Thursday arrived the way Sicheng preferred his days to arrive: quietly, and before anyone else.

He was in the Henderson lab by seven-fifteen, which was early even by his standards. The building was nearly empty at this hour — a few lights on in other offices, the distant sound of the cleaning staff finishing their rounds, the particular hush of a place that had not yet decided to be busy. He liked this version of the lab most: before the day's negotiations began, before anyone needed anything from him, before the work became entangled with all the human machinery that surrounded it. In the early morning it was just the data and the questions the data raised and the quiet satisfaction of being someone who knew how to ask them.

He made coffee in the small kitchen at the end of the hall — the same coffee, in the same mug, that he had been making in this kitchen for four years. He returned to his desk. Opened the data from yesterday's run. Pulled up the presentation file, reviewed it once, closed it again. The talk was at two. The content was sound; he'd been over it enough times to be certain of that. He had no reason to think about it before noon.

He thought about it anyway.

Not about the content. He thought about the room, and the audience, and the specific composition of that audience, which as of approximately seventy-two hours ago had changed to include one additional variable.

Whitmore lab attendance at Henderson seminars was expected. This was standard department practice — interdisciplinary exposure, the university called it, which was a diplomatic way of describing two research groups with competing methodological frameworks being required to sit in the same room twice a month and be civil about it. Song Yutong was a Whitmore lab researcher. Therefore Song Yutong would attend.

This was a simple logical chain. He followed it to its conclusion and then set it aside, deliberately, the way you set aside a variable you've accounted for and don't need to keep examining. He returned to the data. The data did not require him to think about who would be sitting in the fourth row.

He worked until nine-thirty, when the first of his lab members arrived and the morning began in earnest.

---

The seminar room filled in the gradual, organic way that seminar rooms filled when attendance was expected but not enforced: Henderson lab members arriving early and taking the seats nearest the front, researchers from adjacent groups filtering in and distributing themselves according to their level of interest, a few stragglers appearing in the final two minutes and sitting wherever remained.

Sicheng stood at the front, running through his slides one final time, and did not watch the door.

He was aware when the Whitmore group arrived — there were five of them, and they came as a loose cluster, the way people who worked together tended to move through shared spaces. He did not look up from his laptop. He heard them find seats, the specific shuffling and settling of a group negotiating row placement.

He heard, separately and afterward, a single set of footsteps.

He looked up.

She chose a seat in the middle of the room — fourth row, slightly left of center. Not the front, where the most engaged researchers sat and where sitting communicated something. Not the back, where you sat when you intended to leave early or had come out of obligation. The middle, with clear sightlines to the projector and, he noted, to both exits.

She opened her notebook. Uncapped a pen. Settled in with the focused stillness of someone who had come to work.

He looked back at his laptop.

"Two minutes," he said to the room, and the remaining conversations dropped to murmurs.

---

He was a good presenter. He knew this without particular pride — it was simply a skill he had developed, like most skills, through repetition and attention to what worked. He knew how to pace, how to read a room, how to handle the moment when an audience's attention began to drift and redirect it without appearing to. He had given versions of this talk three times in the past year, refining it each time, and the current version was the best.

He distributed his eye contact the way he always did — sweeping the room in a pattern that felt natural to the audience and was, in fact, carefully calculated. Front left, front right, middle, back. He did not linger anywhere. He did not avoid anywhere.

He was on the fourth slide — the one laying out the core assumption of his computational framework, the point on which the entire subsequent analysis depended — when he allowed his gaze to pass across the middle of the room.

She was frowning.

Not a dissatisfied frown, not the frown of someone bored or confused. The frown of someone whose mind had snagged on something and was pulling at it. He had seen this expression before, in a different context, years ago, and he recognized it the way you recognize a word in a language you thought you'd forgotten — not as information exactly, more as a physical sensation, a kind of resonance in the chest.

She was writing something in her notebook. A question, probably. Something she intended to ask.

He moved to the fifth slide and kept his voice even and his pace steady and did not think about the notebook or what was being written in it.

---

The Q&A lasted twenty minutes, which was longer than average and which he took as a good sign. The questions were substantive — his lab members asked about methodology, a postdoc from the chemistry group asked about potential applications, someone from the back asked a broad theoretical question that he answered carefully and completely.

He saw her hand go up.

"Dr.—" she caught herself. "Lin." A fractional pause on his name, the smallest possible acknowledgment that this was not a neutral word between them. "The assumption on slide four. That network stability can be modeled as a function of node connectivity alone — does that hold when you introduce temporal variance? The biological systems I work with show significant instability during phase transitions that doesn't seem to be captured by connectivity metrics."

The room was quiet in the way rooms go quiet when a question lands correctly.

He looked at her directly for the first time since she'd walked in. She was watching him with the particular attention of someone who genuinely wanted the answer — not to expose a flaw, not to demonstrate her own knowledge, but because she had found something that didn't fit and wanted to understand why. He had forgotten, or had worked to forget, how different that looked from every other kind of attention.

"It's a real limitation," he said. "The current model treats stability as a static property, which is a simplification that holds reasonably well for the systems we've been analyzing but breaks down under the conditions you're describing." He paused. "We're working on a temporal extension. It's not in the published version yet."

"Would the temporal extension change the conclusions of the third chapter?"

He considered this for exactly as long as it deserved. "Some of them. The core finding holds. The specific parameters would need recalibrating."

She nodded. Wrote something. Looked up again. "Thank you."

The next question came from the back of the room, and the moment closed, and he moved on.

But he was aware, with the precision he usually reserved for data, that the exchange had lasted forty-seven seconds and that she had asked the one question he had known was coming and had not fully answered in the presentation itself.

---

Afterward, the room dissolved into the informal gathering that followed most departmental seminars — people migrating toward the corridor, conversations forming and reforming. Someone had brought biscuits. The coffee urn was producing a queue.

Sicheng answered questions from his lab members, accepted a brief word of acknowledgment from Henderson, who had appeared at the back of the room sometime during the Q&A and whose approval registered as a small, familiar warmth that Sicheng noted and filed in the category of things he had learned to need. He spoke with the chemistry postdoc about potential collaboration. He drank his coffee.

He was aware of her across the corridor, standing with the Whitmore group, not looking at him, in conversation with the second-year named Marcus who seemed to be explaining something with considerable enthusiasm. She was listening with the polite attention of someone who had decided to be a good colleague even when the enthusiasm exceeded the content. He recognized this expression too. He looked away.

Dr. Osei appeared at his elbow.

"Good talk," she said. She meant it — Patricia Osei was not a woman who said things she didn't mean. "Lin, I don't think you've formally met our newest researcher. Song Yutong — she just started with us this week."

Yutong turned from Marcus mid-sentence.

The introduction happened with the particular formality of two people who were not strangers performing the ritual of strangers. He extended his hand. She took it. Her grip was firm and brief — professional, calibrated. The contact lasted less than two seconds. He was aware of it with disproportionate precision.

"We've met," she said. "Briefly."

"At a conference," he added, for Dr. Osei's benefit. Which was true, and also the smallest possible version of true.

"Yutong's work is in protein interaction networks," Dr. Osei said, with the particular enthusiasm of a supervisor who had recently acquired someone she considered a good find. "There's some interesting overlap with what you're doing, actually. The question she asked—"

"Was a good question," he said.

He felt Yutong's attention shift slightly at this — a small recalibration, quickly managed. He didn't look at her.

"We should get the two labs talking more," Dr. Osei said, in the tone of someone who considered this settled. "The joint seminar series proposal — has Henderson said anything?"

"I haven't seen the final proposal."

"It's going around this week. I think it could be productive." She looked between them with the benign satisfaction of someone arranging furniture whose arrangement she found pleasing. "I'll let you two get acquainted."

She moved away.

The corridor continued around them — conversations, the sound of the coffee urn, someone laughing at the far end. A natural moment to exchange professional pleasantries and disperse in different directions. The kind of moment that resolved itself without requiring anyone to decide anything.

He was aware of the decision point with the same clarity he was aware of most things — precisely, and slightly too late to pretend he hadn't noticed it. The professional pleasantry would have been easy. Four years of distance had given him considerable practice in the architecture of easy exits. He could have offered one. He was choosing, with full awareness, not to.

"Slide four," he said instead.

She looked at him.

"The assumption about network stability. You were frowning at it before you'd asked the question."

"I frown when I'm thinking."

"I know." He said it before he'd decided to say it. Neither of them acknowledged the implications. "The answer I gave in the Q&A was accurate but incomplete. The temporal variance problem is more significant than I made it sound. The extension we're developing would revise approximately thirty percent of the stability predictions in the current model."

She was quiet for a moment. "That's substantial."

"Yes."

"Why didn't you say that in the Q&A?"

"Because thirty percent sounds alarming out of context and I didn't want to spend the remaining time managing that impression instead of explaining the science."

She considered this. He could see her deciding whether to accept it as a reasonable answer or press further. She pressed. "Is that the only reason?"

He looked at her. "What would the other reason be?"

"That you're more comfortable with a version of the answer that doesn't require defending the limitation publicly."

The accuracy of this was, briefly, annoying — not because it was unfair but because it was precise, and he had spent enough time with imprecise criticism to be caught off-guard by the other kind. "That's also true," he said. "Both are true."

Something shifted in her expression — small, quickly controlled, but there. "That's a more honest answer than I expected."

"You asked an honest question."

They stood in the corridor with the conversation between them, which had started as academic and had become something adjacent to personal in the way that conversations between people with history sometimes did regardless of both parties' intentions. He was aware that they had been standing here longer than two colleagues who had met briefly at a conference would typically stand. He was aware that two of his lab members had glanced in this direction. He was aware of all of this and did not move.

At the far end of the corridor, Henderson's voice: "Sicheng — a moment, when you're ready."

He looked toward the voice, then back. She was already repositioning — a small shift of weight, a slight turn, the body language of someone preparing to be somewhere else. Efficient. He had always found her efficiency slightly devastating.

"The argument you were going to make," he said. "About the stability assumption. You didn't finish it."

"You were called away."

"I know. You were going to say something about phase transition modeling — I could tell from how you framed the follow-up."

She looked at him for a moment with an expression he couldn't fully read, which was unusual. He was generally good at reading her. The four years had introduced a variable he hadn't fully accounted for. "Another time," she said.

It was not quite a promise. It was not quite nothing.

She walked back toward the Whitmore group. He watched her go for two seconds — exactly two, he counted without meaning to — and then turned and went to find Henderson.

---

The lab was quiet by nine. Sicheng preferred it this way: the work that happened after everyone left was different from the work that happened during the day, cleaner somehow, more like thinking and less like performing. He had two hours of data analysis ahead of him and a conclusion that was not quite cooperating and the pleasant, focused dissatisfaction of a problem that was resisting but not impenetrable.

His email pinged.

Henderson, forwarding a departmental memo with a single line appended: *Thoughts? Could be useful for the final chapter of your thesis.*

He opened the memo. The joint seminar series — a formal proposal now, with dates and a framework and a list of participating labs that included, of course, both Henderson and Whitmore. Biweekly sessions. Rotating presenters. Cross-lab discussion encouraged. The language was the language of institutional collaboration: productive, interdisciplinary, mutually beneficial. He read it with the attention he gave to everything that arrived from Henderson, which was complete and without filtering.

*Could be useful for the final chapter of your thesis.* This was Henderson's way of saying: I think you should do this. Which was Henderson's way of saying: I expect you to do this. Which was simply the way things worked, and had always worked, and which Sicheng had long since stopped experiencing as a constraint and had recategorized as structure. He functioned well within structure. Structure was predictable. Predictable was manageable. He had built his entire Edinburgh life on this principle and it had served him correctly.

He typed *I'll review the proposal* and sent it before he could deliberate further.

He returned to the data.

He worked for another ninety minutes, and the conclusion cooperated eventually, as conclusions did when you were patient enough and willing to consider that the problem might be in your assumptions rather than in the data itself. He closed the analysis file, cleared his desk with the habitual efficiency of someone who disliked leaving work in disorder, shut down his laptop.

In the taxi home, Edinburgh moved past the windows in its nighttime configuration — the castle lit above the Old Town, the wet streets reflecting streetlamps, the particular dark of the hills beyond the city's edge. He had made this journey hundreds of times. He knew it the way you know a route you've walked so often it requires no attention, which meant his attention was free to go elsewhere.

He thought about the answer he hadn't given in the corridor.

Not the one about slide four — he'd given that one, finally and fully, which had been the right thing to do even if he hadn't planned to. The one that came after, the one that was beginning to form when Henderson called him away. She had been about to make an argument — he'd seen it in the way she'd positioned the follow-up question, the specific angle of approach that indicated she had a larger point and was building toward it methodically, the way she always built toward things. He had been about to hear it. He had been, for the first time in a seminar Q&A in longer than he could remember, genuinely uncertain which of them would turn out to be right.

She would know his answer had been incomplete. She always knew.

He looked out the taxi window at the rain and thought: *another time.*

He wasn't sure yet whether he was dreading it or waiting for it.

He suspected, with the honesty that Edinburgh nights seemed to require of him, that it was both. And that the two feelings, like most of the variables in this situation, were not as separable as he would have preferred. The joint seminar proposal was sitting in his inbox. He would review it in the morning, as he had said he would, and he would respond to Henderson with whatever answer was correct, as he always did.