The Ceiling
There are things the Known World cannot do, and it has known where the line lands for nine hundred years. The line is not drawn around categories; nothing is forbidden as a kind of thing. Steam works, powder works, chemistry works, and mechanisms do whatever springs and gears can be made to do. What fails consistently is quantity and energy, and nearly every limit in the world turns out to be one or the other.
Boilers
A boiler works to a known pressure and fails at anything above it. The figure is recorded precisely in Forgegate and every city builds to it, and it has been tested more thoroughly than anything else in the world, because boilers that fail kill people. Nobody finds this strange; it is a property of boilers the way the melting point is a property of metal. Build to the number and the thing runs for decades. Build past it and it fails, without exception anybody can demonstrate.
The man who did it anyway
He appears every few generations and his story is always the same. He builds past the figure and it works — a week, a month, a season — and then it fails suddenly, and he is left with wreckage, an account, and no way to establish the only fact that mattered. Nobody can see pressure; what his witnesses saw was a machine working, which is consistent with his gauge being wrong, and gauges are wrong constantly. His own guild has the strongest reason to disbelieve him, since if he is right their figure is wrong and every boiler built on it is suspect. So he is disbelieved, and the harder he insists the less he is believed, and anyone who had begun to notice a pattern now has him as their only evidence, which is worse than none.
Powder
Powder works; every city makes it and every Skaerflint fires it. What nobody has is enough — modest batches from many makers, never the quantity a sustained campaign could use, which is why sieges are decided by starvation rather than bombardment. Nor is there anything more energetic than powder. Attempts to make something that burns faster produce batches that do not take, mixtures that fizzle, and on occasion a laboratory that goes up with the chemist in it — rarely enough, and scattered across centuries and cities, that nobody has put the pieces together. Chemistry that heals, preserves, dyes, and makes works beautifully. Chemistry that destroys does not. Nobody has drawn a line between the two facts.
The rain
An army caught in heavy rain is an army using blades. What would fix it is a sealed primer — a shock-sensitive charge in a metal cup that fires when struck and does not care about weather — and the chemistry simply does not work. Batches do not take, and the ones that do behave unpredictably enough that no sensible person pursues it. It is the oldest unsolved problem in soldiering and is treated as a permanent fact of warfare. Nobody has noticed that the sealed primer is the first step of a chain whose rest is a cartridge, a breech-loading weapon, and a repeating one.
Iron
A ship is wood because no one can roll plate. Forgegate's iron and steel are excellent in the quantities a workshop needs; what nobody can produce is plate by the ton, uniform and cheap, because that needs mills at a scale no steam plant in the world will drive. So hulls are timber, and a wooden ship past a certain length works itself apart, and every shipwright knows the length. Somebody, somewhere in nine centuries, has built an iron hull anyway. It worked. What it could not do was pay for itself, since the valuable cargoes are light and a merchant is limited by voyages rather than by hull volume. What an iron hull would change is range — structure enough for a large Fraeska colony and larger tanks, the only thing that limits how long a ship can stay out. A vessel like that could go further north than anything has ever managed.
Why there are no factories
The limit on boilers is the whole of the intervention. Everything that follows from it, people worked out for themselves and stopped.
Ganging small engines does not solve it. A quay already has a dozen donkey engines, each on its own capstan, and that is the normal arrangement. What it cannot do is concentrate power. Coupling several to one shaft sets up a vibration that cracks the shaft at intervals nobody can predict, so a shop that tries it replaces the shaft, blames the metal, and eventually runs four machines on four engines like everybody else.
A factory never pays. Forty machines means forty boilers, forty fires, forty men stoking and forty fuel bills, to beat a skilled hand that costs eight Braell a day and no coal at all. Nothing could move the output anyway: goods travel by cart, caravan and sail, so a building making ten thousand of a thing would fill its own yard and stop. Raw material cannot arrive at that rate either.
And the guilds would not have it. Apprenticeship is the route in, the mark is the guarantee, and a man proposing a hundred unapprenticed hands each doing one motion would find he had no members, no masterpiece, no mark, and nobody who would buy from him.
Railways and harbour steam
Railways fail on the ground rather than on the engine. A line between cities needs iron rail by the thousand Keiloe (roughly a kilometre each), graded, bridged and maintained, produced at workshop scale by guilds working to order. And it would run through country where a Roedhaven exists because people cannot be out there safely. A caravan is a point and can be guarded. A line cannot.
So the world has rails everywhere and railways nowhere: tramways in mines and quarries, tramways along quays, and rope haulage where a stationary winch pulls cars up a slope by cable, which is exactly the sort of thing Clearhold builds into a mountain.
A small plant works perfectly well, and nobody minds what it drives. Steam launches and small tugs do harbour work in every city that can afford them, and the harbourmaster usually owns them. What does not exist is the ocean-going steamship, and not because it is forbidden — a hull that size needs a plant that size, and that plant fails. Every shipwright in the world has done the arithmetic and stopped.
What a good harbour really runs on is machinery set into the stone, because anything stationary can be as powerful as its boiler allows. Steam capstans along the quay bring a hull alongside on lines run ashore. Steam cranes work the cargo. Warping winches on the mole heads haul ships around corners they could not sail. Donkey engines do everything small.
So the Dorath teams are not on the quay. They are in the carting trade, moving goods from the quay into the city, which is where they belong.
Forgegate and Shining Reach have the best-machined quays in the world, and a Stormharbor man arriving at either notices immediately how quiet the work is. Stormharbor's own are a mixture, because the harbourmaster buys what the dues will cover and the whaling berths are older than the machinery. The Iron Coves has almost none of it, which is one reason the archipelago cannot handle real cargo volume and moves everything through fences instead. See Blacktooth Bay#The docks.
Where the limits actually bite
Two different things are happening and the world cannot tell them apart. Most of it is quantity: powder, plate, goods of every kind. A guild can make nearly anything well and no guild can make enough to change how the world works. It is not all quantity. A sealed primer is one chemist in a room, and so is anything that burns hotter than powder. Those fail differently: quantity problems get more expensive as they get bigger, and the others simply do not work, at any scale, for anyone, ever.
The pattern nobody sees
Everything above is recorded somewhere — Clearhold's registrars, Forgegate's guilds, the Academy. What nobody has done is put the records beside each other, because there is no reason to and the work would be enormous. The boiler figure is an engineering fact, the powder problem is supply, the plate problem is economic, the rain is weather, the dead chemists were accidents in different places centuries apart. And the world has developed an explanation — that Reason and the Natural Order have produced a world where such things cannot arise — which is satisfying and has never once been contradicted.