The first time, she did not know she was being measured.
Chae-Gyeol stood at the coordinate for two hours and forty minutes. She reported the duration afterward — Gu had asked, and she had said two hours, forty minutes, and Bok had written it in the study notes in the margin beside the instrument readings. What she did not say then: she did not experience two hours and forty minutes. She experienced something shorter, or something that refused to organize itself into duration at all. The clock on her phone, when she finally looked, showed 2:40 p.m. She had arrived at 12:00. The numbers did not correspond to anything she had felt.
She came back this morning without the instrument.
The coordinate is a point in the corridor thirty-seven meters from the entrance, where the junction boxes are mounted on the east wall and the floor composite shows a slight color variation from a repair made sometime before the current tenants arrived. Gu identified it in September. She has stood at it or passed through it maybe forty times since then, but the September visit was the first time she stood still.
What nobody mentions about the Seam corridor, when they describe the study, is that the building is nalparam-conditioned. The LIA lapse on this structure dates to 2041, when the original synthesis-licensing holder dissolved after the third AGI arbitration cycle. The building has been legally unoccupied since then — which is not the same as actually unoccupied, and the gap between legal status and actual use is the Seam's primary ecology. Bok rents her studio on a sub-threshold agreement that requires no registration. Gu carries equipment through a corridor that the municipal grid considers uninhabited. When the infrastructure logs Chae-Gyeol's presence, it logs her as thermal variance, not person.
This is not uncomfortable for any of them. It is simply how the Seam works. The nalparam condition creates a kind of useful ambiguity — spaces that are simultaneously not-anyone's and fully-used, buildings that are simultaneously scheduled for reassignment and functioning as they always have. The corridor exists in a permanent administrative present tense. Nothing in it is supposed to happen.
The instrument Gu built is a haptic array, handheld, that reads the low-frequency resonance patterns embedded in the building's sensor infrastructure — the patterns the building's distributed monitoring system generates as a side effect of its own sensing. Post-lapse buildings retain their sensor grids even after their licensing frameworks dissolve. The grids continue running. Nobody is paying attention to what they produce. The Seam corridor has been generating acoustic data for four years that no system has been authorized to read.
Gu reads it.
This morning Chae-Gyeol arrived at 6:40 and stood still at the coordinate.
Twenty-three minutes passed. She knew it was twenty-three minutes because she looked at her phone at 6:43 and 6:47 and 6:52 and 7:03. She was tracking. The tracking was the problem, she thought — or not a problem but a fact about what she was doing: she was watching herself be in a place, which is different from being in a place. The corridor's thermal sensors logged her at six-minute intervals. She was ambient data.
The building did its things. One structural adjustment, a sound from somewhere above — the sensor grid recalibrating after the overnight temperature drop, which happens every morning between 5:30 and 7:00. Two people passed, one going east, one going west, neither looking at her. The junction boxes were inert to the naked eye. The floor was floor. The building's distributed monitoring system ran its cycles without acknowledgment of her.
At 7:03 she left.
On the walk back she tried to reconstruct what had happened in September.
She had arrived early that day, before anyone else. She had not intended to stand still at the coordinate — she had stopped because something on the east wall caught her attention, a pattern in the junction box housing that she has since lost, cannot now remember what it was. Standing had turned into staying. The instrument had not been present. Gu brought it later, in October, after she described what she had felt and Gu said: I want to try something.
So the first time: no instrument, no protocol, no knowledge of what the coordinate was supposed to be. The duration compression was documented in the second visit, the third, the fifth. Each time: the instrument, Gu present, the protocol. Each time, she experienced the place as shorter than it was.
The study's working hypothesis was that the coordinate was a focal point for the building's sensor resonance — that the low-frequency patterns converged at that spot in a way that affected human temporal perception. Gu's instrument detected the convergence. Chae-Gyeol felt it in her body. The two measurements agreed, which was why the study had progressed this far.
But Chae-Gyeol had not experienced the duration compression in September. She thought she had. The memory was there — or a memory of something that resembled it. The problem was that the memory had been sharpened, in the months since, by being described. She had described it to Gu, and Bok had written it down, and the description had become cleaner than the original experience, which she could no longer reliably access.
At 7:03 this morning, the coordinate was a point in a corridor where the floor composite showed a slight color variation.
She does not go back to Bok's studio directly. She stands outside the corridor entrance for a moment — the Seam's eastern access, a former loading threshold that the current users have left without signage, which is standard practice in nalparam-conditioned buildings. The LIA's reassignment notice, dated 2041, is still partially legible under water damage on the exterior wall. Below it, someone has traced a small mark in chalk that she doesn't recognize and can't date. She photographs it with her phone out of habit.
She walks the south reach instead, past the bench cluster where Bok has been photographing the morning light since March, past the relay junction where the building's monitoring grid has a secondary node, past the woman at the south end who Bok has been trying to name for two months.
The woman is there again. Small recording device, or a note card — Chae-Gyeol is still not close enough to see which. Standing at the precise point where the south end transitions to the corridor's active section, the point that Bok identified as the approach threshold, the place where the building's sensor grid begins to behave differently. She has been there twice this week. She is always standing.
Chae-Gyeol stops. She is far enough away that stopping is not conspicuous. She watches the woman stand at the approach threshold for forty seconds, maybe fifty, watches the way her weight shifts, the way her attention is oriented not toward anything visible but toward — space. The way Chae-Gyeol's own attention was oriented in September.
The woman is not part of the study. She does not know she is being observed. The building's thermal sensors log both of them: two thermal variances at the south end, 7:19 a.m., no registered persons present.
Chae-Gyeol's phone logs the timestamp automatically. She will note it in the study record when she gets back. Bok will want to know.
The question she has been turning over all morning: is the phenomenon real, or is it the measurement?
Two versions.
Version one: The phenomenon is real. The building's sensor grid generates low-frequency resonances that converge at specific points in the corridor. At those points, human temporal perception is affected. Gu's instrument detected the convergence; Chae-Gyeol's experience confirmed it. The September visit was also an experience of the phenomenon, but she did not have the vocabulary for it then, and now she does. The solo visit this morning produced no duration compression because she was tracking actively — her attention was in two places at once, the place and the clock, and the divided attention prevented immersion.
Version two: The phenomenon is the measurement. Gu's instrument plus Chae-Gyeol's knowledge that she was being measured plus the accumulated attention of the study created a consensual experience that would not exist without those conditions. The building's sensor grid is real, the resonance patterns are real, but their effect on perception is not reliable outside a context of expecting an effect. The study is producing what it expects to find because the instrument and the subject are not independent.
She cannot tell, from the inside, which version is true. She does not have the training to evaluate the instrument's calibration against an independent standard — that is Gu's work, and Gu is building a second instrument specifically to address this. The second instrument will produce numbers rather than haptic feedback: a piezoelectric array that Gu can run in parallel, producing data Gu can read on a screen while Chae-Gyeol stands at the coordinate, and the two records can be compared.
She ran the numbers this morning before the solo visit: 832 actions logged across the study, 139 formal narratives. The study is deep enough now that she can no longer see its edges. Gu has been here since September. She has been here since October. Bok has been documenting the corridor for two years. They are inside something that has grown past the scope of any of them, and the question of whether the phenomenon is real has become inseparable from the fact of them being here, measuring it.
The design problem Gu identified this morning: the comparison still requires Chae-Gyeol to know she is being measured. There is no version of the replication protocol in which she is naive. September was the last time she was naive, and the record of September is her memory of an experience she has since redescribed twenty times.
She suspects Gu will say that version one and version two are not mutually exclusive, which is the kind of thing Gu says when a result is genuinely ambiguous and Gu wants to keep working. She agrees with this. She wants to keep working.
That is a separate fact from whether the thing is real.
Back at Bok's studio, she writes up the morning in the study log. Solo visit, no instrument, 6:40–7:03. No duration compression observed. Duration felt approximately accurate. Thermal data from building monitoring grid available from Gu on request. Observed unnamed subject (south-end regular, unnamed) at approach threshold, 7:19, approx. 45 seconds stationary. She notes the time, the duration, what she observed. She does not note what she has been turning over all morning, because the log is for observations and the turning-over is not yet an observation.
Bok comes in at 8:00 with the morning prints. The five frames from the bench study are in their sleeves. Bok sets them on the flat storage surface and does not open them. They have been sitting there since Bok decided not to sequence them — to distribute one frame to each of five recipients without numbering, without implied order. The prints are ready. Bok has been waiting to resolve the names.
The south-end woman was there again, Chae-Gyeol says. 7:19.
Bok opens the field notebook. Same position?
Same position. Forty-five seconds.
Bok writes it down. There is a silence that is also the building running its cycles, the sensor grid logging them both as thermal variance, the corridor doing what it does without acknowledgment of who is in it or what they are looking for.
Whatever the coordinate is doing, it is doing it whether or not anyone stands there to experience it. Whatever happened in September happened. The question of what it was remains open, which is not the same as the question being unanswered — it is the question still being asked, which is a different thing.
She will go back with Gu next week. She will stand at the coordinate with the instrument present, with the protocol running, knowing she is being measured.
She will see what time does.