Chapter 1

By the third session, the fill compound had reached the viscosity window she wanted: thick enough to bridge the largest loss without slump, thin enough to feather into the surrounding tin glaze before the epoxy's open time collapsed. Sable Yoon checked the mix with the micro-spatula, watched the ridge settle by less than a millimeter, and began.

The delftware plate lay faceup in a foam cradle under neutral light, three losses mapped in pencil on the acetate overlay she had taped above it the previous week. Seventeenth century, Dutch, tin-glazed earthenware, broad rim, blue floral border, central pastoral scene abraded by three hundred years of handling and one bad drop sometime in the last decade. The glaze around the largest loss had started to powder at the edges, which meant the consolidation step had not been optional. She had done that in session one. Session two had been cleaning, stain reduction, and substrate preparation. Session three was fill.

She loaded the spatula and placed the first application into the largest loss with a shallow wrist angle to avoid trapping air against the undercut edge. The compound took. Good. She spread it outward in controlled passes, not trying to finish the surface yet, only to establish coverage and support. The trick with bulking agents was not to trust them. Too little and the fill shrank on cure; too much and it went chalky, lost coherence, chipped under finishing. This batch was balanced. She knew because she had mixed variants of the same formulation for six years and stopped making notes after the first forty.

The room's humidity held at forty-eight percent. Temperature, twenty-one point two Celsius. Stable enough.

She shifted to the second loss, smaller, shallower, trickier because it interrupted the blue line of the border. Color integration later would matter more there than structural fill. For now, geometry. She worked under magnification, feeling the minute resistance change as the compound met ceramic edge instead of void. The fill had to end exactly where the loss ended. Overbuild now, finish later, but not so much that she manufactured unnecessary work for herself in session four.

When she paused, it was only to rotate the plate ten degrees clockwise and change tools. Tool changes were information. A silicone shaper for broad leveling, sable brush dampened in solvent for edge cleanup, sharpened bamboo skewer for extracting one particle of airborne lint that had no business existing inside her controlled workspace. She removed it, checked the surface again, continued.

The pleasure in this stage was not aesthetic. It was sequential. Each step constrained the next; each correct decision narrowed the possible range of error. People who thought restoration was patience misunderstood the mechanism. Patience was passive. This was active control across time.

She finished the first pass in twenty-six minutes. Longer than average by three, accounted for by the border interruption and the fragility of the glaze. She waited four minutes, tested tack, and made the second pass to build the surface just above final plane. Under raking light the filled areas stood out as dull islands against the glaze, temporary and acceptable. Under ultraviolet there was no unexpected fluorescence, which meant no contamination event had occurred between mix and application. She set the plate down and let the room settle around it.

Then she checked everything again.

Fill hardness: pending full cure, surface skin developing on schedule. Edge adhesion: clean. Color base: within expected range before in-painting. Texture profile: rough, intentionally overbuilt, no drag marks visible at low magnification. Cabinet schedule: transfer in seven minutes.

She cleaned her tools in the same order she always did—solvent soak, first wipe, fine brush wash, second wipe, dry cloth, tray alignment—because inconsistent cleanup produced variables that appeared later, falsely, as failures of material behavior. Most mistakes in restoration were not dramatic. They were procedural debt.

The curing cabinet stood against the back wall beside the shelving unit that held reference books, boxed supplies, and one object she did not look at.

She carried the plate there with both hands, one under the foam cradle, one steadying the rim. Cabinet door open. Shelf two, center position, where the airflow was most even. Plate in. Door closed. Latch checked.

To the left, behind a row of conservation binders and a handbook on Asian lacquer systems, the teapot remained where it had remained for eleven years. Stoneware. Four visible repair episodes. Cataloged once. Object present, location stable, no action required.

She turned away and crossed back to the bench.

Her treatment record waited open on the laptop. She entered the session notes while the details were still fresh enough to be exact and not yet abstracted into memory.

Session 3 of 5. Loss fill application using custom bulked epoxy matrix. Three losses addressed. No contamination observed. Environmental conditions stable. Surface build intentionally elevated above final plane for later reduction. Next session: mechanical finishing, texture integration, first color compensation under daylight-balanced illumination.

She saved the record, backed up the file to the external drive, then wrote the session time on the paper log she kept in parallel because digital systems failed in ways paper usually did not. Her handwriting was narrow, upright, nearly architectural. She had once been told it looked impersonal. That had not seemed to her a criticism.

The phone on the far corner of the bench flashed once, then again. Voicemail.

She let it finish, sanitized her hands, and only then pressed play.

A woman's voice, older, controlled, not tentative but not familiar either. “Hello, Ms. Yoon. My name is Helen Tran. I was given your number by the museum on Mercer. I have a bowl that needs restoration. Ceramic, I think stoneware. It's been repaired before more than once, and it broke again last week. I don't know if it's valuable, but it mattered to my aunt, and I was told you're the person to ask. If you're taking commissions, please call me back.”

Sable replayed the message once.

Ceramic bowl. Probable stoneware, though clients misidentified earthenware as stoneware often enough to make first-pass assumptions inefficient. Multiple previous repairs. Recent re-break. Client uncertain about market value, which usually meant sentimental commission rather than collector intervention. “Mattered to my aunt” indicated inherited object, likely domestic rather than decorative. If the prior repairs were amateur, fifteen hours minimum. If one of them involved incompatible adhesives, more.

She opened the schedule file and looked at the next three weeks. The delftware plate would finish on Thursday. Two small porcelain consolidations on Friday. A wooden-and-ceramic mixed-media figure waiting on solvent tests. There was room if the bowl proved structurally interesting enough to justify compression elsewhere.

She typed: NEW INTAKE — H. TRAN — CERAMIC BOWL / MULTIPLE PRIOR REPAIRS / RECENT FAILURE. Tentative diagnostic slot: Tuesday, 9:30.

Then she called back.

Helen answered on the second ring. Efficient. Good.

“This is Sable Yoon returning your call.”

A brief pause, then relief moderated into practicality. “Thank you. I wasn't sure when you would—”

“I have a diagnostic opening Tuesday at nine-thirty. Bring the object unwrapped if possible, or wrapped in layers that can be removed without tape contact. Do not attempt further stabilization.”

“All right.”

“Has it shed fragments?”

“A few small ones. I put them in an envelope.”

“Bring those separately.”

“I will.”

There was a silence that suggested Helen might add context. People often did. Family stories, accidents, declarations of importance offered in the hope that importance altered fracture mechanics. Sometimes it altered treatment priority. It did not alter fracture mechanics.

Helen said, “My aunt would have wanted it fixed.”

Sable looked at the bench surface while she listened. Maple top, sealed. Two scratches near the lamp base from an apprentice period mistake she had never bothered to sand out because the damage was functionally irrelevant.

“I'll assess what is possible when I see it,” she said.

“That sounds fair.”

They confirmed the address. Helen thanked her. The call ended.

Sable set the phone down in its marked place and stood still for two seconds longer than necessary, already building the probable sequence of Tuesday's intake. Visual survey first. Material identification by body and glaze characteristics. Mapping of all prior interventions before any judgment about reversibility. If the bowl had indeed been repaired multiple times, provenance of the repairs might matter for treatment choice. Staples behaved differently from early PVA; cyanoacrylate contamination changed cleaning options; old epoxy created its own weather system.

Interesting enough, then.

She turned back to the bench and aligned the micro-spatulas by size. The smallest one sat a fraction of a degree off parallel to the tray edge. She corrected it.

Outside, somewhere above the basement windows, a bus hissed to a stop at the corner. If it was weekday timing, that would be the 7:14 running two minutes late. Rain later, probably; the pressure had shifted just enough that the building's old pipes were carrying sound differently. She noted the change without moving toward the weather app. If the humidity climbed overnight, she would adjust the cabinet settings in the morning.

The workshop settled back into order around her. Cabinet humming. Laptop sleeping. Tools clean. New intake scheduled.

Behind the row of reference books, the teapot remained on the shelf in the dark.

Sable switched off the bench lamp, and for an instant the reflection in the curing cabinet glass held the room in reverse: the ordered bench, the shelves, the narrow aisle, the place where the teapot sat just outside the brightest field of light. Then the angle changed, and it was only glass again.