Picture the art classroom. Powder paint in little pots, a jam jar of murky water, a brush that was already ruined before you got to it. The teacher writes three colours on the board - red, yellow, blue - and announces that everything else can be made from these. Somehow. If you do it right.
It felt like a magic trick. It also felt slightly dodgy, because no matter how hard you tried, you could never quite make the colours on the board match the colours in your head. But the rule was repeated so confidently, so universally, that most of us just assumed the problem was our technique.
It wasn't entirely our technique.
Where did the three-primary idea come from?
The notion that red, yellow and blue are the three primaries - the building blocks of all other colours - goes back centuries. Apparently it was being taught seriously in European academies by the 1700s, long before anyone had a rigorous scientific framework for colour. It was an observation, not a derivation: painters noticed that mixing these three got you a surprisingly wide range of hues, and the idea stuck.
And it's not wrong, exactly. Mix red and yellow and you get orange. Mix yellow and blue and you get green. Mix red and blue and you get purple. All three together and you get brown - which is harder to control than it sounds. The model works well enough for a classroom, well enough to teach children the basic relationships between colours, and well enough that it survived essentially unchanged for the better part of three hundred years.
That's a very long run for a simplification.
So what does it get wrong?
The problem is the word primary. It implies these are the irreducible originals - that you can't make red, yellow or blue from anything else, but you can make everything else from them. Neither part of that claim holds up cleanly.
Modern colour science - and the printing industry, which has very little patience for artistic sentiment - settled on a different set of primaries for mixing pigments: cyan, magenta and yellow (plus black, for practical reasons). These are sometimes called the subtractive primaries, because mixing pigments works by subtracting light rather than adding it. The CMY model gets you a noticeably wider range of colours than the old red-yellow-blue model, particularly for vivid greens and oranges.
The red-yellow-blue system, by contrast, can't produce a genuinely bright orange from its own red, because that red already leans warm. And vivid greens are famously difficult - the yellow and blue you have in your palette may not be spectrally pure enough to make the green you want. You can get a green. Just not necessarily that green.
There's also a separate system entirely for light - screens and projectors use red, green and blue (RGB), which is additive rather than subtractive and produces white when all three combine rather than brown. You open a different app on your phone and get a completely different primary set. It's not inconsistent; it just reflects the physics of the situation. Mixing pigments and mixing light are genuinely different problems.
Why did the old model stick around so long?
Partly inertia. Partly because it's taught to children before they're likely to question it. And partly - arguably the most important reason - because it's useful. Not perfectly accurate, but useful in the way a simplified map is useful: it gets you oriented, even if it leaves out some roads.
There's something typically British about this, in a way. We love a rule of thumb. We love a system that mostly works and doesn't make too much fuss. Red-yellow-blue is the colour-mixing equivalent of driving on the left: it's not obviously better than the alternatives, but everyone's used to it now and changing would cause chaos.
It also captures something genuinely true about perception. Human colour vision is built on three types of cone cells in the eye, each sensitive to different wavelengths. In that sense, three is the right number - it's just that the specific three used in the classroom model weren't chosen with scientific precision.
Does any of this change how you should think about mixing?
Knowing the limits of the model is actually quite freeing. It means when a mix doesn't come out the way you expected, the problem might not be your technique at all - it might be that the particular pigments you're using aren't spectrally pure enough to behave the way the theory predicts. Mixing paint in practice is messier than mixing it in your head, and that gap is real, not imaginary.
It also means proportions matter enormously. Red and yellow in equal parts gives you orange, yes - but two parts red to one part yellow gives you a deeper, more terracotta shade. One part red to two parts yellow and you're heading towards a warm amber. The primaries aren't switches; they're dials. The old classroom model tended to skip over this, which is why so many people remember colour mixing as frustrating rather than illuminating.
This is, incidentally, the thing that makes Kalabux - a steampunk paint-mixing puzzle game we made - feel so much more satisfying than the art-room memory. The whole point is routing proportional streams of red, yellow and blue through pipework, watching them combine at junctions, and landing on an exact colour. Not approximately orange. That specific orange. It turns out the proportional nuance is where all the interesting decisions live. The simple three-colour rule opens the door; everything beyond it is the actual room.
The three primaries aren't wrong. They're just the start of the story, and for a long time nobody told us there was more.