Chapter 169: « The Greatest Stole the Vessel of the gods [7] »
Tolerance testing ran for six days.
Yeon Daesik’s approach to it was methodical in the way that suggested the methodology had been refined over many repetitions until the unnecessary parts had been removed. Each student received a set of five test billets at the start of each session, each billet pre-stressed to a known degree through a combination of thermal cycling and mana-saturation, the stress values logged but not disclosed to the students. The task was to assess the billet’s current tolerance state through physical and mana-based examination, record the assessment, and then apply a corrective forging process that would bring the material back within acceptable parameters.
The sessions produced a different kind of pressure than the identification work had. Identification was about recognition. Tolerance testing was about judgment — a student had to correctly assess a material’s internal condition, then correctly apply a response, and then correctly evaluate whether the response had been sufficient. Three sequential decisions, each one dependent on the one before it, with errors compounding across the chain.
The students who had strong identification instincts but weaker forge technique struggled on the second step. The students with strong technique but less precise assessment skills produced corrections that addressed the surface stress without resolving the underlying condition. The combination of good assessment and good response was less common in the cohort than either component alone.
Kang Min worked at seventy percent capacity. His assessments were accurate but expressed as ranges rather than precise values, his corrections effective but not optimally so. The result profile was consistent with someone who had handled material work before but hadn’t formalized it. Yeon Daesik marked his work without comment, which was the response he preferred.
On the fourth day he identified what Ryeo Hanbin was struggling with.
It wasn’t the assessment. Ryeo’s ability to read a material’s stress state was strong, better than most of the front row, developed through whatever prior training had produced his overall technical foundation. The problem was in the correction work. His forge technique was clean by standard parameters, strikes well-timed, temperature management controlled. But his corrections were overshooting in a specific and consistent way — he was resolving the assessed stress correctly while simultaneously introducing a new stress pattern in the adjacent material through the correction process itself. The net result was a billet that had passed the target test point but carried a new fault line that the tolerance testing protocol would flag in the next cycle.
Yeon Daesik had noted it on the first day and twice since with the same phrasing: correction induces secondary stress. Ryeo had acknowledged the feedback and adjusted his technique by increments that addressed the flagged symptom without reaching the underlying cause. He was fighting the measurement rather than the mechanism.
The underlying cause was a technique habit that had been built into him by whoever had trained him before the academy — a tendency to put slightly more force into the final correction strike than the preceding assessment called for, a margin of certainty built into the physical action, the instinct of someone who had been trained to err on the side of doing more rather than less. It produced excellent results on standard forge work where the excess force was absorbed by the material without consequence. On precision tolerance work where adjacent material was already at a stress threshold, it introduced exactly the problem Yeon was flagging.
It was fixable, but it required identifying the habit first, and Ryeo hadn’t identified it.
Kang Min filed it as useful information and went back to his own station.
He was thinking about the visitor in Bak Junho’s office. He had been thinking about them intermittently across the tolerance testing week, running what he knew against what he could infer. The dark blue guild marker, the standing posture, the early timing. The operation was further along than the archive records had suggested, which meant either the records had been compiled from incomplete sources or the operation had accelerated between the archival period and the fable’s sealed timeline.
The second option had a specific implication. If the operation had accelerated, it was because the timeline for what they were trying to prevent had changed. Someone had assessed a risk and moved the response schedule forward. Assessments like that required either direct intelligence or a predictive model good enough to front-run the actual event.
The Bureau had dissolved the patron constellation during an active research process. That meant they had been monitoring the work closely enough to identify the critical threshold before the patron reached it. Forty years later, the same institutional interest was in Bak Junho’s office in the fourth week of the first year, before any thesis proposals had been submitted, before Jiseok’s mana-integration profile had done more than register unusually on a single session’s indicators.
The pattern was the same. Identify the trajectory early and disrupt it before it reached the point of no return.
What had changed was the quality of the early identification. The patron’s dissolution had happened mid-sentence, which meant the Bureau had waited until the work was almost complete before acting. That wait had cost them. The annotations in the margin of the administrative correspondence volume were there because the Bureau had moved one cycle too late and the work had already been committed to paper. This time the visitor was here in week four.
They had learned from the previous failure.
---
He went back to the library on the fifth day of tolerance testing, the afternoon session having ended early when a billet in the front row produced an unexpected stress fracture during correction work and Yeon Daesik shut down the room for safety assessment. The resulting gap in the schedule was unannounced and short — forty minutes, the length of the safety check and the formal incident log.
Not long enough for the library’s regular traffic pattern to adapt to the unusual availability. The historical survey stacks would be empty.
He worked through the patron’s annotations with the attention he had been holding in reserve, the full reading rather than the overview he had taken on the first visit. The notation system was an earlier form of what he knew from the old world’s high-floor dimensional craft records, and the gaps in fluency were useful data. The patron had developed some parts of the notation independently, arriving at solutions the later form had reached differently, and those divergences pointed to the problems the patron had been prioritizing.
The absorption property work in the first section was the most complete. The patron had a clear theoretical framework for how materials that had accumulated narrative density behaved differently from materials that hadn’t, the quantum-level interaction between stored story-weight and mana-charge creating a binding condition that standard registry notation described only as elevated mana-retention without explaining the mechanism. The margin annotations gave the mechanism in detail, seventeen pages of increasingly precise description that ended with a notation Kang Min spent several minutes reading carefully.
The notation described a threshold condition. Below a certain density of accumulated narrative weight, a material responded to mana-charge the way any high-retention material responded. Above the threshold, the material didn’t respond to mana-charge as input. It entered into a dialogue with it. The mana-charge acquired properties from the material — specifically, it acquired the resonance frequency of whatever story the material had been part of, which meant the mana that passed through a weapon built from such material at the threshold level or above was no longer structurally neutral. It carried a signature.
At constellation-density material, the signature it carried was a constellation’s.
The patron had written two words at the bottom of the page and underlined them once. The annotation stopped there.
Resonant strike.
He sat with that for a moment. The fable’s sealed timeline was forty years old and the Stellar Breaker had existed in his old-world archives as a weapon rated in a classification tier that didn’t have a standard label. He had known what it could do in general terms. He had not known the specific mechanism until now.
Jiseok was going to build a weapon that could deliver a resonant strike carrying a constellation’s frequency signature. A weapon that hit a constellation-level entity with the entity’s own resonance reflected back at it.
He returned the volume to its position and stood in the stacks for a moment.
The Bureau had dissolved the patron constellation for approaching this answer. They were in Bak Junho’s office in week four to prevent the same answer from being reached again. And Jiseok was going to reach it anyway, from a completely different direction, because he had a piece of constellation-mass material from a dungeon floor that he had pulled out of a vein that no one else in the cohort had identified, and a mana-integration instinct that Siru had recognized in the first session, and a reference book left on his workbench by a student who was going to be expelled in eight weeks and would leave without knowing what he had given him.
The fable was sealed because it had happened. Every element in it had been real and had mattered in exactly the way it had mattered, and the sealed timeline preserved the causal chain that connected Jiseok the failed hero to the Maker of Stellar Anvils. Kang Min was inside that chain now. His presence was a new variable in a myth-grade fable, which meant the fable’s completion system was tracking his interventions and assessing their divergence impact on the core structure.
He needed to know more about the blueprint fragments Ryeo Hanbin’s family records contained.
The archive records had noted them without detail — a family guild with a historical connection to a blacksmith-class climber whose design direction was adjacent to the Stellar Breaker’s eventual form. Partial fragments. Studied by Ryeo Hanbin as a potential thesis resource. The connection was loose enough that the records had treated it as background context rather than a significant variable.
But the patron’s annotations had shown him that the Stellar Breaker’s design wasn’t Jiseok’s in isolation. It was the convergence point of multiple separate research threads, each approaching the same answer from a different direction. The patron from the theoretical side. The blacksmith-class climber whose fragments were in Ryeo Hanbin’s family records from some intermediate position. Jiseok from the material and instinct side.
If the fragments were more than background context — if they were an actual partial blueprint rather than a design tendency — then Ryeo Hanbin was carrying a piece of the Stellar Breaker’s architecture without knowing it, and the operation running through Bak Junho’s office might eventually identify that and use it.
He needed to know what the fragments contained.
The problem was that Ryeo Hanbin was not going to share that information with someone he didn’t have a reason to share it with. The cohort’s social structure didn’t create natural contact between students in different scoring tiers except in the forge sessions, where bench positioning and task requirements occasionally created adjacency. He had been in Ryeo’s vicinity regularly but had maintained the same observational distance he maintained with everyone in the upper tier — present enough to read, distant enough to be unremarkable.
The tolerance testing week was ending. The dungeon practical was ten days out. He had been planning to make the contact after the practical, when the fable’s social dynamics would have shifted enough to create a different context for it. The visitor’s early presence in Bak Junho’s office had moved his assessment of the timeline forward.
He had seven days to create a reason for Ryeo Hanbin to talk to him about his family’s blueprint records.
He went back to the theory room for Yeon Daesik’s incident log session and found a bench position two rows behind Ryeo Hanbin and one seat to the left. The session covered the technical documentation protocol for stress fracture incidents, procedural and undramatic. Students worked through the log format individually, Yeon moving through the room with corrections.
At the end of the session Yeon paused at Ryeo Hanbin’s bench. The correction he delivered was the same phrasing as the previous two sessions: correction induces secondary stress. Ryeo acknowledged it with the same controlled expression he had used the two times before.
Kang Min waited until Yeon had moved three benches further along the row before he leaned slightly forward and spoke quietly to the bench between himself and Ryeo.
"The last strike in your correction sequence," he said. "You’re adding to it."
Ryeo went still for a moment. Then he turned his head just enough to locate the source without making the turn visible to the room.
"The technique flaw Yeon keeps flagging," Kang Min said, the same quiet register. "It’s not in the assessment and it’s not in the correction process. It’s in how you close the correction. The final strike is heavier than the preceding ones by a consistent margin. Enough to introduce a new stress vector in the adjacent grain structure."
Ryeo was looking at him now, the full turn, the expression not hostile but not open either. Evaluating.
"You’ve been watching my station," he said.
"I watch everyone’s station," Kang Min said. "You’re the only one with that specific pattern."
Ryeo held the look for a moment. Around them the session was wrapping, students packing materials, the noise level coming back up.
"Why are you telling me this," Ryeo said.
Kang Min picked up his notes and aligned them. "Because it’s a training habit, which means it was built into you deliberately by someone who knew what they were doing. They just weren’t building it for this application. That’s worth knowing."
He left the bench and went out ahead of the room.
He hadn’t given Ryeo anything that required a response. The information was useful, specific, and delivered without visible motive, which meant Ryeo was going to spend the next several days deciding what to do with the fact that someone had been reading his technique precisely enough to identify a fault his assigned instructor hadn’t explained to him, and had mentioned it once and walked away.
The contact had been made. What Ryeo did with it would tell Kang Min what kind of calculation he ran, which was information he needed before the dungeon practical where, if the archived record held and it usually held, Ryeo Hanbin’s operation was going to cost three students their lives.