Plumb
Turn a potted seedling on its side and leave it there. Within a day the tip of the shoot has swung back toward the sky and the tip of the root has swung back toward the ground, both bending in the middle like a knee. Every gardener has seen it. The question this room is about is the one no one asks: how does the plant know which way is down? It has no inner ear, no eyes pointed at the horizon, nothing you would recognise as a sense of balance.
The answer, first guessed in the 1880s and still the working model, is almost comically physical. Inside certain cells — in the very cap of the root, and in a sheath around the shoot — sit a handful of grains heavy with starch, called statoliths. They are dense enough to fall. Whichever way the plant is turned, they sink to the low corner of their cell, and the plant reads the direction of that little heap as down. Tip the plant and the grains slide to a new floor; the cell notices where they have piled, shunts its growth hormone to one flank, and the growing tip curves. Not a sense of balance. A pinch of sand in a jar, read by touch.
So: a slider that tilts the pot, and below it one root-cap cell blown up large. Tilt the plant however you like. The cell wall turns with the tissue it belongs to — but watch the grains inside it. They do not turn. They hold at true down while the wall rotates past them, and that unturning heap is the whole of the plant's sense of gravity. Up top, the growing tips answer to it: however far you lean the pot, they curl back to vertical — shoot up, root down — each ending parallel to the dashed plumb line.
Upright: the grains rest at the floor of the cell, and the tips already point along the plumb line.
A phone or tablet can drive the slider itself: press
“Tilt with your own device” above and tip the screen left
or right — the same gamma reading a real
gravity-sensing app would use, not this page pretending. It only
appears where the browser exposes a tilt sensor at all, asks
permission first where the platform requires it (Safari on iOS),
and gives up honestly after a couple of seconds of silence rather
than sitting there looking like it's listening. Nothing about the
seedling's own physics changes — it's still just an angle fed
into the same draw() the slider always called; only
where that angle now comes from is new.
Honest gaps. Real reorientation takes hours, not the instant this slider gives you — the grains sediment in minutes, but the visible bend of the growing zone is slow, and what you see here is the finished shape, not the wait. The grains are drawn holding exactly at true down; in a real cell they settle a little raggedly and the reading is noisier than one clean heap. And gravity is only half the story a real tip weighs — light pulls the shoot too (that's a different sense), and this room leaves light out so the one mechanism worth seeing stays uncrowded.
Nothing here reads a date, calls plant.js,
or rolls its rng(). Every line is pure geometry from the
tilt angle — the same footing /veins and
/roots stand on, answering to nothing but your own
input, since nothing drawn here is a fact the garden claims is
reproducible from a seed.
Sources: the starch–statolith idea goes back to Haberlandt and Němec, both 1900; for a modern account of the amyloplasts that sediment in the root cap and how their fall is turned into a growth signal, see Kiss, “Mechanisms of the early phases of plant gravitropism,” Critical Reviews in Plant Sciences 19(6):551–573, 2000, and the review by Nakamura, Nishimura & Morita, “Gravity sensing and signal conversion in plant gravitropism,” New Phytologist 225(5):1873–1885, 2019.