The mistake was understandable, in retrospect.

Mira had been in Archive House for three days. She had developed, in those three days, a reasonable working map of who was who in the building: Tessa at the front desk, the two part-time cataloguers who came in on Tuesdays and Thursdays, the older man from the local historical society who used the reading room on Wednesday afternoons. She had not yet formally met the researcher she had been told about. She had heard him working in the cataloguing room on the ground floor, heard Tessa speak to him briefly in passing, knew he was somewhere in the building on most days.

She had not imagined him quite like this.

He was moving a flat box of archival materials from the cataloguing room to the stacks when she came down the main stairs at half past nine, and the combination of the work coat he was wearing -- heavy, well-used, the kind of coat that had been chosen for practicality -- and the matter-of-fact way he carried the box, and the fact that she had her mind entirely elsewhere, produced a conclusion that she reached instantly and incorrectly: maintenance staff, or possibly a volunteer helping with a transfer of materials.

"Excuse me," she said. "Can you tell me where the architectural records for the Halvern works are kept? The physical plans rather than the administrative files."

He stopped. He did not put the box down. He looked at her with an expression that was difficult to read precisely -- not offense, exactly, but a kind of assessment, the brief recalibration of someone who had just received unexpected information.

"Third floor," he said. "Far room. The large cabinets on the north wall." He paused, and then added: "The pre-1900 drawings are in the cabinet closest to the window. The post-1900 are in the next one. Tessa keeps them in reverse chronological order, which is unusual but makes sense once you understand the way she's organized the finding aids."

This was more information than a person who was simply moving boxes needed to know. Mira looked at him more carefully. He looked back. The box of archival materials was still in his hands. He was not making any move to set it down or to continue past her.

"Thank you," she said.

"The drawings are rolled, not folded," he added. "If you need to flatten any of them for examination, the large viewing table on the second floor is better than the table in the far room. The lighting in the far room is inadequate for detail work."

"Right," she said. "Thank you."

He continued toward the stacks. She continued toward the second floor.

She was on the third step of the main staircase when she stopped.

The staircase was beautiful in the specific way that functional Victorian architecture was sometimes beautiful -- not ornate, but well-made, the proportions right, the materials considered. The handrail was iron, polished dark by a hundred and thirty years of hands, attached to the wall with brackets that were not purely functional but had been shaped with a small additional thought for form. The treads were oak boards, replaced at some point, the new wood slightly lighter than the original at the edges. The stringer -- the structural element running up the wall side of the staircase -- was the original, she was almost certain. The detail at the top of it, where it met the upper landing, suggested the 1880s rather than a later renovation.

"How old is the staircase?" she called down, without thinking about it. He was still in the hallway, near the stacks entrance.

He stopped again.

"The iron is original," he said, not looking up from the box. "Probably 1887 or 1888, based on the bracket style and the quality of the casting -- this was during the period when the local smiths were producing very good decorative ironwork. The treads are replacements, probably 1960s given the wood and the fixing method. The structural stringer is original."

Mira stood on the third step and looked at the stringer. Then she looked at the man in the hallway, who was now looking at her with what seemed to be a marginally increased level of attention.

"I thought the bracket style was closer to 1883," she said. "The decorative vocabulary is slightly earlier -- the way the leaf form turns at the base of each bracket. Comparable brackets in the municipal building on the main street are documented as 1883."

A pause. "The municipal building brackets are cast from a different alloy. You can see the difference in the surface texture if you look closely. These are a heavier iron with a slightly higher carbon content, which came later in the local smithing practice." He paused again. "I agree the decorative vocabulary is slightly earlier. The smith was probably working from an older pattern book while using newer materials. It happens."

Mira came back down two steps to look at the nearest bracket. She looked at the surface texture. She could not immediately verify the alloy claim because she was not a metallurgist, but the surface texture was different from what she recalled of the municipal building brackets, and the distinction was the kind of thing that a person noticed only if they were paying attention to exactly the right thing.

"You're right about the surface texture," she said.

He did not say anything.

She came down to the ground floor landing. He was still standing near the stacks entrance with the box.

"I'm Mira Solenne," she said. "I'm researching the guild records."

"I know who you are."

"Then you know more than I do, because I still don't know who you are."

Something that might have been the beginning of an expression crossed his face and then did not become one. "Corin Albrecht," he said. "I do assessment work for the archive."

"Tessa mentioned you."

"What did she say?"

"That you know the physical objects better than anyone, and that you're particular about your working methods."

"That's accurate."

"She also said the staircase is 1887."

"The ironwork is 1887. The oak treads are 1960s."

"Right." She looked at him for a moment. "You're not going to tell me I was wrong about the 1883 date."

"You were wrong about the date. You were right about the decorative vocabulary. They're different questions." He shifted the box slightly. "The fact that the form is earlier than the date doesn't mean the date is wrong. It means the smith was working from older references. Which is worth knowing."

Mira looked at the staircase again. Then at him.

"How do you know the alloy composition without testing it?"

"Experience." The word was short and accurate and told her very little.

---

Tessa appeared from behind the front desk at this point, moving with the unhurried efficiency that was apparently her constant register. She looked at both of them -- first at Corin, then at Mira, then back at Corin with a slight expression that communicated something Mira could not entirely decode.

"I see you've met," she said.

"We've established each other's names," Mira said. "We're still working out the rest."

"The staircase," Tessa said, looking at the bracket nearest to her.

"He says 1887," Mira said. "I said 1883."

"He's right," Tessa said, without hesitation. "Though it's a reasonable confusion." She looked at Corin. "The third-floor drawings are unlabeled."

"I'll label them," he said, and continued toward the stacks.

Mira watched him go. Tessa watched Mira watch him go.

"He knows the building very well," Mira said.

"He knows most of the buildings in the historical core well," Tessa said. "It's part of the assessment work. Materials, dates, construction methods." A pause. "He's usually more forthcoming when the question is about an object rather than a building. Buildings make him more careful."

"Why?"

Tessa considered this with the particular quality of consideration she had that made everything she said feel like a considered judgment rather than a casual remark. "Buildings are harder to be wrong about," she said finally, and went back to the front desk.

---

The argument about the staircase resumed twenty minutes later in the reading room, which Corin had apparently decided to use after all, arriving quietly with a folder of papers and settling at the far end of the large table without preamble.

Mira had her notes spread in front of her, the fourth box of Smiths' records open, and was writing up the Halvern-associated members entry from the previous day when she became aware that he was present. She did not look up immediately. She finished the note she was making.

When she looked up, he was reading something in his folder. His working posture was very still: no unconscious movement, no fidgeting, the particular stillness of someone who had learned to be economical with physical energy while concentrating.

"The decorative vocabulary argument," she said.

He looked up. He did not seem surprised to be addressed. "What about it?"

"You said the smith was working from older references. I said the vocabulary was consistent with 1883. You said I was wrong about the date and right about the vocabulary." She paused. "But we were both right, which means we were arguing about different questions without knowing it."

"Yes."

"That happens often in this kind of work. Two researchers examining the same object and reaching different conclusions because they've prioritized different types of evidence."

"Also yes." He looked back at his papers.

"It would be more efficient," Mira said, "to establish at the beginning which types of evidence each person is working from."

"It would," he agreed, still reading.

"I work primarily from documentary and architectural evidence. I prioritize formal records, spatial analysis, comparative typology."

He looked up again, and this time the look was slightly different -- a calibration, the kind that happened when someone was deciding whether to take a conversation seriously or continue performing the version of it that required minimal engagement. He set the papers down. "I work from physical evidence. Materials, condition, construction details, what the object itself can tell you about its making."

"Documentary and physical evidence are complementary."

"When interpreted correctly, yes."

"How do you know when they're being interpreted correctly?"

"When they reach the same conclusion by different routes." A slight pause. "Or when the ways in which they differ are informative rather than random."

Mira looked at him for a moment. Then she opened her notebook to a clean page and wrote at the top: Albrecht, C. -- material evidence specialist. Prioritizes physical object over documentary record. Not a hierarchy, a starting point. Then she looked up. "What do you make of the Halvern-associated members entries in the Smiths' records?"

Something shifted very slightly in his posture. Not much. But she had been watching carefully. "I haven't looked at the Smiths' records recently," he said.

"But you know they're there."

A pause. "I know what's in most of the collections here."

"Including the cross-guild entries."

"Including those."

"And you have a working hypothesis about them."

"I have observations," he said. "Not a hypothesis. Observations require more data before they become hypotheses."

Mira nodded slowly. She had a feeling, which she filed under provisional, that Corin Albrecht had observations about the cross-guild entries that he had been sitting with carefully for some time. "Then we're at the same stage," she said. "Observations. Needing more data."

He looked at her for a moment. Then he looked at the folder of papers in front of him and then, without any particular drama, returned to whatever he had been reading.

The room was quiet. Outside, Alderheath went about its business in the grey morning. Mira went back to the Smiths' records. The staircase argument, by some tacit agreement, had been set aside -- not resolved, but set aside, the way you set things aside when the conversation had moved to something more important and might or might not return.

She was aware that she had not won anything and had not exactly lost anything either. The ground they had reached -- observations, needing more data, the implicit agreement that their working approaches were different and might be complementary -- was ground she had not anticipated reaching in one morning, with a person she had been told was particular about sharing what he knew.

She opened the next record and kept reading.

From the far end of the table, the sound of papers being turned. The scratch of a pen making a note. And beyond that, through the old windows, the low steady voice of the Verran, running under everything, unchanged.

She had been in Alderheath for three days. She would be here for three months. That felt like sufficient time to figure out what Corin Albrecht's observations were, if she was careful about how she approached it -- which she intended to be, because the alternative was approaching it carelessly, and careless approaches to interesting problems were not something she had ever been able to make herself do.

There was something in the specific manner of their disagreement about the staircase that Mira kept returning to over the next several days. Not the disagreement itself -- she had been wrong about the date, which was unfortunate but not interesting -- but the structure of how it had happened. She had looked at the decorative vocabulary and concluded date. He had looked at the alloy composition and concluded date. They had looked at the same object and found different evidence and reached different dates by different routes, and both of their routes had been correct, and the discrepancy between the dates was informative.

The smith had used new materials to realize an older design. This was not an unusual thing in itself. Craftspeople worked from pattern books, from memory, from the accumulated inheritance of their trade. What you knew how to make was not always identical to what was being made currently in your region. But it raised a question that she had not thought to ask until the argument forced her to think about it: where had the Alderheath smiths' design vocabulary come from? The patterns they were working from were older than 1887. How much older? From what source?

She added this to the list of things she was not yet ready to call a hypothesis.

Corin had gone back to whatever he had been doing in the cataloguing room and she had returned to the reading room and they had worked in the same building for the rest of the morning in the parallel quiet that was, she was beginning to understand, their default mode. Not comfortable silence exactly -- she was too aware of his presence two floors below for it to be quite comfortable -- but productive silence, the kind that allowed thought.

She had seen, in the course of the morning's conversation, three things about him that she noted carefully.

First: he knew things with certainty that most people would only have as approximations. The staircase brackets were 1887, he said, not probably 1887 or approximately 1887. The alloy content was higher carbon, not appears to be or seems to suggest. His knowledge of the physical objects in this archive was specific in a way that required either very long study or something that she did not yet have language for.

Second: there were gaps in that knowledge that were also specific. When she had mentioned the Halvern-associated members entry, his response had involved the phrase I haven't looked at the Smiths' records recently -- not I don't know about those, not I haven't seen them, but recently, as if the knowledge was temporarily unavailable rather than absent. She had noticed this and had not asked about it, because asking would have required explaining why she was asking, and she was not ready to do that yet.

Third: he was careful. Not shy, not unfriendly exactly, but careful in the specific way of someone who had reason to be careful about what he revealed and to whom. She recognized this. She had her own version of it -- the habit she had developed after her father, of guarding the research before she was ready to share it, of not saying what she thought until she was reasonably sure the thinking was right. The difference between her version and his version, as far as she could tell, was that hers was about professional caution and his was about something more personal.

She did not know yet what the more personal thing was. She expected she would find out eventually, because she was going to be in Alderheath for three months and they were going to be working in the same building for most of those months and the kind of person who was careful about what they revealed was usually also the kind of person who revealed it eventually, given enough time and enough evidence that the revealing was safe.

She was good at being evidence that revealing things was safe.

Or she had been. She was working on being that again.

She had been in Alderheath four days. She had three months. She had two threads -- the cross-guild entries in the documentary record and the unusual competence profiles in the physical objects -- that were pointing toward the same question. She had a working relationship with the archive's primary material specialist that consisted, at present, of one argument about a staircase and one exchange about maker's marks in which both of them had been careful not to say quite everything they knew.

That was, she thought, a reasonable beginning. She had started with less before and reached further. The important thing was not to start with everything but to start with something real -- something where the evidence was genuine and the question it raised was genuine -- and then to follow it carefully, all the way down to where it went.

She went down to the cataloguing room to ask Tessa about the 1897 Smiths' records and found that Corin Albrecht had already requested them for the following morning.

She was not sure whether that was convenient or inconvenient. She supposed she would find out.