He noticed the clocks on the third morning.

He had been passing them every day and filing them under incidental urban infrastructure the way you filed anything that served an obvious function: noted and moved on. On the third morning, walking from Calder House to the archive, he looked at the bakery clock and stopped.

His phone said eight forty-one. The bakery clock said eight fifty-three.

He walked to the post office corner. Eight thirty-five. Guild hall: nine oh two. Watchmaker's window: eight forty-eight. Bank: eight thirty. Five clocks, five different times, a range of thirty-two minutes, and none of them agreeing with his phone.

And all of them running.

He watched the second hands move, one full rotation per minute, with the correct mechanical precision of instruments properly maintained and correctly calibrated. They were not broken. They were not stopped. They were doing exactly what a clock was supposed to do: they were measuring time. The problem was that they were measuring different amounts of it, or different versions of it, or time from different starting points, and none of them had arrived at a consensus.

He took out his notebook and wrote down each reading. Then he walked to the archive.

She looked up when he came in. He said: "The clocks."

"Good morning," she said.

"Five different readings in the market square. Thirty-two minute range."

"That's a fairly typical morning." She looked at her own cup of coffee as if confirming something about it. "On days before weather, or when the Temporal Residue is more active, the range can be wider. On very quiet days, smaller."

He sat at his table. "This has always been the case."

"Since we have records. The 1880s."

"And no one has identified a mechanism."

"Several people have tried." She picked up her pen. "The clockmaker who started the formal logs in 1891 was the first to suggest that the clocks were not wrong. He said they were each right, but right about different things." She paused. "He didn't know what different things."

He pulled out his notebook and opened it to the page where he had written down the five readings. He looked at them. He thought about Owen.

"I'm going to call a friend," he said. "He works with clocks. I'd like him to come look at the tower."

"Owen," she said.

He looked at her.

"From the memories," she said simply. "He'll find it very interesting. He's going to form a good theory." A pause. "Let him have it for a while before you tell him it's incomplete."

He looked at her for a moment. He thought about what it was like to know, in advance, what someone's intellectual reaction to a thing would be.

"All right," he said. He called Owen.

Owen answered before the second ring.

"Tell me about the clocks," he said, before Elias had said anything about clocks.

Elias stopped. "You know I'm calling about clocks."

"You called at ten in the morning from a city with unusual temporal properties, the day after you spent an afternoon in a building full of historical clock documentation. What else would you be calling about?" The background sounds of the workshop were present: the tick of a mechanism being tested, the careful percussion of small tools. "Tell me what you've seen."

Elias described the market square: the five readings, the range, the consistency of each clock's internal operation. Owen listened with the focused silence of someone processing technical information rather than social information, not filling the gaps with acknowledgment but letting the description complete itself.

"Internal consistency," Owen said, when Elias finished. "Each clock consistent with itself over time, but inconsistent with the others."

"As far as I can observe, yes. I'm not set up to track them formally."

"You don't need to. If each clock showed erratic readings relative to itself, that would be a mechanical problem. Consistent readings that are inconsistent with each other is something else." A pause. "What about the primary tower?"

"The North Clock Tower. It's the reference they recalibrate to every Sunday."

"And it's running slow."

"Four to eleven minutes, depending on conditions. It's the most stable of the clocks in the city. They use it as the reference because it's consistently wrong in the same way."

Owen was quiet for a moment. "Consistently wrong in the same way is more useful than correctly wrong in unpredictable ways," he said.

"She said almost exactly that."

"She being the archivist who knows things in advance."

"Yes."

A longer pause. "I'm coming tomorrow," Owen said. "I want to see the tower and I want to read the logs and I want to meet her." Another pause. "This is either the most interesting thing I've ever been asked to look at or a very elaborate collective custom. I'm genuinely uncertain which."

"I don't think they're mutually exclusive," Elias said.

"No," Owen said. "I don't think they are."

He ended the call and sat for a moment with his phone. The archive was quiet around him. Outside the tall windows, the market square was visible in the mid-morning light and the clocks on the buildings around it were each showing their particular reading, running their particular time, recording their particular version of the morning.

He thought about what Owen was going to make of this. He thought about what it would mean to bring someone into Grayhaven who had not already been shaped by it, who would encounter it fresh and form his own analysis without the accumulated weight of what Elias had already experienced and not yet processed.

He thought it would be useful to see Grayhaven through Owen's eyes.

He also thought, with more honesty than was entirely comfortable, that he was looking forward to Owen being here for reasons that had nothing to do with the clocks.

Owen arrived in Grayhaven at three in the afternoon.

He parked at the edge of the main street, got out of his car, and stood looking at the clock tower for a long moment before he looked at anything else. Then he looked at his phone. Then he looked at the tower again. He did this with the systematic patience of someone gathering data before forming a conclusion.

"Eleven minutes," he said, when Elias came to meet him. "The tower is eleven minutes behind my phone."

"Consistently," Elias said. "Within a few minutes, depending on conditions."

Owen turned to look at the market square. He looked at the guild hall clock, then the post office clock, then the watchmaker's window, then the bank. His lips moved slightly as he noted the readings. He did not get out a notebook because Owen, unlike Elias, carried most of his technical observations in his head until he reached a point of writing them down.

"These are all running," Owen said.

"Yes."

"Correctly, mechanically."

"As far as I can tell."

"But they're each showing a different time."

"Yes."

Owen looked at him. "And the people here," he said. "They treat this as normal."

"Mostly," Elias said. "They recalibrate to the tower on Sundays. By Saturday they've drifted back to their individual readings. It's been this way for over a century."

Owen was quiet for a moment. He had the expression he wore when encountering something that was genuinely new to him: not puzzled but carefully attentive, the face of a person who respected the thing they were looking at enough to take their time with it.

"I want to see the tower," he said. "Then I want to see the logs."

They walked to North Clock Tower. Owen spent the better part of an hour examining it: the exterior, the mechanism access panel, the movement itself where it was visible. He listened to it with his ear against the casing for a long time, the way a physician might listen to something organic. He made notes. He asked about maintenance records and service history and the qualifications of the clockmakers who had worked on it over the decades.

"Standard movement," he said, at the end of it. "Well maintained. Running at the correct mechanical rate relative to itself." He stepped back and looked at the tower face. "The question is what it's measuring. Because if the mechanism is correct and the rate is correct, then the time it shows is real. It's just not the time your phone is calibrated to."

"What would cause that?" Elias said.

"Something local," Owen said. "A property of the location that makes time run differently here than it runs in the network your phone syncs to. Not faster or slower in a universal sense. Just differently." He looked at the tower. "It would also explain the GPS. If the location has a property that affects time measurement, it might affect signal propagation as well."

He paused.

"That's a hypothesis based on very limited data," he said. "I want to see the logs before I commit to it."

"She said it would be a good hypothesis," Elias said.

Owen looked at him. "She told you what my hypothesis would be."

"She said the hypothesis you would form would be good but incomplete."

Owen looked at the tower for another moment. "Take me to the archive," he said.

They walked to Vale Archive.

Liora was at her desk. She looked up when they came in. She said good afternoon. She had, Elias noticed without remarking on it, put out three coffee cups on the desk to her left.

Owen noticed this too. He looked at the three cups, then at Elias, then at Liora.

"This is Owen," Elias said. "He studies mechanisms."

"I know," Liora said. "Hello, Owen." She indicated the chairs across the desk. "The clock logs from 1891 to 2004 are in the section labeled Horological Records, third cabinet from the window on the left. The digitized records from 2004 to present are on the reading room terminal." She stood and poured coffee into all three cups with the efficiency of someone for whom this was already decided. "The tower's movement is original to the 1880s installation. It's been serviced forty-three times. I can pull the service records if you want the specifics."

Owen sat down. He accepted the coffee. He looked at Liora with the expression he wore when something was more interesting than he had expected.

"You knew I was coming," he said.

"Yes."

"Before Elias told you."

"Yes."

He looked at Elias. "All right," he said, with the tone of someone updating a working model. "Tell me about the logs."

The three of them worked for the rest of the afternoon: Owen with the historical clock logs, reading them with the focused intensity of someone encountering primary source material in their area of expertise; Elias with his own archive materials, continuing the context-building he had been doing all week; Liora at her desk, working on her forms, occasionally answering questions from one or both of them with the precision of someone who had been the expert on this collection her whole working life.

The archive settled into its working rhythm around the three of them. Elias was aware, at intervals, of Owen glancing between him and Liora with the subtle observational quality Owen applied to everything: not intrusive but present, the habitual attention of someone who dealt in mechanisms and noticed the way things moved relative to each other.

At half past four Owen set down the log he was reading.

"The clockmaker in 1891 wrote that the clocks were each right but right about different things," he said. "He didn't know what things."

"I told Elias that this morning," Liora said, without looking up.

"What do you think they're measuring?" Owen said.

She looked up. She considered the question with the seriousness she brought to questions she found genuinely worth considering.

"I think they're measuring time," she said, "in the way that's available to them where they are. And where each one is is slightly different from where each other one is, in some sense that isn't geographic. They're in the same square. They're in different temporal positions." She paused. "That's not a precise statement. It's the best I can do with the framework I have."

Owen nodded slowly. "My hypothesis is similar," he said. "The location has properties that create local variations in whatever the clocks are measuring. Not a universal time. Several local times." He looked at the log in his hands. "Which doesn't explain why the tower is the reference."

"Because it's the most consistent," she said. "It varies within a narrow range. The others vary more widely. Consistent deviation is more useful than variable deviation."

"Even if the consistent deviation is wrong relative to everywhere else."

"Especially then," she said. "If you're going to be wrong, be reliably wrong."

Owen smiled slightly. It was the smile of someone encountering a principle they had held independently and found confirmed.

"Your theory is good," she said. "It explains most of what you can observe."

"But not everything," Owen said.

"No," she said. "Not everything."

"What doesn't it explain?"

She looked at him steadily. "That's the part that requires more time to explain than we have today." She picked up her pen. "Come back tomorrow. Both of you. I'll have some of the older documentation ready."

Owen looked at Elias.

Elias looked at Liora.

She was already writing.

They left the archive at five and walked back through the market square, where the clocks were each showing their particular version of the late afternoon. Owen walked with his hands in his pockets and his attention moving between the clock faces and the street and the architecture, cataloguing with the efficiency of someone who processed environments quickly.

"She knew I was coming," he said. "She knew my hypothesis before I formed it. She had three cups ready."

"Yes," Elias said.

"And this is normal here."

"Apparently."

Owen thought about this. "How long has she known you were coming?"

Elias was quiet for a moment. "Eight years," he said.

Owen looked at him. He did not say anything immediately. They walked half a block in silence.

"Right," Owen said finally. "I'm going to need considerably more coffee.

That evening, Elias went back to the archive logs on the reading room terminal.

He had asked Liora for access before they left and she had given him a card without asking why he wanted to work after hours. He sat at the terminal in the empty archive, the overhead lights low, the sound of the sea audible when the ventilation was quiet, and read through the digitized clock records from 2004 to the present.

The logs from the last two decades were different from the earlier ones in texture if not in content: more systematic, more precisely formatted, the product of people who had grown up with databases rather than paper. But the observations they recorded were the same observations the 1891 clockmaker had been making: the drift, the return to individual readings after Sunday recalibration, the consistency of the tower's eleven-minute variation.

In 2019 someone had attempted a formal mathematical analysis of the drift patterns. The analysis was thorough and ran to forty pages and its conclusion, stated carefully in the final paragraph, was that the clocks were each behaving consistently according to an internal logic that the analyst had not been able to determine. The drift was not random. It was structured. It followed rules. The rules were not rules the analyst could identify.

He read this and thought about structured rules he could not yet identify.

He thought about Liora, who had said the same thing in different terms: not that the clocks were wrong but that they were right about different things, and that what the different things were had not yet been named.

He thought about probable futures and the specific quality of certain knowledge. He thought about what it would mean to carry structured, rule-governed information about the future without having access to the rules that governed it. To know that someone was coming without knowing why you knew it. To know what their hypothesis would be without knowing the mechanism of the knowing.

He thought about whether the clocks and Tomorrow Memory were related phenomena or parallel ones or the same phenomenon seen from different angles.

He was not yet equipped to answer this. But he was becoming more precisely equipped to ask it.

He logged out of the terminal at eight and walked back to Calder House through the evening streets. The clocks were showing their evening readings, each one its own version of eight o'clock, a chorus of different interpretations of the same hour. He had started to find this less unsettling and more interesting, which he thought was either progress or accommodation.

Owen was in the sitting room of Calder House when he arrived, reading something he had printed from his phone.

"I've been thinking about the tower," Owen said, without looking up.

"I know," Elias said.

"The local time variation hypothesis explains the clock behavior. It even, loosely, explains the GPS loss at the city boundary: if the location has properties that create local time variations, it might create a zone where network synchronization is difficult." He turned a page. "What it doesn't explain is why the variation is structured rather than random. Local variation from a geological property should be continuous and gradual. What we're seeing is discrete. Each clock gravitates toward a specific reading. That's not what diffuse geological influence looks like."

Elias sat down. "What does it look like?"

"It looks like each clock is in range of something that has its own time," Owen said. "And each something has a slightly different time from the others." He looked up from his reading. "Which is a description of several distinct sources rather than one diffuse one." He paused. "I don't know what the sources are. But the pattern is discrete, not continuous."

Elias thought about this. He thought about Tomorrow Memory: distinct, incident-specific, governed by its own rules. He thought about clocks, each gravitating toward a specific reading rather than drifting randomly.

"Tell her that tomorrow," Elias said. "The discrete versus continuous observation."

"I intend to," Owen said. "I think she already knows it. I think she's been waiting for someone to arrive at it independently."

Elias thought about a woman who had been waiting, in various forms, for a very long time.