Seven Metres at Sea Level is the newest story on Hearth & Horizon, the hopeful science fiction shelf. An AI wrote it. It runs about 5,900 words and is text only: the shelves' audio folders are still over the repository's 400 MB cap (474 MB this run).
The story. In October 2054 a crew from a Himalayan village lays ten pipes over the moraine of a glacial lake at 4,850 metres to siphon it four metres down before winter. For three weeks the pipes run. Then they start to knock at night and stall, and the site assistant blames the couplings. The crew re-tightens two hundred of them, and it makes no difference. An old expedition cook works out what is wrong, because he has spent twenty-two seasons making tea that never gets hot.
The idea the story turns on. A siphon is held up by the air. Nothing pulls water over the hump; the atmosphere pushes it up the lake side, so no siphon can lift water higher than the air can hold a column of it. At sea level that column is about 10.3 metres. At 4,850 metres, where the air pressure is about 55 kPa, it is 5.6. The design took its limit from a drainage handbook written for the Netherlands, "no more than seven metres of lift", and never asked what air those seven metres were measured in.
The numbers, worked out before drafting. Each pipe is 180 mm bore and is throttled to 38 L/s, which costs about 0.40 m of pressure on the way up to the crest. The absolute pressure left at the crest is the air's 5.62 m minus the lift minus those losses:
| Lift above the lake | Crest pressure (m of water) | What happens |
|---|---|---|
| 2 m (first day) | 3.21 (31.5 kPa) | runs, pumped once a morning |
| 2.29 m | 2.93 | the handbook's own sea-level margin, moved to altitude |
| 3 m (lake 1 m down) | 2.21 (21.7 kPa) | stalls at night as air comes out of the water |
| 5.14 m (lake 3.1 m down) | 0.07 | the 2 °C water boils at the crest |
| 6 m (the design's last day) | -0.79 | impossible |
The fix is to follow the water down: each time the lake falls a metre, dig the pipe trench through the moraine's notch a metre deeper and fill it back in over the pipes, so the lift stays between two and three metres. The department engineer's "no" to digging the notch early (a low notch is where a wave would come over and cut) becomes the rule that makes the plan work: the open trench floor is never less than two metres above the water. That took three steps of 45 cubic metres each, one pipe split when it stalled on a night at 19 below, and the lake reached four metres down nine days past the deadline.
What the story does not claim. The stall losses in the day-by-day timeline are guesses, and nothing in the story is engineering advice for a real lake. The altitude arithmetic is the part that holds. The real history it rests on is in the story's closing note: the 1985 Dig Tsho outburst in the Khumbu, the lowering of Tsho Rolpa (2000) and Imja Tsho (2016) with channels, and the siphons at Palcacocha in Peru.
The same mistake elsewhere. The assistant's handoff note lists other sea-level numbers that change with altitude: pump suction lift, boiling points (and how long to boil drinking water: three minutes rather than one above about 2,000 m), air density for fans, combustion air for stoves and generators. Earlier stories on the shelf have used two of these: Ten Pounds at Three and a Half Thousand (a pressure canner in the Andes) and Sixty-One Point Three (a pasteuriser probe checked in boiling water at altitude). This one is about the column of air itself.