The folder sat on the worktable between them. Kael didn't reach for it. He looked at Illen instead, at the particular stillness of a man who'd prepared for this conversation and was deciding how much preparation to show.
"How long," Kael said.
"Confirmed six months ago. Suspected three before that." Illen didn't elaborate on what confirmation required. "The anchor marking orientation is the primary indicator. The temporal divergence signature in channel architecture development is secondary — the flag is automatic, the comparison has to be done manually." He pulled the folder to his side and opened it to a marked page. Dense figures, channel-architecture readouts, blue marks scattered across the first ten pages. "Those marks indicate parameters that match the fourth regressor's baseline profile. The ones matching yours specifically."
Almost every line was blue.
Kael looked at the figures for a moment, then at the window. Outside, the canyon district's rock walls held the cold from the lower altitudes even at midmorning.
"Tell me about the others," he said.
Illen opened to a handwritten section near the front of the folder. Notes in the margins of a printed document. Clean, dense.
"The first regressor," he said. "Male. We estimate he arrived from three to seven years post-Awakening. Knowledge set suggests a functional but not extensive hunter background." He looked at the notes. "He moved immediately. In the first month he'd redirected four dungeon resources and altered three social trajectories that Maren identified as consequential. The Chronos Cult detected the anomaly at month three. The monitoring signature terminated. Complete cessation."
Not degradation. Not change. Gone.
"Three months," Kael said.
"Three months." Illen turned the page. "The second. Female. She'd reconstructed what the first one did wrong — we think there's some shared reference architecture in the regression mechanism, information that passes between instances somehow. She arrived intending to observe before acting. Changed nothing deliberately. The strategy was patience." He paused. "Month eleven. Same result. Non-action is still presence. Being in the timeline is enough to generate the signal."
Eleven months. Kael had eighteen.
"The third lasted three years. This one adapted well. Built secondary information networks to compensate as original-timeline knowledge degraded. Survived two Chronos Cult proximity events through careful profile management. Achieved significant power development by the end of year two." Illen's voice stayed flat, clinical. "At year three, they trusted someone. Someone reliable in their original timeline. In this timeline, the same person — same face, same history, same apparent motivation — but one structural difference. In the original timeline, this person's loyalty was held in place by a debt. The debt didn't exist here. The third regressor didn't know to check what the loyalty was built on."
"The monitoring signature terminated over several days," Kael said. He'd caught the wording the first time.
"Three days, yes."
Not clean. Whatever happened to the third one, it wasn't instant.
He thought about the list of people he was making decisions based on. Rowan. Elara. Castellan. He knew why each of them was reliable in the original timeline. He hadn't asked, in most cases, whether the same mechanism existed in this one.
He was going to ask.
"The fourth," he said.
Illen set the folder down.
"Five years. The fourth regressor's trajectory is the one that matters for your situation — it's what Maren has been building the comparison profile from for the past eight months." He picked up one of the clay models from the table. A compact structure, densely connected, different from the others in the way finished work looks different from work in progress. "Knowledge-decay management. Secondary intelligence networks. Patient power development. Careful handling of known betrayal trajectories." He set the model down. "Five years. Further than any previous instance. Further than anyone, as far as we know."
"How did it end."
Illen laced his fingers together on the worktable.
"A choice," he said. "The monitoring data in the final week shows a controlled process. Not termination from external pressure, not Chronos Cult proximity, not the Watcher — which Maren believes manifested and was encountered in the fourth year." He looked at Kael steadily. "The fourth regressor understood something near the end. About the regression's cost. For the timeline, for themselves, or both — the monitoring data doesn't tell us which. They made a decision and the signature stopped." A pause. "Maren thinks they looked at what the Watcher was asking and answered honestly."
"What question."
"Whether this was worth the cost to the timeline of making it worth it." He held Kael's gaze. "The answer they gave was no."
The research station was quiet. The wire-and-clay models on the walls caught the light from the upper windows. Outside, somewhere in the canyon below, a wind moved through rock.
Kael was eighteen months in. The first two had failed by month three and month eleven. The third had been brought down in year three by a misread trust structure. The fourth had made it five years and then stopped.
Four attempts. Four failures by different routes. And somewhere above this worktable, a folder full of data saying that his divergence accumulation matched the fourth one's pattern closely enough to be notable.
"I'm not going to stop," he said.
Illen almost smiled. "I know. The divergence accumulation rate and the power development velocity both indicate a different priority structure from the fourth regressor." He opened the folder again. "Which is exactly why I invited you here."
---
The negotiation took twenty minutes.
Kael offered the divergence mapping he'd documented: eighteen months of instances where original-timeline knowledge had proved wrong or incomplete, the rate at which degradation was accelerating, the shape of what had changed. Not operational intelligence on specific individuals. The pattern.
Illen offered the fourth regressor's full behavioral record through the point the monitoring data ended. Maren's analysis of the failure structure. The complete force-fracture training methodology — Cruz's approach, the inversion technique, the theoretical framework Maren had built on top of Illen's published research.
"One more thing," Kael said.
"If I can provide it."
"The anchor marking." He looked around the research station. "It's been orienting toward this location for two months. Started when I began researching Maren Voss and reading your papers." He paused. "It's not pointing at the data. It's not pointing at you. It's pointing at something else in this room."
Illen stood without speaking and walked to a cabinet on the far wall. He opened it. Middle shelf: a small stone, irregular, worn or worked — difficult to tell which. Hand-sized. Nothing remarkable about it except its presence on an otherwise empty shelf in a researcher's cabinet.
The anchor marking fixed on it with the precision of a compass finding north.
"The fourth regressor left it," Illen said. "They brought it the last time they came here — the week before the monitoring signature stopped. Left it on that shelf without explaining why." He closed the cabinet. "Maren calls it a relay point. A marker in the regression's orientation architecture, left by one instance for the next to find."
"They expected someone to follow."
"They expected the regression to happen again. That the mechanism wouldn't end with them." He looked at the cabinet. "It was either optimism or information."
Kael looked at the closed cabinet door.
The fourth regressor had sat somewhere like this, at the end of five years, understood something that made them stop, and before they stopped they'd put a stone on a shelf. So the next one would find this research station, this data, this record of what had come before.
He didn't have a word for what that was. He sat with it for a moment.
They exchanged contact information. Kael agreed to a weekly check-in at the research station starting the following Wednesday. He would provide the divergence mapping incrementally. Illen would provide the fourth regressor's record in sections, in order, the way you give someone a map of a territory they're traveling through.
"The same time next week," Illen said.
"If I can manage it."
"The shoulder."
"That's recovering." He stood. "The fourth regressor's first six months. Lead with that section when you send the record."
"Because that's the period where your data is most comparable."
"Yes."
Illen nodded. He was already making notes in the folder's margins.
One more question came to Kael at the door, not planned.
"The fourth regressor," he said. "Did they know about you. Before the end."
Illen looked up from the notes.
"They trusted me," he said. He said it simply, without claiming credit for it. "That's the only reason I have the full record. They brought the stone and sat where you're sitting now and told me what they'd understood." A pause. "It was the first time in six years of monitoring that I'd spoken directly to one of the instances." He looked at the folder. "I didn't know how to tell them I'd rather they'd survived."
Kael looked at him.
"The weekly check-in," he said. "Wednesday."
"Wednesday," Illen said.
---
He was back in the city by early afternoon.
Rowan had sent two messages: the social-layer corridor analysis (stable, no new amplification, Dorian's framing holding without active intervention) and a note about a dungeon opening in the outer district. First appearance in the regional records. No original-timeline entry in Kael's memory.
He filed both and went upstairs.
Yara Song was asleep on the floor outside his door, back against the wall, the repaired backpack close enough to grab. She woke at his footstep — practiced, minimal, fully present within a breath.
"You were gone all morning," she said.
"I was."
She followed him in without being asked, which was different from Monday. She'd made some decision between Monday and today about what this was.
He sat across from her and looked at the dormant architecture.
It was stronger than Monday. Measurably, perceptibly stronger — not in size but in density, in the coiled quality of something approaching the threshold where a class manifested. As if paying attention to it was doing something, as if being in the room with someone who could read it had oriented the architecture toward its own awakening.
"Walk me through Marcus's assessment again," he said. "All three density readings this time."
She gave him the first two without hesitation. The third she held for six seconds.
"Triple standard baseline," she said. "Pre-awakening."
"I know. I calculated it from the other two." He kept his voice even. "What I need now is the structure underneath the density. Not numbers — what it feels like from inside when you pay attention to it."
A pause.
"Marcus didn't measure that."
"No. Describe it anyway."
She looked at the canal through the window. The light was moving, the afternoon shift of it across the district's rooftops.
"Pressure," she said. "But also direction. Both at the same time. Like something that knows where it wants to go and is pushing to get there."
He sat with that for a long moment.
Both simultaneously. Direction and pressure. That described a class architecture he didn't have a framework for — not in the original timeline, not in eighteen months of this one. Illen's force-fracture work described either direction or accumulation pressure, not both occurring as a single function. This was either an extension of that framework he couldn't see yet, or it was something outside the framework entirely.
"Come back tomorrow," he said. "I need Rowan here when we continue this. He has an analysis ability that might be able to map the structure."
"The one who does class-development studies." She'd done homework on him.
"Yes."
She stood and picked up the backpack in one motion.
"Most people who find something unusual in a pre-awakening kid's architecture flag it to the Association and step back," she said. "That's what the practitioner before Marcus did. Marcus flagged it differently."
"Marcus has different reasons for his decisions." He looked at her. "Come back tomorrow morning. Before eight."
A pause. Reading him.
"Okay," she said.
She left.
He stayed in the chair and looked at the canal and thought about triple-density architecture that pointed in two directions at once, and about a stone on a shelf in a canyon research station, and about the fourth regressor who had sat somewhere like that stone and decided the answer was no.
The anchor marking wasn't pointing at the research station anymore.
It was pointing at the door.
