Rigged for Red: The 200-Year-Old Eye Trick Behind Submarines, Cockpits, and Night Mode

In the spring of 1819, a medical student named Jan Evangelista Purkyně took his usual walk through the Bohemian countryside at dawn. He noticed something odd about the flowers. The red ones, which had glowed the brightest of anything in the field the previous afternoon, now looked almost black. The blue ones, dull by comparison in daylight, seemed to have gotten brighter overnight. Purkyně hadn't imagined it. He'd stumbled onto a quirk of human vision so consistent that it now has his name attached to it, and so useful that it built the lighting inside submarines, cockpits, and telescopes for the next two centuries.

It also, it turns out, got misapplied for most of that time.

What actually happens when the lights go down

Your eyes run on two separate systems. Cone cells handle color vision in daylight and peak in sensitivity around 555 nanometers, which sits in the yellow-green part of the spectrum. Rod cells take over once the light drops low enough that cones can't fire reliably, and they peak around 500 to 507 nanometers, over toward blue-green. Rods can't distinguish color at all. They only report brightness.

Purkyně's flowers were caught in the handoff between these two systems, a zone vision scientists call mesopic light. In full sun, your cones are running the show and red genuinely does look brighter than blue, because red sits closer to the cones' peak. As the light fades and rods start contributing more of the signal, that ranking flips. Blue starts reading as the brighter color, because it's closer to what the rods are tuned for. Nothing about the flowers changed. The instrument reading them did.

You can watch a fast, dramatic version of this yourself during a total solar eclipse. The drop from daylight to near-darkness during totality happens in minutes instead of the usual half hour of dusk, so instead of a gradual handoff, cones and rods briefly report at once. Reds visibly dull and blues and greens visibly pop, which is why eclipse-watchers in April 2024 kept describing the sky and their surroundings as looking strange in a way they couldn't quite name. Some of that is the retinal shift. Some of it is a genuine atmospheric effect, since indirect light during an eclipse carries more scattered blue than direct sun does. Both were happening on the same afternoon.

Full dark adaptation, the kind that gets you to your eyes' maximum night sensitivity, takes 20 to 30 minutes. Your cones settle in within the first 5 to 10 minutes; then comes what researchers call the rod-cone break, after which the rods keep slowly gaining sensitivity as a light-sensitive pigment called rhodopsin regenerates in them. That regeneration runs at a fixed biochemical rate. You cannot speed it up by trying harder, and a single bright exposure can bleach the rhodopsin back down and send you most of the way back to the start.

Which is exactly the problem someone needed to solve for anyone who has to work in the dark and still read an instrument.

Rigged for red

During the Second World War, US submarines ran on diesel engines and lead-acid batteries that could only be recharged on the surface. Every night, the boat would surface to charge, and lookouts had to climb up to the bridge and immediately start scanning for enemy ships in the dark. There was no time to spend the first twenty minutes of a watch half-blind while their eyes caught up. So the crew compartments below decks switched to red lighting before the boat surfaced. Because rods are barely sensitive to long red wavelengths, a sailor could read a gauge or a chart under red light without bleaching the rhodopsin his eyes needed the moment he stepped outside into darkness. He'd climb the ladder already dark-adapted.

The Navy called the practice "rigged for red," and it did more than one job at once. It also served as a low-tech day/night signal on a vessel where the sun never reaches the crew, and because red light doesn't suppress melatonin production the way white or blue light does, it helped sailors on a submarine's unnatural schedule actually sleep when they were supposed to. Modern boats like the USS Virginia and USS Ohio still rig control rooms and navigation spaces for red during night operations, three-quarters of a century after the practice started as a workaround for slow batteries.

Aviation borrowed the same trick around the same era, for the same physiological reason: a pilot needs to read instruments inside the cockpit and immediately pick out unlit terrain or other aircraft outside it. The idea wasn't new even then. Radiologists had been using red-tinted adaptation goggles since 1916 so they could walk into a dark reading room and see faint X-ray images without waiting out a full dark-adaptation cycle. Aviation adapted the same logic, and by the mid-20th century red cockpit lighting was the default in military aircraft.

Where the red-light logic starts to break

Red lighting has a real cost, and it's one that got more obvious as instruments got more sophisticated. A red cockpit light doesn't just fail to disturb your rods. It also strips color out of everything it touches. Charts and manuals rely on color coding, red warnings against black text on a chart that's printed in multiple inks, and under a red light, red ink and the white paper around it can end up reflecting red in roughly the same proportion. The contrast that made the chart readable in white light collapses. Glass cockpits made this worse: once instrument panels became color LCD displays instead of dial gauges, flooding them with red light meant a pilot genuinely could not tell one color-coded alert from another.

Dr. Anita Rothblum, a human factors researcher who has studied maritime accidents for the US Coast Guard, has been blunter about it: writing on the topic, she's called the traditional red-light rule close to a scientific blunder in practice. Her research found that low-level white lighting supports far better color discrimination and chart reading than red does, without necessarily costing much dark adaptation if it's kept dim enough. Red light, her work notes, also tends to cause more eye fatigue over a long watch and can carry an unwelcome psychological weight during an already tense night operation. Some navies have already shifted bridge and control-room lighting toward dimmable white or blue-green options instead of the old red standard. One low-tech fix from her research: wear an eye patch over one eye before turning on a bright light, then switch it to the other eye when you need to look outside again. One eye stays dark-adapted no matter what the room is doing.

Military aviation ran into an even stranger wrinkle once image-intensifier night vision goggles became standard equipment. NVGs are extremely sensitive across roughly 600 to 950 nanometers, a band that runs from deep red straight through near-infrared. That's almost exactly the band traditional cockpit lighting was designed to use, because that's the band the human eye's rods mostly ignore. Point that same red light at a set of goggles built to amplify exactly those wavelengths and you get "blooming," where the intensifier's gain collapses and the pilot goes briefly blind to everything outside. Modern NVG-compatible cockpits have swapped to blue-green lighting instead, in the 360 to 600 nanometer range, precisely inverting the old rule. The constraint driving the color choice isn't the pilot's retina anymore. It's the spectral sensitivity of the goggles strapped to it.

What this means if you're the one designing the interface

The physiology behind the Purkinje effect hasn't changed since 1819. What's changed is how many devices now expect to be read in the dark, and how badly a lazy hue swap can go wrong on any of them.

If you're building a night mode for something meant to be used outdoors after dark, an astronomy app, a night-hiking tool, a field guide for stargazers, the instinct to reach for red is the correct one, and it's backed by two centuries of evidence. Amateur astronomers and night hikers still buy dedicated red-light headlamps for exactly the reason submarine crews did: a red flashlight lets you check a map or adjust gear without bleaching the rod pigment you spent half an hour building up. But Rothblum's research is a useful check on that instinct. Pure red only works if the rest of the interface still reads clearly under it, meaning enough luminance contrast survives once you strip color discrimination away. A palette that depends on hue alone to separate "warning" from "safe" fails exactly the way a paper chart failed under a red cockpit light, right when someone needs it most.

Below is a palette built around the actual physiology: a near-black base with long-wavelength reds that stay dim enough to spare a dark-adapted eye, useful as a starting point for any night-mode interface meant to be read outdoors after dark.

The eclipse and dusk examples point at a second, quieter trap: colors calibrated for a bright, evenly lit environment don't hold their relative brightness once the light drops. A warm accent color that reads as urgent on an office monitor at noon can go flat and hard to spot on the same screen used outdoors at dusk, while a blue element you meant as secondary suddenly reads as the loudest thing on the page. If a product genuinely gets used across a wide range of ambient light, dashboard displays, outdoor gear, anything meant for dawn or dusk use, test it at the light levels it will actually face, not just in whatever room the designer happens to be sitting in.

For comparison, here's the kind of palette that looks perfectly urgent on a bright office monitor and falls apart once the ambient light drops, the reds and oranges losing their punch while the blue climbs the mesopic ranking uninvited:

The lesson from two hundred years of red cockpit lights isn't that red is right or wrong. It's that "the human eye works this way" is a starting hypothesis, not a permanent spec. Purkyně's answer held up for the wavelength itself. It kept getting revised for everything downstream of it, from paper charts to LCD panels to the spectral range of a night vision goggle, because each of those was a different device reading the same light. Whatever you're designing for the dark is a device too. Test it as one.