Get yourself a new wooden palette and sand down any splinters or imperfections. Make sure it fits your hand well. You will need some fresh shellac; the stuff you get at a furniture store will do. You could also make your own by dissolving dry shellac flakes in denatured alcohol. You will need some denatured alcohol for cleaning up, a one-foot square of lint-free cloth, preferably linen, and a small amount of linseed oil. Fold the cloth into a convenient size—about 2” is good. Dip the pad in shellac and wipe it over the surface of the palette, covering thoroughly and evenly. Pour a small amount of linseed oil onto the surface of your cloth. While the shellac is still tacky, rub the cloth over the surface of the palette in a circular motion. With a little practice, you get a sense of how to do this so that the shellac is spread smoothly and pressed into the wood. Let the palette dry enough so that the shellac is it no longer tacky (15 minutes or so). Apply and then polish another layer, and another, and another, until the palette has a surface that feels sufficiently thick (I apply about 8–10 layers). It is beautifully smooth.

If the surface of your palette eventually gets damaged by repeated scraping, you can remove the old shellac with denatured alcohol and apply a new French polish. In three years of heavy use, I haven’t yet had to do that.

That means that I have a core group of paints that I almost always have on my palette. These include flake white, burnt sienna, raw sienna, yellow ochre, and ultramarine blue. I am very, very familiar with how these paints mix together. I’ve used them over and over; they no longer hold any mysteries for me. Lots of painting problems can be solved with just these paints, because most of the world is pretty low in chroma and is filled with hues and values that can be mixed with these paints. By using low chroma paints, rather than neutralizing intense colors as some painters prefer, it is much easier to avoid accidentally drifting the chroma too high.

When needed, I will add some other low-chroma paints, including red ochre, Studio Products’ Tuscan red, Williamsburg Italian terre verte, Doak French ochre extra pale, burnt umber, raw umber, ivory black, transparent blue oxide. When I want very bright, opaque whites, I add titanium white. When I want very subtle mixtures with white, I add zinc white. I’m pretty familiar with how all of these colors work

That’s 90% of the paint I use. But there are times when I need more chroma. If so, I pull out some of my bigger guns: viridian, Prussian blue, cadmium red, bismuth yellow, cobalt blue, genuine vermilion, pyrol ruby, Doak Florentine lake, Doak Alger blue, Indian yellow, dioxazine purple. I don’t know these colors that well, so when I use them I often need to spend time experimenting with how they mix. Often, I use them to intensify mixtures of my more standard colors. When appropriate, I use them with only slight modification, for those small areas of chromatic color that can really make a painting jump (or fall, if done badly). Lots of my paintings don’t have any of these intense colors, but I like having them there when I need them.

Any small grouping of paint colors will do. Here are a few useful limited palettes:

• Burnt sienna and ultramarine blue.
• Raw sienna and ultramarine blue.
• Yellow ochre, alizarin crimson, and black.
• Yellow ochre, alizarin crimson, burnt umber, and black.
• Yellow ochre, alizarin crimson, ultramarine blue, burnt sienna, and black.
• Cadmium red and black.
• Black (and white).
• Black and burnt umber (and white).

(Note that you can substitute a more lightfast pigment for alizarin crimson, such as pyrol ruby.)

As you can see, with these palettes there are hues and chromas that can’t be mixed, only suggested. Often, by using warm/cool contrasts, you can create the impression of colors that aren’t actually there. The classic example is creating the illusion of bright blue eyes using only a mixed grey from black and white, by placing warm yellows, reds, and oranges nearby. Many master figure painters juxtapose warm flesh tones with colors that look cool, but are actually warm/neutral. In doing so they often make deliberate use of a very limited set of paints. Because you are using so few colors, you become intimately familiar with how each of your few paints mixes with each of the others, and how various mixtures work when set against each other.

The lessened range of hue and chroma that are characteristic of a limited palette create a sense of harmony. Each part of the painting is consistent with every other part, and a group of paintings made with the same limited set of colors makes a series that is obviously related. It is harder to achieve realism with a limited palette, but once you are comfortable with a certain set of colors it can be surprising how seldom that seems like a serious limitation.

When selecting the particular paints to use, be aware that not all “raw siennas” are the same. Paints labeled identically by different manufacturers can have radically different masstone, undertone, color mixing, transparency, or other characteristics. That is particularly true with earth colors, which are often really synthetic oxides these days. So if you get used to having ultramarine blue and burnt sienna work beautifully together, you may find that if you switch brands (or even get a different batch) that wonderful balance of color isn’t quite so perfect. If you find a perfect paint, you may want to get an extra couple of tubes, because that perfection may not be available forever.

My teacher, Dennis Cheaney, uses this approach. It is based on the method advocated by Ted Seth Jacobs, his teacher. Here’s what Ted says about this in “Light for the Artist,” the book I’ve quoted from in a number of posts.

Some painters prefer to work with the fewest possible colors (called a “limited palette”). The disadvantage to this method is that mixed colors are not quite as chromatically intense as their counterparts out of the tube. For example, an orange made of red and yellow loses some chromatic intensity as compared to tube orange. The limited palette reduces our available chromatic range.

Another one of Ted’s students, Tony Ryder, was profiled in a recent article in American Artist. His palette for one painting has 47 paints on it:

flake white, misty blue, zinc white, titanium white, Naples yellow green, jaune brilliant, Naples yellow light, Naples yellow, Naples yellow red, cadmium yellow lemon, cadmium yellow, cadmium orange, coral red, brilliant pink, cadmium red, cadmium red scarlet, alizarin crimson, rose grey, cobalt violet, cobalt violet light, Winsor violet, ultramarine violet, cobalt blue, king’s blue, ultramarine blue, cerulean blue, cobalt green light, viridian, green grey, chrome oxide green, cinnabar green, Bohemian green earth, sap green, yellow ochre light, yellow grey, raw sienna, Old Holland ochre, deep ochre, raw umber greenish, mars yellow, mars orange, burnt sienna, mars violet, burnt umber, Van Dyke brown, Payne’s gray, ivory black.

That’s a lot of different paints.

I don’t claim to know more about painting than Ted Seth Jacobs, Tony Ryder, or Dennis Cheaney. But at my limited level of skill I do see a couple of problems with this approach. One is simply that it is much harder to learn the mixing characteristics of 47 paints as well as you can with, say, 6 paints. When mixed, pigments react in unpredictable ways. If you use a more “limited” palette, you can learn with great specificity the ways that each color mixes with every other color. If you don’t really know your colors, then you’ll often be surprised at the results of any given mixture. What you end up doing is having to fiddle with mixtures. You mix two paints, observe how the color shifts, then add another paint to compensate for the color mixing shift that you didn’t predict, then maybe have to do that once or twice more before the color is exactly right. As that happens, the chroma inevitably goes down. So you might have to then try to add some more of a brighter paint to pull the chroma back up. With this approach, you can spend a lot of time chasing color.

In his book, Ted also points out how mixing colors reduces chroma, but fails to account for that when he is selecting paints containing multiple pigments. In describing the value of a kitchen sink palette, he shows five different greens: gray green, sevres green, cobalt green light, cadmium green light, and olive green. The book was written awhile ago and he doesn’t say what brands he is using, but I think that at least two of those are multi-pigment paints. There is no such pigment as “cadmium green,” for example—it’s usually a blend of cadmium yellow and pthalo blue. Similarly, a number of the paints on Tony’s palette also contain multiple pigments. If the rationale for the large number of paints is to avoid chroma reduction from mixing, then I don’t see how it makes sense to choose paints that the manufacturers have already mixed for you. Paint companies don’t have any special way of mixing paint without the saturation costs that we have to cope with when we do the same thing on our palettes.

That is not to say that any of these guys don’t know how to mix paint. I’ve watched Dennis do it, and it’s impressive. In a few seconds, he’ll pull several colors together to produce a mixture with just the right value, hue, and chroma. But when I put that many colors onto my palette I get all mixed up. I lose track of which colors I’m using for what purpose. When mixing, I find myself either (a) using too many different paints in each mixture, chasing color all over the place, and tossing some mixtures and starting over; or (b) ignoring most of the paints on my palette and using only a few.

Nevertheless, some artists can manage a large palette of paints without apparent difficulty. Ted, Tony, and Dennis do it brilliantly. But it is not an approach that works for me.

Related article

Color and color mixing

One way to choose paint is to use what I will call a “color theory” palette. Such a palette is characterized by a limited number of paints (usually, but not always, between three and six—plus white). These colors are usually selected for high chroma, without regard to value. Their hues are distributed as evenly as possible around the color circle. Color theory palettes vary in how many colors are used and the particular distribution (color space) over which the colors are selected. They can involve a lot of experimentation to find exactly the “right” pigment to fit into a particular theoretical slot. Color theory palette aficionados may agonize over which pigment is just the right middle yellow, for example.

One simple color theory palette uses the traditional primary triad colors: red, yellow, and blue. Intermediate hues are obtained by mixing two primaries. A green is mixed with yellow and blue, for example. Mixtures are lightened with white and darkened or neutralized with mixtures of complementaries. Another simple color theory palette uses the printer’s three primary colors: cyan, magenta, and yellow. A CMY palette (if the pigments are selected correctly) is distributed a bit more evenly around the color circle than the RYB palette.

Either of these palettes can be used to mix just about any hue. Because (almost all) mixtures reduce chroma, they are very limited in which high-chroma colors can be mixed. Because the highest-chroma pigments have very different values, mixing the desired value can also be a challenge. Additionally, because all colors must be mixed from only three, you will probably find yourself doing a heck of a lot of mixing. The advantage to these palettes is that, with only three colors, you eventually learn their properties (however limited) very, very well.

Another approach is to use one each of the four “artist’s primaries.” This method raises green up from a mere complementary to a true primary, so the palette has red, yellow, blue, and green.

There are color theory palettes that involve more than just a three or four pigments. Another approach is to simply have one color for each of the six traditional primary and secondary colors (a “ROYGBV” palette). Yet another is to take each of the artist primary colors and split them into two, one of which is biased toward one of the two secondaries next to that primary, the other of which is biased toward the other of those secondaries. For example, the two secondaries next to blue are violet and green. So you would select a violet blue and a green blue. You’d then find an orange red, a violet red, a green yellow, and an orange yellow, for a total of six paints on your palette. The idea is that this split primary palette makes it easier to hit particular hue/chroma combinations than just a three color palette. And of course it does, to a degree.

One of the big proponents of the split primary palette approach is Michael Wilcox, author of “Blue and Yellow Don’t Make Green.” (His point with the title is that there are no pure primaries, and if there were they would only mix to black.) I find Wilcox hard to take seriously because I simply hate the way he writes. He has an irritatingly smarmy, superior, slappable writing style. He repeats himself over and over. And he doesn’t know the difference. Between a grammatical sentence. And one that is not.

Wilcox recommends two biased pigments for each of the traditional primaries (cadmium red light, quinacridone violet, cadmium yellow light, hansa yellow light, cerulean blue, and ultramarine blue). He also adds five other pigments (plus white) to his recommended palette for no theoretical reason, just because he likes them (pthalo blue, pthalo green, burnt sienna, yellow ochre, raw sienna, and titanium white). Bruce MacEvoy has an excellent criticism of Wilcox’s book, which largely comes down to three points: (1) Wilcox represents his theory as new, but it really rehashes color theories from the middle of the 19th century; (2) he misrepresents the nature of how colors reflect off of pigments; and (3) his pigment choices are not very well distributed around a hue circle. MacEvoy also has a bit to say about split primary palettes in general.

Yet another variant is to use the CMY printer’s primary palette and then add a secondary pigment in between each of them: this palette would have yellow, green, cyan, blue violet, magenta, and red orange, all distributed quite evenly around the hue circle. If I were a color theory palette person, that’s the one I’d probably use, because it’s the one that most accurately reflects actual color and the limitations of actual color mixing.

Plenty of artists successfully use a palette based on one of these approaches. With proper selection of paints, a color theory palette can be useful and flexible. The two problems with such palettes are (1) some high-chroma colors may not be obtainable; and (2) darkening and neutralizing high chroma pigments using only complementary mixtures can be an exercise in frustration, with continual re-mixing to manage hue shifts and saturation costs.