The Root and the Rest of It

The working would either hold or it would not, and she would know which one about four seconds after it was too late to choose differently.

She had said this out loud at the station, in front of Thorne and the Krenyi analyst and the robot, in the flat voice she used for things she did not want misheard as modesty. I do not predict. I attempt, and then I read what the attempt did. Thorne had written it down. That was the correct response and she had thought better of him for it, and it had not made the fact any less true at half past four in the morning with her hands in the wet chalk at the base of the third tree.

Seven trees. The anchor was not in any of them. That had taken her most of the previous day to establish and it was the only part of the work she was certain of: the coupling ran under the root plate, in the space where seven Etheric extensions overlapped and became, for about two metres of depth, a single continuous structure. The entity had not attached itself to a tree. It had attached itself to the overlap — to a thing the grove made together and no single tree could be blamed for.

She had come to admire it, slightly, in the way she admired any well-made joint.

Chalk, thread, and water in a vessel of known volume. The materials were the same ones she had used on the causeway for three years, and the circle she was laying was the same shape as every circle that had failed there for three years. That was the part she had needed to explain twice at the station. The circles had not been wrong. The circles had been correctly rejected by a structure she had been reading through a fog the entity was maintaining. Take the fog away and the same craft, unchanged, met the same structure and was answered.

She set the last thread and put her palm flat against the chalk and did not close her eyes, because closing them added nothing and she had never seen the point of ceremony.

Nothing happened for a while. Then the ground under her hand changed temperature — not warmer, not colder, later, as though the cold had been rescheduled — and she understood that she had four seconds.

He stood where she had put him, at the edge of the root plate, holding a lamp he had been told not to move.

Twelve years of Survey Corps discipline had left him with a reliable instinct for the moment when a procedure passed out of his competence, and this had happened at approximately the point where Elvira knelt down. He knew what he was looking at in the sense that he could have written the incident report. He did not know what he was looking at in any other sense. The distinction had bothered him three days ago. It bothered him less now, which he suspected was itself worth noting somewhere.

She had asked one thing of him before she started, and it had not been what he expected.

"Say the classification out loud when I tell you," she had said. "Not to it. To me."

"Why?"

"Because you have the vocabulary and I have the hands, and last time we tried this the gap between those two things was three years wide."

So he stood with a lamp he was not allowed to move and waited to be asked to say a word he had once withheld from her in a report cleared at a level she had never needed.

"Now," Elvira said.

He said it. Obliger, causeway sector, feeding method and anchor class, the full sentence in the flat committee grammar he had spent a career learning — and the wind, which had been absent all night in the way it was always absent here, came back.

That was all. No sound. No shape between the trunks. Only air moving through a grove where air had not moved, and the lamp flame leaning, at last, in a direction that meant something ordinary.

He did not move the lamp. He noted the time.

LOG ENTRY — Rook-7 Location: Marsh Causeway Dryadic Grove, root-plate perimeter Timestamp: 04:41:19 local / [BOUNDARY DRIFT NOTE: single-source timestamp; confirm with Threshold Station on return]

Primary observation:

The Class II signature is gone.

I want to be precise about what "gone" means, because I have four minutes of record either side of the event and the distinction matters for the file. The signature did not weaken, migrate, or attenuate with distance. It ceased to be present at 04:41:07 local, across all seven specimens simultaneously, in a single sampling interval. My arrays are rated to catch a transition of that magnitude across at least three intervals. They caught it in one.

Elvira's severance was clean. I am recording that judgement in my own voice rather than hers, since she will not claim it and someone should.

Secondary observation, and the one I expect to matter more later: the grove's own Etheric extension is unchanged. The trees are still rooted into the Etheric layer. The overlap structure at the root plate is still a continuous single structure. Whatever was hooked into that overlap is no longer hooked into it, and the overlap does not appear to have noticed.

Adjusted measurement: leaf phase-lag across all seven specimens now reads 0.19 seconds. Pre-intervention figure was 0.22 seconds, held consistently across three nights of observation. The new figure has held for eleven minutes.

I do not have an explanation for the 0.03-second shift. I have logged it rather than accounted for it. Both are permitted; only one is honest here.

Regarding the entity: I have no track. I did not expect one — my arrays read anchor coupling, not motion — but I want the absence stated rather than implied, so that no one reading this file later mistakes my silence for a search that came back empty.

Thorne noted the time correctly and did not move the lamp.

END LOG ENTRY

LOG ENTRY — Rook-7 (SUPPLEMENTARY — MODEL WITHDRAWAL) Location: Threshold Station, Operations Timestamp: 09:02:44 station time

I am withdrawing the two-source boundary model I filed three days ago. I am doing this in the same file where I proposed it, at the same level of detail, because a model that is retracted quietly is a model that goes on being cited.

The prediction was specific. If the grove's permanent Etheric coupling had been maintaining a static perturbation in the local coherence field, and the oscillating boundary anomaly had been superimposed on it, then removing the entity's anchor from that coupling should have produced an immediate, measurable reduction in Threshold Station's chronometer discrepancy. I gave the expected magnitude in the original filing. I stand by having given it; an unfalsifiable model would have been worse than a wrong one.

The discrepancy this morning is forty-two seconds.

It was forty-two seconds yesterday. It was forty-two seconds three days ago, and eleven years ago, and in the six prior postings' reports that Thorne released to the survey archive on Tuesday. Four hours after a clean severance 1.3 kilometres away, it has not moved by the width of my measurement error.

Conclusions available to me:

One. The grove was never a contributing cause of the station drift. My correlation was real and my causal reading of it was wrong — the condensation pattern in the Year 5 report remains consistent with Etheric thermal differential, which now tells us the grove radiates, not that the grove drifts us.

Two. The static component exists and something other than the grove is maintaining it.

I cannot distinguish between these with current instrumentation, and I decline to pick the more interesting one.

What the file should now say, and what I have written into it: the forty-second discrepancy at Threshold Station is unexplained. It was unexplained before I proposed the grove. It is unexplained again. The interval in which it appeared explained lasted three days and produced one correct intervention for the wrong stated reason, which I note is not the same thing as a wasted three days, though it is close enough that I would prefer not to repeat it.

FLAGGING: boundary model reverted to open. Recommend the drift be treated as an active unknown in all forward planning. END LOG ENTRY

DETECTIVE AGENCY LOG — Aldera, Field Unit Assignment: Boundary Anomaly Observation, Marsh Causeway Sector Status: Closing — sector clear

Sector swept at first light. Four hours' observation, full perimeter, including the two places I would use if I were something that had just lost its grip on a grove.

Nothing is using them.

I record this as an absence, not a departure. I did not see it leave. Nobody did — the robot's arrays do not track motion and the humans were looking at a lamp. What I can state is that the causeway's texture, which I have logged nightly for three years and described in this file as a wrongness I could not itemise, is not present this morning, and that the difference is large enough that I noticed it before I reached the stones.

The grove is a separate finding and I am filing it separately, because the two will otherwise get read as one result.

The trees are unchanged in every respect a survey instrument measures except one, and the robot has that figure. What I will add to it is this: they were not friendly before and they are not friendly now. The pale ones grew at a fault line three centuries before anyone laid a causeway across it, and something used the joint they make together as a hook for a long time, and this morning it does not. None of that is the same as the grove being on our side of anything. Field discipline for this sector should not relax. I expect it will.

Case status: the feeding loop is closed and the anchor is severed, by two separate interventions on two separate nights, and I want the record to show they were separate. The first was clarity. The second was craft. Neither would have worked alone and neither was mine.

I have been on this assignment for three years. The part of it I was actually here for ended on Tuesday when somebody asked me a question.

I am not closing my file on the sector. I am closing my file on the entity.

END LOG

She stayed at the grove until it was properly light, because there was no reason to and she wanted to.

The reeds had come back. Not grown back — there had been reeds all along — but come back to being the kind of thing a person walks past without checking. She had not known she had been checking. Three years is long enough to stop being able to tell the difference between a landscape and a habit.

Thorne came to find her at about eight, with a cup of something from the outpost stove and the survey slate under his arm, and asked, in the careful way he asked everything now, whether she wanted the morning's numbers.

"The station drift didn't move," she said.

"No."

"Good."

He looked at her.

"Not good that it didn't move," she said. "Good that you told me at eight in the morning instead of in a report I would have read in three years." She took the cup. "Say the rest of it."

So he said the rest of it: the model withdrawn, the discrepancy back to being an open question, the honest sentence that the thing they had fixed and the thing that was wrong with Threshold Station appeared, on this morning's evidence, to have nothing to do with each other.

"That's worse," she said.

"That's worse," he agreed.

She drank the tea. Out on the root plate the chalk was already going grey at the edges where the dew had taken it, and by evening there would be no sign that anyone had knelt there at all, which was correct — a working that needed a monument was a working that had not finished.

Three years of circles on the causeway, every one of them correctly rejected, and this one had held — and it had held because a man read out a classification and a cat answered a question she had waited three years to be asked. He had called it an Obliger, in the committee's word; she had called it a Court-Fae for three years and would go on doing so, because the word had never been what was wrong. Elvira had done the part with her hands. She was clear about which part that was.

What she was not clear about, and noted in her own record that evening in the plain vocabulary she insisted on, was the question underneath all of it, which nobody had asked yet and which the night's work had not touched.

Something had hooked itself into a joint that seven trees made together, and it had held there for three centuries, and a self-taught practitioner with chalk and thread had taken it out in one attempt.

She wrote: That should have been harder. Find out why it wasn't.