The Predecessor

The case file held three hundred and forty-one entries. Once I mapped the four-component date notation against the Annex's administrative calendar, the span worked out to approximately ninety-one days of active training. Six weeks I'd been at the hall. The predecessor had run this same problem for nearly twice that.

RN-0047-NL was the practitioner's reference number. No name appeared anywhere in the document: not in the header, not in the session entries, not in the administrative stamps at the bottom of each page. The number occupied the top right corner of every entry with the same consistency I'd started using for my own attempt count as a header.

Reading forward from the beginning, the progression was clear enough to follow without working at it, and that bothered me. RN-0047-NL had started with the same baseline problem. Null-state induction that terminated too fast. No stable gap. Premature equilibration presenting as a consistent failure mode. Their early entries read like mine. Short sessions. Methodical notation. Failure modes categorized, method adjustments recorded in plain terms for the session that followed.

By their thirtieth session, they'd developed something they labeled rate-separation, noting in a brief aside that the null state's natural onset was slower than its inverse and that they could use this offset rather than correct it. By their fifty-second session, the entries referenced a concurrent-threshold approach that sounded, in their shorthand, like what I'd been running for the past two weeks. They'd reached forty-three seconds of stable coexistence by day sixty-two. One hundred and twelve seconds by day eighty-one.

The record shifted after day eighty-one. Session entries got shorter and less focused on duration tracking, with more references to external events I couldn't parse. Notations like "administrative exchange complete" and "third interval elapsed, no contact from observer channel" appeared without explanation of what interval or which observer. The methodical tracking style compressed into something closer to reactive notation. Whatever happened outside the training hall during those final weeks had gotten into the entries in a way the earlier sessions hadn't allowed.

The last full entry, dated nine sessions before the file ended, read: "Register plate behavior has changed. Difficult to characterize. Not a reversal of the initial registration but a different quality of response, as if the object is tracking something other than the state I'm producing. Will monitor."

Eight more sessions followed, decreasing in length. The final one ended mid-sentence.

The hall object they called the register plate was my disc. Eight degrees clockwise from its resting position, which it had reached during a session in my second week and had not moved since. The notation RN-0047-NL used for it was almost verbatim to the entry I'd written in my own log. The cord behavior was similarly aligned. The weighted tile they'd classified as confirmed active by day twenty-two had produced the same range of partial registration events that my own tile entries documented.

What specific states or thresholds had produced those results in the predecessor's sessions, the file didn't say. Someone had redacted the classification notations beside each object entry, replacing them with a consistent text string: State Reference Omitted Per Standard Protocol. That information had been removed before making the file accessible at Tier 2, not after.

The linked records tab listed six items. I tried each one and got the same screen back every time: Classification Level Above Tier 2. Observer Authorization Required. No other options.

So the file itself was the complete picture available to me, and the complete picture was a progression that tracked mine closely enough to be a template, ending in a record that stopped rather than concluded. RN-0047-NL had made it to where I was standing before whatever happened had removed them from documentation. The date of that final entry, cross-referenced against the Annex's external date system, came out to approximately forty years before my cohort's intake. Six cohorts, maybe seven at standard intervals, had operated in this facility after them. The file had sat locked at Tier 2 the entire time.

I wrote a list in the session log. What the file contained. What it didn't contain. Where my progress and theirs converged, and where I couldn't yet determine if they diverged. The honest answer to that last part was that I couldn't find a divergence. Object behaviors matched. Failure mode terminology matched. The progression arc, from rate-separation through concurrent-threshold to coexistence, matched across all of it.

I set the session log aside and sat without moving.

Not because I'd run out of things to note. The objects were all where they'd been. Disc at eight degrees. Cord straight. Tile flat. The floor registered the same faint discontinuity under my heel when I pressed down at the right angle. None of that had changed. I sat because I'd reached the edge of what the documentation could do for me tonight. Continuing to write entries past that edge would produce worse data, not more.

The lamp's oil reserve was adequate. The flame burned clean.

Forty years had passed between RN-0047-NL's final session and my cohort's intake. Six cohorts, maybe seven at standard intervals, had rotated through this facility, and not one of them had accessed a sealed case file at Tier 2. Presumably because none of them had reached Tier 2 within their placement cycle. Or they had accessed it and never reported it. Or someone had placed the file under a different classification key than the one my reclassification had unlocked, and I'd stumbled into it sideways.

I logged my working assumptions: six linked records require authorization above Tier 2. State reference data redacted throughout case file. Redaction authority unknown. Someone removed the information before making the file accessible at this tier, not after, based on consistent formatting of the omission text.

Under that I wrote: the register plate behavior notation from RN-0047-NL's final full session is the most significant divergence point visible in the record. Their description of the object tracking something other than the produced state suggests a functional shift that preceded the file's termination. Current status of the disc: eight degrees clockwise. No comparable behavioral change logged in this session record to date.

I underlined that last sentence. Then added: though I have not yet explained the floor discontinuity under the east wall.

That was the honest version. I'd confirmed the texture anomaly with my hand and logged it. Then the review window had closed, the reclassification had arrived, the predecessor's file had opened, and the floor had not moved back up the priority list the way it should have. I added a second line: east corridor texture anomaly and hall floor discontinuity pending comparison. I should prioritize this before the next attempt.

Enough for tonight. I capped the log, checked the lamp, and left the hall.

The common room was nearly empty. Coll sat at the near end of the long table with a cup of something cooling in front of him, the handle turned away from his hand the way it always was when he wasn't actually drinking but wanted the option. He looked up when I came in, then back down at the table.

"Late for a rest night," he said.

"Finishing notes."

"Notes." He turned the cup handle back toward himself, then away. "Fass asked about the observer visit. The formal one."

"What did you tell him?"

"That I didn't know anything about it." He looked up again. "I don't know anything about it."

"The log records what happened."

Coll held the cup with both hands without drinking. When he was working something over, he went very still, his eyes losing focus. That stillness didn't mean he'd stopped thinking; it meant he'd stopped moving. "The reclassification came through this morning," he said.

Not a question. I'd assumed word would move fast. The Annex's administrative traffic wasn't strictly private, and Coll paid attention to more of it than he let on. "Yes."

"Level 4, Tier 2."

"Yes."

He set the cup down. "That's above the review board's authority."

"The two channels are separate, near as I can tell. The board review and the Level 4 designation both apply. One doesn't supersede the other until I have documentation saying it does."

"Tobias thinks the board review is academic now."

"Tobias might be right. I'm not treating it as resolved until one of the channels says it's resolved."

Coll looked at me, then picked up the cup again. "That's what I told Fass you'd say." He drank. The cooling liquid had clearly passed whatever temperature he preferred, and he put it down with a small controlled grimace. "He's going to ask you himself. Probably in the morning."

"He can ask."

"You'll give him the same answer."

"It's the same answer because it's accurate." I pulled out the chair across from him and sat, mostly because standing in the doorway while he held a cup was starting to read as avoidance or interrogation and I didn't intend either. "What does Fass actually want to know?"

Coll considered that with the same stillness. "Whether the council-affiliated accounts watching this cohort are watching it because of you specifically, or whether something larger is developing and you're the first indicator they're tracking."

That was more precise than Renne's version of the question. "And Coll's version of that question?"

"Same question." He said it without any change in expression. "I'm not pretending it isn't."

There was an honesty in that worth not deflecting. "I don't know the answer. The designation message came from a council-affiliated format, which I've confirmed. What produces that format, whether it's automated or requires manual authorization, I haven't been able to determine. The Level 4 observer visit happened because my progress was ahead of standard interval. Neither fact tells me whether the surveillance is cohort-wide or practitioner-specific, and I haven't been able to construct a reliable test for the distinction."

Coll was quiet for a few seconds. "The fact that you have a methodology for thinking about whether it's a reliable test is what's going to unsettle Fass."

"I know."

"He's not wrong to feel unsettled."

"No, he's not."

Coll stood, picked up the cup, and took it to the basin near the far wall. He rinsed it without urgency. "The east corridor," he said, his back to me.

"What about it?"

"The texture on the stone near the junction. Nadia mentioned it to Wynn two days ago. Wynn didn't know what to do with it and didn't log it because she didn't know which channel to use." He turned around, drying his hands on the rough cloth hanging by the basin. "Did you log it?"

"In my own session record. Not in a separate administrative report, because I also wasn't sure which channel it belonged to."

He nodded, confirming something he'd already concluded. "You should probably file a secondary report. Even if it gets routed to a general queue and sits. Having it in the record matters more than having it read promptly."

"Agreed." I hadn't done it yet because the classification question made the routing unclear, and that question had been sitting under four other open problems. I added it to the list I was keeping in my head. "Did Nadia say when she first noticed it?"

"Day twelve, she thinks. She wasn't certain." Coll hung the cloth back. "She went back the next morning and it was the same."

Day twelve was three days before I'd confirmed it with my hand. If she'd noticed it on day twelve and it was stable by day thirteen, it had been present for at least a week before I'd put it in my own log. I filed that under things to correct in the morning.

"Tell Nadia I want to compare notes on it," I said. "If she's willing."

"She'll be willing." He crossed back toward the doorway. "She's been waiting for someone to ask."

He left. I sat at the table a while longer, not because there was more to do tonight but because the lamp in the common room was brighter than the one in the hall, and the room had the particular quality of quiet that comes after a conversation has arrived at its actual subject rather than circling it. Coll had told me two things. Fass would push in the morning. The east corridor anomaly had a longer documented history than my own record showed. Both of those needed addressing before I ran attempt fifty.

I went to bed.

Nadia was at breakfast before most of the cohort, which was how she preferred it. She had a specific seat at the end of the table nearest the window that let in morning light from the east courtyard, and she sat in it with the satisfaction of someone who had made a small territorial claim and won through consistency. I put my tray down across from her.

"The texture on the stone," I said.

She looked up from her food. Her narrow eyes narrowed further as she decided something. "You talked to Coll."

"Last night."

"And you want my notes."

"If you have them."

She considered this for roughly three seconds. Then she reached into the front pocket of her training jacket and produced a folded square of paper that bore the creases of many folds and unfolds. She passed it across without ceremony.

The notes were dense and precise. She'd been logging the anomaly since day eleven, one day earlier than Coll's reported estimate, using a three-point reference system. Distance from the corner junction. Height above the floor. A subjective rating scale for the texture quality that she'd defined in a footnote at the bottom of the first fold. The scale ran from zero, standard stone surface, to four, a discernible difference in surface density that would register to touch without prior expectation. She'd rated the anomaly a two on initial contact and a three by day fourteen. By day nineteen, which was yesterday, the rating was still a three but the affected area had expanded approximately four centimeters toward the south end of the corridor.

I read through the whole thing twice. She ate.

"The expansion pattern," I said. "Four centimeters over five days?"

"That's the consistent measurement. I used the floor grout lines as reference points so the scale wouldn't drift." She pushed a piece of bread around her plate. "It's slow. But it's not stable."

"When were you going to bring this to someone?"

She gave me a level look. "I brought it to Wynn on day fifteen. Wynn didn't know what to do with it." A pause. "Then I waited to see if anyone else would notice independently."

"I confirmed it on day fifteen. I logged it in my session record."

"I know." She said it flatly, a simple acknowledgment. "You pressed your hand to the stone. I was two junctions down the corridor."

So she'd seen me notice it and waited to see what I'd do with it. When what I'd done was log it in a private session record and then let four other problems climb over it in priority, she'd continued waiting. That was a fair appraisal of my handling.

"I'm going to file a secondary administrative report today," I said. "Your notes need to be part of that."

"I was hoping you'd say that." She picked up her cup. "I'll make a clean copy."

"The original is fine. The fold marks don't affect the data."

She looked at the paper. "The fold marks are from carrying it in my pocket for eight days, which does affect the impression the document makes."

She had a point. "Clean copy is fine."

Fass came in while we were still sitting there. He went to the service table, filled a tray with his usual precision, then came directly to where we were sitting, confirming Coll's prediction. He put his tray down next to mine rather than at his usual spot three seats further down.

"The Level 4 reclassification," he said, sitting.

"What about it."

He cut his bread into even pieces before looking up. Fass was careful about things in a particular order. Food first, question second. A preparation habit I'd noticed from the first week of cohort meals. "The observer accounts. Tobias put the number at three, council-affiliated format. Renne's been working a theory about the Primary Attunement Chamber schedule adjusting around whatever your hall is producing." He looked at me. "I want to know how those two things connect."

"I don't know if they do."

"You have a working hypothesis."

"Several, none of them confirmed." I met his eyes. "The observer accounts are real. The format has been confirmed. The Level 4 designation and the classification review are two separate channels that may or may not interact. Those are the established facts. Everything connecting them is inference."

"Your inference tends to be more reliable than most people's confirmed data."

That was a strange thing to say and I didn't deflect it. "The east corridor anomaly is the more immediate priority. It's expanding. It needs a formal report today, and it may need a cohort-level conversation, because it's affecting shared space."

Fass processed that. He picked up adjacent implications before responding to the central point. I'd seen him do it enough times to wait. "If the anomaly is expanding in the east corridor and your hall is on the east wall, and your floor has that discontinuity you mentioned to Coll."

"I logged it. The two may be related. I don't have enough data to say."

"But you're going to try to get that data."

"Today, yes. Before attempt fifty."

He ate a piece of bread. "Okay." Flat, final. He'd gotten what he came for. He didn't ask about the predecessor's case file, which either meant he didn't know about it or had decided not to press. I suspected the latter.

Nadia had been watching the exchange with the same attention she'd given the east corridor. When Fass stood and moved to his usual seat further down the table, she refilled her cup and said, without looking at me directly, "He's not actually worried about you."

"I know."

"He's worried about the cohort being inside something it can't see the edges of." She put the cup down. "That's different."

"Also accurate."

She folded her notes back along the existing creases and put them in her pocket. "I'll have the clean copy ready before midmorning."

I finished breakfast and went back to the hall to check the floor.

The discontinuity was still there. Pressing my heel down at the same angle as before, slightly off-plumb toward the east wall, the floor gave a different quality of resistance. Not soft, not yielding. More like the subtle pressure difference between two panes of glass that are almost the same thickness but not quite. The disc was still at eight degrees. The cord hung straight. The tile was flat.

I pressed my hand to the east wall itself.

The stone surface was cool and uniformly textured for the first two handspans from the corner. Then, at a point roughly forty centimeters up from the floor and ten centimeters from the east wall's junction with the north wall, the texture changed. A fraction coarser than the surrounding stone, with a faint quality of density that my palm registered before my fingers did. The same quality Nadia had rated a three on her scale.

I measured the affected area with finger-widths. The dense region was approximately twelve centimeters across and eight centimeters tall. The boundary was gradual rather than sharp, which made the edge difficult to locate precisely, but the center of the denser region was consistent with what I'd confirmed two weeks ago. If anything, it was marginally larger.

I wrote in the session log: east wall interior surface, texture anomaly confirmed present, estimated area twelve by eight centimeters, boundary gradual, center density consistent with prior confirmation. Apparent slight expansion of affected area, though precise comparison is limited by original measurement's imprecision. Floor discontinuity under heel confirmed at same location, same angle dependency. Spatial relationship between wall anomaly and floor discontinuity requires further measurement.

Then: attempt fifty will wait until I file the secondary administrative report. Report will reference session log entries for floor discontinuity and wall anomaly alongside Nadia's independent measurement record from days eleven through nineteen.

The disc held its angle. The cord held its line. Out in the east corridor, the anomaly was spreading slowly and without explanation, four centimeters wider than it had been five days ago.

I sat down to write the report.