The autopsy began at eleven minutes past midnight.

Lucan preferred this hour for the work that required the deepest stillness -- when Verge Station Nine had settled into its structural quiet, when the ventilation system ran at minimum and the only sounds were his own breathing through the filtered mask and the faint, precise click of his instruments against the examination tray. He had never been able to explain this preference to anyone who asked, and he had eventually stopped trying. Some people required quiet to think. He required a particular quality of quiet: the kind that came only when the world outside had temporarily stopped expecting things of him.

The man on the table had been forty-three years old. Stage IV. He had stopped responding to palliative treatment eleven days ago and had been transferred to the station's small clinical unit -- not for treatment, there was nothing left to treat, but because the station's proximity to the surface kept the Lumen Mycelia in his tissue metabolically active longer after death. More useful for study. Lucan had never been entirely comfortable with how readily that thought came to him, but he had also found no satisfactory alternative framework. The samples taken from the deceased reached people the living could not otherwise reach. He held both of these truths at the same time and did not allow either to erase the other.

He made the first incision with deliberate care, angling the blade to minimize compression of the affected bronchial tissue. The chest cavity opened with the characteristic resistance of advanced fibrosis -- the Lucent Rot in its later stages replaced viable pulmonary tissue with a dense, inelastic material that looked under magnification like frosted glass, rigid where living tissue should have been flexible. He worked around it methodically, separating layers with the practiced patience of someone who had learned a long time ago that hurrying an autopsy was the fastest route to missing the thing you were looking for.

He positioned the examination scope and leaned in.

Blue.

Not the dramatic, pulse-like luminescence illustrated in the older textbooks, drawn by artists working from descriptions rather than direct observation. This was subtler -- a faint bioluminescent threading through the bronchial walls, visible only when he adjusted the scope's filter to suppress ambient light interference. The Lumen Mycelia had colonized the full bronchial tree, from the main carina down through every tertiary branch and into the smallest accessible passages. Under ordinary light you would see only the scarring, the collapsed alveolar structures, the grey-brown texture of tissue that had lost its function long before it lost its shape. Under the scope, the infection revealed itself: a living network, patient and thorough, beautiful in the particular way that things which destroy slowly are often beautiful.

He noted the depth of colonization on the examination form. Cross-referenced the findings against the intake documentation.

Then he went still.

Stage IV, confirmed bilateral, at twenty-two months from initial diagnosis. The standard progression model for Stage II onset to Stage IV threshold was thirty-one months, based on population data from four decades of systematic tracking. This man had crossed that threshold thirteen weeks early. Lucan ran the calculation again, checking for arithmetic error. He ran it a third time with different parameters. The result held.

He set down the scope and picked up his pen.

He did not use the station's voice recorder for clinical notes. He had tried it early in his career and found that speaking findings aloud gave them a provisional quality -- as though voicing a thing made it subject to revision. Written in his own hand, in the small, even script he had developed over years of filling research journals, the numbers were simply true. He wrote:

Bronchial colonization depth: Grade 4, complete bilateral. Stage progression timeline: 22 months II-to-IV (projected: 31 months). Delta: minus 13 weeks versus current standard model.

He looked at what he had written. Then he wrote below it, somewhat smaller:

Mutation rate: plus 23 percent versus projected. Timeline compressed.

He remained at the examination table for a while after that. The lamp threw its clean circle of light over the body, the instrument tray, and his own gloved hands. Beyond the circle, the room was the darkness of a place built underground, without windows, where the concept of night and day existed only as convention and the only natural light anyone had ever seen was through the filtered goggles they wore on the surface. He was accustomed to this. He had been working underground long enough that the absence of natural light no longer registered the way it did for people who had recently come down -- that low-grade disorientation they carried, the restlessness, the excessive attention to artificial lighting levels as though by monitoring the ceiling fixtures they could reconstruct something the rock above them had taken away.

What he was registering was the number.

Twenty-three percent above projected. Applied to the progression model. Applied to the specific patient history he knew better than any other. He did the arithmetic automatically, the way he always did, the way he could not seem to stop doing even when he would have preferred not to know the result. Anwen's Stage II timeline shortened from twenty-four months to approximately seventeen months and three weeks.

Seventeen months was not enough time for the synthesis approach he was currently developing.

He finished the examination with methodical efficiency. Collected the tissue samples he needed, sealed and labeled each vial in his own handwriting rather than using the print system. Washed his instruments in the specified solution and laid them in the correct sequence in the drying rack. Stripped the examination gloves and disposed of them correctly. There was a correct way to do each of these things and he did them correctly, because correctness was something he could control and he was constitutionally unable to leave a controllable thing in an uncontrolled state. This was a quality that had served him in research and occasionally in the rest of his life and that he had long since stopped examining as a character trait and simply accepted as a condition.

The communication terminal across the room chimed as he was sealing the last sample vial.

He recognized the number immediately. Bellhaven General Clinical Unit, Floor Seven -- the long-term care floor. He had had this number stored for seven weeks. He had called it forty-one times. They had called him twice. He stood with his hand near the answer interface and let it ring through three cycles, four, five, until the automated protocol transferred the call to his station's recording system. Then he walked across the room and played the recording.

Dr. Rhyse. This is Dr. Ohen from the Bellhaven long-term care unit. We have Anwen's most recent imaging results, which confirm bilateral Stage II progression consistent with active Lumen colonization across both lower lobes. We'd like to discuss transitioning to the modified palliative support protocol and reviewing long-term care options at your earliest convenience -- if you could contact us to schedule a consultation, we'd appreciate it. Thank you.

He stopped the recording.

Palliative. He understood the word with complete clinical precision and he also understood what it meant in practice, which was that the person using it had concluded that addressing the underlying cause was no longer productive and had redirected their energy toward making the current condition as manageable as possible. He had heard it in clinical corridors before. He had been twelve years old, standing in a different corridor, holding his mother's hand while the doctors spoke to each other in the particular lowered register they used when they thought a child was not listening carefully enough to understand. He had gone home that night and looked up every word he hadn't understood, writing them in a notebook because he was the kind of child who wrote things down, and the definitions had assembled themselves into a meaning much simpler than the medical language: we have stopped trying to fix it.

He was not going to stop trying to fix it.

He opened his Luminase-R research file. Four hundred and sixty-two pages. Notes, failed trials, recalibrations, the dead ends he had followed for months before recognizing them as dead, and the single promising result from three months ago that he had been unable to replicate despite every variation in approach he could design. The theoretical mechanism was sound -- he had verified this through careful, indirect consultation, framing it as peer review on adjacent research to avoid drawing attention to what he was actually pursuing. The underlying biology was correct. The challenge was synthesis. To produce Luminase-R in quantities sufficient for any clinical application, he needed a biological source. Something that produced the protein naturally, from which he could derive both the structural model and the foundational material for replication.

He did not have that yet.

He worked for two and a half hours without making progress, which was most nights, and which he had long since stopped interpreting as failure. Failure implied foreclosure. What he was doing was the accumulation of necessary negative data -- the systematic elimination of approaches that did not work, which narrowed the field, which was its own kind of progress. He had been doing this for sixteen years. He expected to continue until something changed.

At four in the morning he put down his pen and looked at the photograph on the corner of his desk.

Anwen at eighteen. Laughing at something outside the frame, her head turned so the picture caught mostly profile -- the line of her jaw, her hair pushed back, the particular brightness she had when something genuinely surprised her into amusement. She had been well then. Not entirely well -- the first indications had come at sixteen and by eighteen there were monitoring schedules and quarterly imaging and the careful language of early management -- but well enough to look like that. Unguarded. As though something could still catch her off guard and the catching was pleasant.

He did not look at the photograph often. He had placed it on the corner of his desk three months ago as a deliberate act -- a reminder of function, not sentiment -- and had since kept it at the periphery of his attention. Tonight he looked at it for longer than usual. He was not sure what he was looking for. The photograph did not change.

He opened his research journal and wrote a dated entry:

Sample 44, Stage IV, 22 months from diagnosis. Mutation rate plus 23 percent versus projected model. Timeline compressed significantly. Need minimum eight additional samples from varied progression stages before I can establish statistical significance. Cannot afford to wait for eight additional samples.

He looked at the last sentence for a moment. Then, below the clinical entry, in the same handwriting but somehow different in register -- as if written by the same person in a different state, the state that came at four in the morning after an autopsy and a voicemail and two and a half hours of work that had not moved anything forward:

I need more time.

He sat for a while looking at what he had written. Verge Station Nine was quiet around him in the way it was always quiet after midnight -- not the silence of a place where nothing moved, but the particular stillness of a system running below its operating threshold, conserving, maintaining, waiting for the circumstances that would require it to run at capacity again. He had chosen this station for its surface proximity and its equipment, and these were the true reasons. He had also chosen it because the nearest residential unit was four kilometers underground, because the access road was difficult enough that casual visits required effort and advance notice, because when he was here he could work at the hours the research required without the continuous low-level expenditure of social attention that group living demanded of him. He had known, when he chose it, that these were also reasons. He had examined them carefully enough to confirm that they were useful and not examined them carefully enough to ask whether useful and correct were the same thing. Anwen's letter had used the word underneath. He was not yet ready to look at what that meant.

He closed the journal. Put it in the desk drawer. Stood up and stretched until his spine cracked in two places and then a third, which he interpreted as the sound of a body that had spent too many hours in the same position. He made coffee. The pre-dawn collection window was in approximately two hours and he intended to use it, because the sample quality in the early morning hours at Verge Nine's elevation was consistently superior to what he could collect at any other time, and because the coffee was something he could do while he waited, and because waiting without doing something with his hands was harder than it used to be.

---

At half past five he was on the surface in full kit, moving through the grey pre-dawn quiet toward the third collection site northeast of the access point. The ruins of the old surface settlements surrounded Verge Nine in their familiar arrangement of collapsed structures and exposed foundation grids, rendered into strange topography by decades of unmaintained weathering. He moved through this terrain without conscious navigation, the familiar turns and clearances stored in muscle memory that had been laid down over three years of the same walk at the same hours, which left the thinking part of his attention available for other things.

He was six minutes into the collection protocol at the third site when his monitoring equipment flagged an anomaly.

Movement signature. Logged the previous day, at the edge of the sensor network's coverage area. Human-scale stride pattern originating from the northeast quadrant, approximately three hundred meters from the station access point. Irregular spacing -- the signature of someone moving quickly but not efficiently, accommodating either injury or terrain they were not reading well. One individual. Heading toward the old transit infrastructure ruins that occupied the northeast section of the surface survey grid. He had cleared the alert the previous day in the distraction of a sample equilibration that had demanded his attention at the wrong moment.

He collected the remaining material from the current site and sealed the container. Stood and looked northeast.

The transit ruins were approximately three hundred meters away and within his planned route. He had been intending to pass through the western edge of the sector anyway. He adjusted his bearing slightly and followed the northeast bearing at the pace he used when he did not know what he was moving toward but had decided it was worth finding out.

Inside the largest structure, where the roof section was most intact and the interior shadows deepest, he found the line of dried blood on the floor first -- rust-brown, not fresh. A thin line of it leading from the collapsed wall section toward the far end of the space. And at the far end, against the wall where the sightlines to the entrance were longest, a woman sitting with her knees drawn up and a utility knife in her lap, held with the easy grip of someone for whom this was a default posture rather than a readied threat.

Her mask was visibly compromised. He could see the fracture at the left seal junction from four meters away, a hairline crack that would have been allowing ambient air in for hours.

Her chest was rising and falling with an evenness that was completely unremarkable.

He stood in the entrance gap and ran the relevant data through the relevant framework and arrived at a result that the framework did not have a category for. The ambient biopressure at this location and this hour was sufficient to cause observable Stage I symptoms within ninety minutes in an adult with no mask protection. She had been here, by the evidence of the dried blood and the footprint trail, for many more hours than that.

She opened her eyes.

They catalogued him quickly and thoroughly -- exits first, she located them all before her gaze moved to his equipment, then his posture, then his face behind the mask and goggles. She completed her assessment before he had completed his.

From somewhere to the northeast: the unmistakable harmonic of Recovery Corps patrol ventilation systems, moving in a search pattern. He estimated four to five hundred meters, closing.

"There is a research station two hundred meters southwest of this location," he said. "The surface entrance is concealed. I can take you there, or I can leave, and in approximately seven minutes that patrol will complete their sweep and arrive at this sector."

She looked at him. The cataloguing look, not the trusting look -- she was not deciding whether to trust him, she was deciding whether the alternative was worse. He recognized the calculation because it was the same one he made in every situation where the available options were all unsatisfactory.

"Who are you?" she said.

"Lucan Rhyse. Pathologist. I work at the station."

A pause. Three seconds of deliberation that felt like more.

"Move," she said, and stood.

He moved. She followed him, at a distance she chose and maintained herself -- approximately two and a half meters, close enough to keep pace, far enough to preserve options. He did not comment on the distance. He moved efficiently and without unnecessary variation in direction, which was how he had learned to move on the surface: as if the route were already decided and the only task was to complete it.

Behind him, at two and a half meters, he heard her footsteps on the broken surface of the ruins, and he thought about a chest that rose and fell without distress in ambient biopressure that should have been causing distress, and he thought about what that meant, and about seventeen months, and about the protein he had not been able to find a source for.

He thought about these things and he walked toward the station, and whatever was going to happen next, he was going toward it rather than away from it, which was the only approach he had ever known.