The Royal Archive's main hall was the most beautiful room Renji had entered in Elyndor, which was saying something, because Elyndor seemed to approach architecture the way other cultures approached literature: with the expectation that it would outlast everyone who made it and should therefore deserve that outlasting. The hall had ceilings that disappeared into shadow at a height that made it difficult to estimate. The walls were white stone, warm and present, the kind of material that held the accumulated history of all the light that had ever fallen on it. Along the upper registers, between the carved script that ran continuously from entry to far wall, crystallized windows caught the light and scattered it in patterns that moved through the day. The shelves were built into the walls themselves: the architecture had been designed around the books, around the records, around the accumulated documentation of nine hundred years of a kingdom that ran on paper.

Renji stood in the middle of this room for approximately thirty seconds after entering it.

He said: Who organized this?

Lysera, who had been watching him take in the room with the attention of someone who lives in a place and has forgotten how to see it, said: The Archive has been organized by fourteen head keepers over its history. Each modified the system somewhat. The current system reflects all of them.

He said: That explains the inconsistencies.

She looked at him. He looked at the shelves. She said: Most people, when they first see the Archive, remark on its beauty.

He said: It is beautiful. He paused. It is also filing its administrative records in fourteen different formats, three of which are mutually exclusive. The beauty does not fix that problem.

She stood beside him looking at the room. Then she said, in the tone of someone who has been thinking something for a long time and has just heard it said aloud: No. It does not.

He got to work.

The first three days in the main Archive established several things he added to his running document. The public collection held approximately forty percent of the Archive's materials, based on his estimate from the catalog indexing. Of that forty percent, roughly a quarter was related in some way to the Verdant Prophecy or the ceremonies built around it. This was a significant proportion of a public collection. It meant the Prophecy was not just a founding myth but an active administrative framework: property disputes, succession rights, civic role assignments, all filtered through Prophecy interpretation. Modifying the Prophecy was not just a theological act. It was a legislative one. He underlined this twice.

Lysera found him, on the third day, sitting on the floor of the eastern alcove with six volumes open around him and a notebook on his knee. He was aware this was not his best professional presentation. He was also making good progress and had not wanted to stop to move to a table.

She said: There are chairs.

He said: The floor has better light in this section.

She looked at the six open volumes. She said: You are cross-referencing the policy interpretations.

He said: The Prophecy is the legislative framework. I need to understand how the modified version changed the policy outputs compared to what the original would have produced.

She was quiet. Then she sat down on the floor beside him, which he had not expected, and picked up one of the open volumes. She said: This one is from seventy years after the modification period. The interpretations in this section reflect the understanding of the Drakonir as enemy established by the modified text.

He said: What does it say?

She read and translated. The passage described a protocol for handling encounters with Drakonir near Veylwood's borders: classification as hostile incursion, mandatory reporting to the Nightroot Guard, escalation procedures with three levels of response. He wrote it down.

He said: And the comparable passage from before the modification?

She said: I would need to find it. It would be in the restricted section.

He said: What would it say, if you had to guess based on what you know about the period before the modification?

She said: I am guessing.

He said: Guess precisely.

She looked at the passage in front of her. She said: I think the older protocol would have described the encounter as a diplomatic contact. With notification to a records office rather than a military one.

He wrote: the modification changed the classification of an entire people from treaty partners to hostile enemies, which changed the administrative protocols for every encounter with them for three hundred years. This is not a textual discrepancy in isolation. This is a policy architecture built on a falsification.

He said: Do you know how many people have been harmed by that policy architecture in three hundred years?

She said: No. The records of border incidents are in the military archive, which is separate from this Archive.

He said: I would like to see them eventually.

She said: That will require additional authorization. She paused. I can make the request.

He said: You have been making requests on my behalf all week.

She said: Yes.

He said: Why?

She was quiet with the quality of silence she used when deciding how honest to be about her own motivations. Then she said: Because my mother spent thirty years making requests and being denied them, and I have spent sixty more years making smaller requests and being denied those, and I have decided that I am tired of the size of my requests.

He looked at her. She was looking at the volume in her lap. He said: What size of request would you make, if you were not managing the denial?

She said: The size that this situation requires.

He wrote that evening: she is moving. Not quickly. Not with announcement. But the direction has changed. She has been making careful, deniable, strategic requests for sixty years. She is beginning to make the requests that are appropriate to what she actually knows. That shift is significant.

He thought about sixty years. He thought about what it required to sustain the controlled version of yourself for sixty years in a place where you knew, had always known, that the official version of events was wrong. He thought about the Archive's fourteen filing systems and its crystallized windows and the books built into the walls, and he thought about a woman who had come to this place knowing what it concealed and had stayed, for sixty years, keeping the concealment's enemy alive inside her own careful precision.

He thought: she has been doing the most demanding kind of work, alone, without acknowledgment, for sixty years. I should not mistake the precision of her method for the absence of urgency behind it. The urgency had always been there. It had been waiting for the right moment, accumulating pressure behind the careful surface. He thought it was changing now.

The Archive settled into its nighttime quiet around him as he worked. He stayed later than he needed to. He was aware that he stayed later than he needed to. He filed this under: noted. He continued working.

There was a quality to the work, in this room, with its accumulated centuries and its fourteen incompatible filing systems, that was different from the quality of work in any office he had worked in before. The offices had been places of processed information. This was a place of primary information, of original documents, of the first handwriting of people who had been dead for hundreds of years. Reading these records was not reading about things that had happened. It was reading the things themselves. The difference was, he found, significant in ways he could not entirely articulate but felt in the quality of his attention: he was more careful here. More aware that what he was looking at was irreplaceable. More aware that getting it right mattered in a way that transcended professional practice.

He put his pen down at a late hour. He looked at the notes he had made. Forty pages in eighteen days. A more detailed record of a single document's provenance than he had kept in any prior engagement. He looked at the Archive around him, at its sourceless warm light and its carved script and its fourteen inconsistent filing systems, and he thought: someone spent three hundred years building this concealment. Building it carefully, with intelligence and patience and a thorough understanding of how information moves through institutions and how it can be stopped.

He intended to spend somewhat less than three hundred years taking it apart.

The work of the following days established a rhythm that he had not anticipated enjoying but did, in the specific way that any work becomes enjoyable when the quality of attention it requires matches the quality of attention you are capable of bringing to it.

The Archive's public section was vast, and within it the materials related to the Prophecy were both plentiful and strategically incomplete. He developed a practice of reading in layers: first the surface, what the document claimed to say; then the administrative layer, what records existed around the document's production and handling; then the comparison layer, how the document's content related to other documents from the same period; and finally the absence layer, what should be there based on what was there, and was not.

The absence layer was always the most informative. In nine years of audit work he had learned that what institutions removed from their records was often more diagnostic than what they kept. Keeping records was a habit. Removing them was a decision. Decisions had reasons, and reasons could be inferred from the shape of what was missing.

In the Prophecy's case, the absences were clustered. There was very little pre-modification documentation relating to the Drakonir as a political entity. There were references to such documentation in later records, references that assumed the earlier documents' existence and cited them without naming them specifically, in the way that records cite foundational documents that the citer assumes the reader already knows. But the foundational documents themselves were not in the public section.

Lysera confirmed, when he asked, that they were not in the restricted section either. Or rather: no catalog entry for them existed in the restricted section's index. Whether they were there in some unlisted form, or whether they had been removed entirely, she could not say without access to the restricted section's full inventory, which she did not have.

He wrote: the documentation gap around the Drakonir as political entity is complete. Not selective, not partial. Complete. Every document that would have described them as neighbors or treaty partners has been removed from the accessible record. What remains is documentation produced after the modification, which describes them as enemies, and documentation produced before the modification that has been modified to describe them as enemies, and documentation that does not reference them at all because it predates the covenant and the covenant's subsequent corruption.

This was the cleanest institutional cover-up he had ever seen. Not because it was sophisticated in its construction but because it was thorough. Someone had understood that partial removal created comparison points, and comparison points were dangerous. They had therefore removed everything.

He had spent his career finding the things that partial removal left behind: the reference in one document to a document that no longer existed, the administrative note that mentioned a process without describing its source, the gap in a numbering sequence that indicated something had been there. These traces were always present when someone had removed material hastily or without complete authority over the record.

Here the traces were present too, but they were subtler. This had been done slowly, carefully, by someone with complete authority over the record and a thorough understanding of what the traces were and where they would be.

He was grateful for Miron, who had found the traces in the linguistic record, where they were harder to control because language evolved in ways that individual administrators could not fully direct. The word Drakonir had kept its etymology in the academic texts that the Oracle's administrative reach did not extend to. That was the gap. That was where the removal was incomplete.

That was where he was going to build from.

He told Lysera this, late on the fifth day in the main Archive, when they were the only two left in the building and the light had settled into its nighttime warmth. She listened with the quality of attention she brought to things she was going to think about after the conversation had ended.

Then she said: You intend to use the linguistic evidence to establish what the administrative evidence has been cleaned to hide.

He said: Yes.

She said: That is a sound methodology.

He said: It has a weakness. The linguistic evidence is inferential. It tells us what the Drakonir were called, and what that name means, and that the meaning was changed. It does not tell us what the original covenant said. For that I need the original document.

She said: The box.

He said: Yes.

She was quiet for a moment. Then: I have been thinking about how to get to it.

He said: What have you thought?

She said: Elirane has rights to the ceremonial spaces. But her rights are noticed. If she requests access, the request is noted in the ceremonial log. The log is reviewed by the Oracle's office.

He said: So accessing through her creates a record that Caerion will see.

She said: Yes. He will not necessarily act on it immediately. But he will know.

He said: We need to be ready to move quickly once we have the document.

She said: Yes. She paused. I have been considering who else might have unnoticed access to the ceremonial spaces.

He said: Who?

She said: Fen.

He looked at her.

She said: Fen goes wherever it wishes. It is not subject to Archive protocols. I have never understood the nature of Fen's relationship to the physical structure of this building. I have only observed that it appears in places it should not be able to access. She paused. I have been thinking about this for three days and I have not found a flaw in the logic.

He looked at the door of the reading room, through which Fen had exited several hours ago.

He said: You are suggesting we follow the fox.

She said: I am suggesting we consider the fox as a resource.

He said: That is the same thing.

She said: Yes.

He wrote that evening: Fen. I have been treating the fox as a local curiosity and an occasional guide. I should reconsider this. A creature that has lived in this Archive for longer than Lysera has been alive, that moves without restriction through spaces that have every other kind of restriction, that appeared specifically to show me the restricted vault symbol and the misfiled document: this is not a coincidence. This is an actor. I do not understand the nature of the actor. I should find out.

The Archive had fourteen filing systems and three hundred and forty-seven years of accumulated decisions about what to keep, what to move, and what to make impossible to find. He had been in it for five days. He had found more in five days than most people found in a professional lifetime of working with the documents it contained, and the reason was not that he was smarter than the people who had worked here before him. The reason was that he was looking for something specific, and he had the tools to look for it, and he had Lysera beside him who knew where things were and how they connected, and together they were finding the shape of what had been done to the documents and to the kingdom that had been organized around them.

He thought: this is what it looks like when a system has been protected from scrutiny for so long that the first serious scrutiny finds everything immediately. Not because the concealment was bad. The concealment was excellent. But because concealment, however excellent, leaves a negative space. And negative spaces are findable, if you are looking for them.

He was looking for them. He was very good at it. He intended to continue.

He developed, over the first days in the main Archive, a working theory of what he was looking at.

The Archive was not simply a storage facility for documents. It was an argument. Every document it contained, every filing decision, every classification choice, was part of an argument about what Elyndor knew and what Elyndor could access and, by implication, what Elyndor was. The fourteen head keepers had each made contributions to that argument, some of them in good faith, some of them in the service of other agendas, and the result was a room that presented as a comprehensive historical record while actually being a highly curated selection from a more complete history.

He had worked with curated archives before. Every institution curated its records to some degree. The curation was usually done for benign reasons: storage constraints, relevance assessments, administrative efficiency. Occasionally it was done for less benign reasons. The skill of the audit was in distinguishing between the two kinds of curation by looking at what had been curated and in what direction and whether the direction was consistent.

In the Archive's case, the direction was consistent. The curation had removed materials that would have given context to a specific set of events three hundred years ago. The context would have changed the interpretation of those events. The removal of the context had therefore changed the only interpretation available.

That was deliberate curation. That was the kind that mattered. The work would continue in the morning. The Archive would be there, with its fourteen inconsistent filing systems and its three-hundred-year-old carefully curated silences. He would be there too, with his pen and his notebook and the methodical patience he had developed over nine years of looking for the things that institutions did not want found.