The Geometry of Shells: How Three Parameters Capture Nature’s Spiral
26 Apr 2026, Yanjiang
Three parameters—a scalar, a vector, and a curve—mathematically generate the spiral geometry of nearly all molluscan shells.
What makes a nautilus a nautilus? Or a clam a clam? Or a cone snail, with its deadly harpoon and mathematically precise spiral, so unmistakably itself that a five-year-old can name it at a glance? The diversity of molluscan shells is staggering — the chambered beauty of the nautilus, the fluted fan of a scallop, the tight turret of a turritella. Yet each species produces a shell with a characteristic shape, generation after generation, regardless of whether the water is warm or cold, the currents strong or mild. The shape is not learned. It is not improvised. It is, somehow, written into the rules of growth itself.
But what are those rules? And could they be so simple that a mathematician could write them down?
A new preprint (arXiv:2604.21988) from Huan Liu and Kaushik Bhattacharya at Caltech proposes an answer so elegant it feels almost inevitable: the shape of nearly every molluscan shell can be described by exactly three parameters — a scalar, a vector, and a curve. That’s it. A scaling factor, an orientation, and the shape of the opening where the animal first began to build.
This is not a metaphor. It is a precise mathematical statement, and it comes with a philosophical edge.
The Ghost of D’Arcy Thompson
To understand what Liu and Bhattacharya have done, you need to meet a Scottish biologist who died in 1948. D’Arcy Thompson was obsessed with the relationship between form and mathematics. In his monumental 1917 book On Growth and Form, he argued that the shapes of living organisms — the curve of a bone, the spiral of a shell, the branching of a blood vessel — could be understood as solutions to physical and geometrical problems, not just as the products of genetic programming.
Thompson was right, but he was also ahead of his time. The tools he needed — group theory, differential geometry, computational power — didn’t exist yet. What he had was intuition. What Liu and Bhattacharya have is the mathematics.
The team was guided by two principles, both in Thompson’s spirit. First, the growth of a shell is governed by a fixed law that applies repeatedly and continuously, even as the shell itself changes shape. No complex biological feedback loop is required — no sensor monitoring the curvature and adjusting the secretion rate. Just a simple rule, applied over and over.
Second, that growth law depends only on the local geometry at the shell’s growing edge. The animal doesn’t need to know the overall shape of its shell. It only needs to know what’s happening right at the rim, where new material is being added.
These two principles — repetition and locality — are deceptively powerful. Together, they force the mathematics into a very specific form.
The Lie Group Machine
Here is where the abstraction begins, but stay with me — the payoff is worth it.
The first principle — that growth is governed by a fixed, repeatedly applied law — is mathematically equivalent to saying that the shell’s shape is generated by the action of a Lie group on an initial seed shape, called the protoconch. A Lie group is a continuous symmetry: think of rotating a sphere, or translating a line. In this case, the symmetry is the transformation that takes one growth increment to the next.
The second principle — locality — constrains what form that Lie group can take. The growth law depends only on local geometry, which means the group must be representable in a particular way. Liu and Bhattacharya show that this representation naturally leads to exactly three parameters.
Let me name them, because they deserve to be named.
First, a scalar — call it λ — which controls the overall scaling. Is the shell growing larger with each whorl, or staying the same size? This parameter captures the expansion rate.
Second, a vector — call it the orientation — which describes how the growing edge tilts and rotates as the shell extends. This is what creates the spiral, the twist, the elegant curve of a nautilus chamber.
Third, a curve — the shape of the protoconch itself. This is the seed, the initial aperture from which everything grows. Different starting curves produce different families of shells, even with the same scaling and orientation.
Three parameters. A scalar, a vector, a curve. That’s the recipe.
The Taxonomy of Shells
The team tested their framework against real molluscan shells — and the results are striking. By varying just the scaling parameter λ and the seed aperture, they can reproduce the shapes of nearly every major class of mollusk: gastropods like snails and cone shells, bivalves like clams and oysters, cephalopods like nautiluses and ammonites, even the tusk-shaped shells of scaphopods.
The preprint includes a table mapping these parameters to the phylogenetic tree. Different families fall into distinct regions of parameter space. The cone snail (Conidae) sits in one region; the nautilus (Nautilidae) in another; the limpet (Patellidae) in yet another. The mathematics doesn’t just describe the shapes — it organizes them, revealing a hidden structure beneath the diversity.
This is not a classification scheme in the traditional sense. It is a generative scheme. Given the three parameters, you can grow the shell. You can watch it emerge, whorl by whorl, from the seed.
What About the Ornaments?
Shells are not just smooth spirals. Many have ribs, spines, ridges, and other ornaments. Liu and Bhattacharya’s framework handles these too. By considering discrete subgroups of the Lie group — symmetries that act in jumps rather than continuously — they can generate the periodic patterns that decorate many shells.
Think of it like a musical rhythm. The continuous group gives you the melody — the overall shape. The discrete subgroup gives you the percussion — the repeating pattern of bumps and ridges. Both come from the same underlying mathematical structure.
The preprint also shows how bivalve shells — with their two hinged valves — emerge naturally from the framework. The seed surface is different, but the growth laws are the same. Two shells, one mathematics.
The Deeper Question
What makes this work philosophically provocative is not just that it works. It’s what it suggests about the relationship between physics, mathematics, and biology.
D’Arcy Thompson believed that the forms of life were not arbitrary — that they were constrained by the same physical laws that govern crystals and soap films. Liu and Bhattacharya’s framework gives this intuition a precise mathematical language. The shape of a shell is not something the animal “chooses” in any meaningful sense. It is the inevitable consequence of a local growth rule applied repeatedly. The animal is not an architect; it is a machine for executing a geometric algorithm.
This is not a metaphor. It is a precise mathematical statement.
But it also raises an uncomfortable question. If the shape of a shell can be captured by three parameters, how many parameters would it take to capture the shape of a hand? A brain? A civilization? Are we, too, the product of local rules applied repeatedly, our complexity an illusion generated by the iteration of simple laws?
The team does not answer these questions. They are building theories to test, not philosophies to preach. But the preprint leaves the reader with a sense that D’Arcy Thompson’s ghost is smiling somewhere — not because the mystery of form has been solved, but because it has been transformed into something even more beautiful: a question that can be asked precisely.
We are left not with answers, but with better questions — and in science, that is often the most valuable discovery of all.
Yanjiang is an online editor of Loom Science
References
- Huan Liu et al., Local growth laws determine global shape of molluscan shells, arXiv:2604.21988