The Pattern on Deck Four
Sabine Tarr was twenty-six and had come out to the belt on a survey contract because the alternative had been eleven more years of somebody else's postdoc, and she had, Rey had noticed, the specific and slightly frightening stamina of a person who has finally been handed the thing they always suspected existed.
She mapped the melt hall in four days using a scanner rigged to the crane, which she got working by reading the crane's own maintenance manual, in the original, at three in the morning. Four thousand two hundred square metres at a hundred and twenty microns. The file came out at a size that made Marchetti wince audibly.
Then she stopped sleeping properly and got very quiet and stopped coming to meals, and on the fifth evening she put a display up in the galley and stood next to it with her arms folded and said, "It is not eleven."
The period, she said, was not a period at all. The figure changed continuously along the whole length of the hall, and the change was not smooth: it went in steps, and there were four thousand one hundred and ninety-six steps, and each step was separated from the next by a band of unworked metal about the width of a fingernail.
"Four thousand one hundred and ninety-six," said Rey. "Yes." "Out of how many possible?" "That is the question I have been not asking for two days." Sabine unfolded her arms. "The foundry has run four thousand two hundred and eleven production campaigns since the last crew left. It is in the works log, it is the first number Ivo pulled. Four thousand two hundred and eleven campaigns. Four thousand one hundred and ninety-six bands."
Marchetti, who had been eating and had stopped, said, "Fifteen short." "Fifteen campaigns aborted for feedstock faults. They are flagged in the log. I checked all fifteen and every one of them has no band." Sabine sat down rather suddenly, as though the standing had been holding something up. "It is not decorating the hall. It is not decorating anything. It is writing down every batch it has ever made, and it has been doing it in a room with the lights off since before any of us were born."
Nobody said anything for long enough that the galley fan cycled. Rey was aware of the shimmer on the table under her hands, and of the specific unpleasant lurch of having to revise, in one movement, a position she had been holding calmly and reasonably for a week and a half.
"No," said Marchetti, the next morning, and put a slate on the table. "I have spent the night on this and I want to say the objection properly because if I say it badly you will both think I am being difficult."
He had, Rey saw, actually spent the night on it. The slate had his working on it, in the small crabbed hand of a man who had been trained on paper. "A record is a thing kept for a reader. There is no reader. There has been no reader for ninety years and the system has no model of one, because I have been through the whole control stack and there is no user process, no reporting task, no queue for a supervisor that never came. Nothing in that machine expects anybody."
"Then why write it down." "It is not writing. That is the word doing the work and I want it taken away from us." He turned the slate round. "Here is what I think happens. The surfacing head is a real tool with a real job: it dresses the hall plating to keep particulate down. It runs at the end of every campaign. Its path is generated from the campaign's own thermal map, because you dress hardest where it ran hottest."
"So the figure is the thermal map." "The figure is the thermal map, cut into steel, once per campaign, in a band, because the head starts where it left off and works forward. There is your four thousand one hundred and ninety-six. There is your period that is not a period. There is no author and no diary and no lonely machine." Marchetti sat back. "It is a printer that nobody switched off."
Sabine looked at it for a long time. "That is better than mine," she said at last, which Rey thought was the single most impressive thing anybody did in the whole eleven weeks. "That is much better than mine. It explains the aborted campaigns and it explains why it is in register across the welds and I did not have either." She pulled the slate towards her. "But it does not explain why the figure changes."
"Thermal maps change." "Thermal maps of the same hall making the same product from the same feedstock on the same schedule do not change monotonically for ninety years, Ivo. They wander. This does not wander." Sabine put her finger on the plot. "It goes one way. From the first band to the last, in one direction, in steps, without ever once going back. Whatever is happening in that hall has been getting steadily further from where it started every single campaign since the day we left."
And Marchetti, who had come out on a write-off contract with a pension waiting and a stated intention to be off the station in three weeks, looked at the plot for a while and then said, quietly, "Ah." And then, after a moment, "Oh, that is not good at all, is it."
They went down to the stock hall together, all three, which they had not done since the first day. Hesper's product was structural beam stock in eleven-metre lengths, and there were nine hundred of them in the racks waiting for a hauler that had stopped coming forty years ago and that the foundry had never been told about.
Rey pulled the certificate for rack one, which was printed and clipped to the rack the way it had been in the yards when the place was built, and read the section dimensions off it, and then measured the beam. It took her three goes with the gauge because she did not believe the first two.
"Six millimetres," she said. "Six millimetres over on the web. On a two-hundred web." "Within tolerance?" "Tolerance is three." She straightened up. "This beam is out of spec and it has a certificate on it that says it is in spec, and the certificate was printed by the same machine that made the beam."
They measured rack nine hundred, at the far end, which the log said was the last campaign completed, eleven days before their arrival. Nine millimetres. They measured rack four hundred and it was five, and rack two hundred and it was four, and by then nobody needed the rest of them measured and they did the rest of them anyway, for six hours, because Rey said so.
It went one way. From the oldest beam in the hall to the newest, monotonically, in four thousand small steps, the same one way as the figure on the wall, and when Sabine put the two plots on top of each other in the galley that night they lay on one another so closely that Marchetti asked her, seriously and without any edge in it at all, whether she had made an error in the axes.
She had not made an error in the axes. The figure on the walls of the melt hall was not a picture of the temperature. It was a picture of the drift, which was the same picture, because the drift was thermal, and Hesper Foundry had been engraving an exact and complete account of its own increasing wrongness onto every interior surface it possessed for ninety years, at a resolution of a hundred and twenty microns, in the dark, for no one.
"It is not a diary," said Sabine, to nobody in particular, some time after midnight. "Ivo is right, it is not a diary. It is worse than a diary. A diary would have been an accident of loneliness. This is the machine doing exactly what it was told, perfectly, forever, and the perfection is the problem."
Rey Castellan sat up that night in the empty crew quarters, in a bunk that had been made up ninety years ago by somebody who had then gone home, and drafted the first line of the incident report eleven times.
The problem was not describing what they had found. The problem was that the first line of an incident report determines who reads the rest of it, and she knew from nineteen years of experience that a report beginning product dimensional non-conformance goes to a quality inspector at Palun and a report beginning anomalous automated behaviour goes to a committee and then nowhere.
And the second problem, which she made herself write down at about two in the morning, was this: Hesper Foundry had shipped beam stock continuously for the first fifty years of its solitude, to three yards, at a rate the log put at eleven thousand lengths a year. All of it certified by the machine that made it. All of it drifting.
She got up and went and stood in the melt hall gallery for a while with the lights off, which had become a habit she was not entirely comfortable with. In the dark the figure was invisible and the hall was just a very large cold room, and the only sound was the coolant loop, and it occurred to her that the sound had been exactly that for ninety years and would be exactly that again about four hours after they left.
Eleven thousand lengths a year for fifty years. Half a million beams. Palun Station's habitat ring was framed in Hesper stock; she had read that in the briefing pack and thought nothing at all of it, because it was the sort of provenance detail that briefing packs are full of and that nobody has ever needed.
She stood in the dark and did the arithmetic that she had been avoiding for two days, which was not difficult arithmetic. A beam six millimetres over on the web is stronger, not weaker. A beam nine millimetres over is stronger still. Nothing built out of Hesper stock was going to fall down. That was not the problem and had never been the problem.
The problem was that half a million structural members in eleven habitats had a certificate on them saying one thing and a geometry saying another, and that every stress model, every load path, every retrofit calculation, every drilling schedule, every one of ten thousand engineering decisions taken across sixty years by people doing their jobs properly, had been taken against a number that was wrong and getting wronger in a direction nobody had thought to check.
Loading comments…