Tertiary colors are the six hues you get by mixing a primary with the secondary next to it: red-orange, yellow-orange, yellow-green, blue-green, blue-violet and red-violet. That is the answer taught in almost every art class, and it is the answer the search results will give you.
It is also not the original meaning of the word, which is why the question keeps getting asked. Before the six-color version took over, a tertiary was something else entirely, there were only three of them, and one of them was olive. Both definitions are still in circulation, which is how you end up with people confidently telling each other that there are six tertiary colors, three tertiary colors, or, in one of the more popular searches on the topic, five.
Below are both sets, with hex codes. Then the part I have not seen anyone check: whether a tertiary is actually halfway between its two parents. The word says it should be. I ran all six through the color difference formula this site uses for scoring, and not one of them is. The worst offender sits three and a half times closer to one parent than the other.
The six tertiary colors
This is the set people are looking for. Each one is a primary mixed with an adjacent secondary, in roughly equal parts, which is why the names are all hyphenated pairs. Convention puts the primary first.
- Red-orange (vermilion) #ff4000. Red plus orange. L* 57.
- Yellow-orange (amber) #ffbf00. Yellow plus orange. L* 81.
- Yellow-green (chartreuse) #bfff00. Yellow plus green. L* 93.
- Blue-green (teal) #00bfbf. Blue plus green. L* 70.
- Blue-violet (indigo) #5500ff. Blue plus violet. L* 36.
- Red-violet (magenta) #bf00bf. Red plus violet. L* 45.
L* there is CIE lightness, running from 0 for black to 100 for white, built so that equal steps look like equal steps. I have included it because it is the number that decides whether a palette works, and because the spread is worth noticing before you go any further. The six tertiaries run from 36 to 93. Chartreuse is nearly white. Indigo is closer to black than to middle gray. They are presented as six members of one family, and they do not weigh anything like the same amount. If the hex notation is new to you, how to read hex color codes takes about two minutes.
Together with three primaries and three secondaries, these six complete the twelve-part color wheel that most teaching is built around. That wheel is the thing I want to put under measurement, because its central promise is that the twelve steps are even.
The other three tertiary colors
Open a color theory book from the nineteenth century and tertiary means something different. George Field, writing in Chromatography in 1835, is blunt about it: tertiary colours are three only, citrine, russet and olive. They are not primary plus secondary. They are secondary plus secondary.
- Citrine #e4d00a. Orange plus green. A dull mustard yellow.
- Russet #80461b. Orange plus purple. A reddish brown.
- Olive #808000. Green plus purple. The olive you already know.
The logic is tidy in a way the modern version is not. Primaries mix in pairs to give secondaries. Secondaries mix in pairs to give tertiaries. Each level is built from the one above it, and the count stays at three the whole way down. Field also points out that each tertiary can be resolved back into all three primaries, with one predominating, which is exactly why they come out muted rather than bright.
Muted is measurable. Chroma, C*, is the distance of a color from gray on the same scale as L*. Fully saturated hues on the wheel run from C* 50 up to C* 134. Field's three tertiaries measure 83 for citrine, 58 for olive and 41 for russet. Russet is less colorful than any saturated hue on the wheel. That is the real difference between the two definitions, and it is more interesting than the disagreement about which colors qualify.
The modern definition treats a tertiary as a position. The old one treats it as a degree of dullness. Red-orange is a place on the wheel. Russet is what happens when you have mixed enough different things together that the result starts heading toward brown. If you want to see where that road ends, what colors make brown is the same process taken one step further.
My honest opinion on which is correct: the six-color version is the useful one and the three-color version is the coherent one. Six gives you a workable vocabulary for the wheel, which is what most people need. Three gives you a system that actually explains why mixtures get duller as they get more complicated, which is the thing beginners struggle with most. The six-color version won because color pickers are arranged as wheels and wheels want twelve labeled spokes, not nine.
So is olive a tertiary color?
Under Field, yes, definitively. Under the modern six, no, because the modern six contains no olive. Both answers are correct and the person asking is usually being told one of them without being told there is another. Same for the search that asks for five tertiary colors: there is no set of five. It is the residue of two systems, one with six members and one with three, being averaged together by people who half-remember both.
Where the twelve-step wheel stops being even
Now the measurement. The claim built into the twelve-part wheel is that the steps are equal, that moving 30 degrees does the same amount of work wherever you do it. It does not.
The tool is CIEDE2000, which reduces a pair of colors to one number for how different they look. About 1.0 is the smallest gap a trained eye can reliably catch. Past about 50 you are plainly looking at two different colors. I took the twelve fully saturated hues at 30 degree intervals on the digital wheel, the one behind the hue slider in every design tool, and measured each step against the next.
The steps averaged ΔE 23. The smallest was 5.2, from chartreuse #80ff00 to green #00ff00. The largest was 43.9, from cyan #00ffff to azure #0080ff. That is a ratio of more than eight to one between the biggest step on the wheel and the smallest, for two rotations of identical size.
The imbalance is not scattered, either. It is concentrated, and once you see where, the whole layout looks different:
- The green region, hue 90 through 150, covers 60 degrees and ΔE 14 in total.
- The blue region, hue 180 through 240, covers the same 60 degrees and ΔE 73.
Five times the visible change, for the same rotation. The wheel hands out names at a constant rate while the underlying color changes at a wildly variable one. Three of the twelve names get spent inside a stretch of green you can barely resolve, while the enormous gap between cyan and blue gets a single tertiary to cover it. This is a real limitation of the whole framework, not a rounding error, and it is the reason a palette built by rotating a wheel so often needs rescuing afterward.
No tertiary is actually halfway
Which brings me to the part that surprised me. The word tertiary carries a specific promise: it names the color between two others. Halfway. I checked whether any of them are.
Method: take each tertiary position on the digital wheel, measure its ΔE to the primary on one side and the secondary on the other, and compare. If the tertiary is genuinely intermediate, the two numbers should be close to equal. Here is what came back, with the hue of the true perceptual midpoint alongside it:
- Orange #ff8000. ΔE 21 to red, ΔE 42 to yellow. Twice as near red. True midpoint at hue 37, not 30.
- Chartreuse #80ff00. ΔE 19 to yellow, ΔE 5 to green. Three and a half times nearer green. True midpoint at hue 76.
- Spring green #00ff80. ΔE 9 to green, ΔE 27 to cyan. Three times nearer green. True midpoint at hue 164.
- Azure #0080ff. ΔE 44 to cyan, ΔE 29 to blue. True midpoint at hue 205.
- Violet #8000ff. ΔE 9 to blue, ΔE 23 to magenta. Two and a half times nearer blue. True midpoint at hue 280.
- Rose #ff0080. ΔE 20 to magenta, ΔE 26 to red. The closest to honest of the six. True midpoint at hue 334.
Not one of the six is balanced. The best case, rose, is off by about a third. The worst, chartreuse, sits three and a half times closer to one parent than the other, which means that calling it the color between yellow and green is generous to the point of being wrong. It is a green with some yellow in it, and your eye has known that the whole time.
The errors also point the same way. Every tertiary bordering green gets pulled hard toward green, because green occupies a large arc of the wheel in which very little visible change happens. Chartreuse and spring green are both, in effect, dragged into green's orbit by the geometry.
The true midpoints in that list are the practical takeaway. If you want a color that genuinely reads as intermediate between yellow and green, hue 76 gets you there and hue 90 does not. Between green and cyan, use 164 rather than 150. Between blue and magenta, 280 rather than 270. These are not large corrections in the color picker, ten to fourteen degrees at most, and they are the difference between a transition that looks even and one that has an obvious lump in it.
How to mix tertiary colors that hold up
A few things follow from the numbers, and they contradict the usual advice in one specific place.
- Do not mix equal parts. The standard instruction for a tertiary is one part primary, one part secondary. On the evidence above, equal parts lands you at the geometric midpoint, which is not the visual one. For anything bordering green, bias the mix away from the green.
- Check the lightness before the hue. Your six tertiaries span L* 36 to 93. Placed at equal area, chartreuse will bury indigo without either color doing anything wrong. Decide what lightness each one needs to be, then find the hue.
- Use tertiaries where secondaries are too loud. This is what the six are genuinely good at. A tertiary carries the character of its dominant parent while reading as a considered choice rather than a default. Amber instead of yellow, teal instead of blue.
- Reach for Field's three when you need quiet. Citrine, russet and olive are the low-chroma end of the same idea, and they do a job no saturated hue can do. Any palette that feels shrill is usually a palette with nothing in it below C* 50.
- Stay on one wheel. The artist wheel and the digital wheel disagree about where the colors sit, and every measurement above is on the digital one because that is the wheel your software implements. The same split runs through the question of what color two colors make, and mixing the two mid-decision is how palettes that follow every rule still come out wrong.
Where tertiaries sit among the other schemes
Tertiary is a category of color. Complementary, analogous and triadic are schemes for combining them, and the six tertiaries are eligible for all three.
- A tertiary and the tertiary opposite it form a complementary pair, and usually a better behaved one than a primary pair, because neither color arrives with the associations that red or blue drags along.
- Three consecutive positions including a tertiary give you an analogous scheme. Worth knowing that on the green side of the wheel these will be far more subtle than the diagram suggests, for the ΔE reasons above.
- The two tertiary triads, red-orange with yellow-green and blue-violet, and yellow-orange with blue-green and red-violet, are the least clichéd triadic schemes available, precisely because none of their members is a color anyone learned to name in nursery.
That last one is my genuine recommendation if you are choosing a palette and want it to look deliberate. The tertiary triads carry no primary-color baggage at all, and the measured imbalance in them is no worse than in the famous sets.
Turning the numbers into an eye
Every figure above describes something you can already perceive but probably cannot name. The gap between reading that and knowing it closes faster than you would expect, and the fastest way through is to put your eye under a little pressure.
Hue Sort is the direct test for this. It hands you a gradient with the middle shuffled and asks you to rebuild it, which is a browser version of the Farnsworth-Munsell chip test. The green stretch is where nearly everybody stalls, and now you know why: there is barely 14 ΔE of difference spread across those chips. The blues sort themselves.
Color Mixer comes at it from the production side. You build a target color out of red, green and blue light, which forces you onto the digital wheel and makes tertiary positions something you have to find rather than read off a diagram. For the wider set of drills, train your eye for color collects the ones that produce measurable improvement.
The short version
There are six tertiary colors in the modern sense: red-orange, yellow-orange, yellow-green, blue-green, blue-violet and red-violet, each a primary mixed with a neighboring secondary. There are three in the older sense that Field set out in 1835: citrine, russet and olive, each a mixture of two secondaries, and all three noticeably muted. Neither definition is wrong, and the count confusion comes entirely from the two systems circulating side by side.
What no version of the definition survives is measurement. The twelve steps of the color wheel vary by more than eight to one in how much visible difference they cover, the green quarter holding five times less change than the blue quarter for the same rotation. And not one of the six tertiaries is the perceptual midpoint it is named for, with chartreuse missing by a factor of three and a half. Shift ten to fourteen degrees off the nominal positions and you get the colors the names were promising.
Then play a few rounds and find out how accurately you can place one of these hues from memory. Most people are worse at it than they expect, and the places where they fail line up almost exactly with the flat spots measured above.