Skin tone chart: Monk, Fitzpatrick and undertone side by side

What on this chart is measured, sourced or ours

Plenty of skin tone charts are a row of swatches with pretty names and no origin. Nobody can check them, because nothing stands behind them. So here is the status of every column above, and you can hold us to it.

Names like honey and caramel aren’t on the chart at all, since no scale defines them. Our guide to skin tones places those words approximately and says so.

The Monk scale: ten tones made for checking images

The Monk Skin Tone Scale comes from Ellis Monk, a Harvard professor whose research is on skin tone and colorism, working with Google. Its first job was testing whether image datasets and computer vision work as well across skin tones, and it replaced an older habit of using Fitzpatrick types for that. Google is open about what the swatches are: “illustrative references for data annotation tasks”, where the goal “is not to capture a precise numeric value of skin tone”. In Google’s own testing, people across US communities found it more inclusive than Fitzpatrick, and as representative of their skin as a 40-point beauty palette.

Why ten? Monk’s reasoning is about people, not color: past 10 or 12 points, the same person can no longer reliably pick the same tone twice. That choice has consequences you can see in the numbers. The steps aren’t even. The widest gap is between tones 5 and 6, a big drop in lightness. Tones 1 to 3 are nearly level, and tone 3 is actually a touch lighter than tone 2, because what changes there is yellowness. If you feel you fall between two steps, you probably do.

Why the Fitzpatrick column has no colors

Thomas Fitzpatrick developed his types in 1975 to judge how readily skin burns during phototherapy. The original list ran from I to IV. Types V and VI were added later to take in people of Asian, Indian and African origin, as Ware and colleagues recount in Cutis. It sorts people by what happens in the sun, and DermNet’s table is still worded that way.

It drifted anyway. In the same paper’s survey of 140 dermatologists and trainees, 47% said they used it in clinical notes to describe a patient’s baseline skin color and 31% to describe race or ethnicity, though the type correlates poorly with constitutive skin color. An IEEE study reached a similar verdict from the imaging side: Fitzpatrick type is “poorly predictive of skin-tone”. Emoji skin colors grew from it too, with types I and II merged into one.

So when a chart paints type IV a specific brown, somebody made up that brown. We’ve left the colors out on purpose, and the six rows don’t line up with the ten next to them, because there is no honest way to line them up.

The research angle: closest to what a lab would give you

The individual typology angle turns two instrument readings, how light the skin is and how yellow, into one number. Higher means lighter. It traces to Chardon and colleagues, who used this kind of colorimetry in 1991 to classify skin for sun product research, working from a population they described as Caucasian. Del Bino’s group later tied the angle to real biology across a far wider range, reporting that it correlates with how skin responds to UV, in work including 3,500 women across several regions. Their six bands are applied, in their words, “regardless of ethnic origin”.

Three things to know before you lean on the column above:

• It isn’t Fitzpatrick in another form. Osto and colleagues found poor correlation between the six angle bands and the six types, and warn against using them interchangeably.
• It gets used where the stakes are real. A 2026 study in The BMJ used the measured angle to test pulse oximeters and found readings averaged 0.6 to 1.5 percentage points higher on darker skin than on lighter skin.
• A swatch isn’t skin. Run the published hexes through the formula and tones 1 to 4 all land in Very light, while no tone at all lands in Intermediate. That gap is the wide step from tone 5 to tone 6 showing up again, and it’s a fact about the swatches. Your own measured skin could sit there quite happily.

Where is olive on a skin tone chart?

It has no row, because olive isn’t a depth. Wikipedia files it as a tone roughly matching Fitzpatrick III to V, while makeup artists describe a cast that runs from fair to deep. Both can’t be the last word, and the strip on the right of the chart takes the second view: an olive version at every depth, ours and illustrative. For the full chart with its own hex codes, see olive skin.

Which chart should you use?

It depends entirely on the job. Picking the wrong one is where most of the confusion starts.

• Choosing foundation: none of these. Google itself notes that makeup needs the precision of larger scales, up to 40 tones or more. Use a brand’s own range, with your depth and undertone as the starting point. The foundation shade finder gets you that far.
• Talking to a dermatologist: Fitzpatrick, answered honestly about burning and tanning, and let them use it. Google warns that Monk “may not adequately assess skin cancer risk”.
• Drawing, game art or interface design: Monk for a credited reference, then the ramps in our skin color palette.
• Labeling photos for a dataset: Monk, as designed, with at least two annotators per region as Google recommends.
• Knowing roughly where you sit: the skin tone test, which reports the nearest Monk tone and your skin tone in plain words, as an estimate from one picture. A camera reads fair skin darker than tones 1 to 3 look, so the tool moves light readings up the scale, using points that are our own judgment.

Try any color against the chart

Paste a hex from a drawing, a foundation swatch or a photo. This runs on the numbers alone, using the same weighting as our skin tone test, where lightness counts most. A hex sampled from a photo carries that photo’s light with it, so treat the answer as a reading of the picture.

What no chart can tell you

One swatch can’t describe a whole person. Google says as much: skin color varies across people, across areas of the same person, and over time. Xiao and colleagues measured this with a spectrophotometer and found redness differing significantly by body site, highest on the forehead and lowest on the inner arm. Pick one spot and stay with it when you compare.

Nor does the order of a chart mean anything except order. Going from 1 to 10 is a measure of lightness, in the same way a ruler is a measure of length. No tone on it is a better or worse one, and no tone tells you anyone’s race. Google lists that as a plain rule for anyone using the scale: don’t equate skin tone to race. Dr. Monk built it partly to pull those two apart.

Related guides and tools

• GuideSkin color paletteCopyable skin tone hex codes from a published scale, plus ramps by undertone.
• GuideFair skinWhat counts as fair, its possible undertones and how to pick colors that do not drain it.
• GuideCaramel skin toneFour medium skin tone names compared, with reference swatches and likely undertones.
• GuideSkin tonesThe common names, the scales that measure skin and why undertone is a separate axis.