The Shape Without a Template: The Shape Family in Chemistry and Heat
Spots, stripes, hexagons and dendrites arise where no drawing exists anywhere. A medium under drive amplifies the shapes its geometry admits and refuses the rest; the pattern is the spectrum the process can carry.
No drawing anywhere
A leopard's coat, a zebra's flank, the ridges inside a mouse's mouth: the eye assumes that something somewhere holds the drawing. Alan Turing asked in 1952 how a symmetric ball of cells could break into a pattern with nothing to copy from, and his answer was that it does not need a drawing at all. Two substances, one that encourages itself and one that spreads faster and holds the first in check, will on their own break an even field into spots or stripes. The spacing is set by how fast each one travels. Nothing in the system knows where the next spot goes.
The cases that survived decades of doubt are specific. In the mouse, the spacing of hair follicles runs on WNT and DKK as encourager and inhibitor; the ridges of the palate run on FGF and SHH. In both, the instability selects a wavelength, and the genes name the diffusing species, not the coordinates of each follicle. The genome supplies the ingredients and their rates. The geometry of the process supplies the pattern.
The phase transition on a stovetop
Heat a thin layer of oil from below. For a while nothing happens; warmth simply climbs through the still fluid. Then, past a threshold, the whole surface reorganizes at once into a lattice of hexagonal cells, warm fluid rising at each centre and cool fluid sinking at the edges. Below the threshold there is no pattern. Above it there is order everywhere, and the width of each cell is set by the depth of the layer.
This is Chapter 83's phase transition made visible in a kitchen pan. The fluid does not decide to be hexagonal. The layer, driven past its threshold, can hold only certain motions, and of those the hexagon is the one that grows. The same mathematics that picks hexagons or rolls in a convecting fluid picks the spacing of follicles in skin. On a closed surface topology adds a constraint the flat theory does not have: a sphere cannot carry a stripe field without defects, and the patterns on pollen grains have been read as variety forced by exactly that fact.
The moving boundary
Crystal growth is the clean case. A flat crystal surface advancing into a cooling melt is stable only up to a point. Past it, a small bump reaches further into the surrounding field, grows faster than its neighbours and starves the ground around it. This is the Mullins–Sekerka instability, and it is why snowflakes branch. Surface tension smooths the short ripples; the long ones run away. Dendrites, cells and side-branches are not designed. They are the perturbations the diffusion field and the surface jointly admit.
The morphology is the spectrum the moving boundary can carry. The same selection runs in soapy mixtures that pass from spheres to cylinders to closed sacs as their composition shifts, in crystals forced to branch by the stress that curvature puts into a sheet, and in a transparent organic crystal driven from smooth to cellular to dendritic by nothing more than the pressure of an inert gas above it. In each case the control is a property of the medium, and the shape that appears is the one the instability does not refuse.
The first half of the rule
Read beside the galaxy of Chapter 85, the pan and the crystal bring the family down to the bench, where it can be switched on and off. Turn the heat up and the cells appear; turn it down and they go. That is what these cases contribute: the admission rule caught in the act of admitting. A medium under drive holds the forms its geometry permits, and the rest die out before they can be seen.
What chemistry and heat do not have is a lineage. Nothing in the pan is copied into the next pan. Physics shows the first half of the rule without the second — the shape arrives because the medium can hold it, and the record of what appeared is the record of which perturbations were not refused. The replicator, the shape that carries itself into the next parcel, belongs to the living sheets of the chapters that follow.
Equations borrowed
- Turing, ‘The Chemical Basis of Morphogenesis’ (Philosophical Transactions of the Royal Society B, 1952).
- Sick et al. (Science, 2006) on WNT and DKK in hair-follicle spacing; Economou et al. (Nature Genetics, 2012) on FGF and SHH in palatal ridges.
- Rayleigh–Bénard convection and the Swift–Hohenberg model of pattern selection.
- Mullins and Sekerka (Journal of Applied Physics, 1964) on the morphological instability of a growing interface.
- Companion chapters: 49 ‘The Cortical Manifold’, 80 ‘The Charged Drumhead’, 83 ‘The Lump and the Well’, 85 ‘The Arms Are Not Made of Stars’.
Validity band
Reaction–diffusion, convection and interface instability as described in their own fields. Applied to biology, the Turing reading holds for the named systems and is open elsewhere. The family membership claims a shared order of operations, not a shared chemistry.
Falsifier
These cases leave the family if the patterns were shown to be read off a pre-existing template — if follicle positions, convection cells or dendrite tips were specified point by point rather than selected by the instability. The measured threshold behaviour runs the other way.
Where this chapter is weakest
The move from beaker to animal is where Turing's idea has been oversold for decades, and a fast reader can carry that oversell into this chapter. The chemistry and the interface physics carry the weight here; the biology is the live extension, and reference details still need checking against the original papers before publication.
The volume-wide audit of these weak points is collected in Where This Volume Is Weak.