Two thousand people was a number that sounded large until you broke it into its components and distributed it across three shifts, at which point it became a number that sounded exactly right for the problem it was solving and slightly insufficient for the problem it was going to be solving in six months.
Jonas had run the staffing model four times before settling on the structure he presented to Rulf, and the model had been revised twice more during implementation when the reality of the workforce diverged from the projection in the specific ways that realities always diverged from projections — not in direction but in detail, the small frictions of human organization that no amount of careful planning fully anticipated.
The administrative staff occupied the largest of the wooden office buildings in the facility's northern quarter — twelve people at full complement, a number that sounded small until you considered that each of them was responsible for a specific slice of the information architecture the facility ran on. Intake records: what raw material entered the facility, in what quantity, from which supplier, on which date, at what cost. Output records: what finished components left each workshop floor, in what quantity, toward which assembly point. Financial records: wages owed, wages paid, supplier accounts, the Duke's investment ledger against which Jonas maintained a running projection of return timeline that was updated weekly. Quota tracking: each section's daily output against the target rate, the variance flagged and explained in writing, the trend line visible at a glance for anyone who needed to assess whether the training curve was progressing as modeled.
The administrative staff were, without exception, people who had been literate before they arrived. This was a smaller category than Jonas had hoped for and he was already thinking about what it meant for the facility's long-term operational independence that the pool of people who could do this work was constrained by an educational system that had never considered the possibility they might need to. He filed this as a problem with a longer solution timeline than the current one and kept moving.
The logistics staff were the facility's circulatory system — the people who moved material from the causeway inspection point to the appropriate storage location, from storage to the workshop floor that needed it, from one section of the floor to the next as the assembly sequence required. They were the workers who would be least visible in any accounting of what the facility produced and who would be most immediately felt if they stopped functioning correctly, which was a relationship Jonas had noted as a general principle of complex systems and which he had accordingly staffed with more care than the role's surface description suggested it warranted. Forty people per shift, organized in teams of eight with a team lead whose primary qualification was spatial memory and the ability to maintain awareness of multiple concurrent tasks.
The foundry and blacksmithing operations were the facility's anchor. The kilns ran on coal brought in by the logistics teams from the supply chain the Duke's purchasing office had established before the recruitment drive, and the foundry floor was where raw ore became the standardized components the assembly sequence required. This was where the six original blacksmiths from the first steam engine project were deployed as section managers, and this was where the training was most demanding and the timeline to competence was longest — you could train a logistics worker to move material correctly in a week, but training a foundry worker to cast a cylinder bore to the dimensional tolerance the engine required took considerably longer, and the consequences of doing it wrong were not merely inefficiency but a component that would fail under operational pressure at a moment nobody wanted.
The assembly staff and carpenters worked the final stages — fitting the machined components into the engine's structural frame, the sequence that Jonas had designed to be executable by people with basic mechanical training following explicit procedure, and which was accordingly the section where the original skill requirement was lowest and the training period was shortest. The carpenters handled the wooden components of the packaging and transport infrastructure alongside the structural maintenance that a facility of this scale generated continuously.
All of them were currently being trained in the metric system.
This had been the decision Jonas had made in the third week of construction and had not reconsidered since — if he was going to build a production environment that operated on dimensional precision, it would operate on a measurement system that was internally coherent rather than the accumulated historical accidents that passed for measurement standards in this world, where a foot in Munich was not quite the same as a foot in Frankfurt and a pound of iron from one supplier was not quite the same as a pound of iron from another. The metric system was not a preference. It was the foundation that everything else required.
The training was proceeding. Slowly, in places. The concept of a system based on powers of ten was straightforward in principle and required genuine mental recalibration in practice for people who had spent their entire lives in a different system. Rulf had found two workers in the administrative intake who had backgrounds in academic mathematics, and Jonas had pulled them both into the training operation rather than their original assignments, because ten people who understood the system well enough to teach it was worth more than two people running ledgers.
He estimated three to four weeks before the first production cohort reached the competence threshold. He had told the Duke six months to first engines from groundbreaking, which meant he was running ahead of schedule, which was the correct direction to be running.
He trained every morning without exception, as he had since Bavaria.
What the training produced now was not comparable to what it had produced in Bavaria, which was itself not comparable to what it had produced in Munich, which was to say that the compounding nature of the process had continued to compound and the body it was compounding into was not the body he had arrived in.
He was fifteen years old and he could lift approximately one ton. This was not a figure he had tested with a scale — he had no scale calibrated for that range — but the objects he could move without particular strain had allowed him to extrapolate a lower bound with reasonable confidence. The physical development was real, was measurable, and remained entirely invisible to any casual observer because the body still looked like what it had always looked like from the outside: a pale, slightly lean adolescent of average height with dark hair and eyes that had the faint luminescence at the iris that he had noted in the first week and never fully explained.
The ability's development was the more significant change, and it was one he tracked with considerably more attention than the physical metrics.
The absorption radius had expanded. In the early weeks he had needed contact or near-contact to draw energy deliberately. By Munich he had been operating at arm's length. Now, at fifteen, with ten months of continuous development behind the ability's operational range, he could draw from ambient sources at a radius of approximately ten meters — not with the same precision as contact absorption, but with enough control to be intentional rather than reflexive.
The two developments he had been working on for the past two months were different in character from the previous extensions of the ability's range and capacity. They were not about intake. They were about application — specifically, about what he could do with stored energy that did not involve releasing it outward as heat or kinetic force.
The first he had named the kinetic buffer.
The concept had developed from an observation he had made in the training yard at Munich, during a session where Gerhard had landed a strike that connected more fully than either of them had expected. The impact had been absorbed — not deliberately, but reflexively, the ability drawing in the kinetic energy of the blow with the same automatic response it showed toward any incoming energy. What was interesting was not the absorption but the speed of it. The absorption had happened faster than the physical sensation of impact. The body had received the energy before it had processed the force as damage.
He had spent six weeks extending this into a deliberate defensive mechanism. Stored energy, maintained at the body's surface in a thin continuous layer — not visible, not detectable as Flux because it was not Flux, but present, oriented outward, intercepting incoming kinetic force and converting it to absorbed charge before the force could translate to tissue damage. A blade striking this layer did not bounce off. It met resistance that was not physical hardness but energetic conversion — the edge connecting with a surface that absorbed its kinetic energy faster than it could transfer it.
It was not perfect. It required maintained charge. It had a threshold above which the conversion rate could not keep pace with the incoming energy — he had not yet determined what that threshold was, which was itself a variable he intended to establish through controlled testing rather than operational discovery. And it did not work against magical attacks in the same way, because magical attacks carried Flux that the ability absorbed through a different mechanism.
But against conventional weapons — against a blade, a crossbow bolt, a heavy object applied with force — it worked. He had confirmed this with enough controlled testing to be confident of the principle and uncertain only about the edge cases.
The second development he had named molecular binding.
This one was stranger, and he held it with more theoretical uncertainty than the kinetic buffer, because its mechanism was less clearly analogous to anything in the physics he had started from. What he had found, in the third month of working with the ability at the molecular scale, was that stored energy could be directed into the structural cohesion of material — not just his own body's material but matter he was in contact with. Applied inward to his own tissues, it produced a reinforcement effect: the cellular structure of muscle and bone becoming temporarily more resistant to deformation than biology alone accounted for. Applied outward through contact to other material, it could either reinforce that material's structural integrity or, in a different mode he had not yet fully characterized, introduce instability into it.
The practical result of both developments together was that his body, in a prepared state, was not easily damaged by the category of threats this world's conventional weapons represented. He was not invulnerable — he was careful never to use that word even in his private notation, because invulnerability was a claim that invited the specific kind of testing he preferred to avoid. He was, more precisely, significantly more resistant to conventional physical harm than a fifteen-year-old boy had any business being, while remaining entirely capable of being harmed by things that operated at a level the ability's current development did not address.
This was sufficient for now.
He noted it as sufficient and filed the edge cases for future work.
The distillery had been the third project running in parallel with the facility's construction, and it was the one he had been most careful to keep quiet.
Not secret — it was inside the facility's perimeter, conducted with equipment built in one of the smaller workshops, with a staff of seven who had been selected for discretion alongside their practical skills. Not secret, but not announced, because the announcement would raise questions about funding and purpose that he was not yet ready to answer publicly.
The reasoning was straightforward. The steam engine revenue would be significant, but it would flow through the Duke's commercial structure, with the Duke's share distributed according to the fifteen percent arrangement and the operational costs deducted before Jonas's portion was calculated. That was a legitimate and sufficient arrangement for the facility's ongoing needs. What it did not provide was capital that Jonas controlled independently — capital that could fund projects the Duke did not know about and had not approved and might, if asked, have opinions about.
He needed that capital.
The German Empire's drinking culture ran on two things: ale and wine. Ale in the cities and working settlements, wine among those who could afford the preference and the storage. Both were produced widely, consumed consistently, and purchased without particular enthusiasm by people who had no alternative to compare them against.
Jonas was going to give them an alternative.
Three of them, in fact.
Beer first — a bottom-fermented product that the brewing traditions of this world had never developed, the cold fermentation process producing a clarity and a crispness that ale's warm fermentation did not. He had studied the existing ale production methods in the Duke's library and identified the specific process differences that would distinguish what he was making from what the market already had. The result was not merely better ale. It was a categorically different product, with a different taste profile, a different appearance in the glass, a different relationship to temperature and storage. A person who had only ever drunk ale would not know what to make of their first beer except that it was not what they were used to, and not what they were used to in a direction that was better.
Whiskey second — a grain distillate that this world's distillation tradition had not produced in this form, the aging process in charred wooden casks developing the amber color and the complex flavor architecture that no existing spirit in the German Empire resembled. It was a long-term investment. The aging had to start now for the product to be commercially viable in two to three years. He had started it.
Vodka third — a high-purity grain spirit, the distillation process refined toward a clarity and neutrality that this world's rougher spirits did not approach. It had two uses and he was pursuing both simultaneously: a commercial product for a market segment he was confident existed, and a base for the medical-grade pure alcohol he was producing in a parallel operation for the facility's own needs. The facility's medic received one hour of instruction from Jonas weekly on antiseptic application, wound treatment, and the framework for using pure alcohol in ways that this world's medical practice had not systematized. The pure alcohol output from the vodka distillation went directly to the medical stores.
He did not drink any of it. He had not drunk alcohol with any regularity in his previous life and he had no particular interest in starting. He could assess the output by its clarity, its smell, and the reactions of the workers he used as practical calibration instruments — people whose palates gave him the consumer response data that his own detachment could not.
The assessments were encouraging.
He was not selling any of it yet.
This was the discipline the operation required. The product was ready — the beer in particular had been ready for three weeks — and the revenue was sitting in barrels accumulating while he waited. But the stockpile was the strategy. Not a month's supply entering the market but a volume large enough to sustain consistent presence once the product was introduced. A new beverage that appeared in Munich's taverns and then ran out taught the market not to rely on it. A new beverage that appeared and remained available, at consistent quality, taught the market to prefer it.
He was building the second outcome.
The glass workers were part of this. Six of them, producing standardized bottles he had designed for each product line — consistent volume, consistent shape, a sealed stopper mechanism that improved on the wax and cork combinations currently in use. The bottle was a communication before the contents were tasted: here is a product that someone thought carefully about. In a market where ale was sold from barrels into whatever vessel the buyer presented, a consistent, sealed glass bottle was itself a quality signal.
He checked the stockpile inventory at the end of each week and noted the accumulation against the target volume he had set for market entry.
Not yet. But the line was moving correctly.
He filed this as proceeding to plan and went back to the facility floor.
There was always more to do.
He had found, somewhere in the past six months of doing it, that this was not the constraint it would once have appeared to be.
He noted this as well, without pursuing it further, and got to work.