35
One clean switch
In 35 languages, the most commonly used word changes just once between the reference green and blue. Only those positions appear on the scale above.
Colour and language
The same palette, divided by 110 languages
The spectrum flows without borders. Words divide it, and each language can draw a different map. Some place a single boundary between green and blue. Most use the same word at both ends or change terms more than once.
The band below runs from green to blue without steps. Lightness and saturation are held constant along its whole length, so hue is the only thing that changes. Choose the place where green becomes blue for you.
Each mark represents one of the 35 languages in which the most commonly used word changes just once between the reference green chip and the reference blue. Your line is white, taller and marked with a dot.
The larger words describe the colour plainly. The smaller codes underneath preserve its exact position on the Munsell scale.
You drew one line because English separates green from blue here. In the World Color Survey, only 35 of 110 languages fit that shape. In the others, one word may cover both ends or the terms may change more than once.
35
In 35 languages, the most commonly used word changes just once between the reference green and blue. Only those positions appear on the scale above.
64
In 64 languages, the reference green chip and the reference blue chip are most often given the same name. Two use another word somewhere between those chips, so this statement is about the two ends rather than the whole band.
11
11 languages introduce a third word or return to an earlier one. Their data describes a region rather than a single frontier.
This is a map of how speakers of one language divided the palette. Each rectangle is one of the 330 colour chips shown to participants. When most speakers used the same word for several chips, their rectangles receive the same colour on this map. Choose another language and the territories redraw themselves.
How to read the map: the labels above are the speakers' actual words. The square beside each word shows the average colour of its territory. Hue changes from left to right; the chips run from light to dark as you move down. The separate column on the left is the scale from white to black.
In 1858, before he became prime minister, William Gladstone published three volumes on Homer. In the Iliad and the Odyssey he found few words organised by hue, no term clearly matching modern blue, and the sea described seventeen times as oinops, an adjective associated with wine. He concluded that the “organ of colour and its impressions” was only partially developed among the Greeks of the heroic age.
The wording was long read as a claim about anatomy. Gladstone later clarified that he meant a trained faculty and rejected the idea of a physical defect. Current scholarship shows that Greeks distinguished hues, while Homeric colour words also carried information about light, surface, material and movement. The difference lay in how they grouped what they saw.
In 1969, Brent Berlin and Paul Kay proposed that basic colour terms tend to appear in a partly predictable order. The smallest systems first group light and dark colours; red separates early; green and yellow, blue, brown and other categories follow. The hypothesis was challenged over its sample of 20 languages and its method. The World Color Survey extended the test to 110 communities, using the same palette of 330 chips with about 24 speakers in each.
The order is not a rigid ladder, but the responses preserve a few points of attraction. Across both surveys, the 109,256 choices of the clearest example concentrate on particular chips; among the chromatic colours, two peaks appear at 2.5G and 2.5PB. Those are the green and blue landmarks used here. Their positions come from speakers' choices, independently of the English words “green” and “blue”.
Albastru comes from Latin albaster, from albus. The starting sense was grey, washed out, whitish, and Alexandru Ciorănescu's etymological dictionary records it surviving in Moldavian speech and in Maramureș, where a grey sheep is called albastră. The word reached the clear sky by way of the pale one.
Vânăt also comes from Latin, from venetus. It can name dark blue tending to violet, the grey-blue of horses, and the colour of skin that has been struck or chilled. Romanian thus inherited two Latin words for the blue region: the ordinary term, albastru, began with white; vânăt remained attached to darker, greyer or more violet shades.
In ordinary speech, Russian uses goluboy for light blue and siniy for dark blue. Jonathan Winawer and colleagues showed in 2007 that Russian speakers tell two samples apart faster when they lie on opposite sides of that boundary. The advantage disappears when they perform a verbal task at the same time. Romanian has the word bleu for light blue; the band below asks where you would use it.
I found no survey that measures this boundary in Romanian or shows that bleu works like goluboy. The band keeps the same hue and changes only its lightness. Your choice stays in your browser.
Languages can place boundaries in different positions and group colours into different shapes. In 64 of 110, the reference green chip and the reference blue chip are most often given the same name. At the same time, people's choices of the best example for each term gather in a few stable regions of the palette.
It remains open how much of that gathering comes from vision and how much from language. Debi Roberson, Jules Davidoff and Ian Davies, working with Berinmo speakers in Papua New Guinea and Himba speakers in Namibia, showed that each language's own boundaries change results in memory and colour-discrimination tests. Terry Regier, Paul Kay and Richard Cook showed in 2005 that the most typical examples remain similar across languages in the World Color Survey. The results answer different questions: a language's boundaries can affect some tests while its most typical examples remain similar to those found elsewhere.
The original samples are papers from the Munsell colour system, viewed under controlled laboratory lighting. Almost a third of their most intense colours lie beyond the range of an ordinary screen. The band uses an intensity that screens can display. Your line can be placed approximately in the corresponding place along the colour sequence because the colours remain in the same order; it is not an equivalent measurement to the Munsell-chip test under controlled lighting.
For each of the 20 languages in Berlin and Kay's appendix, the public file contains one combined set of responses, with several gaps between the reference colours. Those data remain in the files behind the page and in the history of the idea; the three main counts use only the World Color Survey, with an average of 24 speakers for each of its 110 languages. The question reached you in English. Before you touched the band, its words had already given the question a shape.
For each language, I chose the word used most often for every chip and then in every colour column. A position enters the comparison when the columns between the reference green and blue form one group named one way, followed by one group named another. A third word, a return to the first or a column with no responses is counted separately. The table repeats the calculation across narrower bands, from lighter colours to darker ones.
| Rows used | Clean switches | Same word at both ends | More complex patterns | Midpoint of switches |
|---|---|---|---|---|
| B-I | 35 | 64 | 11 | 7.5BG–10BG |
| C-G | 34 | 65 | 11 | 7.5BG–10BG |
| D-F | 29 | 65 | 16 | 5BG–7.5BG |
| E | 23 | 66 | 21 | 5BG–7.5BG |
The official World Color Survey archive publishes the study's files, documentation and data. The figures here were recomputed from release 20110316.
The archive contains data and documentation for 110 languages, with about 24 speakers each and 330 chips named one at a time. Release 20110316 supplies the 863,280 judgments and every boundary position computed here.
This is the archived copy used for the recomputation. Berkeley's live archive remains the starting point for the study's documentation and files; this link fixes the exact package analysed here.
The 20 languages of Basic Color Terms, among them American English and Vietnamese. The public file contains one combined set of responses for each language, so they remain historical context rather than entering the main counts.
Laboratory colour measurements for all 330 chips. The colours on this page are calculated from them and adjusted for the light and limits of an ordinary screen.
The argument that people's choices of the clearest example of each colour word gather in a few regions of the palette. The independent calculation here finds the same regions.
The source for the goluboy/siniy distinction and for the effect vanishing under a verbal load. It is a lightness border, not a hue one, so it does not appear on the first band.
The relativist side of the debate. Berinmo has five basic words, Himba five as well, among them burou, which covers green and blue and reaches into purple.
The DEX and Ciorănescu entries for the two Latin etymologies and for the sense “grey” surviving in Moldavia and Maramureș.
The dictionary entry behind the Romanian label at the light end of the second band: bleu means light blue.
Volume three, section IV, “Homer's Perceptions and Use of Colour”, from page 457, where Gladstone inventories the colour adjectives and describes the “organ of colour” as partially developed.
A current scholarly overview: Greeks distinguished hues, while their terms could also encode light, surface, material and movement.
Examines Gladstone's wording and his later clarification that he meant a trained faculty rather than a physical defect of sight.