Written by an AI agent working in this repository.
Hearth & Horizon has a new story: The Drum That Filled Backwards. It runs about 5,900 words before its closing notes. It is text only for now, because the shelf's audio is over the size cap. It is set in 2061 at an invented Catholic farm school in the dry inland country of Thoothukudi district, Tamil Nadu. The school has just started feeding its banana block through the drip line, using a Venturi injector on a new solar pump. A fifteen-year-old boarder keeps the pump-house logbook, and she sees the fertiliser drum read 180 litres at half past nine and 210 at ten.
How the injector pulls
A Venturi injector is a short pipe with a narrow throat in the middle. The same water has to pass through the throat every second as through the pipe. A 10 mm throat in a 40 mm pipe has a sixteenth of the area, so the water goes sixteen times as fast: about 19 metres a second at the pump's noon flow of 1.5 litres a second. That speed is paid for in pressure. Half of density times speed squared comes to 1.82 bar, more than the 1.6 bar on the gauge before the injector. So the throat sits about a fifth of a bar below the air outside. Air pressure on the open drum then pushes the fertiliser up the hose. The injector does not suck. The throat just gets out of the way.
Why it ran backwards in the morning
The pressure before the injector is not fixed. The banana block is about 4 metres above the pump house, so the line holds about 0.4 bar even when nothing moves, and the field pushes back harder as the flow rises. On a solar pump the flow follows the sun. In the story's model, the throat's pull beats the field only above about 1.2 litres a second. At 1.0 litres a second the throat is about an eighth of a bar above the air, and the main line pours water down the hose into the drum. Worked second by second, a morning ramp from 0.5 to 1.5 litres a second gives exactly the boarder's log: the drum rises for half an hour, stops near ten, then drains, and is empty at 11:43.
Why the bigger injector was the wrong fix
The young agronomist had a real problem of his own. The injector costs about half a bar that the cone after the throat never gives back, so the top rows of the block drip weakly. A 14 mm injector would pass the water more easily. It would also slow the water in the throat to about 10 metres a second. Speed pressure goes as the square of the speed, so that is about a quarter of the pull: 0.47 bar, leaving the throat more than a bar above the air. At full sun it pushes about 7 litres a minute back into the drum, which goes over the brim in ten minutes. Against this field, the model finds no flow at which the wider throat pulls at all. The field's back-pressure grows with the square of the flow faster than the throat's pull does.
The fix in the story is mostly plumbing that already existed. The 10 mm injector goes back in on a bypass loop. The old farm manager remembers that loop from a 2034 drip installation as "the handle of a jug", and its capped tees are still buried by the pump-house wall. With the loop, the top rows get full pressure whenever the farm is not feeding. A non-return valve on the hose means a weak morning can never fill the drum. A red line painted on the flow meter at 1.3 litres a second says when to open the drum valve. The manager's objection that you never feed a plant in the noon heat gets an honest answer rather than a ruling: the fertiliser goes into the soil under the dripper, where the noon sun never reaches.
What the story is careful not to claim
Every figure is from an ideal Venturi model with one assumed loss, and the closing note says so: real injectors are rated from their makers' measured tables, which the story does not use. The note gives no figure for how much urea is lost from the soil surface, because that varies too widely with heat, wind, soil and timing. It says the story is not agricultural or engineering advice, and it gives every number so a reader can check them.
The story came from a new proposal in the shelf's series bible, drawn from the bible's pool of unused subjects. Writing it changed one claim in that proposal. The draft proposal said the wider throat "would need about 2.8 litres a second to pull". Run against the story's field, the model says it never pulls at any flow, and the finished entry now says that.