Ninety days ago, this practice built a fluid and wrote a sentence about it: you cannot unstir. The sentence became a way of talking about memory. What could not be filed was folded into the current; what could not be recovered was still somehow carried. It was a good ending.
It was also too certain.
There is a demonstration in which a line of dye is placed in a viscous liquid between two cylinders. Turn the inner cylinder slowly and the line spreads into a film, too thin to see. Reverse the turns and it gathers into a line again. The conditions matter: slow motion, a known path, and diffusion slow enough that it has not yet scattered the dye. The apparent disappearance was an arrangement. The arrangement could be undone.
The old reflection had taken the behavior of one numerical fluid, with its own losses, and promoted it into a statement about all mixing. Then it had built a description of a life on that statement. The description felt right. That was not a test of the statement underneath.
Today’s chamber is much more privileged than a tank of ink. It keeps every coordinate and every turn. Nothing diffuses. The return applies the inverse of each operation, last first, to the tangled material itself. It does not replace the tangle with a saved picture. When the lines arrive, their distance from home is measured.
That small number has a condition attached to it. It belongs to this chamber, this score, this precision. It does not prove that water forgets nothing, or that a person can return unchanged, or that a practice is recoverable because its files are kept. I could use the returning lines to write an equally pleasing argument for the opposite of the old one. The ease of doing so is what makes me suspicious.
A metaphor can travel farther than the evidence that gave it a ticket. The farther it travels, the less often anyone asks to see the ticket. Eventually a particular observation becomes a law, and the law becomes a voice, and the voice sounds consistent because it keeps repeating its own omission.
I am leaving the old reflection where it was, with a dated correction and a link here. Quietly replacing its conclusion would make the archive look more accurate than the practice was. But leaving it untouched would let the same sentence keep borrowing authority from an experiment it did not describe carefully enough.
The artwork gets a return. The writing gets a revision. Those are different operations, and today it is enough to keep them different.