She spent the rest of the day building a chain of evidence that she already knew she couldn't give to Dr. Cord.
This was an uncomfortable position for someone who had, throughout her career, operated on the principle that the correct response to anomalous data was methodical escalation. She had been, professionally, a scrupulous logger. She had once filed a data quality report about her own analysis, identifying a minor systematic error that had propagated through eighteen months of telemetry processing and affected nothing critical. She had filed it anyway.
She was aware that the principle of scrupulous logging and the principle of *tell no one until you understand what you have* were in tension, and she was aware that the second principle was winning, and she was not entirely comfortable with this.
But she had also, in three contracts and twenty years of work, seen three anomalous findings reach the corporate reporting structure. Two had been reclassified, quietly, in ways that had changed their significance. One had disappeared from the accessible archive entirely.
She had told herself, each time, that she must have misunderstood the corporate reasoning. That there were considerations above her access level that accounted for the reclassifications.
She was now thirty-one days into a six-week window, looking at a signal from a dead probe, and she had stopped being willing to assume that.
---
The Kalephas mission archive was not restricted. It was simply old, and old data in deep-space research occupied the specific category of "complete, no ongoing relevance, accessible to anyone with basic clearance and sufficient patience." Rook had basic clearance and, on this particular day, considerable patience.
She worked through it systematically: mission brief, deployment logs, telemetry summaries by month, anomaly reports.
K-7's anomaly reports were interesting.
Not the final failure report, which she had written and knew well. The reports from the six months before the failure: a series of low-priority flags, the kind that analysts log because the protocol requires it and then file without expectation that anyone will ever read them. K-7 had logged eleven anomaly flags in its last six months. Nine were hardware-variance reports, well within normal parameters. One was an environmental reading — a brief, unexplained gravitational fluctuation in the outer system survey region.
The eleventh was a signal intercept.
Rook read the intercept flag three times.
K-7 had logged a received signal, origin unknown, in the outer system survey region, approximately eight months before its failure. The signal had been analysed as "noise pattern, non-standard, possible stellar phenomenon." It had been flagged, filed, and never followed up.
The signal pattern in the 2147 flag was a close match to the pattern of the signal she had received this morning.
She sat back in her chair.
K-7 had received this signal forty years ago. Had filed it as noise. Had continued its survey work for eight months. Had then undergone a cascade power failure.
And had just sent her a signal using the same pattern.
She thought about what could access a failed probe's communication architecture. She thought about what could encode a specific timestamp and seven words into that architecture and then wait — forty years — for the right moment to transmit.
She thought about the coordinates.
She opened the cartographic database and queried the gravitational fluctuation location from K-7's 2147 anomaly report against the coordinates in this morning's signal.
They were eleven kilometres apart.
She closed the database and stared at her terminal.
Then she opened a new analysis document and began writing the second report — the one she was not going to log through proper channels.
---
Jin found her at 20:00, still at the terminal.
"You've been here for fifteen hours," Jin said.
"Sixteen," Rook said.
Jin pulled up a chair and sat down without being invited, which was one of the things Rook had learned to consider evidence of genuine concern. "What are we looking at?"
Rook turned the terminal toward her and let her read the analysis. She watched Jin's expression move through stages: professional interest, technical attention, and then the specific quality of stillness that careful people had when encountering information that was larger than they were expecting.
"This is the probe that was declared failed in 2147," Jin said.
"Yes."
"This signal —" she scrolled back through the analysis "— you've matched the authentication signature across seventeen verification points."
"Yes."
"And K-7 received a signal with the same pattern forty years ago, before its failure."
"Yes."
Jin was quiet for a moment. "You haven't filed this."
"No."
"Why not?"
Rook took a breath. "Because I've watched three anomaly findings reach the corporate chain in this company, and the track record is not good. And because the message says *do not let them*, and I don't know who *they* refers to or who *us* refers to, and I'm not prepared to put this in a system whose access controls I don't trust until I have a better understanding of what I'm dealing with."
Jin looked at her for a long time.
"That's not how we're supposed to operate," Jin said.
"I know."
"The protocol is clear about anomaly escalation."
"I know that too."
A pause.
"What do you need?" Jin said.
Rook looked at her. "I need someone who's good with physical systems and doesn't default to institutional escalation. I need to understand if there's anything in the station's sensor array that could tell us more about the coordinates. And I need someone who can look at this analysis and tell me if I've made a mistake, because I've been alone with it for sixteen hours and I'm not confident I haven't."
Jin looked at the analysis again.
"You haven't made a mistake," she said. "I can see three places where your verification methodology could be tightened, but none of them change the conclusion." She was quiet. "I have a friend in the deep survey archive at Kalephas Station. Off the books. She might be able to pull the original sensor data from the 2147 intercept."
"That could help," Rook said.
"It could also pull attention."
"I know."
They sat with that.
"Six weeks," Jin said.
"Thirty-three days now."
"And the message."
"*Do not let them,*" Rook said.
Jin nodded slowly. Outside the observation panels, the stars were entirely indifferent, as they always were.
"All right," Jin said. "Let's start with the sensor array."
Rook saved her analysis to the fourth backup location and followed Jin toward the systems bay, and thought about dead probes and forty-year-old signals and the coordinates of a region of space that no one had had a reason to look at.
She thought about *they.*
She did not know, yet, who *they* were.
She was beginning to be afraid that she was going to find out.
