3 : hansa yellow (PY97), quinacridone rose (PV19), phthalocyanine blue GS (PB15:3) The "primary" triad palette has been anointed by artistic tradition as the chalice of "color theory" dogma. But it is important to see it in the context of other palette choices described here. The primary triad palette rejects the single dark paint used to create a complete value range, or the three paints of the Velázquez palette that balance hue variety with value range, for three very saturated paints that produce a complete range of hues at the highest possible chroma that is, the largest chromatic gamut possible with three paints.
Before you get too far into "color theory" you should be aware that all "primary" colors are either imaginary or imperfect. That is, if the "primary" colors are real (visible) colors then they cannot mix all the colors in their medium (no matter which medium, whether it is watercolors or monochromatic lights); and if they can mix all colors then they are actually imaginary (invisible) concepts used only in the specialized field of colorimetry. In other words, there is nothing inevitable or superior about the "primary" triad palette: it is just one of many possible three paint palettes.
Primary triad mixing principles are clean and simple. Any two of the "primary" paints mix a range of hues at the limit of the palette's gamut. These hues can be lightened by diluting with water, but they can only be darkened by adding the third paint, which also pulls the mixture toward a neutral gray. Thus, cyan darkens and grays the warm mixtures of magenta and yellow, and magenta darkens and grays the green mixtures of yellow and cyan.
In contrast, yellow is light valued, pulling a duller, lighter brown or green from the dark purple mixture of magenta and cyan. This inextricably links yellow to the symbolization of light. And with three paints in almost every color mixture, shadows are no longer "more dark paint": they must be carefully mixed as colors of a particular hue and value. These mixing dynamics force the artist to analyze exactly the chromatic content of every surface and shadow as a specific balance among three different paints, not just as a value between light and dark. The primary triad palette exercises your skill in color analysis in the same way the value design palette exercises your eye for value structure.
The "primary" colors produce a full range of hues but it is important to understand that this comes at the price of an uneven reach in color saturation. A limited gamut is the price you pay for any limited palette: the "primary" triad palette just makes this limitation more obvious and more of a painting challenge. Thus, yellow and magenta are used to mix intense warm hues, which also create the full range of "earth" tones when dulled with a touch of cyan, but the orange mixtures are noticeably duller than paints such as cadmium scarlet or pyrrole orange. The greens and violets are even less saturated, because the hue circle distance between magenta and cyan, or cyan and yellow, is much larger. This causes greater saturation costs in green or purple mixtures.
This imbalance creates a tension into the selection of "primary" colors, as these determine which mixtures will be intense and dull, and by how much. Grastorf's choices bear study:
Yellow. The main way to resolve the saturation balance is in the position of the yellow paint, which can be either an orange yellow or a green yellow, brightening the green while dulling the deep yellows, oranges and reds. An orange yellow restricts the saturation of green mixtures quite a bit and actually makes cool greens more attractive. The conceptual "primary" color wheel suggests a cool yellow, such as hansa yellow light (PY3) or cadmium lemon (PY35). Grastorf has chosen hansa yellow medium (PY97), a transparent yellow that does not lean too far toward warm (orange) or cool (green). This choice of yellow bathes the entire painting in a bright, balanced but warm sunlit glow.
Magenta (violet red). With the yellow decided, the next decision choice is the "red" paint which in fact is usually not pure red at all but either a dark deep red (such as quinacridone pyrrolidone, PR N/A) or a bluish violet red (such as quinacridone magenta, PR122). The choice of an orange yellow emphasizes the darker warm values, while the choice of a green yellow creates bright "earth" browns and ochres. Most artists prefer the magenta and yellow to be as far apart as feasible, to distribute saturation costs as evenly as possible, and do this by choosing quinacridone rose (PV19) or quinacridone magenta. This ensures the warm mixtures will be saturated, without impairing too much the saturation of violet and blue violet mixtures. Even so, the color effects should be used subtly: compare the rosy shadow against the foreground tree trunk with the cinnamon oranges in the sunlit foliage behind.
Cyan (green blue). The choice of blue determines whether the saturation costs will fall mostly on the mixed greens or the mixed violets. Unfortunately, there is a dearth of lightfast blue pigments, so most artists end up choosing a blue rather than a green blue paint: Grastorf has chosen the dark, staining phthalo blue (PB15) which is significantly on the red side of cyan. This choice keeps the mixed violets from being too dull (red blues and violets are unsaturated anyway in realistic landscapes) and shifts the chroma in the mixed greens so that yellow greens are saturated but blue greens are dull. Grastorf's makes this work in her landscape by using the desaturated blue greens as shadow greens, and painting the sunlit foliage greens, which would normally be a saturated yellow green, as lightened (desaturated) by the intense light. Because blue is usually the darkest color in a "primary" palette, it also tends to determine the palette's value range. An even darker and greener phthalo turquoise (PB16) would produce even more dramatic saturated and dark greens, with very dark violet grays.
The best "primary" triad selection typically consists of (1) all transparent paints (as in Grastorf's palette), or (2) opaque paints for the yellow and/or magenta, and a transparent paint for the cyan or green blue. An opaque cyan (such as cerulean blue) produces whitish, foggy shadow colors, while a palette of three opaque paints produces a duller, smaller gamut.
It's easy to get too theoretical when choosing "primary" colors, trying to find the "objectively" best paint for each of the three hues. This seems to me an intellectual exercise that ignores the testing practice of actual painting. A palette is meant to be used, so the paints you choose depend on the effects you want to achieve, and nothing else. Clearly, for the subject Grastorf has chosen, and at her level of mixing and painting skill, her "primary" triad selection is faultless.