The transcription work began twelve days after the fire.

Raeha had spent the intervening time organizing. This was not a simple task. The archive had been organized according to a system her grandfather had developed and her father had maintained, a system of lineage-based categorization that made sense if you knew the lineage history well but was, she had always thought, suboptimal for practical retrieval. She was rebuilding from memory, and memory did not naturally arrange itself in the same order as physical shelves, so she had spent twelve days creating a new system before putting a single word down on the new pages her father had obtained from the merchant in the valley town.

The new system organized by technique type first, lineage second. She explained this to her father and he looked at it for a long time and said nothing and she took the silence to mean that she could proceed.

She worked in the entrance room of the main building, which had been cleared and reorganized as a combined study and administrative space while the archive rooms dried out enough to be used again. She sat at the writing desk from first light until the light was gone, with a break for the midday meal and occasional walks around the compound that she had learned were necessary for the kind of sustained concentration her work required. She had never been good at sitting still. She had learned to structure her restlessness rather than fight it.

By the end of the third day, she had reconstructed four complete manuals and the outline of a fifth.

By the end of the first week, twelve.

By the end of the second week, twenty-one. She had also, during this time, begun to notice something.

---

The notice had begun with a small inconsistency.

She was working on the third form manual for the Crane sequence and had reached the section describing the third transition, a movement she had watched performed dozens of times in the training yard. She wrote it from memory, as she had written everything: the angle, the weight distribution, the foot placement, the timing. She set down her brush and read back what she had written.

It looked correct.

She read it again.

She called for Insu, who was still at the compound and had been, since the fire, considerably less inclined to make observations about her time on the walkway. He came to the entrance room and she showed him the page and asked him to demonstrate the third transition.

He demonstrated it.

She watched.

She read her written description again.

There was a discrepancy. Not a large one. The foot placement she had written was four or five degrees off from what Insu had just demonstrated. A small error. The kind of error that might not matter in twenty repetitions but would compound in two hundred, drifting the technique gradually off-true the way a ship drifts from a miscalculated heading.

She thought about it.

The discrepancy could have come from two places: Insu was doing it wrong, or she had written it wrong. To determine which, she needed a second data point.

She asked one of the other remaining students to demonstrate the same transition. He did. His foot placement matched Insu's.

So: two practitioners, consistent with each other, inconsistent with her written description. The most likely explanation was that she had written it wrong. The next most likely explanation was that both practitioners had been taught the same incorrect version. The third explanation, which she sat with for a while and found disturbing, was that she had been watching incorrect versions performed for years and had memorized the incorrect version with perfect fidelity.

She added a note to the written description. She wrote: unverified, possible discrepancy, source unknown.

---

She spent the next three days systematically testing her reconstructed texts against physical performance.

She chose techniques she was most confident about: the foundational forms she had watched performed hundreds of times, the ones described in multiple overlapping texts, the ones she could verify from multiple angles of observation across many years. She asked students to demonstrate. She compared the demonstrations to her written descriptions. She noted discrepancies.

By the end of the three days, she had a list.

The list was longer than she had expected. Not catastrophically long, not a list that suggested her memory was fundamentally unreliable, but longer than she would have liked. Twenty-three techniques out of the approximately two hundred she had tested carried some form of discrepancy between her written description and the physical demonstration. Of those twenty-three, she estimated that twelve were cases where the demonstrator was likely performing incorrectly. The remaining eleven were less clear.

Eleven techniques where she could not determine whether her memory was the source of the error.

She stared at the list for a long time.

The problem, once she had identified it, was not difficult to articulate. Her memory recorded what she had seen. If what she had seen was correct, her record was correct. If what she had seen was a flawed performance, her record was a perfect copy of a flawed performance. Her recall had no mechanism for distinguishing between the two. It recorded visual and kinesthetic information, not the underlying biomechanical truth that the information was supposed to represent.

She had a library of images. She was not certain, any longer, that all of the images were accurate.

She went through the list a second time, more slowly. The eleven problematic techniques all shared a characteristic: they were complex, multistage techniques that required specific physical prerequisites to execute correctly. They were not beginner forms. They were not forms that had been demonstrated for her frequently at the basic level. They were techniques she had watched performed by practitioners whose own skill level she could not independently assess.

The conclusion she did not want to arrive at but could not avoid: she needed someone who understood the physical execution at a level beyond observation. Someone who could look at a written description and say: this is correct, or this is wrong, and here is why. Someone who understood the body the way she understood the text.

She picked up the list and walked to her father's room.

---

He was at the window when she came in, looking out at the courtyard where the new outdoor training area had been marked with rope since the covered walkway's destruction. His leg was healing slowly but he was walking carefully on it, and the slowness of the healing seemed to have given him a particular quality of patience she had not seen in him before, a willingness to sit and look at things rather than move through them.

She placed the list on the table beside him.

He read it without touching it. His eyes moved down the page at the pace of someone reading carefully rather than quickly, which with Garin meant he was taking in each entry individually rather than processing the list as a whole.

When he finished, he was quiet.

"Eleven," she said.

"Yes."

"I can't determine the source of the discrepancy."

"No."

"I need someone who can. Someone who knows execution well enough to look at a written description and evaluate it. Not just perform it, evaluate it." She paused. "That person isn't here."

He turned from the window and looked at her. "No," he said. "They are not."

She had known this, had known it since she had made the list, but hearing him confirm it was different from knowing it herself. It meant something that he agreed. It meant she was not inventing a problem that did not exist. "Seven Roads Assembly," she said.

He said nothing.

"If I go, I'll be around practitioners at multiple levels from multiple schools. I could verify the disputed techniques by watching comparable forms performed by multiple practitioners independently. I could cross-reference." She looked at the list. "I could also learn to perform them myself, which would give me the physical understanding I'm currently missing."

"You cannot perform most of them now," her father said.

"No. That's why I need to go."

He looked at her for a long time. She stood still and let him look. She had long ago learned that the appropriate response to her father's considering silences was to remain present and let the considering happen.

"You would be gone for months," he said.

"Yes."

"You have never competed."

"I know."

"The Assembly is not a school. It does not accommodate people who need to learn as they go. It assesses what you already know."

"I know what I know," she said. "The problem is that I don't know what I know accurately. The Assembly would tell me."

He turned back to the window. She watched his reflection in the glass.

After a long while: "The school cannot continue without qualified practitioners."

"I know."

"If you completed the Assembly and registered as a practitioner under the Reedfall name, the school would have its credential."

"Yes."

He did not say anything else. She did not say anything else. The conversation had reached the place where the logic had completed itself and what remained was the decision, which was not hers alone to make but was not entirely his to prevent.

She left the list on the table and went back to the entrance room.

---

The registration was done in the valley town, at the clerk's office that handled Seven Roads administrative matters during the year prior to each Assembly cycle. She went on a morning when her father was resting and the students were occupied with their practice. She carried the methodology volume, which she had retrieved from the archive and kept in her room, and the registration forms she had obtained from the town's posting board three weeks ago.

She signed her name. She paid the registration fee, which was the largest single expenditure she had made in years, drawn from the account her father maintained for school expenses. She told herself she would explain this to him.

She walked back up the valley road with the registration confirmation folded in her inner pocket and thought about eleven techniques and what it would mean to know, finally, whether her memory of them was accurate.

When she arrived at the compound, her father was sitting in the courtyard.

She took out the confirmation and placed it on the stones beside him.

He looked at it for a long time without picking it up. She stood nearby and waited.

Then, without comment, he went inside.

That evening, when she came back from the village, there was something on her sleeping mat: a folded piece of paper. She opened it.

A map of the Assembly route, hand-drawn in her father's careful script, each of the seven Road locations marked with a brief notation about terrain and typical conditions. The paper was not new. The ink was not fresh. He had drawn it before she had registered.

She held it for a long time.

Then she spread it out on her desk and began to study the route from the beginning.

Seven Roads. Seven places she had never been. Between here and the first one, the question she had been carrying since the fire would begin to find its answer. She would learn what she actually knew and what she only remembered. She would learn the difference, in her own body, between the image and the thing.

She would also need to find, somewhere on that route, the person who could teach her.

She folded her hands over the map and looked at the first destination, marked in her father's handwriting with a single word: Darye.

The work of reconstruction was not like the work of reading.

She had always known they were different, the way she knew that understanding a map was different from walking the terrain. Reading required her to hold still and let the text come through her. Reconstruction required her to hold still and let the text come out of her, which used the same faculty in a different direction and the different direction was harder.

The specific difficulty was this: when she read, the text organized itself according to its own structure, and she absorbed that structure along with the content. When she reconstructed, she provided the structure herself, which meant she was making choices about organization that the original text had made for her, and making those choices differently than the original author had made them was a kind of loss, even if the content was preserved.

She had been thinking about this for two weeks and had not resolved it. She had decided to note it in the margin of her reconstructions, a system of marks indicating where she was certain she had the original structure and where she was working from inference. The marks were accumulating. She found this uncomfortable in the way she found all uncertainty uncomfortable: as a condition to be addressed rather than accepted.

The former student was not someone she had expected to see again.

His name was Dohan, and he had trained at Reedfall for a year before leaving to study with a school in the capital region. He had come back for other reasons, a family matter in the valley town, and had heard about the fire and had stopped at the compound out of, she thought, a combination of genuine concern and the slightly guilty quality of a person who left something and is now being reminded of it.

She had welcomed him with courtesy and not with warmth, because warmth was not her natural register and she was tired and focused on the work, and he had understood the courtesy for what it was and accepted it.

She had shown him the third form manual for the Crane sequence, specifically the passage she had been uncertain about. She had asked him to demonstrate the third transition.

He had demonstrated it.

The discrepancy was small and specific and exactly the kind of discrepancy she had been afraid of finding.

---

After Dohan left, she sat with the discrepancy for a long time.

She was not distressed by it. She had always known this was a possible outcome of the reconstruction work, and she had structured the work to allow for discovery and correction. What she was doing now was the work itself: identifying where her record was imperfect and understanding why.

The difficulty was that she could not, from the discrepancy alone, identify the source. Dohan had trained at Reedfall for a year and then at another school for several years. His execution of the Reedfall form was likely influenced by both traditions. His four-degree variation from her written description might be a Reedfall error in her record, an influence from his other training, or a personal variation he had developed over time. She could not distinguish between these possibilities with a single data point.

She needed more data points.

She worked systematically for the next three days, testing twenty techniques against four practitioners: Dohan, the three students still at the compound, and Baekhwa, who agreed to her request without asking why she was making it. She tracked each test carefully, noting not just the discrepancy itself but the possible sources, the confidence level of her assessment, and the cross-references to the original texts she had memorized.

By the end of the three days, she had the list.

Twenty-three techniques with some form of discrepancy. Twelve where she was confident the practitioner was the source of variation. Eleven where she could not determine whether her record was the problem.

She looked at the list for a long time.

---

The problem, when she articulated it fully, was elegant in its difficulty. She had a perfect record of what she had seen. She could not determine whether what she had seen was a perfect record of the thing itself. The gap between observation and reality, which she had thought was a gap she could close by being careful enough and attentive enough, had turned out to be structural. It was not a gap in her attention. It was a gap in the nature of what observation could provide.

Her grandfather had written about this in the methodology text, in a passage she now retrieved from her memory with the specific feeling she always had when a text she had read abstractly became concrete: the feeling of a key finding its lock.

He had written: The practitioner who believes she can learn solely from watching has confused the map for the territory. Maps are essential. Maps are accurate. But the territory has a dimension the map cannot carry, the dimension of the body moving through it, and this dimension cannot be added to the map by making the map more accurate. It requires a different kind of knowing.

She had read this at eighteen and thought she understood it. She had thought it meant she needed to practice more.

She understood it now as something larger: she needed someone who could translate between the map and the territory. Not a teacher who would explain the map more clearly. A practitioner who had walked the territory enough times that they could tell her what the map was missing.

She wrote the list out twice. The second copy, she folded and kept in her inner pocket. The first, she brought to her father.

---

He read it standing, in the doorway of the room where he had been working on the administrative documentation of the reconstruction. He read it with the careful attention he brought to things he knew were important. When he was done, he handed it back to her and said: "That is why written transmission has limits."

"Yes," she said.

"Your grandfather's text addresses this."

"I know. I've been thinking about the passage in the methodology chapter."

He looked at her with the sideways quality that she associated with him thinking about how to say something he was uncertain about saying. She waited.

"He could not solve it," her father said finally. "He could document it. He could name it. He could not create a written solution for a problem that was definitionally outside of writing."

"I know."

"The solution he proposed was: proximity to practitioners who had integrated the knowledge in their bodies. Not instruction. Proximity. The distinction matters."

She held this.

"I am going to find the solution," she said. It was not a declaration. It was a statement of what she had decided, which was different. Declarations were addressed outward. Decisions were their own thing.

Her father looked at her.

She put the list back in her pocket.

"I'm going to Seven Roads," she said. "I need to be in proximity to practitioners who have what our archive described but could not transmit. The Assembly is the only context where that many practitioners from that many traditions are in the same place."

Her father was quiet.

She waited.

He looked out the window at the courtyard.

"Your grandfather would have agreed with that reasoning," he said.

She nodded. Then she went back to her desk and continued the reconstruction work, because the reasoning was settled and the work was not.