The Other Seven

The morning had been productive enough that returning to the courtyard was a small demotion.

My session notes were solid. I had logged the cord's response patterns in detail, and I had identified the tile's remaining untested entry, which waited on a board event two days out. Everything was in order. The work had a shape. Then the meal bell rang somewhere in the corridor and I had to stop being a person alone with useful data and go be one of eight people at a stone table.

The courtyard had a specific quality in daylight it did not have during the rest intervals, when the overhead angle was more forgiving. At midday the open top admitted direct sun, and the stone held it, so the temperature at the table ran several degrees warmer than the surrounding annex. Oren had set out food without ceremony: bread in sections, a soft grain preparation in a covered clay pot, dried fruit on a flat dish, and two ceramic pitchers of something warm that smelled faintly of bark.

Tobias, Fass, and Coll had arrived first. Tobias sat near the middle of one bench, the twins across from each other at the far end, with a gap between the two groups slightly too deliberate to be accidental. Nadia occupied the far end of the opposite bench, one hand wrapped around a cup, her attention directed somewhere to the left of everyone else. Renne kept her status window open, angled so the panel's surface caught the sun, reading with the economical stillness of someone who did not need to track the room to know what was in it.

Wynn and Petra sat together near the middle of the second bench, close enough that their shoulders were almost touching, with a half-meter of empty bench on either side of them. The deferred attunement made that gap visible in a way no one intended but still sat there.

I took the remaining stretch of bench near Nadia's end and reached for the bread.

Fass was talking low and fast to Coll, gesturing with one hand, and Coll kept interrupting with short responses that made Fass restart. Twice during this, Coll looked up with an expression adjacent to alarm, the look of someone who had encountered a result no one had told them to expect and was working out whether it was good or bad. The linked amplification loop they had demonstrated three days ago in the common corridor had apparently produced new data in today's session.

Tobias ate without apparent investment in the food, his attention moving across the table in the measured way he managed most things. He moved as though he had set a daily budget for extraneous motion and tracked against it. He had closed his panel, which he did at meals while Renne did not.

Renne's bread sat without a provision cloth, without any arrangement showing what she had taken versus what she had left, and yet she had already sorted her portion at the plate's edge: dried fruit off to one side, bread arranged to pick up in a specific order. She had done the same at the two prior meals I had observed closely enough to notice. Whatever her sessions pulled from her, she compensated with minor logistical control over the things she could arrange.

Wynn said something to Petra in a tone too low to carry across the table. Petra nodded without looking up from the clay pot she was ladling from.

Tobias looked at me.

"The evaluation annex," he said. Not a question, not quite a statement, and he let it sit in the middle of the table like something he had set down and was waiting to see if I would pick up.

"Morning session," I said.

"Not the assessment. The pre-assessment."

The distinction mattered to him or he would not have made it. "There's a private addendum in my file. Report to the evaluation annex one hour before the cohort assessment starts."

"Modified parameters," he said.

"Modified parameters," I confirmed.

He turned his cup in a small circle on the stone, just a quarter rotation, then back. "Do you know what the modification involves?"

"No."

"Do you know whether it affects how you participate in the cohort portion?"

The grain preparation was still hot enough to steam. I gave it a beat before answering. "I don't know yet what the modified parameters involve, so I don't know the answer to that question."

He looked at me steadily. "That's the problem."

"Yes."

He did not press it. He was not the kind of person who repeated a point once it had landed, which made the exchange cleaner than most of what passed between people who did not fully trust each other but worked in the same constrained space. He picked up his cup, drank, and returned to the food.

The silence that followed was not hostile. Both of us had said the true thing and arrived at the same answer: not yet, nothing more available.

Fass stopped talking mid-explanation and looked at Coll with a helpless expression. Coll shook their head, made a small gesture with two fingers that continued the demonstration they had been running in the hall all morning, and Fass exhaled through his nose and looked at the ceiling before writing something on his panel.

Petra filled her bowl and set the ladle back in the pot at the exact angle she had found it. Wynn watched this without commenting. Neither of them looked at the other, both of them working through the same unspoken question and neither one willing to be first to raise it out loud.

After the meal wound down, the table cleared mostly through people taking their own plates and cups to the stone ledge at the courtyard's east side, an informal arrangement the annex did not formally require but everyone had converged on. I stayed at the bench a beat longer than necessary, and Nadia came to stand at the table's edge near my end.

She did not sit. She set her cup on the bench across from me and kept her hands free, her habit when she was about to say something she had already thought through.

"Have you read the noticeboard listing? The one for partial-attunement candidates in mandatory assessments."

"No."

She pulled her panel up without making a production of it and read directly. "Partial attunement candidates undergo assessment on theory application and classification-state demonstration rather than affinity-framework output. The assessment will record contribution separately from standard track evaluation."

The grain in my bowl was cooling. I set the spoon down and let the sentence sit.

"Theory application," I said. "That's what the restricted index has been covering."

"Yes. And classification-state demonstration is exactly what your hall behavior is. What the disc is doing, the cord response, the methodology. That's demonstrable output."

"If the board has acknowledged it."

"The session log is the record. If it's reviewed, it's acknowledged." She did not mention Hael's role, and I did not either. We both knew that was the hinge.

"The listing also says the contribution appears as a separate line."

"Which means it sits visible as a distinct line in the cohort record. Not hidden, not averaged in."

"So everyone in the assessment sees there's a separate evaluation track."

"If they're paying attention to the cohort documentation, yes."

Most of the cohort had drifted back into the annex corridor or gathered near the archway, waiting for the end-of-break chime, and the table sat empty between us. "That's the argument for disclosure," I said.

"I'm not telling you to disclose anything," she said. "I'm telling you what the documentation structure produces whether or not you do. If the other candidates can read the cohort record, they already know the assessment will place your contribution in a different category. Whether they know why is a separate question."

That distinction was worth sitting with. The assessment would document the difference regardless. The only variable was whether the cohort went into it knowing what that difference meant.

"It's the structural case," I said.

"Yes."

She picked up her cup and went back toward the archway, unhurried, finishing the conversation the same way she had arrived.

Renne was at the stone ledge stacking cups into the rough holder someone had built from smaller stones in the first few days. I brought my bowl over, and as I set it down she glanced at her panel, not at me.

"The evaluation annex pre-assessment. Does it sit in the cohort's shared record or only in your individual file?"

I opened my status window. The addendum sat in the individual file section, below the primary status entries, in the administrative tier that did not propagate to shared documentation. "Individual file only."

She stacked the last cup. "Good to know."

She did not explain why. She picked up her panel and walked back toward the corridor entrance, unhurried, and I stood at the stone ledge with the cold bowl in my hands, working out that Renne had been tracking which pieces of the cohort's information remained visible to all eight of us and which were not, probably since day one. She had not asked the question to learn the answer. She had asked to confirm data she had already derived.

I set the bowl down and followed the rest of the cohort back inside.

The notification arrived during the return walk, pushing through the status window while I was still in the annex corridor. The panel indicator pulsed once and settled.

COHORT ROSTER: ALL CANDIDATES ACTIVE. DEFERRED ATTUNEMENT SCHEDULING CONFIRMED. SESSIONS WILL TAKE PLACE PRIOR TO COHORT INTEGRATION ASSESSMENT.

I stopped moving.

The corridor was wide enough that stopping did not block anyone. Fass and Coll went past without noticing, mid-sentence again. Tobias passed me with a glance at my panel, which told me he had received the same notification and had already read it.

Wynn and Petra were ahead of me by three or four meters and both had stopped at the same moment I had. They were not looking at their panels. They looked at each other, Wynn's jaw set tight, Petra's hand pressed flat against her own sternum as though checking that something under it still moved the way it should.

Wynn said something to Petra, short, and Petra made a small sound that was not a word. They resumed walking.

My panel was still open. Below the cohort notification, the PARTIAL ATTUNEMENT STATUS entry sat unchanged. Below that, INTERIM TRAINING ACCESS (LIMITED). The individual file addendum: ASSESSMENT PARAMETERS MODIFIED. REPORT TO EVALUATION ANNEX ONE HOUR PRIOR.

The cohort notification had not produced a parallel update in my restricted status tier: no scheduling confirmation, no parallel path moving from partial to incoming. Whatever the board's convening process resolved, it resolved differently for compound-primary restricted candidates than for deferred attunement candidates. My delay was not a scheduling gap waiting to close. The board needed to determine what my classification actually was before it could decide what hall I belonged in.

Wynn and Petra had been waiting on a door that was slow to open. Whatever door existed for my case did not have a frame yet, let alone a scheduled opening.

The corridor let out into the branching point near the orientation wing, and I took the left branch toward the theory track room. The hallway was empty in that direction. The lamp at the far end was unlit at this hour, and the only light came from the high slot windows running the length of the outer wall, thin rectangles of afternoon sun falling across the stone floor in evenly spaced intervals.

The theory track room had the restricted index sitting where I had left it, the page dog-eared at the observer confirmation subsection. I sat down and did not open it. There was nothing new to find in that section, and I had already mapped the cord's behavioral parameters from the morning session.

What the morning had produced was a cleaner version of what I had suspected going into it: three objects, three tiers of my classification file, each one physically representing a different category of what the board was working with. The disc for primary status, current and unresolved. The cord for logged inferences, specifically the compound-primary propagation entry and its interaction properties. The tile for something the file had not yet generated, probably the board convening date confirmation, though that was still a prediction rather than a verified mapping.

Two days remained to find out if that prediction held.

What I had not fully accounted for, sitting in the quiet of the theory room, was what Nadia's noticeboard reading had clarified. Theory application and classification-state demonstration constituted the modified assessment parameters. The cord response and the disc position counted as classification-state demonstration, both observable and already logged, while the restricted index work covered theory application on its own. The real question was whether the board had formally acknowledged the demonstrable components through a Council-level observation, not simply whether I had them, because without that acknowledgment the session log would remain just my own record of events in an unlabeled room, evidence that would not survive a formal board review on its own terms.

Three conditions had to hold. A Council-level observer needed active read access to the file. The session log needed a documented review timestamped after the inference entered the record. A formal notation needed to enter the file within the same administrative cycle as that review. Hael held the first condition. The session log, which I had kept open during both the cord response and every prior disc observation, supplied the raw material for the second. The third required Hael to act on what she had reviewed within the same administrative cycle, and that part of it remained unresolved.

The afternoon break chime would ring in approximately forty minutes. Before that, I needed to run the tile against the one remaining untested entry, CLASSIFICATION REVIEW: BOARD CONVENING DATE CONFIRMED, which did not currently appear in my panel, only anticipated.

The alternative was that the tile test would need to wait for the board convening confirmation to populate, and if that happened after the review window closed, the tile's response or non-response would become data without context, a result outside the framework I had built to interpret it.

I closed the restricted index and left it on the desk.

The hall was where the answers were. The theory track room was where I understood what the answers would mean. Both had been useful. The distinction between them, which I had written down in the session log around day eight as the index requiring prior knowledge before returning information while the hall supplied information through direct physical response, still offered the clearest description I had of what I was working in. The index told me what to look for. The hall showed me what existed.

That meant the cohort assessment, whatever its modified parameters produced for my evaluation, would go better if the hall had already demonstrated the full range of its registration behavior before the board convened. The tile was the remaining open variable. Tobias's question from the table still sat at the back of my attention, unanswered.

He had asked if I knew what my modified parameters involved. I did not, not fully. What Nadia's reading had supplied was the category: theory application and classification-state demonstration. What the tile's response might supply, if the board date confirmation populated before the window closed, was the third tier of the classification record, the one that would tell the board what my configuration actually contained.

The hall did not argue for my case so much as run parallel to it, reflecting the board's own case back in stone and cord fiber and a disc held at eight degrees. Whether anyone with the authority to read that parallel record had looked at the session log was the only part I could not control.

Back in the hall, the lamp sat at mid-reservoir from the morning session. The cord rested at its middle tension when I settled across from the pedestal. My panel sat open to the full status window. The CLASSIFICATION REVIEW entry existed only as a timer line: no confirmed date, no populated flag, the same as it had been that morning.

The cord registered the COMPOUND-PRIMARY RESONANCE PROPAGATION entry the way it had all morning: tension increasing over three to four seconds, holding, then releasing when I moved to a different entry, the pattern repeatable across every attempt. The disc sat undisturbed at eight degrees.

The tile gave nothing, no response to AFFINITY: UNRESOLVED (ACTIVE), no response to the inference log entry, no response to PARTIAL ATTUNEMENT STATUS or INTERIM TRAINING ACCESS (LIMITED), just a smooth grey square on a pedestal that had not moved once across thirteen days of testing.

The board convening date confirmation was not there yet.

I wrote the afternoon session record, dated it, noted the cord's continued response pattern, the tile's continued null result, the disc's unchanged position, and closed the log. The meal had been two hours ago. The rest of the afternoon was mine, restricted index or hall or the theory track room, all of them available and none of them able to produce the one thing the situation needed: a Council-level observer who had read a log currently closed for the night.

Out in the corridor, Fass and Coll went past again, still talking. Coll's voice dropped into what sounded like a calculation, and Fass responded with something that drew a short laugh. They were deep enough into their linked amplification work that the assessment probably served more as a presentation than a test. They had a result to show.

Tobias's door was closed. Renne's sat open a few centimeters, enough to hear pages turning.

My room was at the end of the wing. I went to it, sat at the desk, and opened the restricted index to the section on compound-primary behavior during unresolved classification intervals, not because there was anything new in it, but because the architecture of the thing was still worth sitting with. The section had unlocked on day nine, and whatever the trigger for its availability had been, the index had decided I was ready for the material nine days into a fourteen-day window, which meant the threshold had resolved somewhere between day eight and day nine.

Day nine was the morning after I had first confirmed the cord's null result and logged it with methodology. The session log had been open, the methodology note sat in the record, and the index had unlocked the compound-primary section twelve hours later.

I had noted that coincidence at the time and flagged it as potentially indexical, meaning the index might be reading the session log rather than just the status panel. With Nadia's noticeboard reading still sitting in my head, the flag was worth picking up again. If the index drew on the session log, it read a larger body of information than the panel supplied, which meant the cord response, timestamped in the closed log, might have already triggered the next unlock somewhere in the index's architecture, one I had not found yet.

Eleven sections in the restricted index remained inaccessible. Four displayed as locked. Seven displayed as not yet accessible, a distinction the index made without explaining the difference. The cord response had happened yesterday afternoon.

I went through the table of contents slowly, section by section. Ten of the eleven unavailable sections looked the same as two days ago: locked or not yet accessible, no change in display state. The eleventh was different.

Near the back of the index, in the section labeled CANDIDATE CONFIGURATION: ADVANCED COMPOUND STATES, a subsection that had previously displayed with the same grey shading as the other inaccessible content now read normally, fully accessible, unlocked sometime between the cord test yesterday and this moment.

The section title read: TEMPORAL AFFINITY INTERACTION: PHYSICAL REGISTER OUTPUTS AND BOARD DOCUMENTATION STANDARDS.

The session log had been open during the cord test. The index had unlocked a section on temporal affinity interaction and physical register outputs within one administrative cycle of that test, which was either coincidence or it was not.

I turned to the first page of the subsection and read.

The opening paragraph used administrative register, the same flat institutional language the rest of the index employed when it was conveying framework rather than content. It established that temporal affinity interaction, when present in a compound-primary configuration alongside an unresolved primary designation, produced physical register outputs that did not conform to standard hall behavior profiles. Standard hall behavior responded to single-affinity or dual-affinity configurations with resolved primary classifications. Temporal interaction introduced a third variable: the candidate's current position within a time-bounded administrative process could itself function as a registration trigger, distinct from affinity flag states and inference log entries.

I read that sentence twice: a candidate's position within a time-bounded administrative process could itself function as a registration trigger.

The disc had moved when the AFFINITY: UNRESOLVED (ACTIVE) flag opened in my status window. The cord had responded to the COMPOUND-PRIMARY RESONANCE PROPAGATION inference entry. Across thirteen days of testing against every entry currently present in my panel, the tile had shown nothing. If the tile's trigger category was administrative time position rather than a static flag, it would not respond to any entry that currently existed in the file. It would respond to an entry that reflected where I was in the fourteen-day window itself.

The second paragraph confirmed the mechanism in the flat declarative way the index used when it was done hedging. Temporal affinity interaction produced register sensitivity to countdown interval states: specifically, the ratio of elapsed window time to total window time, rendered as a threshold-crossing event when the ratio passed defined markers. The subsection did not specify the markers. A cross-reference pointed to a section still displaying as locked.

That was the threshold-crossing event. The tile was not waiting for the board convening date confirmation. It was waiting for the window ratio to cross a marker.

The third paragraph: physical register objects responsive to temporal affinity interaction shifted at each threshold crossing, not accumulated over time. Each shift reflected the current interval state, not the cumulative history of the window. This distinguished temporal register behavior from primary and inference-tier register behavior, which reflected persistent flag states and held their positions until the flag changed.

That meant the tile might have already moved. The threshold markers did not appear in this subsection, cross-referenced to a locked section, but if one of the markers fell at a ratio I had already passed, the tile would have shifted and then returned to baseline as the next interval began. I had thirteen days of null results, and I had been treating null as absence of response rather than as an interval state the tile had already transitioned through.

My notes from the first twelve days sat on the desk, stacked under the index. I pulled the stack out and went through the daily cord and tile observations in order. The tile entries were consistent: no movement, no positional change, observer confidence high, methodology noted. On nine of the twelve days I had specifically checked at multiple points during the session. On three days I had checked once, at session open, because the tile had become the lowest-priority pedestal after the cord and disc established their response patterns.

Day seven was one of the single-check days.

That was also the day the index had shifted in a way I had not tracked carefully, two sections moving from not yet accessible to locked, which at the time I interpreted as the index tightening access rather than reclassifying availability. If the index and the hall both drew on session log data, a simultaneous reclassification in the index and a potential tile threshold event on the same day was not a coincidence I could dismiss.

I could not verify what the tile had done on day seven. The observation was a single check at session open, tile in neutral position, no movement noted. If it had shifted at some point during that day and returned to baseline before I re-entered the hall in the afternoon, the record would look exactly the way it already did: a null result.

The fourth paragraph of the subsection addressed what it called re-observation protocols for temporal register objects. Because temporal affinity register objects reflected interval states rather than persistent flags, the protocol advised candidates who had recorded null results at prior threshold-adjacent sessions to conduct a full re-observation sequence during the final interval before window closure, with the session log open throughout. Threshold markers near the window boundary typically appeared within the final forty-eight to seventy-two hours of the review period.

Forty-eight to seventy-two hours. By my own count, closer to a day and a half remained in the fourteen-day window than a full two, unless the board pulled their convening date forward, which the documentation had already suggested possible but rare.

I had been running the wrong test on the tile for thirteen days, testing it against static flag entries when the trigger category was something the documentation had not described until now, because the subsection on temporal affinity interaction had not unlocked until tonight. The index had withheld this information until I had produced the cord response, because the cord response was the prior knowledge threshold the index required before returning the relevant framework.

The tile had not fallen behind. It had been running exactly on schedule, just not the schedule I had been checking it against.

The afternoon light through the high slot windows had gone flat and grey, which put the time past fourth bell. One day and some hours remained until the re-observation window the subsection described. The session log needed to stay open. The observation needed to occur during the final interval. The cord response already sat in the record, and so did the disc position.

If the tile crossed its final threshold marker while the session log remained open and Hael reviewed the log within the same administrative cycle, all three objects in the hall would sit in the observer's formal notation pathway. The disc, the cord, and the tile represented three tiers of the classification record, and all three would now appear in the documentation.

Renne's pages had stopped turning sometime in the last few minutes. The corridor was quiet.

I set the notes back under the index, the stack aligned with the desk edge the way it had been. The oil tin was still in the cupboard below the shelf. Tomorrow's session would run long, and the lamp reservoir from this morning would need refilling before I opened the log.