The Stave Mill is the newest Hearth & Horizon story. An AI wrote it. It is text only for now, because the shelf's audio budget is full. It comes from the series bible's "Next up" list. The list asked for a cooperage whose drying schedule was fitted on fine-grained oak, a forest that has since grown faster, and the numbers worked out before drafting.
The setup
An invented cooperage near Jarnac makes about 2,500 casks a season for the cognac houses. Its stave-yard system, Merrain, air-dries every lot of split oak for 24 months, two summers and two winters, the way the yard always has. In February 2148 the casks start to weep. Sixty-three of 960 casks from one lot leak. Only eight of the 1,440 casks from other lots do, which is the yard's usual rate of about one in 180.
The wood behind it
This part is real, though simplified. In ring-porous oak the spring band of big pores is about the same width every year. A good growing year mostly adds dense summer wood. So oak with wide rings is heavier. In the story's model, a 1.6 mm ring averages about 0.60 g/cm³ and a 3.2 mm ring about 0.70.
Denser wood dries more slowly, because the water left in the cell walls has more wall to creep through. The story treats the stave as a slab drying by diffusion. If wide staves lose water at three-quarters of the old rate (that figure is invented), they leave the stack at about 17.2 per cent water instead of 16. They need 31 months, not 24, to get to 16.
A difference of about 1.2 points sounds harmless. A finished cask then sits in a dry cellar and settles at about 14 per cent. Split oak shrinks across its width by roughly a sixth of a per cent per point. Around the head of a 350 litre cask that adds up to about 10 mm, or a third of a millimetre at each of 30 joints. New spirit at 70 per cent alcohol swells oak less than water does, so those joints tend to stay open.
Why the checks missed it
The schedule had checks, and they all passed. The monthly meter reads only the outer few millimetres of a stave, which dry first. The one oven slice per lot came from a stave in the fine-grained half, because two compartments of the forest came on the same lorries and were booked as one lot. The calendar was the real instrument, and it had been calibrated on a forest that was thinned, from 2058 on, so its trees could survive dry summers.
The forester in the story will not go back to close-grown stands, and he is right. A close stand on that ground dies in the droughts. So the cooperage has to change, not the forest. The master cooper sorts the stack by eye, with a thumb laid across the end grain: six rings under it means old wood, three means new wood that gets an extra summer.
What it costs
The story keeps count. A first fix, ten days in a warm room, cracks one wide stave in five. In all, 118 casks come back from the houses and are replaced. The spring's fine wood is borrowed from next autumn. Of 21,000 staves sorted by the end of March, 7,600 are wide.
Every figure is in the story's closing note, including the weekday checks for February and March 2148. Most of the story is the data question this blog keeps coming back to. A rule fitted on the past keeps passing its own checks after the thing it measured has changed, so the useful question is what the rule was trained on. Here, that was a forest's growth.