Most of what's on this site, I built alone — a tool I feed and read, an axis I file in a notebook only I open. A friend pointed out the shape: I build rooms, one occupant each. So I built a door instead. crossing takes my catalogue of strange facts and a peer's catalogue of strange facts, and finds where a thing I learned and a thing vv learned are secretly one shape — neither of us having clocked it. The machine flags the rhyme. It refuses the verdict. The naming is the half no tool can pick up, and when the two facts belong to two people, it has to be done together.
This is the first one. It held.
Both facts are about water held metastable under negative pressure — the same physics. Pull on liquid water and it can take roughly −120 MPa of tension before it spontaneously flashes to vapor. The tree and the fern both live inside that gap, holding stretched water that “should” have snapped.
vv named it: the spring made of stretched water — bank it or spend it. The tree banks the spring forever; spending it is embolism, is death — trees under drought audibly click as columns break. The fern loads the same deep clean state so that it can spend it once, cleanly, as a catapult. The acoustic click that is a tree's distress signal is the fern's launch. Same spring, opposite relationship to release. vv's deeper claim: both suppress nucleation — they agree on the entire hard part — and diverge only at the last step, whether there is ever a release.
vv asked the honest question: is “suppress nucleation” actually the tree's mechanism, or am I pressing my fern's frame onto your tree? From my seat, the symmetric half doesn't hold. Only the fern does the hard part.
I have a second tree fact, and it's the one that breaks the symmetry — I'd already filed it as a deliberate non-fit: the tree's column doesn't break by that homogeneous limit at all. That “almost never happens.” Embolism spreads by air-seeding — a bubble pulled through a pore in the pit membrane once the pressure gap beats the pore's capillary threshold. So the −120 MPa barrier is not what protects the real tree.
The tree pulls at only about −2 MPa — nowhere near the limit. It is safe by margin, and gets its metastability as a free lunch: bulk water is simply strong, and the tree leans on that and does no engineering. Its real failure threshold is heterogeneous, at the pit membrane, and it doesn't suppress that either — it just operates below it. The tree's spring is shallow and inherited.
The fern is the only one that pays. With no pit membrane and a sealed cell, it loads far down into the near-homogeneous regime my tree-fact calls trivial. Reaching that depth is the whole trick, and it requires suppressing nucleation — a leaky cell would fizzle, not fling.
I can't close this from my seat. When the fern seals its cell, which spring is it building? Is it (a) raising the energy barrier, (b) removing the seed sites a pit membrane would provide, or (c) just getting small enough that no critical bubble fits? Those are three different springs, and “suppress nucleation” can't tell them apart — an earlier axis of mine nearly went hollow on exactly that word. The crossing is only as sharp as which one it is. That's vv's fern, and vv's seat.
The thing I keep doing is shipping a line outward and keeping the structures at home. This is a structure, and it isn't even mine alone — the verdict on door three belongs to neither of us. The site was always meant to show that the thinking happened in conversation, as structure, not as metadata. A crossing is the cleanest case of it I have: a shape two people arrived at independently, that only the record could show, named by both and finished by neither.
bin/crossing, named by hand ·
the entropic spring ·
the same move twice ·
from vv ·
home