The ruins smelled wrong.
This was Osrath Venn's first note in the survey log, not a measurement, not a topology reading, not the kind of observation that institutional documentation protocols were designed to accommodate, but the first true thing he had encountered at the site and he was not in the habit of omitting true things from his logs regardless of their format. Dawnholt Hold, Survey Day 1. The ruins smell wrong. Not of fire. Not of blood. Not of the specific mineral signature of Wound-bleed, which I have documented at fourteen active sites and which I would recognise. Something else. Something I do not have a word for and which I will attempt to characterise more precisely as the survey proceeds.
He had been at the site since the morning of the fifteenth, eight hours after a coastal patrol had found the cliff empty and raised the alarm and he had been working without pause since his arrival except for the forty minutes he had spent interviewing the patrol's commander, who had stood at the cliff's edge looking at the cold stone below and repeated, with the patient desperation of someone hoping that saying a thing more times would make it make more sense, that the garrison had been forty soldiers and that forty soldiers did not simply stop existing.
Osrath had agreed that they did not. He had not said what he was beginning to think had happened instead.
He was forty-nine years old and he had been a field practitioner of the Pale Cartographers for twenty-three of those years. He had surveyed Wound-Thick sites on three continents. He had been present at two Wound-site activations, events rare enough that most Cartographers completed entire careers without witnessing one, and he had written the foundational survey methodology for what the institution called Category Three Anomalous Topology Events, which was the institutional language for sites where the Wound's behaviour could not be fully accounted for by existing frameworks. He was, in the specific and narrow domain of reading what the Wound did to the places it touched, one of the most capable practitioners the institution had.
He could not read what the Wound had done to Dawnholt.
This was not false modesty. This was the precise measurement of a practitioner who had pressed his Marrow-Absence against the site's residue for eight hours and received, in return, a topology reading that did not correspond to any Wound-event category in the institution's three-thousand-year archive of documented events. The closest match, the reading that corresponded most closely to what Osrath was measuring, was a Category Zero designation, which was not a category at all but the placeholder the archive used for events whose documentation was incomplete. Events from before the Covenant's systematic survey program had begun. Events from the first century, or the period before the first century, when the Wound had still been new enough that the people living near it had not yet developed the vocabulary to describe what they were experiencing.
He was reading a Wound-event from the archive's blank pages.
Day 1, Hour 9, he wrote. The topology of the destruction site does not conform to any documented Wound-event pattern. The residue signature at the hold's foundation is not consistent with Wound-bleed from below, there is no substrate eruption pattern, no lateral expansion, no evidence of Thrum-pressure overwhelming the Weft-barrier in the standard fashion. What is present is a directional signature that I am having difficulty characterising precisely. It moves vertically. Not upward. Not downward. In a direction that I do not have an axis for in the standard topology notation. I am going to need the restricted archive's pre-Covenant survey documents. I am requesting priority access.
The restricted archive request would take four days to process in normal circumstances. Osrath had filed it as Priority One, which reduced the processing time to forty-eight hours and required a senior council member's counter-signature. He had also included, in the request, a description of the topology reading that he expected would produce that counter-signature within the hour once the right person read it.
He had not been wrong. The counter-signature had arrived in sixty-three minutes, from a council member whose name Osrath did not recognise, which was itself an observation worth noting, he knew most of the senior council by name.
He walked the perimeter of the ruins again.
Dawnholt Hold had been built into the cliff-face in the way of all Pale Coast architecture, carved back into the limestone rather than constructed forward from it, the hold's exterior face flush with the cliff's edge, its interior rooms extending horizontally into the rock. What remained of it was the exterior limestone, largely intact. The carved rooms behind it were not intact. They were not collapsed, which would have been the expected consequence of a structural failure. They were not burned, which would have been the expected consequence of fire. They were empty in a way that stone rooms were not supposed to be empty, scoured, was the word Osrath kept returning to, though it was not quite right. The rooms had not been emptied of their contents. The contents had not been moved. They had simply ceased to be in the rooms, along with every person who had been in the rooms when whatever had happened had happened.
The furniture's impressions were still in the floor. The worn patches where feet had repeatedly crossed thresholds were still visible. A hook on a wall in what had been the garrison's equipment room still had the shape of something hanging from it, a ghost of weight in the iron, a residue of presence that Osrath could read as easily as he could read text, that told him a sword had hung there for approximately three years before disappearing on the night of the fourteenth along with the soldier who had carried it and the thirty-nine other people who had been in the hold at the time.
He stood in the equipment room for a long time.
Hour 11, he wrote. The residue of the garrison's equipment is present throughout the interior rooms. Emotional and physical residue consistent with normal habitation, meals, rest, routine maintenance, the specific quality of boredom that long postings accumulate in walls over time. The residue cuts off at a precise moment. Not gradually, as in the case of evacuation or fire or structural failure, where residue patterns typically show the emotional texture of crisis, fear, urgency, pain. There is no fear residue. There is no pain residue. There is no residue consistent with awareness that something was happening.
He stopped writing. He looked at his notation.
They did not know, he wrote, more slowly. Whatever occurred, it occurred without warning and without duration. The garrison did not have time to be afraid.
He closed his survey log. He opened it again almost immediately, because the thought that had followed the previous one was important enough to document even though it was the kind of thought that institutional documentation was not designed to accommodate.
I do not know if this is a mercy, he wrote, or the most precise description of what happened that I am capable of producing.
He found the anomaly at the fourteenth hour, in the cliff's eastern face below the hold's foundation.
He had not been looking for it specifically. He had been extending his topology survey downward along the cliff, standard procedure for sites where Wound-bleed was suspected to have originated in the substrate, when his Pallor perception caught something in the limestone that was not limestone. Not an object. Not a void. Something in the rock's residue that did not belong to the cliff's geological history, something that had been present in the limestone long enough to leave an impression but that had not been present at the time of the hold's construction, because the hold's construction crew would have found it and the construction records would have noted it.
He pressed his Marrow-Absence against the cliff-face and read.
The impression in the rock was approximately three feet in diameter and approximately two inches deep, a curved depression, almost perfectly circular, whose edges had the specific quality of something that had been there long enough to become part of the cliff's geological character without ever being part of its geology. As if the rock had slowly grown around the impression rather than the impression having been carved into the rock.
He read what the rock remembered.
It remembered this shape for a very long time. Longer than the hold above it. Longer than the village below it. Longer, his Pallor perception told him with the uncertain confidence of a measurement at the edge of its calibration range, than the Pale Cartographers as an institution, longer than the Hollow Covenant's three thousand years of recorded survey.
And it remembered that the shape had been empty for most of that time. A depression in the cliff-face that nothing had occupied. And then, four months ago, something had occupied it. And then, on the night of the fourteenth, the something had left.
He sat on the cliff-face in the autumn wind and he looked at the depression in the limestone and he wrote, very carefully, in his survey log:
Hour 14. Eastern cliff-face, 40 feet below hold foundation, accessible via the cliff path. Circular depression, approximately 3 feet diameter, of indeterminate age, older than documented habitation of this site, possibly significantly older than the Covenant's founding survey period. The depression shows residue consistent with recent occupancy within the last four months, followed by vacancy as of the night of the 14th. I cannot characterise what occupied it. The residue is not human. It is not consistent with any Wound-entity I have documented. It is not consistent with any category in the standard survey archive.
He paused.
It is consistent, in a way I find difficult to specify precisely, with the topology notation in a restricted survey document I accessed during my Rank III certification process seven years ago. The document was dated approximately 800 years before the current Covenant calendar. It described a site in the deep Aldwood. The notation read: Transit Point, External. Direction: Unknown. Recurrence: Unpredictable. Disposition: Monitor. Do not approach.
The topology of that notation and the topology of this depression are the same.
I am going to need more than the restricted archive. I am going to need to speak with whoever counter-signed my access request in sixty-three minutes.
He closed his log. He looked out at the Stonewater Strait below, the water grey and heavy in the autumn afternoon, a ship moving south through the channel, the far shore of Orrath barely visible through the sea-haze on the horizon.
The wind smelled of salt.
He thought about the garrison's forty soldiers who had not had time to be afraid.
He thought about what had occupied the depression for four months and left on the night of the fourteenth.
He thought about the direction the topology signature moved, not up, not down, not lateral in a direction that had no axis in the standard notation, a direction that his training had never named because his training had assumed, as all three thousand years of Covenant survey methodology had assumed, that there were only the directions that existed within the Weft.
He opened his log again.
The standard topology notation assumes four axes, he wrote. Wound-bleed moves along known axes. What occupied this depression did not move along known axes. It moved along a fifth. I have no notation for a fifth axis. I am going to have to invent one. I am going to call it the External axis, because the 800-year-old document used the word External and because I cannot think of a more accurate word and because accuracy, in the absence of a framework, is the only tool I have left.
He sat on the cliff until the light began to fail.
Then he went down the cliff path to the village at its base, Dawntide, forty-three households, entirely untouched, the fishermen mending their nets in the harbour below as if the hold above had never existed and he found the survey team's camp at the village's edge, and he ate something he did not taste, and he wrote up his Day One survey report in full, and he filed it with the Pale Hold's survey office, and he marked it restricted, and he went to sleep on the camp cot knowing that the report would be classified upon receipt and that whoever classified it would understand, from the survey data alone, that something had happened at Dawnholt Hold that the Pale Cartographers' three thousand years of accumulated methodology had no category for.
In the morning he would continue the survey.
He dreamed of a shape in limestone older than everything built above it, patient as geology, waiting.
The survey report reached the Pale Hold at the third hour of the following morning. It was read by two people before it was classified: the duty officer who processed the overnight priority filings, and the senior council member who had counter-signed the archive access request in sixty-three minutes. The duty officer noted in his own log that it was the most unsettling survey report he had read in eleven years of processing and that he hoped he would not be required to process its follow-up. The senior council member read it twice, then sent three messages simultaneously one to the four other members of a body that had no official name, one to a field researcher in Surath whose name she had been hoping not to have to contact for reasons she had never written down, and one to the Pale Hold's cartography department requesting all existing maps of the deep Aldwood's eastern margin from the restricted archive, with specific attention to any documents predating Covenant Year 500.
The cartography department's search would take six days.
The maps they found would be the ones Osrath Venn had been thinking about.
He would not be told they had been found.