Is “split primary” the best starter palette?

I’ve read a lot of beginner watercolor advice, and I’d estimate that 80% of books/blogs/videos recommend starting with a “split primary” palette. Common starter kits also follow this blueprint, such as Daniel Smith’s Essentials set, which I started with. But is the “split primary” palette really the best set of six colors for a beginner?

What is a split primary palette?

When people talk about a split primary palette, they typically mean a set of of six colors, with two versions each of yellow, red, and blue: one warm and one cool of each. You end up with six color slots:

Cool	Warm
Cool yellow (lemon yellow)	Warm yellow (orange yellow)
Cool red (magenta or crimson)	Warm red (scarlet or middle red)
Cool blue (cyan)	Warm blue (violet blue)

The Daniel Smith Essentials kit is an example of a split primary palette because it contains a warm and cool yellow (Hansa Yellow Light and New Gamboge), a warm and cool red (Pyrrol Scarlet and Quin Rose), and a warm and cool blue (French Ultramarine and Phthalo Blue GS).

What the split primary palette does right

A positive that I notice right away about the split primary palette is that it contains yellow, magenta, and cyan – the “modern primaries” which mix the most vibrant secondary colors. This triad gives you a great basis for mixing nearly any hue.

A six-color palette feels like a good size to get a beginner started on color mixing and exploring. Only having three minimal primary colors can be annoying, as many hues take forever to mix, and you don’t have enough slots to really learn about how different pairs or paint qualities mix differently. On the other hand, a very large palette may discourage color mixing, as beginners will tend to use colors unmixed, or become overwhelmed by learning too many options at once.

What the split primary palette does wrong

I’ve ranted before about how arbitrary I find the “warm and cool” dichotomy. The reason the modern primaries are so useful isn’t because they’re “cool”, it’s because of the way our eyes work. There’s no special need for the other three colors to be “warm versions” of those colors. Almost any other three colors would do just as well to add more variation.

Moreover, the split primary palette only concerns itself with hue and doesn’t address a lot of other aspects of palette-building, such as:

• Value variation. Because the split primary palette is concerned only with hue, it doesn’t ensure that your palette will include dark colors, or colors that mix darks easily. In my opinion, my biggest fatal flaw is filling my palette with bright colors and then having trouble mixing the dark colors that give a painting depth and definition.
• Chroma variation. The split primary palette also lacks variety in chroma or saturation – all the colors are bright.
• Complementary pairs. Complements are opposites on the color wheel; if you mix these, you get grayed-out, muted colors. In this way, you can get muted colors even if all your paints are bright. But the split primary palette, by design, tends to avoid complements: all the colors are supposed to be variations on primaries.
• Variation in other qualities, such as granulation and opacity.

The split primary system doesn’t prevent you from adding this variation, it just doesn’t build it in. If you’re canny about palette-building, it’s possible to add some of this variety on top. For example, the Daniel Smith Essentials kit:

• Includes value range because some of the colors (like Phthalo Blue) are very strong, and because Ultramarine Blue also happens to be a great dark mixer. This isn’t necessarily true of every violet blue; e.g. Cobalt Blue, which is almost the same hue, does not mix as dark.
• Includes granulation because Ultramarine does triple-duty as an earthy granulator.
• Adds opacity in the form of Pyrrol Scarlet.
• Incidentally contains a complementary pair, Phthalo Blue GS and Pyrrol Scarlet (this is not necessarily true of every cool blue/warm red pair – most would mix dull violet, not complementary black)

I would submit that these strengths of the DS Essentials Palette are because the folks at DS are good at picking colors that work together and have thought about adding all this additional variation. It doesn’t succeed because it adheres to the split primary system; if anything, it succeeds in spite of it.

Some Alternatives

So, if we know we want a six-color palette that includes yellow, magenta, and cyan, is there a better system for choosing the remaining three colors?

Alternative #1: Secondary or Complementary Palette

Bruce MacEvoy of Handprint.com has a whole rant about the split primary palette, and one same-sized option he offers is a secondary palette where the 3 primary colors yellow, magenta, and cyan are joined by hues equidistant from them on the color wheel. For a six-color palette, this would result in:

Primary Color	Secondary Color
Yellow (lemon yellow or middle yellow)	Between yellow + magenta: Scarlet
Magenta	Between magenta + cyan: Violet-blue
Cyan	Between cyan + yellow: Green

You will notice that this is functionally the same as the split primary palette above, with one exception: orange-yellow is dropped in favor of green.

I do think the theory of “space along the color wheel” makes more sense than the theory of “warm and cool.” For one thing, this system contains mixing complements by design. In fact, you could simply rearrange the table above and think of the same colors as a “complements” palette:

Primary Color	Complement
Yellow (lemon yellow or middle yellow)	Violet-blue
Magenta	Green
Cyan	Scarlet

With all that said, in practice, I do not think that violet-blue, green, and scarlet are the next-most important colors I would really choose to expand the modern primaries.

I have both orange-yellow and green in my extended palette, but I’m not sure I’d nominate either one for inclusion into truly functional a six-color palette. Both hues are fairly easily mixed from cyan, yellow, and magenta, unless one wants the brightest possible version; which I sometimes do, but typically only seasonally (orange-yellow for fall leaves and green for spring buds).

In actuality I do think that secondaries tend not to be as useful as variations on primaries (yellow, magenta/red, and cyan/blue.) What if split primary idea works, but the way to “split” it isn’t by color temperature or color wheel position, but something else?

Alternative #2: Split Primary, but Bright/Earth

An alternate split primary palette system builds in chroma and granulation variety, and offers an earthy version of each primary.

Bright Primary Color	Earth Equivalent
Yellow (lemon yellow or middle yellow)	Earth Yellow, e.g. Raw Sienna or MANS
Magenta	Earth Red (Magenta), e.g. Indian Red
Cyan	Earth Blue (Cyan), e.g. Cerulean Blue

This is already a lot closer to my preferred palette, but there are some potential issues.

1. The theory here demands an “earthy” version of magenta, implying a violet-red hue, but I think earth orange is actually more useful, because it makes it easier to mix brown with blue. Similarly, the theory demands an “earth cyan” but I think Ultramarine would actually be more useful. Though this might be due to the value range, which brings me to my next point.
2. It doesn’t enforce something I think is important, which is value range; a lot of the “earth blues” (that is, granulating blues) are opaque and light-valued, and would not give you as good options for mixing dark values as my fave Indanthrone Blue.
3. It is probably most attractive for landscape painters, who paint a lot of earthy stuff. (Though I still think “ease of mixing brown” is an important quality for many types of subject – including portraits and animals – and is generally ignored by the split primary palette.)

Alternative #3: Complementary, but Bright/Earth

I wonder if it’s key to strive for complementary pairs, instead of getting them by accident. This palette addresses problem #1 above.

Bright Primary Color	Earth Complement
Yellow (lemon yellow or middle yellow)	Earth violet or violet-blue, e.g. Ultramarine Blue, Ultramarine Violet
Magenta	Earth green, e.g. Viridian, Chromium Oxide
Cyan	Earth Orange, e.g. Burnt Sienna/TRO

This is closer, I think; you get the all-important earth orange in here.

Earth green isn’t a bad idea, as a granulating green can be useful for foliage, but in reality I actually think earth yellow – which is dropped from this system – is more useful. I prefer the earthy greens mixed from cyan + earth yellow compared to those that start with an earth green as a base and adjust from there. But this may be personal preference.

The earth violet/blue slot is interesting, because the theory suggests it doesn’t really matter whether you put Ultramarine Violet or Ultramarine Blue in there, but I think it really, really matters. Ultramarine Violet is weak and lacking in value range, and I don’t consider it a key color in a small palette. But Ultramarine Blue would be super-useful! Again, though, I don’t think it’s so much because of the hue, but because it can mix dark; it is just about the only color in this palette that can.

Alternative #4: Split Primary, but Bright/Dark

This leads me to my next idea, which is to lean into value variety and propose a dark version of each primary. This goes back to the split primary system, but centrally addresses the problem of value range.

Bright Primary Color	Dark Color
Yellow (lemon yellow or middle yellow)	Gold, e.g. Quin Gold or Nickel Azo Yellow
Magenta	Crimson, e.g. Alizarin Crimson or Pyrrol Crimson
Cyan	Dark Blue, e.g. Indanthrone Blue

I can see a lot of possibilities here! Like the earth split primary palette, I think this is imperfect, but still much more usable than the traditional warm/cool split primary palette.

Positives: it makes it so easy to get those all-important dark values.

Negatives: various other types of variation are ignored (including granulation), and it lacks mixing complements and obvious easy brown mixers.

Alternative #5: Complementary, but Bright/Dark

Just to round things out.

Bright Primary Color	Dark Complement
Yellow (lemon yellow or middle yellow)	Dark violet, e.g. Dioxazine Violet
Magenta	Dark green, e.g. Perylene Green
Cyan	Maroon, e.g. Perylene Maroon

This is interesting. It kind of looks like a botanical palette.

But for me, personally, as a primarily landscape artist, I feel like this is straying further from usability. Maroon is possibly a good inclusion, as I think it can make extremely nice grays and neutrals with cyan (though browns are less possible than with Earth Orange). The others are not really colors I would tend to nominate as crucial – I don’t even have them in my larger palette.

What are my top six colors?

Switching from a top-down, theoetical lens to a bottom-up, practical lens, what are the actual top six color slots I use most often in painting? I estimated this by looking at the most commonly used colors in my painting notes and which tubes are most used up/how frequently I refill my pans, making allowances for how strong the paints are and how many options I have each in slot. Here is what I came up with:

Color I use often	Associated slot
Indanthrone Blue (PB60)	Dark Blue
Transparent Red Oxide (PR101)	Earth Orange
Monte Amiata Natural Sienna (PBr7)	Earth Yellow
Some kind of middle yellow; PY154, PY151, PY150, and PY97 have all had their day in the sun.	Yellow
Quinacridone Rose or similar – generally prefer Red Rose Deep or Quin Red PV19	Magenta
Some kind of Phthalo Blue – generally prefer Phthalo Blue Red Shade	Cyan

The gang is all here in terms of the modern primaries – but if anything, more used are these other, more muted colors!

You could argue that DS Indanthrone Blue is a violet-blue, that TRO is an orange, or that MANS is, in some way, an orange-yellow. But if you ask me what I find most useful about these colors, it’s not their position on the color wheel really – it’s other characteristics:

• Indanthrone Blue is dark, which makes it easy to mix dark values. If I couldn’t use IB anymore, I would look replace it with another equally dark color rather than one that has a similar hue.
• TRO also gets pretty dark and has a muted chroma that makes it similar enough to brown to easily mix a wide variety of browns, which is useful in landscape painting. It also has great granulation.
• MANS also has the granulation going for it as well as the gentle brown-ness that means it mixes pleasant browns.

I was hoping one of the above systems would suggest a six-color set like this or as useful as this, but none of them did. Beyond CMY primaries, it feels like the other colors I actually like are a hodgepodge, with different ones serving the roles of dark, earth complement, and earth equivalent. But maybe that is what is so useful about them: that they each serve different roles.

Given that I know what colors I like, it’s starting to feel silly to try to back my way from my favorite colors into a system. Why not just use what I like? I do wish there were some coherence to it, though, so it could be turned into a simple rule of thumb for teaching purposes. It’s easy to say “screw the rules, paint with the colors you like,” but that gives beginners nowhere to start.

Maybe it’s enough to say that my recommended starter palette is CMY + dark blue, earth orange, and earth yellow. But is that applicable to anyone else, or just me?

What are your six most-used colors? If you’ve used a split primary palette, what did you like and dislike? Did you end up swapping any colors?