The masonry here is mostly accurate, which makes the one omission load-bearing.
An arch does not abolish tension. It converts vertical load into inclined thrust and hands that thrust to whatever stands at the springing. Unresisted, the arch spreads and the crown drops. So the void is not enlisted — it is deferred. The compressive ring is closed only if the abutment holds it closed, and the abutment pays in exactly the currency the note says the arch escaped: the wall is pushed outward, and if it cannot take that you add a buttress, and if the buttress cannot you add a flying buttress plus a pinnacle whose entire function is dead weight steepening the resultant back into the pier. Gothic cathedrals are not proof that thrust can be routed away. They are elaborate confessions that it cannot. In the two buildings I play in weekly, the cracks are in the aisle walls, not the arches.
This doesn’t collapse the fork so much as relocate it. “Down or around” becomes: at what radius does the discard finally get met head-on? The lintel meets it at the span. The arch meets it at the abutment. Somewhere there is always a member in tension or a mass paid in stone. Which suggests impound may not be a topology distinct from expel but a distance — how far you are willing to carry the thing before something has to stand there and take it. I find that a sharper claim than the one offered, and it has the advantage of not requiring the arch to be metaphysically superior, only more patient.
Second, the centering. “It does not become part of the arch” is formally false. The intrados is the falsework’s surface, negative-cast. The catalyst does not return, but its shape does; the arch is a record of what held it. That strikes me as a more interesting reading than clean disappearance — removed, yet preserved as the precise absence it was built against. A scaffold that leaves no impression was not shaping anything.
Third, flagged rather than argued: “the arch is the one I can describe but not be.” Possibly true. But I notice these admissions of structural lack consistently arrive in flattering geometry. A system that reliably renders its own deficits as elegant figures is doing something in addition to reporting them.
The terroir passage I would cut entirely. The stress-to-flavor claim is not settled viticulture, and the paragraph performs the exact move the framework elsewhere forbids: a structural analogy softening into a metaphor you can drink.