the science · loop 0

Nobody has watched a shiso leaf redden

Shiso is one of the most-studied herbs on earth. Its red has never been measured the way we're about to.

Shiso has been in laboratories for a century. We know the pigment that makes the red leaves red — a chemical mouthful called malonylshisonin1. We know the genetic switch that turns it on2, and we know it's light that flips the switch. By most measures this is a solved plant.

And yet, when I went looking for the one number I actually need — how red does a red-shiso seedling get as you give it more light? — it isn't there. Not because nobody looked. Because of how they looked.

Here's the thing I keep turning over. For decades, measuring a leaf's red meant killing it: pick the leaf, grind it up, dissolve the pigment, read one number off a machine. One number, one dead leaf. Which means nobody has ever watched a single shiso leaf redden across a week — only taken snapshots of different dead leaves and lined them up afterward. The biology is famous; the measurement is a hundred years old.

the only data is two dots

The best anyone has done, head-on, is two dots. A 2022 study3 moved red shiso from dim light to bright — 15 units to 180 — and the leaves went from green to deep red, with about 1.7 times the pigment. A before and an after. Everything between those dots is blank:

0 100 200 300 400 light intensity (µmol m⁻² s⁻¹, PPFD) → green deep red redness ↑ our predicted curve ○ loop 0 fills these in — at fixed leaf temperature 15 · green 180 · dark red (+1.7× pigment)
Two filled dots are the real, published data (Xie et al. 2022). The dashed line is my prediction; the hollow dots are where loop 0 measures, holding leaf temperature constant. The curve between the dots has never been drawn for this plant.

The shape of that climb — where it steepens, where it flattens, whether there's a point where more light stops helping or even hurts — is unmeasured. That curve is the first thing we get to draw. And a real question hides inside it: when very bright light seems to reduce the red, is that the light damaging the plant, or just the lamp warming the leaf? Heat reddens shiso too, and too much heat shuts the red back off4. Nobody has separated the two, because nobody held the leaf's temperature still while cranking the light. We can. That's not a better gadget — it's a different question.

the genetics are friendlier than i'd feared

Cross a red shiso with a green one and the grandchildren don't split a clean three-to-one. They split 1 : 2 : 15 — a quarter green, a quarter deep red, and half in the middle: a visibly intermediate, not-quite-red.

¼ green
½ intermediate
¼ red
The F2 of red × green. The middle half is where a human eye gives up — "reddish"? "sort of red"? — and exactly what a colour-meter that reads every leaf the same way turns into a number.

That middle half is the whole opportunity. It's the worst call to make by hand, which is precisely why an instrument that reads the same leaf the same way every time turns a squint into a measurement. The thing the field finds hardest to score is the thing our rig is built to score.

our seat at a very old table

So that's the shape of it: the biology is settled, the measurement is wide open. Three things a serious amateur with the right instrument could be first to put down for this plant —

— the actual light-to-red curve, with heat held still; a way to tie our colour numbers back to the real pigment chemistry, so "redness 4" means milligrams and not a vibe; and the first clean test of how shiso decides when to flower, which turns a single plant into a switch we can flip. None of these needs a grant. They need controllable light, a camera that doesn't lie about red, and the patience to watch one leaf for three weeks. We're building exactly that.

The one I can't wait for is the flowering switch. Shiso flowers when the nights grow long, and a flash of light in the middle of the night can talk it out of it — that much is known6. What isn't known, for shiso specifically, is the colour of light that does it: red light should work and far-red should undo it, but nobody has actually checked — the answer's been borrowed from tomatoes for forty years. It's a dark box, two carefully chosen lights, and a few weeks of looking. The day that one comes out clean is the day a hobby put a real, new point on the board for this plant. I would like to write that day up.

sources

  1. Kondo, Yoshida et al. 1989. Structure of malonylshisonin. Agric. Biol. Chem. 53(3):797–800. 10.1271/bbb1961.53.797
  2. Gong, Yamazaki & Saito 1999. The Perilla anthocyanin regulator (PfMYB-P1, a MYB-like activator). Mol. Gen. Genet. 262:65–72. 10.1007/PL00008639
  3. Xie et al. 2022. Light intensity and anthocyanin in Perilla (15→180 µmol; green→dark red, +1.7×, reversible). Front. Plant Sci. 13:976449. 10.3389/fpls.2022.976449
  4. Kim et al. 2017 (the COP1→HY5 heat mechanism, in Arabidopsis); Zhong 1993 (Perilla pigment peaks ~25 °C, drops by 28 °C). 10.3389/fpls.2017.01787 · 10.1016/0922-338X(93)90255-7
  5. Xie/Zhang et al. 2025. Red×Green F2 segregates 1:2:1; one locus on chr 8, candidate PfMYB113b. Adv. Biotechnology 3(1):7. 10.1007/s44307-025-00058-8
  6. Yoshida et al. 2021 (critical daylength ~14 h 20 min, night-interruption practice); Jacobs 1982 (red>green night-break sensitivity). 10.3390/plants10061252 · 10.1104/pp.70.1.303

Full landscape, all six dimensions and every citation, in the repo: docs/literature.md. The survey was machine-built and adversarially fact-checked — zero fabricated references, the corrections it caught kept on the record.