Rate Separation

The notice was still there when I looked at it a second time, same administrative typeface, same line breaks. WYNN ALDGATE AND PETRA CARREN: ATTUNEMENT SESSION RESCHEDULED. DATE: DAY SEVENTEEN OF ASSESSMENT PERIOD. LOCATION: PRIMARY ATTUNEMENT CHAMBER, VERATH REGIONAL.

That made it public knowledge. Whatever the scheduling change meant, the whole cohort now had it. Behind me, Fass said something clipped about harmonic decay, and Osku's response came back fast and dismissive, the two of them locked into a rhythm they had been running for longer than this morning. Renne's cup clinked against the table. Tobias had returned to his notes, his pen still.

The format code at the bottom of the posting identified it as cohort-wide rather than individual-file. I stepped back from the board.

The scheduling authority for the primary attunement chamber sat with whoever managed that chamber's calendar, not with the Annex administration, which meant the change had not come from Hael, or at least not from Hael alone. Why Wynn and Petra's sessions had moved to day seventeen was genuinely unanswerable without more information than any of us had. Standing in front of the board was not going to produce more of it.

I went to find water.

The common room's secondary shelf had a ceramic pitcher that someone refilled each morning, but it was empty. The courtyard well was the next option, which meant going outside, which I had been planning anyway.

The courtyard sat between the residential wing and the training halls, an open stone square with a low wall on the south side and a well at the center. The morning air was cold in the way that settles into stone overnight and does not leave until the sun has been on it for hours. The rope on the winch had the right kind of wear, fiber smoothed at the grip point and fraying where it ran over the lip. The stones around the well's base were damp in the way stones get when water spills on them regularly, a working well rather than a decorative one.

Tobias was already there, filling a cup from the bucket he had drawn. He did not look up when I came through the courtyard door. He took half a second to read the situation before acknowledging it. His notes were not with him. He had come down with a cup and nothing else, not what Tobias did in the mornings. From what I had seen, his mornings involved the notes.

"Pitcher's empty," I said.

"Noticed. This is easier anyway."

He stepped back from the well to let me draw my own bucket. I worked the winch. The rope was rough under my palms, the kind of rough that comes from fiber absorbing months of moisture and never fully drying. The bucket rose with the uneven cadence of a winch that had not been oiled in some time, jerk and pause, jerk and pause.

"The Wynn and Petra notice," Tobias said, while I was still cranking. "You saw it."

"Just now."

"I saw it yesterday."

That was interesting, though not immediately informative. The notice had posted system-wide. When he had seen it versus when I had did not tell me much unless he was about to explain how he had come across it early. He was not. He let that sit and looked out over the south wall at whatever was visible beyond it, a section of tiled roof belonging to the next building over.

The bucket came up. I poured into my cup and set the bucket on the well's edge.

"Have you been using the courtyard for exercises?" Tobias asked.

There it was. The courtyard had enough open space for certain types of physical attunement exercises but clearly was not designed for anything requiring a controlled environment. Nobody in the cohort had used it for exercises because it was not an exercise space. We both knew it.

"Not yet," I said, and drank.

"It has some utility for kinetic work. The dimensions are reasonable." He turned back from the roofline. "I have been thinking about whether the cohort is making efficient use of the available infrastructure."

Technically, it was a sentence about courtyard logistics. It was also the opening move of an entirely different conversation, and the way he had packaged it was careful enough to respect the craft while the meaning underneath it was clear.

"What kind of kinetic work?" I asked.

He looked at his cup. "Force propagation exercises. The ones that require spatial distance between the candidate and the focus point."

"And your hall does not have the room?"

"My hall has adequate room. The issue is different." He paused, clearly having decided in advance what came next. "There is a propagation behavior that my exercises have been producing that I cannot account for through the standard framework. The force dissipates at a rate inconsistent with what the documentation predicts for my attunement profile."

"What is the dissipation pattern?"

"Faster than expected in the first quarter of the propagation path. Then slower. As if the force is encountering a medium transition that is not physically present in the space."

He said "as if" the way someone says it when they think it is actually true and are waiting to see how the listener responds before committing.

"Is the documentation accounting for null-space interaction?" I asked.

The pause before his answer was one beat longer than the previous ones. "No. The standard propagation framework treats null-space as an absence rather than a medium."

"Right."

He looked at me, not sharply, just steady. Waiting without expression for me to stop deflecting.

"What is your current exercise producing at the quarter-path mark?" I said.

"A forty percent deviation from projected output. Consistent across twelve sessions."

"And at the halfway mark?"

"Recovered to within nine percent of projection."

Twelve sessions of consistent forty-percent deviation that mostly recovered by the midpoint, that was not a measurement error or a documentation gap. It read like a real interaction between his force propagation and something in the space itself, something that attenuated and then released. Null-space interaction would do exactly that if the propagation was passing through a region where the affinity field had a local null component. It would attenuate kinetic force in the first quarter and allow recovery as the force exited the null field and re-entered standard medium.

Whether his hall had a null-field component depended on things I did not know: how the system configured the training halls, and whether my unlabeled hall's null-affinity orientation had any lateral bleed into adjacent spaces.

"Does your hall share a wall with the unlabeled hall?" I asked.

Tobias's expression did not shift. "The third wall, yes."

"Okay," I said.

Neither of us spoke. A door opened somewhere in the residential wing, a brief sound that closed again.

"If the null-field has lateral extension," Tobias said, carefully, "the attenuation zone would be consistent with what I have measured."

"It might. I do not have documentation on the lateral extent."

"But you would expect some extension."

"I would expect some extension," I agreed.

What I had just confirmed for him did not violate any instruction I had received. The privacy instruction covered my compound-primary inference and the specifics of the session log. It did not cover general properties of null-space interaction. Tobias had worked out the null-space hypothesis on his own. I had confirmed it without giving him anything he did not already have, which was an honest exchange and also a calibrated one.

He nodded once, without explaining why, and that was answer enough.

"If you run into documentation that is relevant," he said, "sharing it would be useful."

"Same."

That was the arrangement. Nobody had stated terms or a timeline, and nobody had defined the scope. Both of us were aware that what we would each give would depend on what we thought the other had and what we were willing to expose in exchange. It would be uneven from the start, because my situation was more complicated than his and he probably knew it, but the arrangement still had value on both sides.

Tobias refilled his cup and set the bucket back on the well's edge.

"One thing," he said, in the tone people use when they have decided to add something they had been holding back. It came out controlled and casual, a register of its own.

"Sure," I said.

"My training interface has a system log visible in the configuration tab. Most of it is administrative standard: initialization dates, cohort file numbers, module access states. But there is a section toward the bottom that I noticed during the first week and only checked properly yesterday."

He stopped. Letting the silence stretch to see if I would fill in the rest. I did not.

"Three galactic observer credentials registered to the Arch-Tier Annex at initialization," he said. "Not candidate accounts. Observer credentials. Galactic Council affiliated, based on the credential format."

The cup in my hand stayed at roughly the same angle. The well's edge was still under my elbow. Another door opened somewhere inside the building, footsteps moved through a corridor, and then the building was quiet again.

"Read-access accounts?" I asked.

"That is what the credential type indicates. Read access to Annex activity." He watched me as he said it. "You had not seen this."

He had not phrased it as a question. The fact that he had named it told me he had already decided how to read my response and was checking whether what he got matched his prediction.

"Not in my interface," I said, which was accurate. My training interface gave limited access. Most of the configuration layer was not visible to me in the way it apparently was to fully attuned candidates.

Tobias accepted that without pressing it. He picked up his cup and stepped back from the well. "Thought you would want to know," he said, and that was the closest he came to naming why he had arranged to be at this well, at this hour, with no notes.

He went back inside.

The water in my cup had gone cool while we talked. The system had registered three galactic observer accounts at initialization, meaning they had existed since day one of the cohort's placement at the Annex, with read access to everything that happened here: all session logs, all training records, all administrative activity. Everything from the first disc registration on day one through the tile entry from last night.

If those accounts were actively monitoring rather than passively registered, then every entry I had made in the session log since the cohort session had already reached someone with Galactic Council affiliation. The first condition of the observer-confirmation protocol required a council-level observer with active read access to my classification file. What Tobias had described was exactly that category. I had no way to know whether any of them had opened my file specifically. The first condition alone did not satisfy the protocol; all three conditions had to meet within the same administrative cycle. But the possibility that condition one was already structurally in place without any action on my part changed the probability calculation by a substantial amount.

The cord response was in the closed session log. The tile entry sat in the open log. The disc registration had been in the record since day one. I had logged the compound-primary inference and its two unconfirmed properties. If an observer had been reading any of it, they had seen all of it.

The courtyard door opened.

Nadia came through with her own cup, unhurried, and crossed to the well without acknowledging either that I was standing there or that Tobias had just left. She drew her bucket, filled her cup, set the bucket back with the efficient movement she applied to most things.

She tasted the water. She looked at the cup, not at me.

"This tastes like the system wants us to think it is good water," she said, to no particular addressee. She set her cup down, adjusted her grip, and walked back through the door.

The water in my own cup was flat and slightly cool. The neutral non-taste of stone-filtered water from a depth that kept it consistent regardless of season. It tasted exactly like water. Whether that was a point she was making or not, I had no way to resolve it without knowing more about Nadia than I currently did.

I set the cup on the well's edge and went inside.

Back in the training hall, the air smelled the same as always: stone and lamp oil and the faint particular dryness that came from the pedestal room's poor ventilation. The lamp was still at the quarter-mark I had left it. The session log was open on the writing shelf, my last entry from the previous night visible at the bottom of the page.

The tile sat flat on the third pedestal. Disc at eight degrees, unchanged. Cord hanging straight from its pin.

Standing in the center of the room, I tested my own state. Six hours of sleep. The residual pressure from last night's partial resonance, the memory of the tile's response. The null-state induction work from the previous night had left something I could not name precisely: a memory in my chest and throat of where the partial resonance sat, the lateral pressure with no directional push that I had only ever managed to recall before, not actually experience. Last night had changed that. The tile's response had come while I was holding the partial resonance state in observation, and the relationship between the two, if there was one, was something I had not worked out yet.

The null-state induction exercise required holding two opposing attunement states simultaneously without collapsing them into each other or canceling them. The partial resonance I could access was the null-primary state, the absence of directed affinity. The opposing state I had been trying to introduce was minimal resonance engagement in any one of the three secondaries, sustained at low enough intensity that it would not immediately overwhelm the null state. Every previous attempt had ended in premature equilibration: the two states resolving into each other within ninety seconds, producing a flat null output with no gap between them.

I settled into the standing position I had been using. Located the partial resonance state.

It came up faster than it had a week ago. Within thirty seconds rather than the four or five minutes it had taken initially. I held it, found the pressure point, let it stabilize.

When it was stable, I introduced the inverse.

Temporal was the affinity that correlated with the disc's eight-degree position, the cord's positive response, the tile's interval behavior, the whole cluster of temporal-register phenomena I had been logging since day one. It was also the affinity I had the least active experience with, since the documentation required the restricted index subsection I had only just accessed, and that subsection described observation and interval-capture rather than active engagement.

The states met.

The inverse introduction produced what it always produced: rapid stabilization, both states converging toward a center point rather than sustaining their separation.

This time I held it past the first convergence. Premature equilibration meant the states merged early, a failure of maintenance rather than a failure of the states themselves. I had been treating the equilibration as a boundary condition, something that happened when the states came close enough to interact. A different reading: it was a rate problem. The states did not fail to hold their separation. They failed to hold it at a speed faster than equilibration could occur.

The flat merged state already formed, both affinities sliding toward center. Instead of releasing, I pulled each state toward its own edge, deliberately, not harder but faster. Working the separation the way you would work a fraying seam before it could close on itself.

The merged state paused.

Both states held for a count of ten, then twenty. The lateral pressure in my chest did not increase, but it did not dissolve either. At forty, the gap still held: two distinct attunement states occupying the same space without resolving, a distinction I could mark as a physical difference between left-side chest and right-side throat, each maintaining its own quality.

At fifty-eight, it dissolved.

Still no resonance gap. No active null-state induction in the sense the documentation described. But fifty-eight seconds was longer than anything I had managed before. Previous attempts had topped out around thirty. Fifty-eight was nearly twice the duration, a meaningful change in the pattern.

I logged it: time, duration in seconds, attunement states involved, the rate-rather-than-boundary hypothesis, the method modification. All in language I kept accurate and restrained. "Duration of partial concurrent maintenance extended to fifty-eight seconds before equilibration. Previous maximum: approximately thirty seconds. Method: active rate separation rather than passive boundary maintenance. Hypothesis: concurrent threshold maintenance is a rate-maintenance function, not a proximity-limit function."

Then I sat down at the writing shelf with the restricted index subsection on temporal affinity interaction and physical register outputs and read it again. This time looking for anything that touched on rate dynamics rather than proximity conditions.

The subsection had four main sections. The first three covered observation protocols for temporal-register objects, administrative documentation standards for board-level review, and the physical register outputs associated with different classes of temporal affinity interaction. The fourth section appeared under the heading CANDIDATE OPERATIONAL NOTES and was the shortest, three paragraphs.

I had gone through this section twice already. The first paragraph described the physical sensation associated with temporal affinity engagement at partial resonance, which matched the sensations I had been logging. The second distinguished between interval-capture, passive and observational, and interval-engagement, which required sustained concurrent maintenance and produced a different class of output.

The third paragraph was the one I came back to now.

"Concurrent threshold maintenance in temporal affinity candidates records an inverse relationship with equilibration rate: candidates who demonstrate faster rate-separation during initial induction attempts typically achieve longer maintenance durations earlier in the training progression. Rate-separation capacity is trainable within a single assessment period. The documentation recommends logging progressive duration increases for board review as evidence of active engagement with the compound-primary configuration."

Rate-separation capacity is trainable. The documentation had just told me that what I had been treating as a hard limit was a progress variable. It also specified that documenting progressive duration increases would serve as evidence for the board.

The session log was open. The board review window had one working day remaining. Thirty seconds to fifty-eight seconds was a meaningful increase, and if the rate-separation hypothesis held, continued practice through the rest of the day could produce additional documented increases that the subsection specified as relevant evidence for the board.

The lamp was burning steadily. Outside the slot windows, the morning light had moved from pale to something warmer, mid-morning, the day already running. Somewhere in the building, a door opened and closed and footsteps went past in the corridor without stopping.

I stood up. Went back to the center of the room.

Located the partial resonance state.

This time it came up in twenty-five seconds. I introduced the inverse. The states met, started their familiar slide toward center, and I worked the active rate separation before the convergence could settle. Fast and deliberate. Holding both states at their own edges before they could close on each other.

The equilibration came. I pushed through it faster than before.

The gap held at thirty seconds. Then forty. Then past fifty. The two states separate and distinct, the lateral pressure not doubling but differentiating, each quality occupying its own position in my chest and throat. More intentional than the previous night's resonance-adjacent state during the tile observation. More structured. With a clarity to the distinction between the two that I could hold and examine at the same time. I could feel the boundary between them as a defined edge, the way you can feel the seam between two different temperatures of water when you hold still enough.

At sixty-three seconds the equilibration finally won and the states collapsed flat.

Sixty-three seconds. Better than the fifty-eight before it.

I went back to the shelf and logged it.