Walking around Berkeley at dusk last week, we saw a hydrangea that almost seemed to glow. It was an example of the Purkinje shift. The effect is named after Johannes Purkinje, a nineteenth-century Bohemian physiologist who discovered the Purkinje cell and the Purkinje fiber; Purkinje also gave blood plasma its name and was the first person to classify fingerprints. Purkinje noted the shift when looking at an Oriental rug one evening; as dusk settled, some colors appeared to grow relatively brighter. In low-light conditions, the rod receptors in your eye (scotopic sensitivity) take over from the cone receptors (photopic sensitivity). Rods and cones are most sensitive to different wavelengths of light, so as it gets darker, we perceive colors as changing in brighteness as reds and oranges grow relatively dimmer and greens and blues grow relatively brighter; this Purkinje shift demonstration can be done with your computer in a dark room. Unlike many optical illusions — the waterfall effect, for instance, or relative length and angle tricks or this horrid thing — the Purkinje shift is not based upon fooling the brain. It’s a result of the mechanics of the eye. The eye doesn’t work the same way as mental models of the eye, as telescopes or cameras. It’s a slightly eerie notion; upon his discovery of the blind spot in 1668, Edme Mariotte was disturbed by the conflict between what he had just observed and Kepler’s model of the eye as a natural lens. It wasn’t until 1819 that scientific exploration of the blind spot really took off, both because nerves were poorly understood and because no one had a model of the eye good enough to displace Kepler’s that also accounted for the blind spot and the weird way it seemed to flow into the background. Nineteenth century philosophy, of all things, becan to provide this model. Schopenhauer sums it up at the beginning of On Seeing and Colors: We see nothing, save through reason. None