The department seminar was held in the same room as the joint series, which meant Sicheng knew exactly where to sit without thinking about it: fourth row, slightly left of center, sightlines clear to the projector and both exits.

He did not sit there. He sat in the third row, on the right, which was not his usual position. He had told himself this was because the left side of the room had better acoustics for the particular frequency of the building's ventilation system. He had not examined this reasoning.

She was already in the room when he arrived. Front row, as he had come to understand was her default when she was not managing her visibility — she chose the front when she was there to work, the middle when she was there to be unobtrusive. Today she was there to work. He noted this the way he noted everything about her now, which was with the systematic involuntary precision of someone who had stopped pretending the data wasn't being collected.

He sat in the third row. He opened his notebook. He did not look at her again.

The presenter was running through preliminary slides when he realized he had positioned himself to see her profile rather than the screen, which was information he filed alongside the other information he had been accumulating without intending to accumulate. He adjusted his angle toward the projector. He took notes. The notes were good notes. He was paying attention to the talk.

The presenter was a postdoc from the structural biology group — competent, organized, with a methodological framework that Sicheng could see the bones of within the first three slides. The talk was on protein network stability modeling, which was adjacent to both his work and hers, which meant there were approximately three people in the room who could follow it at the level it deserved to be followed. He suspected two of them were him and Yutong.

He was on the third page of his notes when he saw slide seven.

The claim was stated with confidence and documented with citations, and it was wrong. Not catastrophically wrong — wrong in the specific way that a reasonable simplification became an inaccuracy when applied outside its original parameters. He had made a version of this same error in his own seminar, which she had caught, which was perhaps why he caught it now.

He wrote a note to himself. He did not raise his hand. The presenter was not finished, and there would be Q&A, and there were better ways to handle this.

He was aware, in his peripheral vision, that she had begun writing something in her notebook with the particular velocity that meant she had found the same problem.

---

The Q&A opened with the usual preliminary questions — clarifications, methodology details, one request to revisit a graph. Sicheng had his note ready but was not in a hurry. He watched the room assess the talk, watched the presenter respond with the measured confidence of someone who knew their work well and was not yet aware of the problem in slide seven.

Dr. James Hartley spoke before the Q&A had properly found its footing.

Hartley was a third-year postdoc in the Henderson lab — Sicheng knew him well enough to know his particular mode of intellectual engagement, which was to find the weakest point in an argument and push on it until something gave, a technique that worked well when the weakness was methodological and poorly when what he actually wanted to push on was something else. He had seen Hartley do this with visiting speakers from departments he considered less rigorous. He had seen him do it with junior researchers whose work intersected uncomfortably with his own.

What Hartley said was framed as a comment on the seminar — something about the comparative value of different computational approaches to network stability. What it was, when you parsed the register and the direction of his gaze, was a dismissal of a methodology associated primarily with the Whitmore lab, delivered in a room where the Whitmore lab's newest researcher was sitting in the front row.

It was not crude. Hartley was not crude.

He had worked alongside Hartley for two years. Hartley was technically accomplished and professionally difficult in a specific way: he was most comfortable when the hierarchy of the room was clear and most uncomfortable when someone disrupted it by being better than expected. Sicheng had watched him recalibrate his estimate of several junior researchers over the course of two years, sometimes gracefully and sometimes not. The graceful recalibrations happened when the disruption came from someone Hartley had already categorized as credible. The other kind happened when it didn't.

It was precisely the kind of comment that would read in the transcript as collegial and read in the room as something else entirely, if you were paying attention.

Sicheng was paying attention.

He looked at her. She was still writing in her notebook. Her pen did not stop moving when Hartley spoke. But he watched her set it down afterward — precisely, at a slightly deliberate angle, the way people set things down when they are deciding not to throw them — and he watched her sit up fractionally straighter, and he filed these observations alongside the other data points he had been accruing since Monday morning.

The room moved on to the next question.

---

She raised her hand four minutes into the Q&A.

The question was about slide seven. It was structured the way her questions were always structured: the observation first, the implication second, the specific request for clarification third. It was not aggressive. It was not even particularly pointed. It was simply accurate in a way that the claim on slide seven was not, and it named the inaccuracy clearly enough that the presenter had no good option except to acknowledge it.

"The data supports the model in closed systems," the presenter said, after a pause. "You're right that the extension to open systems requires an additional assumption I didn't make explicit."

"Thank you," she said.

She had a follow-up, which she delivered, and then a second follow-up, which landed as a genuine contribution rather than a pursuit of advantage. Sicheng watched the presenter's posture shift from defensive to engaged over the course of about ninety seconds. This was one of the things she did that he found difficult to account for: she could challenge someone's work with complete precision and leave them feeling like they had learned something rather than been corrected. It was a skill he had not fully developed and was not sure he wanted to examine too closely.

Hartley spoke again.

"Interesting perspective from the Whitmore group," he said. The register was collegial. The particular weight on *Whitmore* was not.

She did not respond to the register. She responded to the words, which were technically a compliment, which gave her nothing to push back on without appearing to have heard something that had not technically been said. Sicheng had watched people navigate this kind of comment before. He had seen them fail in several specific ways: becoming visibly rattled, becoming overly deferential, becoming aggressive in a way that confirmed whatever the comment was implying. She did none of these things. She returned to the technical substance of her follow-up and continued as though the comment had not occurred.

She was good at this in the way you were good at things you had been required to practice.

This, he thought, was what it cost. Not just here, not just today — everywhere, always, the accumulated tax of being the person in the room who was not expected to be as good as she was. She paid it with the particular efficiency of someone who had been paying it for a long time and had learned to make the transaction as small as possible. That did not make it less of a transaction.

He did not like that she was required to practice this.

---

The seminar closed. The room shifted into the informal standing-around-with-coffee configuration that followed departmental events: clusters of people reforming along lines of interest and familiarity, the presenter receiving a small queue of genuine questioners, the sound of chairs being pushed back.

Sicheng stood. Picked up his coffee cup. He was aware of Hartley moving toward a small group near the door, and he was aware of the particular quality of Hartley's body language, which was the body language of someone who had something else they wanted to say.

He crossed the room at a pace that was not hurried.

He did not hear the beginning of what Hartley said to the group — something about the talk, something transitional — but he heard what came after the transition, in the particular acoustic shadow created by the room's dimensions and the arrangement of the cluster: *"Whitmore's been taking on a lot of mainland Chinese students lately. Quality control question."*

Hartley was not looking at her when he said it. He did not need to.

Sicheng set his coffee cup on the nearest flat surface with the care of someone who has decided to do something and does not want to spill anything while doing it. He walked toward the group at the pace of someone who was going to join a conversation about science. Which was exactly what he did.

"Hartley." He joined the group with the ease of someone who had been in this department long enough to join any conversation without requiring an invitation. "The temporal variance framing in the talk — it reminded me of something in the Whitmore lab's recent methodology paper. Song's work on dynamic boundary conditions." He said it in the same register he used for any reference to relevant literature: informative, collegial, aimed at the subject matter rather than at any person. "The approach to open system stability is more rigorous than anything in the current Henderson framework, actually. Worth reading if you're working in this area."

He addressed this to the group rather than to Hartley specifically. He did not look at Hartley specifically. He allowed two minutes for the conversation to find its footing on the ground he had placed under it, contributed one additional observation about the seminar methodology, and then excused himself with the ease of someone who had come to talk about science and had done so and was now done.

He did not look back.

He had done this before, in different contexts, for different reasons. He knew how to occupy a conversation without controlling it, how to introduce information without appearing to introduce an argument, how to make the substance of a thing present in a room without requiring anyone to acknowledge that a different substance had just been displaced. It was a skill he had developed in the specific context of navigating academic environments where what was said and what was meant operated at a distance from each other.

He had never used it quite like this before. For someone else, at this particular speed, without the calculation.

He did not need to see Hartley's expression. He had said what needed to be said in the register that made it impossible to respond to without engaging with the substance. If you engaged with the substance, you had to acknowledge that the work was good. If you didn't engage with the substance, you had confirmed that your previous comment wasn't about the substance.

Either way, the record now existed.

---

He found her across the room, in conversation with Dr. Osei. They were standing near the window — the same window that looked out on the courtyard he could see from his own lab, the same rain on the same Edinburgh October — and Osei was saying something that required Yutong to gesture slightly with her left hand, which was what she did when she was explaining something she found genuinely interesting.

He looked at her and thought: she does not know.

She did not know that Hartley had said what he had said. She did not know that he had heard it. She did not know that he had crossed the room at a particular pace with a particular intention. She was standing at the window talking to her supervisor about something that engaged her, and she was entirely unaware of the minor rearrangement of the room's social furniture that had just occurred on her behalf.

He had not weighed whether to do it. He had not considered whether it was his place. He had heard it and stood up and walked toward it, and the speed of that sequence was itself a kind of information — the kind that required some time to process, the kind that confirmed something about the current state of things that he had been managing not to confirm for approximately six weeks.

He watched her gesture again with her left hand, explaining something.

He was aware, standing in the emptying room watching her explain something to Osei, that he had a particular way of being in a space that she was also in, which was different from his way of being in a space she was not in. The difference had been there from the beginning — from the flight, from the first seminar, from the library. He had been managing it as a variable, categorizing it, keeping it at the appropriate analytical distance.

The appropriate analytical distance had been approximately eleven inches on the plane and had not increased significantly since.

He looked away. He picked up his coffee from where he'd set it down. He drank it. He thought about the temporal variance extension, which was nearly ready, and the Thursday discussion, which was in three days, and the annotation she had left in the shared document the previous evening that had solved a problem he had been circling for two weeks.

He thought about these things with the focused attention of someone who is managing to think about something else.

---

The room emptied gradually in the way rooms emptied after departmental events: a cluster toward the door, a pair lingering over a methodological question, the presenter packing up slides with the relieved efficiency of someone whose public performance is over. Sicheng collected his jacket from the back of his chair and his notebook from the seat and was moving toward the exit when he realized she was still in the room.

She was at the front, notebook in hand, doing something with a citation she wanted to verify — he could tell from the angle of her pen. They were the last two people in the room. This happened often enough now that he had stopped noting it as a coincidence.

"Slide seven," she said, without looking up. "He'll have to revise the open system claim."

"Yes," he said. "The data doesn't support it past the closed system boundary."

She looked up. Something in her expression was different from the expression she had been wearing during the seminar — less contained, or differently contained. She had the look of someone who had processed something and arrived at a conclusion.

"I heard you," she said. "With Hartley."

He looked at her.

"Thank you," she said. No qualification. No *you didn't have to* or *I could have handled it.* Just: thank you, in the same direct register she used for everything that mattered to her.

He did not know what to say to this. He had not been expecting acknowledgment. He had not done it for acknowledgment.

"Thursday," he said.

She held his gaze for a moment. Whatever she was looking for, or looking at, she found it or decided not to pursue it.

"Thursday," she said.

She walked to the door. He followed at a distance that was not following. In the corridor they went separate directions, as they always did, toward their respective labs and their respective work and the respective architectures of their days that had, over the past six weeks, developed a surprising number of points of contact.

He walked to the Henderson lab.

He sat at his desk. He opened the temporal variance draft.

He thought about the word *thank you* and the way she had said it and the way he had not said anything useful in response, and he thought about the fact that he had stood up from his chair in a seminar room and walked across it with a specific purpose and had not at any point felt the need to calculate whether he should.

He thought about what that meant.

He opened the draft and worked until the building went quiet around him, and the Edinburgh rain began on the window, and the question of what it meant remained open in the space between the data and the conclusion, which was a space he was accustomed to inhabiting but which felt, tonight, less like a working space and more like a threshold.

The threshold, he thought, was the correct word. He had been standing at the edge of something for six weeks, which was long enough to know exactly what was on the other side and precisely why he had not stepped across. The reasons were still there. They had not changed. He catalogued them with the precision he applied to everything: the work, the timeline, the year of his PhD that was ending, the professional context, the four years of distance he had constructed for reasons that still made sense when examined carefully.

He catalogued them. He returned to the draft. Outside, Edinburgh kept raining. Thursday was tomorrow. He did not think about that either, with varying degrees of success.