In a tube the standing members lose

Standing up beats lying down by eight put two ways of filling a space against each other and found a ratio with no size in it. Members standing across a clearance each have the whole of it and compete only for somewhere to stand; layers lying along it share the depth and pay for the sharing. The quotient was eight at every clearance and every sheet thickness, and leaning the members over turned out not to change it either.

Every drawing of that argument has a flat base under it, and a flat base is the one thing no body has. A gut is a tube, an airway is a tube, a vessel is a tube, and a lining that stands inward from the wall of one is not standing on a plain.

Curving the base does not weaken the eight. It reverses it.

The room a fin needs is at its tip #

On a flat base two members are parallel slabs and the gap between them is the same everywhere along their length. Inside a tube they are not parallel: standing on the wall at radius $R$ and pointing at the axis, they converge, and the gap between two neighbours shrinks as they rise.

At the wall the pitch is $2\pi R/n$; at the tips of fins of height $h$, which sit at radius $R-h$, it is $2\pi(R-h)/n$. The binding constraint is the second, so the count is set by the radius the fins reach rather than the radius they stand on:

$$n \le \frac{2\pi (R-h)}{\tau}$$

Each fin still has two faces of height $h$, so the surface per unit length of tube is

$$f_{\text{fins}}(h) = \frac{4\pi h (R-h)}{\tau}$$

which is a parabola in $h$ with roots at $0$ and $R$. Its best height is exactly half the radius, and there is nothing else in it — no thickness, no supply, no property of the material. Swept numerically at two radii the maximum sits at $0.5000R$ in both, and what it reaches there is $\pi R^{2}/\tau$.

A ceiling that holds for any arrangement whatever #

The number to compare that against is not another arrangement. It is a bound.

A flat sheet of thickness $\tau$ has two faces, so a piece of it occupying a volume $V$ carries $2V/\tau$ of surface — whatever shape it is bent into, and however it is arranged. A lining of a tube occupies at most the tube’s cross-section, so no lining at all can hold more than

$$f_{\max} = \frac{2\pi R^{2}}{\tau}$$

per unit of length. That is not a property of fins or of layers; it is arithmetic about sheet.

Concentric layers at pitch $\tau$ reach it. Summing $2\cdot 2\pi r$ over radii from $R$ inward in steps of $\tau$ gives $2\pi R^{2}/\tau$ in the continuum, and the discrete sum exceeds it by exactly the half-layer a whole count leaves over — 2% at radius 0.5, 1% at radius 1, 0.5% at radius 2, 0.25% at radius 4, halving each time the tube doubles.

The fins at their best reach exactly half of it, and the reason is visible in the volume rather than in the surface. At $h = R/2$ the fins occupy $2\pi(R-h)h = \pi R^{2}/2$ of a cross-section of $\pi R^{2}$. They fill half the tube and leave half of it empty, and the half they leave is the set of wedges between converging fins, which widen exactly as fast as the fins themselves narrow the space.

A ceiling the flat model never had #

The bound deserves a moment on its own, because a flat base does not have one and that absence is why every earlier answer grew without limit.

On a plain, a comb of walls of height $c$ reaches $2c/\tau$ and a taller wall always holds more. There is no ceiling because there is no fixed volume: raise the clearance and the box gets bigger. The only thing that ever stopped the growth on that line of argument was the material — how much surface fits in a body found a corrugation’s curve turning over because the sheet’s own thickness fills the box it folds into, and set the thickness to zero and the turn disappears.

A tube is different in kind. Its volume is fixed by its radius, so the bound $2\pi R^{2}/\tau$ exists before any arrangement is proposed, and it survives setting the thickness to zero in the only sense that matters: the surface and the bound both go to infinity together and their ratio stays put. The ceiling here is made of the shape of the space and not of the material, which is the opposite of every earlier limit on this line of argument and is why the two must not be read as the same kind of result.

It also explains why the fins have a best height at all. A quantity bounded above and zero at both ends of its range has an interior maximum; a quantity that rises without limit does not. The flat comb has no optimum because it has no ceiling, and the tube gives it one for free.

What the eight was really about #

It is worth being exact about which sentence survives.

The eight came from a difference in what the two architectures are charged for: standing members do not share the clearance, so they compete only for footing, while lying layers share the depth and pay a quadratic penalty. That sentence is still true on a flat base, and it is still why standing members win there.

Inside a tube, standing members do share. What they share is not the clearance — each fin still reaches whatever height it likes — but the room at the far end of the reach, which is the very thing the flat model had no name for because on a plain it is the same as the room at the near end. Converging members compete twice: once for footing at the wall, where they are far apart, and once for clearance at the tips, where they are close. The second contest is the one they lose.

So the advantage of standing was never about standing. It was about the members’ two budgets being charged in different directions, and curvature is the operation that collapses them into one.

The same reversal, read as a division of a budget #

There is a second way to see the factor of two, and it connects this to a result already on the record.

Two surfaces in one box found that dividing a depth between several surfaces looks free and is not, because the area one surface reaches goes as the square of the depth it has. The tube does the same arithmetic to a single lining. A fin of height $h$ is charged for the depth it reaches — the count falls as $R-h$ — and paid for it in proportion to $h$, so its total is a product of a rising factor and a falling one, which is the signature of a shared budget.

Concentric layers have no such product. Each layer is paid its own circumference and charged its own $\tau$ of radius, and the two are charged against different things — one against the perimeter, one against the radius — so nothing multiplies and nothing turns over. The layers win because their two budgets stay separate, which is precisely the property that made the standing members win on a plain.

The ranking inverts because the property changes hands. That is a more useful statement than the factor, because it says what to look for in a third setting rather than which of two arrangements to prefer.

What fins can do about it #

The fins have an answer and it is worth measuring rather than dismissing, because it is what a real lining looks like.

The room a fin needs is the room at its tip, so a shorter fin fits where a longer one does not. Filling the gaps between long fins with shorter ones recovers part of the wedge. Measured at radius 1 out of a sheet 0.01 thick: one length at $0.5R$ holds 314, which is 50 per cent of the ceiling; adding a second at $0.25R$ takes it to 393, a third at $0.125R$ to 412, a fourth to 417. A different schedule — $0.6R$, $0.3R$, $0.15R$ — reaches 443, which is 70 per cent of the ceiling.

Each extra length is worth less than the one before, and the series does not converge on the ceiling. A fin of any length still leaves a wedge behind its own tip, and a hierarchy of fins is a hierarchy of wedges.

That last point is the one worth carrying out of the measurement. Four lengths at successive halvings hold 417; three lengths at $0.6R$, $0.3R$ and $0.15R$ hold 443. More lengths is not better than better lengths, and the reason is that a schedule which starts too short gives away the long fins’ contribution without buying enough count to pay for it. The schedule is a small optimisation problem of its own, and nothing here solves it — the five drawn are five guesses.

What can be said without solving it is where the gap comes from. A fin of height $h$ leaves a wedge behind its tip whose area is a fixed fraction of the disc it reaches into, and filling that wedge with shorter fins leaves smaller wedges of the same shape. It is a self-similar shortfall, which is why the recovery is quick at first and then slow, and why no finite schedule closes it.

Half, and why it is exactly half #

The factor deserves one more sentence, because a ratio of two in a geometry problem is usually approximate and this one is not.

The fins’ volume at height $h$ is $2\pi(R-h)h$ — the count times the height times the thickness, with the thickness cancelling. Maximised over $h$, that is a parabola with the same roots as the surface, so the arrangement that holds the most surface is also the one that occupies the most cross-section, and both peak at $h=R/2$. There the volume is $\pi R^{2}/2$, against a cross-section of $\pi R^{2}$.

So the half is not a coincidence of the optimum: the fins’ surface and the fins’ volume are the same function up to the factor $2/\tau$, so whatever maximises one maximises the other, and the fraction of the cross-section the best arrangement occupies is the fraction of the ceiling it reaches. The layers occupy everything and reach everything; the fins occupy half and reach half. One statement, read twice.

What the tube assumes #

Fins are flat radial slabs of constant thickness. Each lies in a plane through the axis, so the gap between two of them shrinks monotonically from the wall inward and the tip is where the constraint binds. A fin that tapered would relieve exactly that, and a fin that spiralled would not converge at all.

Layers are concentric and complete. The count is $R/\tau$ and each is a full cylinder. A layer with a gap in it, or one that spirals from the wall to the axis, holds the same to within the discretisation, which is why the ceiling is reached by a family rather than by one arrangement.

Counts are treated as real numbers. The half-layer excess is what that costs, and it is reported rather than rounded away because it is the whole of the difference between 2.0400 and 2.

And nothing here is supplied, and nothing flows. The channel grows with what it feeds charges a channel against the pitch, and in a tube the pitch depends on height, so a supplied version has both effects at once and is not computed. A tube something moves along is a different problem again, and the layer arrangement that wins here is the one that obstructs it most.

What the numbers cannot settle #

They do not say a lining should be concentric. Surface is one quantity among several, and the arrangement that maximises it fills the tube solid, which is useless for anything a tube is for. The result is a bound and a ranking, not a recommendation.

They do not cover a cross-section that is not a circle. Fins standing inward from an ellipse reach toward the inward offset of an ellipse, which has corners, so the count is limited by different amounts at different places round the wall and the best height becomes a function of position.

They say nothing about what happens between the members. A fin comb leaves wedges and a layer stack leaves annuli, and transport through those two spaces has nothing in common. Every quantity here is geometry at rest.

And a described structure is not a measurement. No organ is measured here, and nothing grown has a seam is the standing reminder that a lining is reached and built rather than placed. What the model gives is a way of asking which economy an arrangement is in — and the answer fins now carries a specific cost, where before this it carried an advantage.

Nor does it say a lining is one thing or the other. A structure with layers near the wall and fins further in is a third arrangement with its own accounting, and so is a fin that carries a corrugation of its own — which is the price a nest pays applied to a curved base, and is not computed anywhere.

Still open: the fin that tapers #

One assumption is doing all the damage and it is the easiest to relax.

A fin of constant thickness needs its full $\tau$ of room at the tip, where room is scarcest, and none of its length needs the room it has at the wall, where room is abundant. A fin that narrows as it rises is spending its material where the material is cheap, and the constraint it must satisfy is a relation between its profile and the radius rather than a single number.

The computation is well posed. Let a fin have thickness $\tau(r)$ at radius $r$; the count is bounded by $2\pi r/\tau(r)$ at every radius, so the binding constraint is the tightest of those, and the surface is the count times twice the length. Asking which profile maximises it, subject to a fixed amount of material, is a one-dimensional problem with a closed-form answer — and the answer would say whether a tapered fin can approach the ceiling that a constant one falls half short of, or whether the wedge is irreducible.

That question has a second half worth stating with it. If tapered fins do close the gap, the factor of two here is an artefact of an assumption rather than a fact about the geometry, and the honest headline becomes constant-thickness fins lose. If they do not, the wedge is a property of radial arrangement and the ranking inverts for good.

There is one more thing the taper question would settle, and it is about supply rather than surface. The surface has to be supplied charges a channel against the same budget the members compete for, and in a tube that budget is a function of position: there is room to spare at the wall and none at the axis. A lining whose members thin as they rise is also a lining whose channels have somewhere to run, and the two economies point the same way for once — which is unusual on this line of argument and worth checking rather than assuming.

Sideways from here, the shape of this result belongs beside the one it overturns. A ratio computed inside one shape of container is a fact about that container, and the flat base was never stated as an assumption because a diagram has to draw something. The eight was measured correctly, checked at four sizes, and found insensitive to the clearance, the thickness and the angle of the members — and it was a fact about a box the whole time.

The habit worth carrying: vary the thing the figure had to choose. Not the parameters it exposes, which are the ones somebody already thought were free, but the ones it had no way of not committing to — the shape of the frame, the flatness of the ground, the straightness of the edge.