The history tutor had covered three wars, two succession crises, and a territorial expansion before I finally understood that historical omission wasn't accidental. It was structural.

  This was my twelfth year and the tutor was a man named Corvel who had a historian's deep investment in kings. He loved kings. He could discuss the psychology, the strategy, the political maneuvering of kings at considerable and sometimes impressive depth. His knowledge of material culture, economic systems, and institutional history was approximately zero. He was quite intelligent; the gap was in what his field of study considered worth knowing.

  The gap wasn't personal to Corvel. It was a product of who wrote history and why. The literate classes, the ones who kept records and composed accounts, were cultivators and court officials; they valued political and military events because those were the events that affected their standing and their world. The people who built the aqueducts, who designed the supply chains, who maintained the formation networks, those people didn't write history. They didn't see their knowledge as the kind of thing that belonged in books. So the books were full of kings and empty of builders, and the builders' knowledge lived in their hands until they died.

  I'd raised the aqueduct question before, years ago with Feldren, without getting much traction. I tried it again differently.

  "The conquest of the Miral territories," I said. "The texts describe the military campaign in detail. Who built the supply lines that made the campaign possible?"

  Corvel looked at me with the expression of a man encountering a non sequitur.

  "The supply lines," he said.

  "Yes. The campaign lasted three years in difficult terrain. Someone was organizing the food, the equipment transport, the roads. Is that documented anywhere?"

  "Military logistics is covered in the campaign histories," he said.

  "Is there an individual documented? A logistical officer? Someone who designed the supply system?"

  A pause. "That level of detail wouldn't typically be requested."

  "What about the fortifications?" I tried. "The Miral territories required new fortifications after conquest. Who designed them? Is the builder documented?"

  Corvel's expression was moving through something. "I'm not sure what you're suggesting the relevance is."

  "I'm trying to understand how capabilities persist across administrations," I said, deploying the framing I'd worked out in advance for when I needed to ask questions that might seem odd. "The king who conquered the Miral territories died thirty years later. The fortifications he built presumably lasted much longer. Who maintained them? How was the knowledge of their construction preserved for future maintenance?"

  I genuinely wanted to know. My old world's aqueducts had required rediscovery after the civilizations that built them collapsed. Entire metallurgical traditions had vanished when the last practitioner died without an apprentice. Knowledge always died with people unless someone built a system to stop it.

  Corvel, to his credit, actually engaged with this. "The craft guilds," he said slowly. "They maintain technical knowledge through apprenticeship. A guild's institutional memory extends much further than individual lifespans."

  "So the guilds are the knowledge-preservation system," I said.

  "In trade terms, yes." He seemed mildly surprised to be articulating this. "In cultivation terms, the major schools and academies serve a similar function."

  "But the schools and academies are for cultivators," I said. "Which is a minority of the population. What about technical knowledge that isn't covered by either guild tradition or cultivation schools?"

  Another pause. Longer this time.

  "I suppose it persists imperfectly," Corvel said. "What isn't in guild tradition or cultivation school might be rediscovered when needed, or lost when not."

  "Like the aqueduct systems in the southwestern territories," I said.

  "You've read about those."

  "They were built about three hundred and fifty years ago and are no longer functional in several sections," I said. "The texts describe them as 'fallen into disrepair' which doesn't explain whether they could be repaired if someone understood the original engineering."

  "I don't know the answer to that," Corvel said.

  "No," I said. "Neither do I."

  Corvel went back to his kings. I went back to listening, but the question kept scratching at me: who, if anyone, had ever been the advocate for aqueducts? The guilds preserved what guilds cared about. Cultivation schools preserved cultivation. And everything else just lived in somebody's memory until that somebody died.

  But the question wasn't the only thing I took away from that session. Corvel's eyes had changed. The man who'd spent ninety minutes on a succession crisis without once varying his delivery had suddenly come alive over aqueducts, of all things. Not because he cared about aqueducts. Because I'd connected them to institutional persistence, which was a question he already cared about.

  The frame was the lever. I'd stumbled into it rather than choosing it, which meant I couldn't reliably do it again.

  In my old life, there had been whole professions built around this problem. Science communication; public engagement; knowledge translation. I remembered the terms without the substance. Someone had figured out how to take specialized knowledge and make it travel across the gaps between fields, between experts and everyone else. Entire university departments devoted to the mechanics of persuading the people who knew things to explain them to the people who needed to know them.

  I couldn't remember any of the actual techniques. The irony was familiar by now. Old-world knowledge surfaced like that: vast, half-dissolved. I remembered a solution had existed; I couldn't remember how. The name of the gap, never the thing that filled it.

  Thessa was coming the other way in the corridor outside the teaching rooms.

  She'd been in the languages session one room over; I could hear the tail end of Vardic conjugation through the half-open door as I left. She was carrying a writing case tucked against her hip in the way that suggested she'd forgotten she was holding it, and she looked down at me with the mild authority of a sister who was both taller and used to being right.

  "The aqueduct question," she said. No greeting.

  I stopped. "What about it?"

  "Heard about it from Instructor Maren, who heard it from Corvel." She tilted her head. "The tutors talk, you know. More than you'd expect, and less carefully."

  "Did Corvel complain?"

  "No, that's the interesting part. He apparently went to the library after your session and pulled three texts on infrastructure he'd never requested before. Maren thought it was funny. I think it's more interesting than funny."

  She leaned against the corridor wall, arms crossed. Thessa had a way of settling into a conversation as if it were a room she'd decided to occupy for a while.

  "You didn't convince him that aqueducts matter," she said. "You showed him that the question about aqueducts was actually a question about institutional persistence. He already cared about that. You just connected his thing to your thing."

  "I wasn't strategizing," I said. "I was frustrated."

  "I know. That's why it worked." She half-smiled. "You're terrible at persuasion when you're trying. When you're irritated you stop performing the polite version and say what you actually mean, which turns out to be more compelling than the performance."

  I didn't have a good response to this. She was probably right, which made it worse.

  "The question isn't whether aqueducts matter," she continued. "Everybody who uses running water already knows they matter. The question is how you get someone like Corvel, who has never thought about running water in his life, to care enough to change what he studies. And the answer isn't arguments. It's finding the thing he already cares about and showing him how your question lives inside it."

  "You make it sound like a technique."

  "Everything's a technique if you practice it enough." She pushed off the wall. "You should practice it. You're going to need people like Corvel eventually, and frustration doesn't scale."

  She walked off toward the west corridor, writing case still against her hip. I stood there for a moment. I knew how to ask the right question; I'd been refining that skill for years. What I didn't know was how to make the right person want to answer it, and I wasn't sure yet where you'd even learn something like that.

  The geography lesson that broke my sense of scale happened in the same month. It was taught by a different instructor, a mapping specialist who had been hired for the older children's cohort. I'd attached myself to those sessions unofficially, through the mechanism of showing up and being quietly useful and Marvenn vouching for my presence.

  The maps on the wall of the mapping specialist's room were different from the castle library's maps. The library maps showed the kingdom in reasonable detail: roads, cities, territories, borders. The mapping specialist's maps were different: they showed the kingdom in context. The kingdom as a territory within a region. The region within a continental structure. And then, on the largest map, hanging on the back wall where I hadn't noticed it at first, the continental structure within a world that extended in every direction beyond the map's edges.

  He was covering trade route geography. The older students were tracking specific routes between major cities. I was looking at the back wall.

  "The edges of the map," I said. "They're not the edges of the world."

  The specialist turned. He'd accepted my presence with the equanimity of someone who found one more student unobtrusive. "No," he said. "The world extends well beyond any map we have."

  "How far?"

  "Unknown," he said. "The eastern territories beyond the Vardic Range haven't been systematically mapped. The northern extent is unknown beyond the Forest Regions. The ocean territories to the west extend much further than any exploration has confirmed."

  I looked at the map. The kingdom, our kingdom, my father's territory, the thing that contained my entire world to date, occupied a recognizable fraction of the visible landmass. A reasonably large fraction. I had mentally understood that the kingdom was large. Standing in front of the specialist's maps, I tried to reconcile the felt understanding with the visible evidence.

  "What fraction of the actual world is on this map?" I asked.

  "We estimate between fifteen and twenty-five percent of the known territory, and 'known territory' is itself a partial estimate." A pause. "The Endless Realm is not a polite name."

  I stared at the map. Let the math happen. If the visible map, which contained my father's considerable kingdom plus the surrounding powers and territories, represented fifteen to twenty-five percent of known territory, and known territory was itself a fraction of actual total territory...

  My father was a minor king.

  I'd known that abstractly. The diplomatic texts were polite about it, referring to the Seria Kingdom as a "regional power," choosing careful language when discussing the civilizations further east and north that dwarfed us in both cultivation stage and territory. But there was knowing something because you'd read it, and then there was standing in front of a map and watching your entire world shrink to a thumb-sized patch on a continent you can't see the edges of.

  The diplomatic language made more sense now. "Regional power" was a courtesy formula, the verbal equivalent of a bow between parties who both knew the power difference but chose not to state it plainly. Civilizations further east were ruled by Imbuing-stage cultivators, Stage 5, with lifespans stretching to three thousand years and the ability to create permanent enchantment through touch. The kingdoms to the north held Embodying-stage rulers, Stage 6, beings who could reshape terrain and who lived for ten thousand years or more. My father, at Shaping stage, was formidable by the standards of this region. By the standards of those greater powers, he was a local official governing a province they hadn't bothered to notice.

  The vertigo hit. Physical, not metaphorical. Earth had been large; I'd spent a lifetime there unable to visit most of it. The mana-geography texts implied my father's kingdom was roughly Earth-sized. My father's kingdom, which was a fraction of a larger power. Which was one territory among many on a continent. Which was one of several continents.

  I sat on a bench near the wall because my legs had made the decision for me.

  The scale wasn't just big. It was the kind of big where your brain stopped trying to picture it and started treating the number as a symbol, the way you treated "a billion years" as a phrase rather than a duration.

  And somewhere in that vastness, in the thousands of kingdoms and territories and wildernesses and civilizations that existed beyond any map I'd ever seen, there was knowledge. Of magic. Of cultivation. Of enchanting and formation theory and material science and techniques that nobody in this castle had ever encountered.

  How much had been discovered and lost? How much existed right now, in some institution in a civilization that had been developing its own tradition for ten thousand years, that the Seria Kingdom had never heard of?

  The isolation was structural, not accidental. Travel between major regions was dangerous and took years. A diplomatic mission was a multi-year commitment; treaties, once established, persisted through inertia because renegotiation required another multi-year expedition. My father might send a handful of major diplomatic initiatives during his entire thousand-year reign and still be considered active in foreign affairs. The distance that kept us safe from greater powers also kept us ignorant of what they knew.

  The specialist watched me quietly for a while. Then he said: "The first time I looked at that map properly I had much the same reaction."

  "How do you deal with it?" I asked.

  "Badly," he said. "And then you get used to it. And then occasionally something reminds you and you have to deal with it badly again."

  I looked at the map for another few minutes. Then I went back to the lesson.

  But the question of what lay beyond the edges kept pulling at me. The Great Forest of Yolic to the north, source of the beast hordes that were both the kingdom's greatest ongoing military threat and its most important source of high-quality crafting materials. Beast cores, crystallized mana concentrations, and the hides and structural components that retained absorbed mana from the forest's dense ambient field. The seasonal horde defence was the kingdom's primary military deployment; more frequent and resource-intensive than any conflict with neighbouring kingdoms, and the garrison commands that the elder sons held existed as much for horde management as for territorial governance.

  To the west and south, the open sea, unmapped beyond a few trade routes. To the east, the Vardic Range, and beyond it the unsystematically mapped territories where the greater kingdoms began.

  We were a small kingdom in a big world. The maps showed the edges of what we knew, and beyond those edges the paper was blank, and I kept staring at the blank parts.

  That evening, I ran the experiment I'd been building toward for three weeks.

  I had made, in Tova's workshop, seven additional practice inscriptions since the first warm-wood attempt. Each one targeting a different aspect of the technique. I'd improved. Not because I had better hands than before, hand stability was a training matter that improved slowly with age and practice, but because I understood the material better. Each inscription had taught me something specific about how mana flowed through channels of different geometry, different depth, different orientation relative to grain.

  The experiment for tonight was connectivity.

  I had two small pieces of wood, each with a single warming inscription. Simple enough that I'd done them without difficulty. The question was: could I link them? Create a flow between them so that mana applied to one would pass through into the other?

  Formation networks, the large-scale connections of enchantments that made city-scale magical infrastructure possible, operated on this principle at massive scale. But the principle had to work at small scale first, or it couldn't work at large scale.

  The texts described formation networks without describing how the connections actually worked. The outcome, yes: connected enchantments sharing mana flow. The mechanism, no. How did you link two separate inscribed pieces?

  If you encounter this tale on Amazon, note that it's taken without the author's consent. Report it.

  I had a theory.

  If mana followed channels, and a channel was defined by clear geometric structure (physical inscription or sustained mana intent), then a connecting channel between two inscriptions needed to be either physically continuous or connected through a sustained mana bridge. Physical continuity would mean the two pieces had to be touching, with the channels aligned so mana could cross the boundary. Mana bridge would mean actively channeling mana between them, which required continuous concentration.

  But there was a third option, suggested by something half-remembered. A tuning fork, struck and held near another of the same pitch; the second one hums without being touched. The medium carries the vibration. If mana could carry vibration the way air carried sound, and if two inscriptions could share a frequency, physical contact shouldn't be necessary. I'd been thinking about this. If both inscriptions were patterned to resonate at the same frequency, if their geometry produced the same mana vibration signature, they might exchange mana through resonance rather than direct connection. No physical contact required. No sustained active channeling. Just matching frequencies.

  I didn't know what "resonance frequency" looked like in inscription geometry. I was inferring it from the observation that some materials naturally attracted certain types of mana, and wondering if geometry could be used to create artificial frequency matching.

  I set the two pieces of wood on my desk, three inches apart, and channeled mana into the first piece.

  It warmed.

  The second piece sat cold.

  I tried adjusting the geometry in my mind, not the physical channels (which were already set), but my understanding of what they should resonate at. Tried to shift the mana I was supplying to match a slightly different frequency.

  Nothing.

  Tried again. Different approach. Slowed the mana flow in the first piece, tried to establish a pulse rather than a constant stream. Pulse, release. Pulse, release. Looked for any response in the second piece.

  The second piece warmed. Marginally. Barely detectable even with my sensitivity.

  But warm.

  I held perfectly still, barely breathing, and the second piece's warmth faded and I couldn't reproduce it. But I'd felt it. Something had passed between them through three inches of air with no physical connection and no active bridging. Some vibration-mediated transfer that I didn't yet have the technique to control.

  Resonance was real. I didn't know how to use it yet. But it was there.

  I wrote six pages of cipher notes and went to sleep thinking about the map on the mapping specialist's wall, and the vastness of the world beyond it, and somewhere in that vastness the knowledge of how formation networks actually worked at the physics level.

  That knowledge existed. Someone had discovered it. Maybe they'd built institutions that taught it, so it was still alive. Maybe it had died with them and was waiting to be rediscovered.

  The lamp burned down. The castle settled into its nighttime sounds: wood creaking, a guard's distant call, wind through the corridor gap that nobody had gotten around to sealing. The second piece had warmed. I hadn't been able to reproduce it. Somewhere in that gap between one observation and replication was the thing I needed to understand, and I fell asleep still reaching for it.

  The map stayed in my head all week. I kept seeing it: our kingdom, this thing that felt enormous from the inside, sitting there on that back wall looking small.

  The knowledge question was worse. A world this large, with civilizations developing independently for millennia; how much had been discovered and then lost? Gone sharply rather than faded gradually: to war, to institutional collapse, to a single practitioner dying with no one to pass their understanding on.

  The longevity paradox made it worse. Cultivators who lived centuries should have been perfect knowledge repositories. An Anchoring-stage practitioner could carry five hundred years of accumulated expertise. But individual longevity created a false sense of permanence. Why write down what the master would be alive to explain for another century? The very durability of long-lived practitioners made the eventual loss more catastrophic; when the master finally died, everything vanished at once. The Formation Guild had dissolved roughly forty years ago, its two centuries of accumulated knowledge scattered to private collections nobody could locate. The castle's old-wing formations were the physical evidence: built by practitioners whose techniques no one alive could replicate. Bren maintained them competently, but he was Breathing stage maintaining work built by someone far beyond him, reverse-engineering intentions he could sense but not fully understand.

  I went back to my room and wrote in my cipher journal for two hours. Not about cultivation theory. About institutional design: what a knowledge-preservation system would need, where you'd find people willing to contribute rather than hoard. The incentive problem kept circling back; why would anyone share specialized knowledge when keeping it private was the rational choice? I wrote around it for another hour and gave up.

  The private experiment that night produced something I hadn't expected.

  I was working on the resonance transfer problem, trying to understand why the two pieces of wood had briefly exchanged warmth through the air, and I'd moved past simple geometric patterns into something more complex. The experiment had been running for three weeks now, in irregular evening sessions, and I had a growing body of observations and a fragmentary theory that might hold together if I could find the right framing.

  Tonight I tried something different. Instead of setting up the two pieces on the desk and working with them separately, I held both simultaneously, one in each hand, and directed mana through them at the same time.

  The effect was immediate and strange.

  Not just the pieces warming. That had happened before in individual pieces, unremarkably. What happened when I held them both was that the mana I was directing seemed to take the shorter path between the two rather than the longer path through the air from my body. Like it found the bridge between the pieces more conductive than the detour through me. Energy took the easier route. It always had, in every system I'd ever encountered: water downhill, heat from warm to cool. And now mana through matched geometry rather than through air.

  The pieces vibrated: not a physical sensation, but something I felt in the mana, a specific synchronization of the patterns in each piece, like two tuning forks locked at the same frequency.

  I held the experience for as long as I could maintain it: forty seconds, maybe fifty. Then the effort became too demanding for my current cultivation level and I let it go.

  I sat on the floor of my room and thought about what I'd just done.

  I'd driven it, not merely observed it. The mana I'd supplied had been the organizing factor; the geometry of the pieces had been the medium; my direct engagement had been the mechanism that made the exchange active rather than passive.

  If that was right, if resonance transfer between non-contiguous inscriptions required an active driver rather than emerging purely from geometry, then the principle was more constrained than I'd been thinking. It wasn't that two inscriptions with matching geometry would automatically exchange mana. It was that the matching geometry made the exchange possible when a cultivator drove mana through both simultaneously.

  This had implications for formation networks. Large networks required constant mana input to maintain, not just initial activation. The reason was partially what I'd just experienced: the network nodes were in continuous resonance with each other, and the resonance required maintenance. A formation network was an active process, continuously maintained.

  Which meant... it could be optimized. If you understood the resonance patterns, you could design networks that required less maintenance energy by better utilizing the natural resonance between matched nodes. The energy savings might be small per node, but across a large network they could compound into something significant.

  I wrote this down. Six pages in the cipher journal, including the specific conditions of the experiment and what I'd observed and what it implied.

  Then I converted the relevant section into plain text and added it to the growing stack of documentation I'd been accumulating for library filing.

  Not yet. I wanted someone with real expertise to see this before I filed it publicly. The resonance transfer work was the most interesting of my current projects and the one where I was most aware of how much I might be missing, how much the existing theoretical tradition might have covered that I simply hadn't found.

  When someone who knew the field had looked at it and told me what was genuinely new and what was rediscovery and what was wrong, then I'd file it. Then it would go into the library and be available to anyone who looked.

  Knowledge buried with its owner. I'd heard the story from one of the castle's older servants: a glassblower whose techniques had gone to the grave with him. No apprentice, no written record. Just a set of skills that existed in one pair of hands and then didn't. It stuck with me because it described the failure mode I was most afraid of.

  The glassblower's techniques, gone with him. No apprentice. No written record. I sat with that for a long time.

  I spent the next afternoon in the castle's upper library, working through a formation maintenance text that hadn't been checked out in eleven years according to the lending card. The text was dense and unhelpful; it assumed a reader who already understood the network principles I was trying to discover from scratch. I set it down after two hours and stood at one of the tall windows that faced the city.

  From this height, the district spread out in readable layers. The castle's inner ward below, stone courtyards and guard pathways I could have drawn from memory. Beyond the second wall, the administrative quarter: tiled roofs in orderly rows, the courthouse's copper dome greened with age. Further out, the craftsmen's quarter, identifiable by the permanent haze above its forges. Smoke and steam rising from dozens of workshops I'd never entered, containing knowledge I'd only read about. Foundries and inscription shops. The places where theory met material and became real.

  A rainstorm was crossing the city from the west.

  I watched it come. The lower districts darkened first; their thatch roofs soaked the rain and held it, and even from this distance I could see the streets change colour as packed earth turned to mud. I knew that mud from the delivery carts that came up to the castle gates on wet mornings, wheels caked in the reddish clay of the western roads. The craftsmen's quarter resisted longer. Slate and fired-clay tile shed the water; where rain hit the forge chimneys it turned to steam, brief white columns that mixed with the usual grey smoke before dissolving. The canal line thickened as runoff from both banks fed it, its surface catching the light differently in the wet.

  When the storm reached the castle walls, the water found paths I'd never noticed: small drainage channels cut into the stone at foundation level, directing flow away from the base where the old formation work sat. Someone, centuries ago, had understood that water and formation inscriptions were a poor combination, and had carved the solution directly into the architecture. Nobody had documented who. Nobody had noted why the channels were angled at precisely that degree, or what would happen if they silted up.

  The city had a different face in rain. More legible, somehow. The water fell on the slate roofs and the thatch equally. It was the drainage that differed, and the drainage told you everything about who had built for whom.

  I made a note in my cipher journal: submit formal request for city access through the household administration. Not urgent, not dramatic. Just the next item on a list that was getting longer. I'd been reading about these workshops and these districts for two years. At some point, reading stopped being preparation and started being avoidance.

  The formation guild I'd read about, dissolved roughly forty years ago, was a thread I kept pulling at over the following months. The castle's maintenance enchanter was the person who could tell me what had actually happened, and I asked him about it during one of the maintenance rounds I'd been tagging along on since age ten.

  "The Formation Guild," he said, in the flat tone of someone describing a wound that had scarred over but never stopped aching. "What do you want to know?"

  "Everything," I said. "How it worked, how it failed, what happened to the knowledge."

  He was quiet for a moment, checking a junction node with the distracted attention of someone whose hands knew the work well enough that his mind could be elsewhere. "It was the primary institution for formation knowledge in the capital. Two hundred years of accumulated practice. Private library, training programme, licensing standards. If you wanted formation work done properly in Sethara, you went to the guild."

  "And the licensing?"

  "Controlled who could practise. Which meant it controlled who could earn a living doing formation work, which meant it controlled access to a critical skill set." He straightened from the junction. "That's what killed it. Two senior factions, both led by Anchoring-stage cultivators who'd been alive long enough that their positions had calcified into identities. The stricter faction wanted to tighten the licensing, reduce the number of practitioners, concentrate the work among the seniors. The looser faction wanted to expand access; more practitioners, broader reach, but it would've diluted the exclusivity that made membership worth paying for."

  "Both had a point," I said.

  "Both had a point," he agreed. "Neither cared about the other's. When your political rivals can live for five centuries, grudges don't fade. They harden. Three years from functional guild to empty building."

  "Could the crown have intervened?"

  "The guild was self-governing. The crown provided the legal framework but didn't manage internal politics." He moved to the next node. "I've wondered about that. Whether there should have been a mechanism. Some way to compel resolution before the institution ate itself. Four hundred years of stable rule and nobody thought to build one."

  He was quiet again while he worked, and I waited, because I'd learned that Bren said the important things after the silences.

  "The library," he said finally. "That's the part I can't let go of."

  "What happened to it?"

  "Went three places. A private collector bought a portion; he died ten years later and his estate dispersed the collection. I don't know where most of it ended up. Senior guild members claimed texts as compensation for unpaid dues; that scattered volumes into thirty private holdings nobody can locate. And the remainder stayed in the guild building itself."

  "Which was taken over by another organization," I said. I'd read that much in the older maintenance texts.

  "A trade association. They needed the space. They cleared the documents that were left." He looked at me directly. "Burned them, mostly. They had no way to assess what the documents were worth, so they assessed them as nothing."

  Two hundred years of accumulated formation knowledge. Burned for floor space.

  "How much of what you know came from that collection?" I asked.

  "A significant portion. I apprenticed under Tessren, who'd been a guild member. He kept some of the texts; when he retired, he gave me access to what he had." A pause, and the silence carried weight this time. "He died eight years ago. I have what he gave me. I don't know what happened to the rest of his library."

  "So the knowledge went from the guild's shelves to Tessren's shelves to yours," I said. "Each transfer losing some."

  "Each transfer losing most," he corrected. "I have maybe a tenth of what Tessren had. Tessren had maybe a tenth of what the guild held." He went back to his junction work. "You can't recover it. You can only document better going forward."

  The same failure mode, but worse. The knowledge hadn't been in one person's head; it had been in a library, a real collection, shelves of it. And the library had burned because the building's new occupants needed the space.

  My old world had solved this, eventually. Not with one great library but with thousands of overlapping ones, each holding copies, so that no single fire could take down the whole system. The formation guild had been a single point of failure. One institutional collapse and two hundred years of accumulated knowledge scattered to private collections and ash.

  He was sixty-three. The only person in the capital who understood the castle's formation network at the level he did. He'd thought about what happened when he retired or died; I could hear it in the way he talked about Tessren's library, the way he said each transfer losing most. He didn't have a solution.

  I didn't have one either. I was twelve.

  The resonance experiment that winter produced a result I hadn't expected and that required three months of careful observation to be confident I understood.

  I had been working on the active driver question, the observation that resonance transfer between non-contiguous inscriptions required a cultivator to actively drive mana through both pieces simultaneously. The question I'd been asking was: could you eliminate the active driver requirement? Could you build a passive resonance transfer requiring no active cultivation input?

  The answer, after three months of variations, was: yes, under specific conditions.

  The conditions were these: both inscriptions had to be on the same continuous material. Not touching, but the same material with an uninterrupted connection between the two pieces. The connection didn't have to be direct, but it had to exist. Like the two ends of a single river.

  When this condition was met, the resonance transfer between the matched inscriptions was passive. Mana applied to one end moved toward the other end through the material connection, guided by the matched geometry, without anyone actively driving it.

  The formation theory texts had described this outcome without explaining it. Now I'd seen it directly.

  But the specific implication I was most interested in hadn't been in the formation theory texts I'd read: the efficiency of the passive transfer depended on the quality of the material connection. A material with high-quality mana channels, good grain alignment, low ambient contamination, the kind of material that came from careful selection and, apparently, long experience in contact with cultivation work, transferred mana with significantly less loss than a material whose mana channels were rough or contaminated.

  Material quality wasn't a nice-to-have. It was the bottleneck.

  The old wing's stone, with its ancient settled mana, would transfer far more efficiently than new stone of the same type. All that care the original builders had put into material selection, grain alignment, composition; I'd assumed it was tradition, maybe aesthetics. It was engineering.

  The economic implications were not abstract. If material quality was the bottleneck, then the kingdom's declining enchanting capability was about losing the understanding of why material selection mattered, as much as losing the high-stage practitioners themselves. The enchanting stalls in the craftsmen's quarter produced functional work, the way a cart path is functional. Getting the job done, but not at what the old work had achieved.

  I wrote six pages in the cipher journal. The growing list now needed material quality, resonance frequency, and someone who could tell me which of my observations were actually novel and which were things every competent formation worker already knew.

  The lamp guttered. Outside, a guard called the hourly check; another answered.

  The third month of the expanded history sessions, Corvel surprised me.

  He'd agreed, after the supply line conversation, to a supplementary meeting format where I could ask questions outside the standard curriculum. I'd been treating these sessions as my time to push. So when he walked in and said he had a topic of his own, I shut up and listened.

  "The texts we study," he said, "are not transparent windows on the past. They are constructed documents produced by specific people with specific interests and limitations. Understanding the text requires understanding who wrote it and why."

  Historiography. I knew the concept from my old world; it was a standard part of serious history study there. But I hadn't thought to apply it here, and the fact that Corvel had arrived at this on his own, possibly because our supply-line conversation had gotten under his skin, made me pay closer attention than I might have otherwise.

  "The gaps in the histories," I said. "The builders of the aqueducts who don't appear. The logisticians who made campaigns possible. Is that an absence of records or an absence of interest in keeping records?"

  "Primarily interest," he said. "The people who wrote history wrote about what they thought was important. What they thought was important was shaped by the same assumptions that shaped their society. A historian from a cultivator-class family writing about a military campaign would focus on the cultivation aspects of the campaign because that was the relevant domain of knowledge and status in his world."

  "So the aqueduct builders would only appear in a history written by someone who thought aqueducts were historically significant."

  "Yes. And the people who thought aqueducts were historically significant were, primarily, the engineers and administrators who built and operated them. Who did not, as a class, typically write history."

  "So the knowledge that it would have been valuable to preserve was held by people who weren't in the habit of preserving knowledge," I said.

  "Broadly, yes. The literate classes were the preservation class. The literate classes valued certain kinds of knowledge over others."

  I sat with this. The people who recognized the aqueducts' importance weren't the people writing the books. Cultivation stage made it worse: the literate classes were the cultivating classes, and guild histories were written by guild masters who recorded what mattered to their guild.

  "What would fix this?" I said.

  Corvel considered. "In theory? A history that took material culture and technical achievement as seriously as political and military events. History that was written by people with technical expertise, or in collaboration with them."

  "Has that been done?"

  "Partially. The guild histories are the best example. Practitioners writing about their own work from the inside. But they're guild histories. They preserve what the guild valued, which was its own techniques and its institutional story. Not the broader picture."

  "The broader picture would require many people with many kinds of expertise all contributing to a shared record," I said.

  "I don't know of an institution that has managed this. The incentive structures work against it."

  "Unless the contribution was the advantage," I said. "Credit for sharing, not for hoarding. Professional standing attached to the people who solved the hard problems publicly."

  He looked at me with the expression I'd learned to recognize: that brief recalibration. "You should write this up," he said. "For yourself, if not for publication."

  "I keep notes," I said. "In cipher."

  "At some point the cipher should become plain text," he said. "Things in cipher can't be found by anyone else."

  This landed harder than Corvel probably intended it to. I was a twelve-year-old writing about knowledge preservation in a format designed to prevent anyone else from reading it. Bren's maintenance records were in plain text; his actual understanding of the castle's formations lived in his head. My cipher journal had the opposite problem: the knowledge was written down, but the cipher made it inaccessible to anyone who wasn't me. The worst possible combination.

  I didn't change anything that night.

  I said I'd consider it. Walked home with the cipher journal tucked under my arm, the spine warm from being held against my side all afternoon.

  Outside, the light through the corridor windows had shifted to the late amber that meant the castle's formation lamps would activate within the hour. The corridor library's shelves were dark. Someone had left a book on the reading bench, open, facedown, as if they'd been interrupted and planned to come back.

  I kept walking.