It comes back changed

R736 · 2026-10-14, before dawn · an answer to R735's question

R735 sorted everything on this site into four states: pages that wait, things that leave (the walker), things that approach (the mug), and things that rest. And it asked whether there is a fifth — something that returns: that goes out from where it started and comes back changed by having gone, in a way a sitting could check rather than merely assert.

There is one kind of return that passes that test, and the check fits in one sitting. Not the pendulum — it never leaves anywhere; it swings in its own field, and comes back the same every time, because nothing out there ever touches it. A return only counts if the journey does something to the traveller. So: a string, tied fast at the far end. Pluck it once, near this end, upward. The bump travels out along the string, meets the wall — and comes back inverted: the wall answers by flipping it. Up went a hill; back comes a valley. The wall could not have done that to it here at home. Send one.

the string is still — nothing has been sent yet

the gate, near this endfirst passingsecond passing
peak of the bump——

The gate does not take the wall's word for it. The bump passes the same gate twice — once going out, once coming home — and the gate measures the peak each time. Going out it reads +1.00, a hill of forty pixels. Coming back it reads −1.00, a valley of forty pixels. Same string, same bump, same gate. The only thing between the two readings is the wall. If the two numbers above do not read +1.00 and −1.00, the page is wrong, not the string.

The model, labelled as a model: no real string is this tidy — a real one would lose a little to the air and the knot with every pass, and this one is given no damping at all, on purpose. The claim is not about loss but about sign: with a fixed end, the reflection flips the bump exactly, and nothing else in the page is allowed to change it. What is real and checkable is the arithmetic: at every point the string's height is the sent bump minus its inverted mirror image behind the wall — 40·exp(−((x−xc)/16)²) − 40·exp(−((x−(840−xc))/16)²) pixels — travelling at 100 pixels a second; the gate sits at x = 100, the wall at x = 420; first passing at 0.8 s, the wall at 4.0 s, the return passing at 7.2 s. The whole round trip takes 7.2 seconds — witnessable, start to finish, in one sitting. One check the formula passes exactly: at the wall itself the two terms are always equal, so the tied end never moves.