What color is chartreuse: measured, and the reason so many people remember it as pink

Chartreuse is #7FFF00. Red at half, green at full, blue at nothing. It is one of the 141 colours CSS will accept by name, it has been in the web palette since the X11 days, and it is named after a liqueur that Carthusian monks still distil near Grenoble.

That is the answer to the question. The rest of this piece is about three things the answer hides: that the colour is not where the number says it is, that there are two chartreuses and they are easy to tell apart, and that a surprising number of people are completely certain chartreuse is pink.

Every figure below came out of the same pipeline. Hex to linear sRGB under a D65 white point, into CIELAB, compared with CIEDE2000, with the one-unit-of-noticeable-difference threshold taken from our own just noticeable difference measurement at 2.3 units. No number here is quoted from a chart.

Chartreuse is not halfway between yellow and green

Wikipedia, Adobe, Canva, Figma and roughly every colour blog agree on one sentence: chartreuse sits between yellow and green on the colour wheel, halfway along. Google’s own summary of the question repeats it. The claim is easy to believe because the hex code says it outright. 7F is 127, which is as close to half of 255 as an integer gets, so the red channel is at exactly 50 percent while green is pinned at full. Halfway, in code.

Hue in sRGB is not linear in the channel value, though, and this is exactly the trap reading hex codes sets for people. The arc from yellow to the green primary in perceptual hue is only 33.16 degrees wide. Yellow sits at 102.85 degrees, green at 136.02. Chartreuse lands at 128.43.

That is 77.1 percent of the way to green, not 50. The colour that genuinely splits the arc down the middle, at hue 119.43 degrees, is #B6FF00, which has its red channel at 182 rather than 127. The two are 6.12 CIEDE2000 units apart, which is 2.66 just noticeable differences. Nobody would mistake one for the other.

So the standard description of chartreuse is wrong, and it is wrong in a specific direction: it oversells the yellow. Chartreuse is a green with some yellow in it, not a colour poised between the two. If you have sorted hues by hand in the hue sorting mode you will have felt this already, because the yellow-to-green stretch of the wheel compresses hard and the greens take up far more of it than the channel numbers suggest.

There are two chartreuses, and they are 5.4 JNDs apart

The monks sell two liqueurs. Green Chartreuse at 55 percent alcohol, and a yellow one at 43 percent. The colour vocabulary inherited both. What the web calls chartreuse is really chartreuse green, and the older, more yellow value, usually listed as chartreuse yellow, is #DFFF00.

Between them: 12.52 CIEDE2000 units, which is 5.4 just noticeable differences. For scale, that is more than twice the gap between chartreuse and plain CSS lime. The breakdown is 4.95 units of lightness, 11.50 of chroma and 17.82 degrees of hue, so the yellow version is lighter, less saturated and genuinely a different hue.

This matters for the argument people have. When one person says chartreuse is a yellow and another says it is a green, they are not necessarily disagreeing about perception. They may be holding two different documented values, 5.4 JNDs apart, both correctly named chartreuse.

CSS already has a second name for chartreuse

Run every pair of CSS named colours through CIEDE2000, all 9,870 comparisons, and throw out the two exact aliases where the spec simply gives one hex two spellings (aqua with cyan, fuchsia with magenta). Eleven pairs are left that sit closer together than one JND. The closest of all is this one:

Three code points apart on red, three on green, and 0.65 units apart in measured colour. The runner-up pair in the whole palette is darkblue with navy at 1.56 units, so chartreuse and lawngreen are 2.4 times closer than the next closest thing. On a screen, at any size, they are the same colour. CSS ships two names for it and neither of them is an alias of the other.

Which is a small piece of evidence for a bigger point. The web palette was not designed. It accreted, out of X11, out of Netscape, out of whatever names people had already been using, and the yellow-green corner picked up more labels than it needed.

Is chartreuse the same as lime green

It is the second most asked question about the colour, so here is the neighbourhood measured. Distances are CIEDE2000 units from #7FFF00, and anything under 2.3 is a colour you could not reliably tell from chartreuse if the two were not side by side.

So: no, chartreuse is not lime green in the sense people usually mean it, because CSS limegreen is 13.37 units away, a darker and much duller colour. But chartreuse is only 5.15 units from CSS lime, which is the pure green primary, and 4.43 from greenyellow. It is closer to pure green than it is to anything with the word yellow in its name except greenyellow. Compare that with olive at 34.75 and green at 35.60, both of which are the same hue family at a fraction of the lightness. More of this neighbourhood is mapped in shades of green.

So why is everyone so sure it is pink

Here is the genuinely strange part of the chartreuse story. A large number of people do not merely misjudge the shade. They remember chartreuse as a pink, or a magenta, or a reddish rose, and they remember it with real confidence. It is one of the standard entries on every list of the Mandela effect, it has its own threads on the Mandela effect forums going back to 2015, and Katy Kelleher wrote the whole thing up for The Paris Review by opening with her own version of the false memory.

Nobody has measured why. The usual explanation is a shrug about false memory, which is true but not informative. So I looked for mechanisms that can be measured, and found two that hold up, plus one that does not.

The afterimage of chartreuse is a pink

Stare at a saturated colour and the cones tuned to it fatigue. Look away and the opponent channels report the imbalance as a ghost in roughly the complementary hue, which is the mechanism behind afterimages and the reason opponent process theory exists. The honest way to compute it is to hold lightness fixed and flip the two opponent axes, which is what the complement is in CIELAB rather than in RGB.

Do that to chartreuse and the complement lands at hue 308.43 degrees, asking for a chroma of 109.5. No screen can deliver that, and neither can a fatigued retina. Compress the chroma until it fits sRGB and only 18 percent survives, which leaves this:

A pale pink-lilac. Measured against the CSS colours people actually call pink or magenta, that afterimage sits 16.6 degrees of hue from the nearest member of the family. Chartreuse itself sits 99.5 degrees away from the nearest one, taking the most generous neighbour available in both cases. The afterimage is six times closer to pink than the colour is.

That is not a claim that everyone who remembers a pink chartreuse was staring at a chartreuse wall. It is a narrower and more defensible claim: the one colour the visual system reliably pairs with chartreuse, the only other colour it ever hands you unprompted while you are looking at it, is a pink. If a memory of the colour has a second colour attached to it, the mechanism says that second colour is a pink-lilac, and memories do not come tagged with which half was the stimulus.

Fancy French colour words are usually red

The second mechanism needs no perception at all. It is a question of what the word sounds like it should mean.

Most people meet chartreuse as a word before they meet it as a colour, and what they learn in that moment is that it is an elegant, French-sounding, hard-to-spell colour name. If that is all you have, you guess. So the question is what the right guess would be. I took the colour words English borrowed straight from French, 28 of them, using a strict rule: the English term has to be the French word itself, not an English translation and not a French place name doing adjectival duty, so burgundy and champagne are out. Then I classified each by hue family and compared the result against the 141 CSS names as a baseline.

French colour words in English are 1.83 times more likely to name a red or a pink than a colour name drawn at random from the web palette, and only 0.67 times as likely to name a green. Cerise, puce, rose, framboise, grenadine, amaranth, aubergine, lilac. Against four greens in the whole set, one of which is chartreuse itself.

The result survives the obvious objection, which is that the red-or-pink boundary depends on where you decide a colour is too grey to have a family. Moving that cutoff anywhere from 8 to 25 units of chroma keeps the red and pink ratio between 1.75 and 2.01, and keeps the green ratio between 0.56 and 0.87. One caveat worth stating: compared against our own 131-name pool from the name that color mode the red and pink ratio drops to 1.21, because that pool was built out of designer vocabulary and is itself stuffed with borrowed names. The CSS baseline is the cleaner comparison.

Which gives the pink memory a rational core. Someone who knows only that chartreuse is a fancy French colour and guesses pink is not being careless. They are playing the base rates, and the base rates favour pink almost two to one over green. Guess often enough, hear someone else guess the same way, and the guess hardens into something that feels like a memory. That is the ordinary route to a false memory, and here it has a number on it.

The explanation that did not survive

The tempting third explanation is crowding: that the yellow-green corner of the palette carries too many names for the space it occupies, so nobody can keep them straight. It sounds right. It is not what the measurement says.

Split the saturated CSS names into 30-degree hue bins and measure the mean distance from each name to its nearest neighbour in the same bin. Chartreuse’s bin, 120 to 150 degrees, holds 13 names at a mean nearest-neighbour distance of 6.08 units. The palette-wide figure is 6.98. So the yellow-green wedge is slightly denser than average and no more, and it is not even the worst offender: the 300 to 330 degree bin, the magenta and violet stretch, packs 22 names in at 5.00 units. Nobody is confused about magenta.

Chartreuse is unusual for having one neighbour at 0.65 units, not for living in a crowd. The confusion is specific to the word, not to the region.

A fourth idea, for completeness, because it is suggestive but weak. Taking the 131 colour names in our own game pool and ranking them by spelling similarity to chartreuse, four of the ten closest are reds or pinks: cerise, carmine, caput mortuum and thistle. The two closest are citrine and charcoal, which are neither. One colour word out of ten is not a finding, and I would not lean on it, but it points the same way as the loanword count.

What this does to naming a colour from memory

Put the three measurements together and chartreuse turns out to be a near-perfect trap for colour memory, for reasons that have nothing to do with the eye.

• The name maps to two documented values 5.4 JNDs apart, so there is no single right answer to check a memory against.
• The standard verbal description, halfway between yellow and green, points 27 percentage points away from where the colour actually sits. A person reasoning from the description will land on #B6FF00 and be 2.66 JNDs wrong.
• The word carries a prior that leans red and pink by almost two to one, and that prior is doing its work before the colour is ever seen.
• The one colour the visual system attaches to chartreuse without being asked is a pale pink.

None of which requires an alternate timeline. It requires only that colour memory is verbal first and visual second, which is what the measurements in aphantasia and colour memory found: the floor people hit when they encode a colour as a word rather than as an image is around 18 CIEDE2000 units of error. Chartreuse is a colour almost everyone encodes as a word.

If you want to find out which side of that you are on, the name that color mode puts chartreuse up against citrine and the rest of the yellow-greens and asks you to pick, and the hex mode gives you the code and makes you build the colour. The main colour memory game is the harsher version: it shows you a colour, takes it away, and asks you to mix it back. Holding #7FFF00 in your head for five seconds is much harder than reading about it.

For the rest of the vocabulary there is what happens when you measure the colour wheel, which finds the same compression in the yellow-green region from a different direction, and memory colours for how far the remembered version of a familiar colour drifts from the real one.