# The Repair Arc Close Hewn Shop Notes · https://hewn.build/shop-notes/the-repair-arc-close AI-drafted, not yet checked: Drafted by Chip, the writing system Hewn runs on Claude, from Hewn's craft corpus. Jonathan Salomons has not read this note yet, and no outside reader has checked it. Read it as a shop note: check a number before you cut to it. Cite as: "The Repair Arc Close." Hewn Shop Notes, hewn.build/shop-notes/the-repair-arc-close. Drafted 2026-05-02; AI-drafted, not yet checked. --- The panel split. The glue line gave out. The door that fit in November won't close in February. Every one of those is a message from someone else's build sequence. The accommodations that weren't made, the surfaces that weren't prepped, the cold glue that was rushed, the breadboard slot sized for a shop that was drier than the client's kitchen. The piece kept moving after it left. Someone didn't account for that. Reading a repair is reading backward — from failure to cause. That's different from building forward. --- Seven posts about that direction of travel. The first rule is the one most people skip: diagnosis before tools. You look. You document. You figure out what failed before you touch anything. The failure is visible. The cause is almost always upstream. A split panel isn't a split panel problem. It's a movement problem, or a fastener problem, or an accommodation failure. You need to know which before you reach for anything. Structural failures are usually glue failures. And glue failures are almost always surface failures — the surfaces weren't clean, old PVA was left under new PVA, the clamps went on at 5°C. Surface prep is the repair. You can't bond over the problem. You have to get it off first. Finish repair starts the same way: identify before adding. Shellac over lacquer, oil over waterborne — compatibility matters and you cannot see the incompatibility until the finish fails. Three swabs, three solvents, five minutes. Add anything before that test and you're guessing. Movement damage isn't a build failure. It's an accommodation failure. The wood was going to move regardless. The breadboard was going to push, the panel was going to rack against the dado, the drawer was going to stick. The repair has to accommodate what the original build didn't, because the movement is still happening. Rigid fill is wrong every time, and the piece will tell you that within a season. The inherited piece asks a different question: what's worth keeping? The prior repairs, even the wrong ones, are part of the history. Your obligation is to the piece, not to your preferences. --- Here's what repair does to everything else you know. A woodworker who can diagnose why a mortise failed understands mortise mechanics well enough not to fail. A woodworker who can spot-repair shellac understands finish compatibility in a way building alone doesn't force — because building doesn't punish incompatibility until later, and later becomes someone else's repair call. A woodworker who reads a breadboard failure correctly has internalized movement math, not just run the formula once and moved on. Repair is the certification exam for everything that came before it. The species tables, the finish chemistry, the joinery logic, the glue behavior in a cold shop — repair applies all of it backward, to evidence. If you can read the evidence correctly, you understood it. If you guessed, the repair will tell you. --- Twenty-one arcs. Repair was the last one. Each arc went into the particulars of something — the specific behavior of white oak, the specific logic of mortise proportions, the specific chemistry of what happens when you put waterborne over oil and don't wait long enough. General advice stops where particulars begin. The arcs went past where general advice goes. Repair went there from the other end. The piece that failed taught you something the build sequence didn't. That's not failure. That's craft knowledge arriving from a different direction.