She had reorganized the four versions of the technique overnight.

Not physically: the pages were the same pages, the notation the same notation. What she had reorganized was the way she was holding the problem, the structure she had been applying to the divergences between versions. In the previous weeks she had been organizing by where the versions diverged. What she was now organizing by was where they converged: the specific points in the sequence where all four versions agreed completely, and the specific points where they split.

The convergence points were not interesting. Every version agreed about the opening stance, the first six steps, the return position at the end. These were the fixed poles of the technique, the parts that had survived transmission intact regardless of which school or lineage carried the version.

The divergence points were two: steps thirteen and seventeen.

She had not previously noticed that both divergences involved the same structural element: the relationship between the rotational force and the joint position at the moment the force peaked. In step thirteen, the peak was in the elbow. In step seventeen, the peak was in the wrist. Both peaks were at maximum extension.

She was turning this over as they walked north toward Road 2, and when she had turned it enough she told Ilyun.

He listened without speaking until she had finished. Then: "You are saying the injury is not in step seventeen specifically."

"I am saying the injury is at maximum extension under peak rotational load. Which happens at step thirteen and step seventeen both. Step seventeen is where it appears because that is the final and most intense occurrence. But step thirteen prepared the joint for failure."

He was quiet for a moment. "My school taught the sequence through to completion without stopping at thirteen."

"Yes. Which means by seventeen the joint was already compromised."

The road was good, the morning cool, and neither of them spoke for a while. She was thinking about the shape of the thing they were looking at: not a single error but a cascading one, each step's damage accumulating in the joint until step seventeen was the final increment past the limit. This was a different kind of design problem than she had been treating it as.

---

In the small town three hours from the Road 2 venue, Ilyun took them to a house at the edge of the residential district where a retired practitioner named Bak Sungwan had lived for the past twelve years. He had not corresponded with Sungwan recently but knew of him through the network of teachers and former students that Ilyun had navigated for years without mapping it explicitly.

Sungwan was in his sixties and received them with the courtesy of someone accustomed to practitioners appearing at his door with questions. He made tea. He asked what they were working on.

She told him about the technique. She named it and described the specific problem they were investigating: the version divergences, the injury mechanism, the structural question about steps thirteen and seventeen.

He looked at her for a long moment. Then he looked at Ilyun's hand.

Then he said: "I have a version. Different school. I learned it from a practitioner who studied in the western highland tradition."

She had not found any documentation of the western highland tradition in the Reedfall archive. This was a fifth version.

He demonstrated it slowly. She watched with the specific quality of attention she had been building for months, the watching that moved toward understanding rather than recording, and she understood several things simultaneously: the western highland version agreed with version two and three at step thirteen, disagree with all four previous versions at step seventeen, and introduced a modification at step fifteen that none of the other versions contained.

"Step fifteen," she said.

"Yes," Sungwan said. "The modification at fifteen reduces the accumulated load at seventeen. It distributes some of the force earlier."

She sat with this. The western highland version had the same understanding of the structural problem that her grandfather had identified and had attempted to address it at step fifteen rather than by modifying the peaks at thirteen and seventeen. It was an incomplete solution: the modification reduced the risk but did not eliminate it, and it altered the force output of the complete technique in ways that might or might not be acceptable depending on what the practitioner needed the technique to do.

She described this analysis to Sungwan.

He nodded slowly. "The practitioners in that tradition knew it was incomplete. It was the best they could do without abandoning the technique entirely."

"Who built the technique in the first place?" Ilyun asked.

Sungwan looked at him. "That is the question, isn't it."

---

They worked through the fifth version for the next hour, sitting around the table in Sungwan's front room with pages spread across the surface and Sungwan demonstrating sections when her notation required physical clarification. Ilyun contributed the lineage knowledge he had accumulated from his own years of investigation: the approximate dates and locations where each of the versions had first appeared in the documented record, which schools had been in contact with each other during those periods, which practitioners had been mobile enough to have carried knowledge between traditions.

She had not known the extent of what he had been researching independently before their agreement.

She thought about this while she was writing. He had been pursuing the same problem from the external-research angle that she had been pursuing from the archive angle. Their combination had produced more than either would have reached alone not because their knowledge overlapped but because it was shaped differently, fitting together at the edges like two pieces of a larger thing.

This was what working with someone meant, she thought. Not duplicating effort. Covering the ground the other person could not reach.

She wrote the fifth version into the documentation and added a new section to her comparative analysis: attempted resolutions, incomplete. The western highland modification represented the first evidence she had found that someone in the historical chain had identified the structural problem and tried to address it. This was important. It meant the knowledge of the flaw had not been entirely lost.

It also meant the flaw had been known and had still propagated. Schools had continued teaching versions of the technique despite the documented risk. She noted this without fully unpacking it. There would be time for unpacking later.

---

Walking away from Sungwan's house in the late afternoon, she added the fifth version to her organized mental archive of the sequence. Five versions now, across five distinct lineage branches, with one documented attempted repair. The divergences at steps thirteen and seventeen held across all five. The convergence points held across all five. The problem was real and had been real for long enough that at least one tradition had tried to fix it.

She told Ilyun: "The person who originally designed this technique either did not know about the joint vulnerability or knew and did not correct it."

"Or knew and could not correct it without destroying the technique."

She considered this. "If the full force output requires the joint to be at maximum extension under peak load, then reducing the load at that point reduces the force output. The technique may only produce what it is supposed to produce at a structural cost that cannot be eliminated without changing what the technique is."

"A technique with a built-in price," he said.

"Which may be why there are five versions. Everyone who inherited it understood the price and tried to find a version that charged less."

He was quiet for a while.

She thought about the agreement at Road 1, the exchange she had offered and he had accepted: her knowledge of his injury for his knowledge of the body. She thought about how far that exchange had traveled from its original terms, how much larger and stranger and more important the investigation had become, and how neither of them had planned for this when they had shaken hands at the qualifying venue.

She did not say any of this aloud. She was still getting used to the category of things that were true and did not need to be said immediately.

She filed the fifth version and walked north toward Road 2 and thought about a technique with a built-in price and what it meant to keep teaching a thing you knew cost something.

She thought about what Sungwan had said after they had worked through the fifth version and were preparing to leave: that the practitioners in the western highland tradition had known their modification was incomplete and had kept it anyway. She thought about what it meant to preserve a flawed thing because no better version was available.

This was, she recognized, the situation the entire practitioner community had been in regarding this technique for an unknown number of generations. They had inherited something that did not work perfectly, had understood at some level that it did not work perfectly, and had continued to teach it because the partial version was better than nothing. The damage it caused was not a secret, exactly. It was a known cost that had been accepted without being explicitly accounted for.

She thought about Ilyun's five years.

She thought about every practitioner in the five lineages she had now documented who had attempted the technique and experienced the injury and been told they had done something wrong.

The weight of the thought was considerable. She walked with it for a while without trying to put it down.

"The sixth manuscript," she said.

He looked at her. They had been walking in silence for nearly an hour.

"Gyeorun described a sixth manuscript. From the period before the school consolidations. He believes it is the original version, or closest to the original." She paused. "He cannot access it."

"Why."

"He did not say. But if the consolidation involved suppression of the absorbed school's documentation, the sixth manuscript may be in the possession of the school that absorbed it. Which means accessing it requires either the cooperation of that school or working around the cooperation."

"Do you know which school it is."

"Not yet. But the lineage chain I am building from the five versions should narrow it significantly. The convergence points tell me which versions are closest to the original. The version closest to the original was taught at the school most recently in contact with the source."

He was quiet for a moment. "That is why he came to you."

"Yes." She thought about this. "He knows I have the five versions in sufficient detail to do the lineage analysis. He does not have the same quality of access to the versions because he is working from written documentation that may itself be incomplete."

"And you are not working from written documentation."

"I am working from what I witnessed demonstrated and what I read and retained. The quality is different from a copy of a copy."

He nodded.

She thought about the sixth manuscript and what finding it would mean for the investigation and what the investigation finding its answer would mean for the longer question of what came after the investigation, which she had been approaching indirectly and was beginning to approach less indirectly.

She thought about the folder.

She set these thoughts in order and walked north.

---

That evening she wrote a new section in the investigation notes: lineage convergence analysis. She mapped the five versions against each other and identified the points where they agreed: the opening twelve steps, every version identical. The return position after step seventeen: every version identical. And then the divergences, which clustered tightly around steps thirteen and seventeen and which, when mapped against the lineage chains, showed a specific pattern.

The two versions that diverged most dramatically from each other at step seventeen were also the two versions that shared a common ancestor most recently. This was counterintuitive: you would expect closely related versions to be more similar, not more different. But if the ancestor had introduced a specific modification and two subsequent traditions had independently tried to reverse that modification in different ways, the result would be two versions that were close in lineage and far in execution.

She wrote this observation down and looked at it.

The ancestor had made a change. The change had been recognized as problematic. Two traditions had tried to fix it independently. Neither fix was complete.

Who had made the change in the ancestor version, and why?

She did not have enough information yet. But the question had a shape now, which was more than it had had before.

She looked across the table at Ilyun, who was reading a text she had lent him: one of the secondary sources on the highland traditions that she had collected at Road 1. He was reading with the complete attention he gave everything. She watched him read for a moment and thought about the question he had answered and not answered over the months of investigation, the question of why he had been pursuing this, and whether his answer had changed as the investigation had grown into something larger than either of them had originally mapped.

She thought about asking. She did not ask.

She went back to the lineage analysis and worked until the lamp burned low.

She ran the lineage analysis forward from the convergence point she had identified.

If steps one through twelve were identical across all five versions, those twelve steps were the stable core of the technique, the part that had survived five different lineages without alteration. This meant the original designer had made twelve steps that were correct and two steps that were not, and the two incorrect steps had been the focus of all subsequent modification.

She thought about a designer who had built twelve correct steps and two incorrect ones and whether that was the profile of someone who did not know about the incorrectness or someone who accepted the incorrectness as the cost of a specific benefit.

The benefit of the technique, as she understood it from the combined documentation, was the force output at the completion of step seventeen. The technique produced an unusual amount of force at the end of the arc because of the specific way the load was distributed through the preceding steps, culminating in the extension at seventeen. The load pattern that made the technique effective was the same load pattern that made it dangerous.

She thought about this as a design problem rather than a transmission problem. The designer had not made a mistake. The designer had made a choice.

"It was deliberate," she said.

Ilyun looked at her from across the table.

"The load placement at thirteen and seventeen," she said. "It is not a transmission error. It is the mechanism. The technique is designed to generate maximum force through maximum load at maximum extension. The joint failure is a consequence of the design, not a corruption of it."

He was quiet for a moment. "Which means there is no version that eliminates the danger."

"There is no version that keeps both the force output and the joint safety. You can have one or the other. The western highland modification reduced the force output to reduce the danger. Every other version kept the force output at the cost of the danger."

She thought about what this meant for the investigation. They were not looking for an uncorrupted original. They were looking for the designer's original document, which would tell them what the designer had known and accepted about the technique. That was a different question.

"The sixth manuscript," she said. "If it is the original, it may tell us what the designer intended. Not what the technique was supposed to do. What the designer knew it cost."

He looked at her.

She thought: the designer knew. The designer accepted the cost. And then the technique was transmitted without the documentation of the accepted cost, and five generations of practitioners paid it without knowing they were paying something the designer had calculated.

She wrote this in the investigation notes with the specific clarity of someone who has finally understood the shape of a problem they have been circling for months. She thought about the implications for the reconstruction project. She had been building the archive as a preservation effort, ensuring that the knowledge that had burned would not be permanently lost. The designer question complicated the preservation frame: if the technique had a known flaw that the designer had accepted, preserving the technique without preserving the designer's reasoning was incomplete preservation. The reconstruction needed to include not just the technique but the documented cost of the technique, which was what the sixth manuscript presumably contained. She added a new section to the reconstruction notes: techniques with documented known costs. She had three entries for this section already, from smaller texts in the archive that had warned about specific physical risks associated with specific movements. The sixth manuscript, if found, would be the most significant entry. She thought about the difference between knowledge that was preserved and knowledge that was complete, and whether her grandfather had understood the distinction and whether the reconstruction was honoring the distinction. She sat with the design problem as a problem about knowledge and responsibility: what you know and what you choose to pass on. The designer had known the cost. Whether the designer had passed on the knowledge of the cost was the question that made the sixth manuscript essential. She had been treating the investigation as a question about what went wrong in the transmission. She was now understanding it as a question about what was chosen in the original. These were different questions with different answers. She added this to the investigation notes and felt the shape of the problem shift in her mind, settling into something more complete.