Color Anamolscope

Two small windows glow on top of a black box: a yellow on one side, a red-and-green mixture on the other. The device is calibrated and tests run from the website at https://anomaloscope.cairn.com/ . Drag the slider on the 'calibrate' page until the mixture looks exactly like the yellow. For most of us a whole range of mixtures will pass as a match. For a rare few who carry a fourth kind of cone, that range narrows to a slot, because their eyes separate what ours blur together. The width of your match is the measurement, and the machine is hunting for the people who see more color than I do.

I'm color weak, so this is the instrument I had no business building and couldn't put down. It does the most democratic thing I know: sit two people at the same box and they will disagree about when red and green become yellow, and neither of them is wrong.

This device is a fork of the Brainard Lab's Penn Anomaloscope, rebuilt to need no MATLAB and no lab. The Arduino only holds the LEDs steady; a web page drives the whole test over a USB cable (the laptop in the photo is running it), so the color math, the calibration, and the data all live in the browser. An orange filter and a scrap of diffuser turn three crude LEDs into a real Rayleigh match.

Online it's the start of a citizen-science project: build the box, run the test, and, if you like, add your result to an open record of how differently people see. Build it, or just read how it works, at anomaloscope.cairn.com.

(image: A working prototype, driven by this page over Web Serial. (Orange filter not yet fitted, so the fields aren't a calibrated match. Code forked from the BrainardLab Penn Anomaloscope — Keesing & Brainard, University of Pennsylvania.)