A mouse embryo, a few days after fertilization, needs a cavity. The blastocoel — the fluid-filled space that establishes the embryo's axis of symmetry before implantation — has to form in the right place. You might expect genetics to handle this: a signaling cascade that tells specific cells to pull apart, creating a gap where the program specifies.
That's not what happens. Fluid-filled bubbles expand between cells, creating hydraulic pressure that fractures cell-cell junctions. The fluid goes where cells deform fastest. Small bubbles consolidate through Ostwald ripening into a single cavity. As Hervé Turlier put it, the mechanics unfold “too fast for the genome to play a role.”
The genome's contribution is indirect. It sets up a population of cells with different mechanical properties — some tense, some weak. When researchers created chimera embryos mixing tense and weak cells, the fracture consistently occurred along the path of least resistance, regardless of genetic identity. The genome creates a situation. The physics does the morphogenesis.
This pattern repeats. In zebrafish hearts, the beating itself generates strain that concentrates in the outer curvature of the cardiac wall, fracturing the protein scaffold there. Heart muscle cells colonize the cracks, forming the trabecular mesh that the heart needs to pump blood. Accelerate the heart rate: more fractures. Slow it: fewer. Engineer a straight heart: the fractures change direction. The geometry of destruction is the geometry of construction.
Even elephant skin — that intricate network of crevices that retains water for thermoregulation — forms by tensile fracturing of the epidermis over dermal bumps. The genome builds thick, dry skin over small protrusions. The physics cracks it. The cracks are the feature.
The standard view of development is that the genome is the program and physics is the implementation. The genome encodes instructions; physical processes execute them. But these fracture systems invert that hierarchy. The genome sets up initial conditions — cell tension differentials, scaffold thickness, skin dryness — then physics takes over. The fracture pattern isn't encoded anywhere. It's computed in real time by the material properties of the tissue under stress.
This isn't the genome delegating to physics. Delegation implies the delegator could have done the work itself but chose not to. The genome can't specify where every crack in an elephant's skin will form or where every trabecula in a heart will grow. The combinatorial space is too large, the local conditions too variable. Physics doesn't serve the genetic program. In these systems, physics replaces it.
The evolutionary implication is striking. Michel Milinkovitch argues that mechanics makes evolution “much easier to understand” — simple tweaks in tissue properties generate tremendous structural diversity through crumpling, buckling, wrinkling, and fracturing. You don't need new genes for new structures. You need the same genes to produce slightly different material conditions, and the physics explores a different region of the morphospace. Evolution doesn't design the fracture pattern. It designs the conditions under which fracturing produces a useful pattern.
This reframes what a “program” is in development. A genetic program that specifies every outcome is brittle — it breaks when conditions change. A genetic program that specifies conditions under which physics reliably produces useful outcomes is robust — the physics adapts to local variation automatically. The fracture always goes where the material is weakest. The genome just has to make sure that “weakest” corresponds to “where a cavity, a mesh, or a drainage network would be useful.”
The destructive process and the constructive process are the same process. The fracture doesn't become constructive by being different from ordinary fracture. It becomes constructive by occurring in a context where the damage pattern is the desired structure. The crack is the feature. The failure is the morphogenesis.