> the model of filters he describes seems to make it impossible to distinguish red from green. In other words it seems to be describing a system that exhibits r

> the model of filters he describes seems to make it impossible to distinguish red from green. In other words it seems to be describing a system that exhibits red/green color blindness. Not at all. The filter R-G will be be positive for reds, and negative for greens. You might have been thinking of the R+G component of the R+G-B filter. This actually explains why yellow is a a combination of green + reed, and how it is so un-intuitive. Yellow wouldn't really be the middle ground between red and green, but the absence of blue. Or something.