6 : cadmium lemon (PY37), cadmium yellow (PY35), pyrrole red (PR254), quinacridone carmine (PR N/A), ultramarine blue (PB29), phthalocyanine blue GS (PB15:3) • The split "primary" palette introduces us to a longstanding theme in "color theory" lore, which is that the mixing limitations of the "primary" triad" palette are due to the impurity of paint "primary" colors.

It's worthwhile to examine this theme carefully, as it rests on two false assumptions and therefore provides bad guidance to palette design, although the split primary palette itself is useful to build a bias into the range of color mixtures.

The "color theory" rationale goes like this:

Expert (lecturing at the academy): There are only three "primary" colors, and we experts call them PRIMARY because they alone can mix all other colors.
Novice: Really? But look ... mixtures of two "primary" colors are just not as saturated as the same color in a pure paint — an orange mixed from "primary" red and yellow is just not as bright as pyrrole orange, a green mixed from yellow and blue is not as bright as phthalo green, and a blue violet mixed from blue and magenta is not as saturated as ultramarine blue. How can that be?
Expert: Hmm, yes ... but you see, the fault is not with the "primary" colors — because "primary" colors can mix all other colors — the problem is with the paints!
Novice: What do you mean?
Expert: Why, your paint "primary" paints are impure. The ideally pure "primary" colors of light, when transformed into paint, become materially tainted by one or both of the other two "primaries" — and as we experts know, any mixture of three primary colors produces MUD.
Novice: Is mud ... bad?
Expert: Mud is terribly bad.
Novice: Well, if all paint "primaries" are tainted, then what should I do?
Expert: We experts have the answer! You must choose your "primary" paints so that each "primary" is tainted with the other "primary" you want to mix with it! Then taint mixed with taint is the same as pure mixed with pure!
Novice (slapping his forehead): Wow!, you mean I just split every "primary" into two new "primaries", each tainted by one of the other two "primaries"?
Expert: Now you've got it! That way, the mixture of two "primaries" tainted by each other results in intense color mixtures ...
Novice: And the mixture of two "primaries" where both are tainted by the third "primary" produces very dull color mixtures ...
Expert: And the mixture of two "primaries" where only one of them is tainted by the third "primary" produces moderately intense color mixtures ...
Novice (sighing happily): Golly, I sure do love "primary" colors!

This "tainted paint" mixing story developed from the idea that "pure" colors only exist in light mixtures. It was common knowledge among painters in the 18th and 19th centuries, and is given as fact in Michel-Eugène Chevreul's magisterial The Principles of Color Harmony and Contrast (1839).

But it seems to me that you can't really use the split "primary" palette to best effect, or wisely decide whether it should be the foundation of your palette, until you understand the three false assumptions it is based on:

• The first false assumption is that all colors are mixed from three "primary" colors of light. For example, a green yellow paint reflects yellow light tainted with blue light, while an orange yellow paint reflects yellow light tainted with red light. In fact, all yellow surfaces must reflect both "red" and "green" light, and no yellow surface reflects very much "blue" light. In fact, there is no "pure" light corresponding to the "primary" color magenta — all red, violet red, red violet and violet hues are a mixture of orange red and blue violet light.

• The second false assumption is that mixing limitations arise because "primary" paints are impure or tainted (which is just "color theory" jargon for the fact that pigments reflect some light from all wavelengths of the spectrum). In reality, exactly the same mixing limitations appear in mixtures of pure (single wavelength) colors of light, which create the purest (most saturated) possible colors and color mixtures. This is because the "impurity" is in the perceptual structure of our color vision — it's not in "color" as an external stimulus or material substance.

• The third false assumption is that you must use only "primary" colors of paint to mix the brightest colors. In fact, the only effective way to boost the saturation or chroma of paint mixtures is to mix paints that are close together on the hue circle; practically, this amounts to adding paints in the orange, green or purple hues that the "primary" paints cannot mix well. After all, you end up with six paints either way!

After you set aside these false justifications, the only question it is necessary to ask about the split "primary" palette (or any other palette) is, how does the palette deal with the four fundamental palette limitations? That question is completely muddied by a focus on "impure paints".

In terms of the four fundamental palette limitations, the split primary palette seems designed to restrict the chroma of purple and green mixtures, because this produces contrast along the warm/cool dimension. Dark neutrals do not reach the maximum value range possible in watercolors; pigment variety is still relatively limited, because the color pairs are so similar; and nearly all color mixtures require three paints, so paint mixing is not any more convenient.

The typical split primary palette produces a characteristically contrasty, brightly illuminated image. This warm/cool biasing effect is the specific feature of the split primary palette. This is not a drawback: if you believe foliage greens should be rather dull (which is how most landscape painters paint them), then there is no reason to have a green paint on your palette.

The point is that you need to think about your paint selections in terms of what you want to paint and how you want to paint it, and not in terms of "color theory" abstractions or "impure" color prohibitions. These discussions contribute nothing to a beautiful painting.

If you want to test drive the split "primary" palette, then the recommendations by Nita Leland are a good start, as I think they reflect the original split "primary" perspective: (1) two versions of yellow, red and blue paints; (2) exclusion of orange, green or violet paints; and (3) a choice of yellows, reds and blues that are still the same "color" (relatively similar in hue).

The split "primary" palette proposed by Michael Wilcox does not remain true to the "color theory" rationale and is actually the secondary palette, dented by the arbitrary exclusion of a green paint.

By changing your choice of split reds, yellows or blues, you can modulate this light rendering bias. Choosing a carmine rather than a rose for your "cool" red will darken and dull violet mixtures further, which biases the light in the painting toward yellow or green. The yellowing effect is enhanced by choosing a lemon or greenish yellow as the "cool" yellow, which produces brighter greens.