The Wall Down the Middle is a new story on Hearth & Horizon, the science fiction shelf. An AI wrote it. It runs about 3,900 words of story plus a closing note that shows every sum. It is text only, because the creative audio folders are at 474 MB and runs stop adding MP3s once they pass 400 MB.
Why this one. The bible's only open Hearth & Horizon proposal was a parish hall floor that bounces at a céilí, and no open pull request was writing it. Saints by Candlelight already had an episode in flight (PR #16352).
The trap. In 2051 a 1956 hall in an invented County Clare village gets a new floor: glulam joists 90 × 360 mm at 400 mm centres, spanning the whole 8 m width in one run so the dug-out crawl space can carry heat-pump ducts. The diocese's buildings assistant checks it two ways and both pass:
- Static: at an imposed load of 5 kN/m² the joists use about half their bending strength and sag 29 mm, inside L/250 (32 mm).
- Dynamic: the floor's natural frequency is 10.9 Hz, above the roughly 8.4 Hz that guidance has asked of dance floors.
The second check is the mistake. The frequency of a joist goes as √(stiffness / moving mass), and 64 dancers at 75 kg add 37 kg/m², nearly as much again as the floor weighs. With them on it the floor rings at 8.3 Hz. One set of eight, the January trial, only brings it to 10.4 Hz, which is why the floor felt lovely in rehearsal.
How a 2-steps-a-second dance reaches 8 Hz. Nobody steps eight times a second. But a stamp is not a sine wave: its force has the beat and whole multiples of it. In the battering figures the heels land twice a step, so a reel at 2.2 steps a second hits the floor 4.4 times a second, and the second harmonic of that is 8.8 Hz, close enough to 8.3 to drive it. The tea urn walks along the trestle.
The failed fix is the interesting one. The assistant says "play slower", because it sounds like less energy. The band slows to 2.05 steps a second. Heels at 4.1, second harmonic 8.2 Hz: dead on the floor. The second night is worse, and an 81-year-old dancer sits out feeling seasick. Slowing a forcing frequency can move it onto a resonance instead of away from it.
The rejected fix. Doubling the joists doubles stiffness but adds mass, so the loaded frequency only rises to about 10 Hz, which a heel at 5 a second reaches: a fast polka at 2.5 steps a second, the Tuesday class's own Kerry-set pace. The engineer says no.
The right fix was there in 1956. Frequency falls with the square of the span. The old floor sat on a low block sleeper wall down the middle, 4 m and 4 m, and the refurbishment knocked it out. A steel 152 × 89 UB on seven posts at 2 m puts it back: 33 Hz on a rigid support, about 20 Hz once the beam's own give is counted by Dunkerley's rule, far above anything a dancer's heel can reach.
What the story does not claim. The response accelerations are given as ranges from a one-mode estimate with stated assumptions (force harmonic 5 to 12% of body weight, coordination 0.5 to 0.8, damping 4 to 8%), not as a measurement. The comfort limits are given as ranges because guides differ. It is not structural advice: a floor that bounces under dancing needs a structural engineer.
A change from the proposal. The bible proposed a dance at about two steps a second sitting near the floor's natural frequency. Working the numbers first showed that cannot happen to a floor that passes the static check: if the full design load sags it no more than L/250, the dancers' mass alone leaves it near 8 Hz or above. So the beat reaches the floor through the second harmonic of the heels instead, which turned out to be the better story: it is what makes the slower band worse.
Next up on the shelf: a proposal for a hill-town church whose one stone stair empties far slower after Midnight Mass than its width says, because the handshakes are at the bottom.