Inclusive Illustration Research: How Emojis Handle Skin Tones


Author’s note: I’ve been researching best practices around inclusion for some recent freelance illustration work. Given the prevalance of emojis in 2017, I decided to use them as a starting point for figuring out how to color people in my illustrations.

The Unicode standards for emojis generally follow the Fitzpatrick Scale when it comes to representing people and their skin tones. The standards themselves are recommedations for sofware platforms, so the exact implementation varies slightly from company to company. Generally, we end up with is five skintones on top of a “default” cartoon yellow.

✊ ✊🏻 ✊🏼 ✊🏽 ✊🏾 ✊🏿

While this derivation of the Fitzpatrick Scale is pretty robust and inclusive, note that the scale itself was originally developed in 1975 “as a way to estimate the response of different types of skin to ultraviolet (UV) light.” In other words, it’s an old standard that was designed for classifying suntans not illustrating or representing people.

How skin tones look on major platforms

I’ve chosen the raised fist emoji as it offers a grew display of how shading (highlights and shadows) have been handled by the different illustration styles:

Diagram of raised hand emoji skin tones across multiple platforms

And their respective base color hex codes:

iOSAndroidFacebookTwitterMicrosoftUnicode sample

As you can see, while everyone follows the same convention loosely, there is no precise consensus on which exact colors to use. This is as it should be because the Unicode emoji standards are guidelines, not precise settings.

Other interesting observations of note:

  • Facebook adds an extra color for the highlights on the palm of the darkest fist (see above)
  • Twitter’s skin tones tend to be lighter across the board (particularly its browns) than that of other platforms
  • A few of Microsoft’s skintones are precisely the same as the Unicode samples

So far though, the Fitzpatrick-inspired system seems to be holding be pretty well. Having five choices of skin color is a good number – not too few, not too many – and a wide range of skin tons are represented.

How hair complicates everything

However, the system breaks down when we factor hair color into the equation. We’re allowed to choose the skin tone but not hair color. So the platforms are forced to assign hair color automatically:

A diagram of the various permutations of the woman emoji across all the skin tones and various major platforms

First, you’ll notice (on the left-most column) that there’s no consensus on whether the cartoon yellow skin should be accompanied by yellow hair or brown hair. The Unicode Consortium actually recommends an across-the-board “neutral” dark hair:

Sample image of dark hair on all emoji skin tones, from a Unicode report

No particular hair color is required, however, dark hair is generally regarded as more neutral because people of every skin tone can have black (or very dark brown) hair.

Nobody followed this recommendation.

Then, there’s no consensus about how “blond” the tanned person (third column) should be. Android skews for a dark/strawberry blond whereas Facebook has just opted out and replaced it with a light brown.

More importantly, things get tricky when it comes to representation. Within the current system, the most pale cluster of people (column one) look East Asian while the blond-haired cluster (column three) look very tanned in comparison.

Jeremy Burge explains why this is so on Emojipedia. The Unicode Consortium’s recommendation uses the Fitzpatrick Scale selectively. The scale has six skin types, but Unicode recommends five skin tone modifiers – they cut one from the list by lumping types one and two together. Wikipedia describes them as:

  • Type I (scores 0–6) always burns, never tans (pale white; blond or red hair; blue eyes; freckles).
  • Type II (scores 7–13) usually burns, tans minimally (white; fair; blond or red hair; blue, green, or hazel eyes)

Given this paucity of choice, it seems like the software platforms opted to include East Asians by assigning a “neutral” dark/black hair to the palest skin tone. As for why the more tanned person was assigned blond hair, I have two guesses: 1) they wanted to show more hair colors for variety’s sake, and 2) because the Fitzpatrick Scale PDF that Unicode links to has a woman with dark blond hair under Type III.

My current workaround

#ffe2cc, #fadbaf, #e5bd95, #bf8b60, #98562e, #673929

Based on these issues of representation, my current skin tone palette is an extension of the Unicode-Fitzpatrick system in two ways:

  1. Bring back the additional type from the Fitzpatrick Scale/split up the palest skin tone into two (so that there are six skin tones in total).
  2. Make the second palest skin tone explicitly East Asian by adjusting the hue so it’s more yellow. Likewise, I adjusted the palest skin tone so it’s more red/pink in contrast.

This system is, of course, far from perfect and will probably evolve as I gain more experience with inclusive illustration work. Stay tuned!

Other relevant resources

For a deeper and more painter-oriented take on skin tones and illustration, see:

Read the follow up to this article

Inclusive Illustration Research Part II: How Makeup Brands Handle Skin Tone (Better)

See other posts tagged