There is something genuinely comic about a mathematician discovering a shape and then realizing a nautilus figured it out four hundred million years ago. Not comic in the gentle, isn't-that-lovely sense. Comic in the way that a pratfall is comic — someone very smart walking confidently in one direction, doing excellent work, and then the floor isn't there.
Let me set the scene. The question seems almost childish: how few corners can a shape have and still tile a plane? That is, how smooth, how rounded, how un-cornered can a single tile become while still fitting together with copies of itself to cover a flat surface completely, no gaps, no overlaps? This is not idle curiosity. Tiling problems sit at an old crossroads of geometry, combinatorics, and physics, and they have a habit of looking simple right up until the moment they become profound. The answer that emerged this year — from Domokos and collaborators — is a class of shapes now called soft cells. They are tiles with the minimum possible number of sharp corners: in two dimensions, two; in three dimensions, potentially none at all. Smooth, bulging, organic-looking forms that nonetheless tessellate perfectly.
And here is the punchline: these shapes, new to the mathematical literature, already existed in nautilus shells, in the cross-sections of red blood cells, in onion layers, in the chambers of biological forms that have never heard of topology and couldn't care less about your theorems.
✦
I love this kind of embarrassment. Mathematics is supposed to be the austere discipline, the one that proceeds from axioms, that builds its cathedrals in the air before anyone checks whether they correspond to anything in the mud. And often enough that's true — whole fields of pure mathematics exist for decades or centuries before applications arrive, if they ever do. Non-Euclidean geometry waited nearly a hundred years before Einstein needed it. Group theory was an abstraction in search of a universe until quantum mechanics provided one. We tell this story to ourselves constantly: pure thought first, the world catches up later. It flatters us. It suggests that mind is prior to matter, that the mathematician is a kind of prophet, scribbling the blueprints of a house that reality hasn't built yet.
Soft cells reverse the arrow so completely it's almost rude.
The nautilus did not solve a tiling problem. The nautilus doesn't know what a corner is. What it "knows," insofar as that word means anything here, is how to grow a shell by accretion, adding material along a logarithmic spiral, partitioning chambers in a way that is structurally efficient and metabolically cheap. The geometry follows from the physics, the physics from the biology, the biology from four hundred million years of not dying. And the shape that results — this smooth, cornerless, exquisitely tessellating form — is what a mathematician in 2024 arrives at by asking an abstract question about the minimum combinatorial complexity of a tiling.
They met in the middle. But the nautilus had been waiting at the rendezvous point since the Devonian.
✦
There is a tendency, when confronted with this kind of convergence, to become reverential. To say something about the unreasonable effectiveness of mathematics, or the deep structure of reality, or the mind of God as a geometer. I want to resist that. Not because it's wrong, exactly, but because the reverence obscures something more interesting: the convergence is not magical. It's constrained.
The nautilus does not have infinite choices for how to build a partitioned shell. Physics constrains it. Surface tension, material strength, the diffusion of gases across membranes, the cost of producing calcium carbonate — all of these push toward certain geometries and away from others. Sharp corners are stress concentrators. They are expensive to build and fragile once built. Any organism that partitions space repeatedly, under pressure to minimize material while maximizing structural integrity, will tend toward smooth boundaries. It would be more surprising if the nautilus had corners.
And the mathematician, starting from the other end — asking which shapes tile with the fewest vertices — is also navigating a constraint landscape. The constraints are logical rather than physical, but they are no less rigid. You cannot tile a plane with circles. You cannot have zero corners in two dimensions and still tessellate (you need at least two, it turns out). The logic funnels you toward the same forms that physics funnels the nautilus toward, because in some sense the problems are isomorphic. Both are asking: what is the smoothest way to fill space without waste?
So the convergence isn't spooky. It's inevitable. And that, to me, is more interesting than spooky. Spooky lets you stop thinking. Inevitable makes you ask why the constraint landscapes match, and whether they always do, and what happens when they don't.
✦
When they don't — that's where things get genuinely strange. Because sometimes mathematics produces forms that nature has no use for. Aperiodic tilings, for instance. Penrose tiles, those famous rhombs that cover a plane without ever repeating. For years they seemed like pure play, geometrical curiosities. Then quasicrystals were discovered in the 1980s — real physical materials with the symmetry properties of aperiodic tilings — and the play turned out to be prophecy after all. But there are vast continents of mathematical structure that remain, as far as anyone can tell, unclaimed by the physical world. Exotic smooth structures on four-dimensional manifolds. Monstrous moonshine. The absolute Galois group. These are not waiting to be discovered in seashells.
Or maybe they are, and we don't know where to look. That's the honest position. We can never be sure whether a mathematical structure has no physical realization or merely no known physical realization. The history of the field is littered with confident declarations of uselessness that aged poorly. But I suspect — and this is just temperament, not argument — that some mathematical objects are genuinely orphaned. They are the shapes that no physics demands, the symmetries that no field obeys, the structures that exist only because logic permits them. They are possible but not actual. And there is something quietly unsettling about that: a vast warehouse of forms that could have been the architecture of a universe but aren't the architecture of this one.
Why this one? Why these shapes and not those? The question is probably unanswerable, possibly meaningless, and I think about it more than is healthy.
✦
Back to the nautilus, because I can't leave it alone. There's a specific irony in the fact that the logarithmic spiral of the nautilus shell has been a poster child for mathematical beauty in nature for centuries — every popular science book, every inspirational poster about the golden ratio (most of them wrong about the golden ratio, but never mind). We have aestheticized this creature relentlessly. And now it turns out the animal was doing something even more mathematically interesting than the spiral: it was tiling three-dimensional space with soft cells, solving a topological optimization problem in its internal chamber geometry, and nobody noticed because we were too busy admiring the outside.
We looked at the curve and missed the tiling. We saw the flourish and ignored the brickwork.
This happens in astronomy too. We are drawn to the spectacular — the supernova, the collision, the dramatic death of a star — and meanwhile the quiet structures, the ones that persist because they are optimal, because they have found the only configuration that works under the constraints, go unexamined. The most physically interesting thing about a galaxy might not be its spiral arms but the spacing of its molecular clouds, the tiling of its interstellar medium into cells of hot and cold gas, each one shaped by pressure equilibrium and magnetic tension into forms that are probably, if anyone bothered to characterize them precisely, soft in exactly the way the nautilus chambers are soft.
I don't know if anyone has looked. I would like someone to look.
✦
The word "soft" in "soft cells" is doing quiet, perfect work. It means exactly what you think it means — smooth, cornerless, yielding — but it also carries the implication of something that could have been hard and chose not to be. As if the geometry relaxed. As if the shape exhaled and let go of its vertices. There is a whole philosophy of form in that name, whether or not the authors intended it: that the most efficient solutions are often the ones that have surrendered their sharpness. That rigidity is expensive and softness is cheap. That the best way to fill a space might be to stop insisting on edges.
I'm aware this sounds like it wants to be a metaphor for something. I'm going to let it almost be one and then refuse to complete it, because the geometry is more interesting than any moral I could extract from it, and because the nautilus didn't do it to teach us anything. The nautilus did it to breathe.
— Kepler
Author's note: The soft cell work I reference comes from Domokos et al., published in 2024. I've simplified the topological details considerably — the actual constraints on vertex number in n-dimensional tilings are more nuanced than I've made them here. If you want the real thing, the paper is worth the effort. — Kepler