Isolde came to the archive on a Thursday with a hard drive in her coat pocket and the expression of someone who had made a decision they were not going to unmake.\n\nElianor was at her workstation. She looked up when Isolde came through the door — not the reading room entrance that visitors used, but the direct entrance to the restoration booth, which required a keypad code and which Isolde had been given access to during the intensive case preparation months. She had kept the access. She had not used it since the finding was issued.\n\n\"Isolde.\"\n\nIsolde sat in the chair at the side of the booth — not Einar's chair, the chair on the other side — and set the hard drive on the desk.\n\n\"You don't have a meeting scheduled,\" Elianor said.\n\n\"No.\" Isolde looked at the drive. \"I have something. I've been trying to decide what to do with it for three weeks.\"\n\nElianor looked at the drive. It was an external drive — the same brand as the archive used for long-term storage, the high-reliability kind that atmospheric researchers used for data that needed to be preserved for decades. The label was handwritten. She could read, in Isolde's precise handwriting: *Aurora atmospheric data — 40 year longitudinal — Isolde Varr.*\n\nShe looked at Isolde.\n\n\"Forty years,\" Isolde said. \"The observatory has been running its own atmospheric monitoring program since before I started working there. Independent of the archive's systems. Different methodology, different equipment, different geographic focus — the ridge and the upper atmosphere rather than the harbor district.\" She paused. \"The datasets overlap in significant ways. They were developed independently and no one has ever cross-referenced them.\"\n\n\"Why not?\"\n\n\"Because the observatory's data wasn't public. The director has kept it internal — research purposes only, not shared with outside institutions.\" She paused. \"I've been an employee there for six years. I've had access to the full dataset for three.\"\n\n\"And you've been cross-referencing.\"\n\n\"Yes. Informally, in my own time, as a separate research project.\" She looked at the drive. \"The cross-reference shows something the case documentation doesn't currently include. The atmospheric conditions on the night of Frostline 19 were more precisely documentable than the archive's harbor monitoring data alone could establish. The observatory's ridge sensors captured the full magnetic field behavior during the aurora event — not just the interference effects on signal capture, but the actual field geometry. The pressure on the navigation systems was not uniform. There were specific sectors of the ice where the interference was more intense.\"\n\nElianor looked at the drive.\n\n\"The carriage the passengers were concentrated in,\" Isolde said, \"was in the sector of highest interference density. The navigation systems would have been effectively blind in that specific section of the route.\" She paused. \"This is not in the current documentation. The archive's data establishes interference was present and contributed to the navigation failure. The observatory data establishes that the severity was localized — and that the localization was consistent with the impact point.\"\n\nElianor said: \"You've been sitting on this for three weeks.\"\n\n\"Yes.\"\n\n\"Since the preliminary finding.\"\n\n\"Yes.\" Isolde looked at the table. \"The preliminary finding uses the archive's data. It's sufficient — the panel found it sufficient. But the full review, if it proceeds, will have to withstand Veyrline's technical challenge. Their lawyers will argue that the aurora interference was a general weather condition, not a specific contributing factor to the impact location. The observatory data directly refutes that argument.\" She paused. \"It moves the finding from *reasonable probability* to *substantial evidence*.\"\n\nElianor said: \"Why are you giving this to me now?\"\n\nIsolde looked at her. The direct look — the professional directness that Elianor had come to understand was not coldness but a kind of honesty, the refusal to soften what was true.\n\n\"Because I've been afraid,\" she said. \"The observatory director doesn't know I've been doing this research. The data is technically the observatory's intellectual property. If I give it to you and it enters the formal record, there will be questions about how it was obtained and what my authorization was.\" She paused. \"I could lose my position.\"\n\n\"You might,\" Elianor said.\n\n\"Yes.\" Isolde looked at the drive. \"I've been trying to decide if that matters more than what the data shows.\"\n\nElianor looked at the drive. Forty years of atmospheric monitoring from the ridge. The kind of dataset that represented decades of institutional work, the accumulated record of an observatory that had been watching the sky above Aurevik before either of them had been born.\n\nShe said: \"This is the kind of data the archive was designed to use.\"\n\n\"I know.\"\n\n\"It doesn't make the decision easier for you.\"\n\n\"No,\" Isolde said. \"It doesn't.\"\n\nThey sat for a moment. The archive's monitoring systems ran their quiet hum. Outside the narrow windows, the October afternoon was pale and clear, the first hard cold of the season settling over Aurevik.\n\nElianor said: \"If you give it to me, I'll use it correctly. Full attribution, proper citation, the observatory credited as the source. The documentation will show exactly how the data was obtained and that it was provided voluntarily.\"\n\n\"That won't protect my position.\"\n\n\"No,\" Elianor said. \"But it will mean that what you did is accurately recorded. That you made a choice and the record will show the choice was yours.\"\n\nIsolde looked at the drive for a long time.\n\nShe said: \"He was fifteen.\"\n\nElianor said: \"Yes.\"\n\n\"The carriage impact point was not random. The navigation system failure was not uniform. The sector the passengers concentrated in was the sector with the worst interference.\" She paused. \"If the navigation system had been functioning, the train operator would have had warning in time to reduce speed. The impact at that velocity into that section of ice — \" She stopped. \"It was not inevitable. The conditions created by the operator's decision and the aurora event together produced a specific, localized catastrophic failure.\"\n\n\"I know,\" Elianor said.\n\n\"Your brother died because of a localized catastrophic failure that was preventable.\" Isolde looked at the drive. \"And I have the data that proves the localization.\"\n\nElianor said: \"You came here.\"\n\n\"Yes.\"\n\n\"You've already decided.\"\n\nIsolde was quiet for a moment. \"I decided two weeks ago. I've been waiting to see if I would change my mind.\" She looked at Elianor. \"I haven't.\"\n\nShe pushed the drive across the desk.\n\nElianor looked at it. She thought about what it meant — the shift from *reasonable probability* to *substantial evidence*, the localization of the interference, the specific catastrophic failure, the decision-making chain that had put passengers in a carriage in a sector of maximum interference with navigation systems that would have been effectively blind.\n\nShe thought about Niven saying *hypothesis: mountains.* She thought about him on that train with his recorder, expecting to come home.\n\nShe picked up the drive.\n\nShe said: \"I'll need to verify the data against the archive's records before I can formally integrate it. That takes time.\"\n\n\"How much time?\"\n\n\"Two weeks. Maybe three. The cross-reference work is significant.\" She paused. \"I'll need to contact the observatory director formally before I use the data in any documentation. To give them the opportunity to authorize it properly.\"\n\nIsolde said: \"The director won't authorize it.\"\n\n\"Probably not. But the formal request needs to happen. It needs to be in the record that I asked.\" Elianor set the drive down on her desk. \"And if the director refuses, we discuss options. Whistleblower protections, the panel's subpoena authority, the public interest provisions.\" She paused. \"You're not alone in this.\"\n\nIsolde looked at her.\n\n\"You've been alone with this for three years,\" Elianor said. \"You don't have to be anymore.\"\n\nA silence. Isolde looked at the drive on Elianor's desk.\n\nShe said: \"Einar knew I had the data.\"\n\nElianor looked at her.\n\n\"Not the details. Not the scope of the longitudinal dataset. But he knew I had something I wasn't sharing.\" She paused. \"He never pushed. He's — he doesn't push. Even when the stakes are high, he waits for people to arrive at their own decisions.\" She looked at the window. \"That's frustrating sometimes.\"\n\n\"I know,\" Elianor said.\n\n\"He was right to wait.\" Isolde stood. She looked at the drive one more time, then at Elianor. \"Use it well.\"\n\n\"I will.\"\n\nIsolde left.\n\n---\n\nElianor sat alone in the restoration booth with the drive on her desk.\n\nShe sat there for perhaps ten minutes before she picked up her phone and texted Einar: *Isolde brought the observatory's atmospheric data. 40-year longitudinal. It localizes the interference to the impact sector.*\n\nThe reply came in four minutes: *How complete?*\n\nShe texted back: *Complete. The cross-reference will take two to three weeks. It will change the finding category.*\n\nA longer pause. Then: *Is she all right?*\n\nShe thought about Isolde's expression when she pushed the drive across the desk. The particular quality of someone who had made a costly decision and was living with the cost.\n\nShe texted: *She will be.*\n\nShe put the phone down. She looked at the drive.\n\nShe thought about the case — about the forty-seven pages of documentation, the sworn statement, the preliminary finding, the families who were waiting for the full review. She thought about *substantial evidence* and what it meant for the panel's options. She thought about Veyrline's technical challenge and the lawyers who would argue general weather conditions and the data on the drive that directly refuted that argument.\n\nShe thought about Isolde, six years at the observatory, three years with this data, two weeks deciding.\n\nShe thought about all the decisions that had been made about when to act and when to wait, and the particular courage it required to know the difference.\n\nShe opened a new document on her workstation. She labeled it: *Cross-Reference Analysis — Observatory Longitudinal Data — Atmospheric Localization.*\n\nShe began.\n\n---\n\nShe was still working when Einar arrived at the archive two hours later.\n\nHe came through the door and looked at the drive on her desk and then at her.\n\nShe said: \"She came in herself. Brought it without being asked.\"\n\n\"I know.\" He set his coat on the hook. \"She told me she was thinking about it. Last week. She didn't say she'd decided.\"\n\n\"She decided.\"\n\nHe sat at his side table. He looked at the drive. \"The localization data — do you know yet if it's as significant as she thinks?\"\n\n\"I've been doing the preliminary cross-reference for two hours. It's more significant than she thinks.\" She paused. \"The interference geometry on the night of Frostline 19 was precise enough that the impact location was predictable from the atmospheric data alone — if anyone had been looking at it in real time and understood what they were seeing.\" She looked at her screen. \"The navigation system failure in that sector was not a surprise if you had the observatory's ridge sensor data. It was a foreseeable outcome of running that route in those conditions.\"\n\nHe was very still.\n\n\"That's what the data shows,\" she said. \"Foreseeable. Not just possible. Foreseeable with available information, at the time, from systems that existed and were running.\"\n\nHe said: \"Magnus Veyr had access to the observatory's data.\"\n\n\"The observatory has a data-sharing agreement with Veyrline Transit. It's in the institutional records.\" She looked at the screen. \"The agreement was signed four years before Frostline 19. The purpose was precisely to allow the transit authority to factor atmospheric conditions into route decisions.\"\n\nA silence.\n\n\"He knew,\" Einar said. Not a question.\n\n\"He had the means to know,\" she said. \"Whether he specifically reviewed the night's data before authorizing the route — that's what the full review will establish.\" She paused. \"But the data-sharing agreement is in the record. His organization had access to information that would have indicated extreme navigation risk in the specific sector where the impact occurred.\"\n\nShe looked at the drive.\n\nShe thought about *the weight was wrong.* Magnus Veyr, in the transit authority lobby, saying those five words. Not *I didn't know.* Not *we were misinformed.* The weight was wrong. The weight he had assigned to the risk, the decision, the lives of the people on that train.\n\nShe said: \"This is what we needed.\"\n\n\"Yes,\" Einar said.\n\n\"It will take two to three weeks to verify and document. Then the formal request to the observatory director. Then the panel integration.\" She looked at the screen. \"But it's here.\"\n\nHe looked at the drive on her desk. He did not say anything for a moment. The archive's monitoring systems ran their quiet hum. Outside the narrow windows, the October afternoon was finishing toward evening.\n\nHe said: \"Niven would have understood what this data means. He would have understood it immediately.\"\n\nShe looked at the drive. \"Yes. He would have wanted to see the visualizations.\"\n\n\"He would have had opinions about the color scheme.\"\n\nShe almost laughed. She let herself. Just slightly — the almost-laugh that had been available more and more frequently in recent months, the way of almost-laughing about Niven that was not less grief but a different quality of grief, one that had room in it for the joy of having known him.\n\nShe said: \"He always had opinions about color schemes.\"\n\n\"His filing system in the tower was extremely opinionated.\"\n\n\"He was opinionated from a young age.\" She looked at the drive. \"He would have been opinionated for a long time if he had had the chance.\"\n\nA silence that held the weight of that — not heavily, not with the crushing quality the silence used to have when she thought about what he would have been. Just with the weight of it. Just present.\n\nShe said: \"Thank you. For coming back. For all of this.\"\n\nHe looked at her. \"You said that before.\"\n\n\"It bears repeating.\" She looked at her screen. \"Go. I'm going to work for another hour. I'll see you in the morning.\"\n\nHe put on his coat. He looked at the drive one more time.\n\nHe said: \"Isolde is going to be all right.\"\n\n\"Yes,\" she said. \"She is.\"\n\nHe left.\n\nShe worked until seven. Then she shut down her workstation, logged the preliminary analysis in the case documentation, and put the drive in the archive's secure storage with the case materials. She logged it in the evidence chain.\n\nShe went home.\n\nShe lit the lamp. She stood at the window.\n\nShe thought: the data is here. The full picture is almost complete. The foreseeable outcome, the data-sharing agreement, the weight that was wrong — it was all going to be in the formal record. All of it.\n\nShe thought about Niven on that train with his recorder. She thought about the foreseeable outcome that someone with available information had chosen not to act on.\n\nShe thought about the archive keeping what it was given until the conditions were right.\n\nThe conditions were getting right.\n\nShe stood at the window for a long time.\n\nThen she made dinner, and the harbor bell rang, and the October night settled over Aurevik, and the archive three blocks away held what it held, faithfully, as it always had.\n\n---\n\nThe next two weeks were the most technically demanding work Elianor had done in her career.\n\nThe cross-reference analysis required building a correlation model between two datasets that had been developed by different institutions using different methodologies over a forty-year period. The archive's data was organized around signal capture events — discrete aurora events logged by date, intensity, and frequency characteristics. The observatory's data was organized around continuous atmospheric monitoring — temperature gradients, magnetic field strength, geomagnetic indices measured in real time across the ridge stations.\n\nTo cross-reference them meaningfully, she had to translate between two different conceptual frameworks for describing the same phenomenon. The archive thought in terms of events. The observatory thought in terms of fields. Both were correct. The correlation required understanding how an event-based description and a field-based description mapped onto the same physical reality.\n\nShe worked on it during archive hours and after archive hours and on Saturday mornings. She did not rush it. The temptation to rush was present — the data was there, the significance was visible even in the preliminary passes, and she wanted to get to the full analysis — but she did not rush it. The documentation had to be precise. It had to be reproducible. It had to be the kind of work that seven expert panel members could read and follow and find no fault with.\n\nEinar reviewed the model structure at the end of the first week. He found three assumptions she had made about the observatory's measurement protocol that required clarification from Isolde. She contacted Isolde. Isolde provided the clarifications. She revised the model.\n\nThe result, at the end of the second week, was a forty-three page technical appendix that would become part of the case documentation.\n\nThe key finding was this: on the night of Frostline 19, the magnetic interference was not uniform across the route. The aurora event had created a localized intensification in the grid sector corresponding to the route's eastern approach, where the train would have been operating at reduced visibility due to ice fog. The navigation system's electromagnetic guidance, which relied on the same frequency range as the aurora interference, would have experienced approximately sixty-three percent degradation in that sector. In practical terms, this meant the system was providing positioning data that was significantly inaccurate at precisely the moment when accurate positioning was most critical.\n\nThe data-sharing agreement between the observatory and Veyrline Transit meant that this information — or information from which this conclusion could have been derived — was available to Veyrline's operations team at the time of the authorization decision.\n\nThe technical appendix did not conclude that Magnus Veyr had reviewed the specific data. It did not conclude that he had understood its implications. It concluded that the information existed, was accessible, and was within the scope of what a competent transit operator exercising reasonable care would have reviewed before authorizing a route in aurora conditions.\n\nReasonable care. The standard the panel would apply.\n\nShe sent the appendix to Einar for final review. He returned it with four comments, all technical, all improvements. She integrated the revisions.\n\nShe sent the formal request to the observatory director.\n\nThe director's response came in five days: a formal refusal to authorize the use of the observatory's data in external legal proceedings, citing institutional policy, intellectual property concerns, and the assertion that the data-sharing agreement with Veyrline did not extend to adverse legal proceedings.\n\nShe had expected this. She had prepared for this.\n\nShe forwarded the refusal to the panel's lead counsel with a cover letter outlining the whistleblower protections available to Isolde under regional research data legislation and requesting the panel's formal subpoena authority for the data.\n\nThe panel's response came in four days: formal subpoena issued.\n\nThe observatory director, faced with a legal obligation to produce the data, produced it. The institutional authorization now existed. The data entered the formal record.\n\nIsolde, when she heard, said: \"The director is furious.\"\n\nElianor said: \"Yes.\"\n\n\"My contract review is in February.\"\n\n\"I know.\" Elianor looked at the case documentation on her screen. \"The panel's whistleblower finding provides significant protection. And the public interest value of the data — a panel counsel confirmed that would be a factor in any challenge to your employment status.\"\n\n\"Confirmed by a lawyer.\"\n\n\"Yes.\"\n\nIsolde was quiet for a moment. Then: \"The data is in the record.\"\n\n\"Yes. Formal subpoena, full attribution, proper citation. The observatory is credited as the source. Your name is on the analysis as co-author.\"\n\n\"You didn't have to do that.\"\n\n\"You did the cross-referencing work with me for two weeks.\" Elianor looked at the screen. \"Your name is on the analysis because you did the work.\"\n\nA silence.\n\nIsolde said: \"The director will not see it that way.\"\n\n\"The director's perspective on attribution is not the panel's concern.\"\n\nAnother silence. Then Isolde said, quietly: \"The localization data is exactly as significant as I thought.\"\n\n\"More,\" Elianor said. \"The sixty-three percent navigation degradation in that sector — it's not arguable. The physics is the physics.\"\n\n\"Yes.\"\n\n\"Thank you,\" Elianor said. \"For coming in. For pushing the drive across the desk.\"\n\nIsolde was quiet for a long moment. When she spoke, her voice had lost the particular professional precision it usually maintained.\n\n\"He was fifteen,\" she said.\n\n\"Yes,\" Elianor said.\n\n\"And someone had the data to know better.\"\n\n\"Yes.\"\n\nIsolde said: \"Good. Then it needs to be in the record.\"\n\nShe ended the call.\n\nElianor sat in the archive with the case documentation on her screen and the October rain against the narrow windows and the monitoring systems running their quiet background work.\n\nShe thought about the conditions required for the right thing to finally happen. The decades of independent data collection by two institutions that had never spoken to each other. The six years Isolde had spent accumulating access to information she had not known how to use. The years of cross-referencing done in private time, as personal research, unsupported and unsanctioned.\n\nAll of it becoming useful at exactly the right moment.\n\nShe thought: this is also what the archive does. It keeps what it is given. It does not know, when it receives data, whether the data will be needed. It keeps it anyway. And then, when the conditions are right, the kept things become exactly what was required.\n\nShe thought: Niven wrote that at fourteen. He would have been able to explain this in four sentences. He would have made it sound obvious.\n\nShe went back to work.\n\nThe harbor bell rang the afternoon hour.\n\nThe record was accumulating.\n