w = \mu (\partial u/\partial y)_w$ — a gradient at the wall, not a mass. And as the layer thickens, that gradient gets shallower: the same velocity difference is spread over more distance. So the drag per unit area falls as the layer grows.

Reading down that table at fixed chord, a thinner layer carrying less air has ten times more friction. The hauling story predicts the opposite of every row.

There is a cleaner way to say why the two are unrelated. The air carried is the first moment of the velocity deficit; the drag is the wall gradient, which is a local slope. A profile can be made to carry more deficit while having a shallower slope at the wall, or less deficit with a steeper one, and both happen: a laminar layer far downstream does the first, and a turbulent layer does the second. There is no arrangement of the two quantities that makes one a measure of the other.

What no slip actually says #

Stripped of the blanket, the condition is a statement about a limit rather than about a substance.

The tangential velocity of the fluid at the surface equals the surface’s. That is all. It says nothing about a millimetre away, it does not assert that any particular air stays anywhere, and it does not by itself imply a drag — an inviscid fluid with no slip imposed would have an infinitely thin shear layer and no force at all.

What produces the force is the combination of that condition with viscosity: a velocity difference across a finite distance, giving a finite gradient, giving a stress. The condition alone is an extra thing a wall may say because the equation has room for it, and the layer is what the fluid does to make it true.

Where the picture came from, and what it gets right #

The blanket is not a foolish idea and it has an ancestor with a good pedigree: it is what a Couette flow looks like. Two plates, one moving, a linear profile between them, and the fluid genuinely being dragged. That geometry has no free stream, no entrainment and no growth, and everything the story says about it is true.

Carried to an external flow, each of those absences becomes a difference. There is a free stream to entrain from; there is a length along which to grow; and the drag is a local gradient rather than a property of the gap.

What the picture does get right is the sign of the effect on the outer flow. The displaced air is real: the outer inviscid flow behaves as though the body were thicker by $\delta^{*}$, which is why an aerofoil at high incidence effectively has less camber than it was drawn with, and why the pressure recovery on a real section is weaker than ideal flow predicts. The blanket exists as a displacement. It does not exist as cargo.

What the picture cannot show #

These are laminar numbers. A turbulent layer on the same plate is thicker, its shape factor is nearer 1.4 than 2.6, and it carries proportionally less deficit for its thickness — so the ratio of carried air to layer depth is even smaller, and the friction is several times higher. Every qualitative conclusion above is stronger in the turbulent case and none of the numbers survive.

No slip is empirical. It is not derived from the equations of motion; it is what ordinary fluids are observed to do, and it took a century of argument to settle. In a gas rarefied enough that the mean free path is comparable with the geometry, a slip condition replaces it — which is a fact about the state of matter and belongs to a different collection than this one.

And the frame matters more than usual here. The profile drawn in the wing’s frame and the same profile drawn in the ground’s frame invite opposite intuitions, and every number in this essay has been computed in one and stated in the other with the transformation written out. That is worth care: a reader who mixes them will conclude that the layer’s air is moving backwards, which is true in one frame and nonsense in the other.

What the layer does carry, which is momentum #

Dropping the blanket leaves a question worth answering: if the drag is not the price of hauling air, what is it the price of?

It is the price of momentum given away. The layer leaves the trailing edge with a velocity deficit, and that deficit is a momentum deficit — air that was moving at flight speed relative to the ground and now is not. The quantity that measures it is the momentum thickness, and von Kármán’s momentum integral says the drag is exactly $\rho U^2\theta$ per unit span, with $\theta$ evaluated at the trailing edge and nothing else in the expression.

That is a genuinely different accounting from the one the blanket suggests, and the difference shows up where it matters. The momentum thickness at a Reynolds number of a million is 0.66 millimetres per metre of chord against a displacement thickness of 1.72 — the two are in the ratio of the shape factor, 2.59, and they measure different things. Deficit of mass is what the outer flow sees; deficit of momentum is what the drag is. A wing that displaced a great deal and gave away no momentum would push the outer flow aside and cost nothing, which is very nearly what a well-designed laminar section does.

So the honest replacement for the blanket is a wake: a thin ribbon of slower air stretching behind the aircraft, spreading and weakening, containing exactly the momentum the engines had to supply.

The air a body really does carry #

There is a sense in which a body does take fluid with it, the quantity is not small, and it is nothing to do with no slip. Setting it beside the blanket is the cleanest way to see what the blanket was reaching for.

Accelerate a body through a fluid and the force needed exceeds its own mass times its acceleration, because the fluid around it must be accelerated too. In potential flow that surplus is exactly a mass: the mass a body has to borrow, $\rho \pi a^2$ per unit span for a circular cylinder, which is the mass of the fluid the cylinder displaces. For a thin plate of chord $c$ accelerating broadside it is $\rho\pi c^2/4$ — for a one-metre chord in sea-level air, 0.96 kilograms for every metre of span, against the 1.4 grams per metre of span of velocity deficit that the boundary layer holds. The genuinely carried air outweighs the blanket by a factor of 685, and it is carried by an inviscid mechanism, in a calculation with no viscosity, no wall gradient and no no-slip condition anywhere in it.

And it is not a quantity at all — it is a tensor. The same plate accelerated edgewise, along its own chord, borrows nothing whatever: a flat plate has zero thickness in that direction and the ideal flow round it is undisturbed. So how much air a wing carries depends on which way it is shaken, and the answer ranges from a kilogram a metre to exactly zero for the same body in the same air. No cargo behaves like that, and the fact that this one does is the surest sign the quantity is not freight but a property of the field the body sets up.

This is why the borrowed mass is invisible in cruise and dominant in a manoeuvre. Flying straight and level, nothing is accelerating and nothing is borrowed; the only air the wing has any claim on is the 1.7-millimetre deficit, and the drag is a wall gradient. Pitch into a gust and the wing must accelerate a mass comparable with a small car before it has changed its own velocity at all — which is the term that makes a light aircraft’s response to a sharp-edged gust smaller than the steady lift curve predicts, and which is an impulse rather than a momentum, because the momentum integral over the whole fluid does not converge.

Two quantities called the air a wing carries, differing by nearly three orders of magnitude, and the smaller one is the one the folklore is about. The blanket picture takes a viscous condition at the surface, infers a mass, and attributes a force to it. The real carried mass comes from the pressure field, has no surface condition in it, and produces a force only when the velocity changes. Getting the two confused is not merely an error of size; it puts the mechanism in the wrong half of the subject.

Where the phrase does real damage #

A memorable wrong picture is not harmless, and this one costs three things.

It makes the boundary layer sound passive. A blanket is something a wing has; a boundary layer is something a wing is continuously doing, and the doing is where every interesting phenomenon in this collection lives. Separation, transition, the drag crisis, the stall — none of them is a property of an amount of air, and all of them are properties of a profile’s shape and of what the pressure outside is doing to it.

It makes viscosity sound like stickiness. The word suggests adhesion — air clinging to a surface — where the mechanism is momentum diffusing sideways through the fluid, a molecular transport that has nothing to do with the wall except that the wall is where the velocity difference is anchored. A reader who has the blanket picture will expect a smoother surface to carry less air and therefore less drag, and will be surprised that polishing a wing past a certain point does nothing at all while a fraction of a millimetre of roughness can be invisible to the flow entirely.

And it hides the wake. The layer’s whole output is a momentum deficit that leaves at the trailing edge and stays in the air behind the aircraft; the blanket picture has the air arriving and staying, which is precisely backwards. What a wing leaves behind it is the whole of what it paid.

The one-sentence version #

A wing carries a millimetre or two of air per metre of chord, it is different air every instant, and the friction it pays falls as the amount rises. The no-slip condition is a boundary condition, not a cargo manifest, and the layer it produces is a place where a velocity difference is arranged rather than a load being hauled.

Where the ladder goes next #

The rung above is what happens when the layer is deliberately not allowed to grow — sucked away through a porous wall, which this collection has already solved exactly and which produces a layer with no distance in it at all. Suction removes the displacement and raises the friction, in exactly the ratio this essay’s table predicts.

The one beside it is the other thing the second boundary condition does. The blanket picture asks what a wall takes from the fluid; the sharper question is what it gives — vorticity, at a rate the pressure gradient sets, with no viscosity in the answer at all.