How to make sense of ancient Greek colours

When people want to talk about how language affects colours, ancient Greece is a handy thing to point at. Fundamental errors pop up again and again, though.

1. Popular treatments (and scholarly discourse, too) rely far too much on dictionary definitions. Languages divide up the available colour palette in different ways. A colour term in one language does not necessarily correspond to a single term in another language. This is totally normal. It happens in modern languages too, and it has nothing to do with the physiology of the eye or the optic nerve. So a translation that works in one situation won’t work in others. Basically, for the purposes of scholarship, never translate colour terms.
2. Because of the reliance on dictionary definitions, 19th-20th century discussions often take an essentialist view, that there’s something real and objective about linguistic labels for colours. Historically, this view goes back to Isaac Newton’s canon of seven ‘primary’ colours — ROY G. BIV (red, orange, yellow, green, blue, indigo, violet).

No word for ‘blue’? Homer would have called Alexis Tsipras’ jacket kyaneos. The shirt is glaukos. The catch is, Tsipras’ hair is kyaneos too.

3. More specifically: yes, you can say ‘blue’ in ancient Greek. More precisely, Greek has words for the area of the colour palette that English calls ‘blue’. But English ‘blue’ covers a huge region of the palette. Greek splits it into multiple smaller regions: glaukos for lighter, non-vivid shades; kyaneos for darker non-vivid shades ranging to black; porphyreos for vivid shades ranging from blue to violet to ruby, but also for less vivid shades in the middle of that range (light magenta, pink); lampros for metallic-silvery-azure. Yes, ancient sources do mention sky colour: it’s glaukos or lampros. It’s just that Homer doesn’t mention the sky’s colour (and why would he). For example: Cornutus, Compendium 10.18-20, compares sky colour to olive-tree foliage, because both are glaukos: glaukos covers a much larger area of the palette than ‘azure’ does. But Homer does refer to kyaneos clouds, and glaukos eyes and sea.
4. It’s often claimed that ancient Greek colour distinctions are based mainly on brightness. Well, it’s true that Homer mentions ‘dark red, brown, black’ (melas) much more often than most other colours. And it’s true that brightness and saturation matter for some colour terms. But the same is true of English. ‘Pink’, ‘brown’, and ‘olive’ are low-saturation or low-brightness versions of red, orange, and yellow; then there’s ‘azure’, ‘navy’, ‘crimson’, ‘scarlet’, ‘lime’, ‘indigo’, and so on. Don’t take it for granted that there’s something methodologically different about how ancient Greek organises its colour terms, just because one individual (Aristotle) happened to like sorting by brightness.

I have covered colour terms in Homer previously: (1) The bronze sky, (2) The wine-dark sea, both from January 2016. I feel the need for an update: The second one had a bit too much speculation in its last section, and a topic that’s under such a constant barrage of misinformation and confusion (1, 2, 3, 4, 5, 6) is always worth revisiting.

Methodology

We can’t interview ancient Greek people, and we can’t conduct experiments to see where they would draw colour boundaries on a Munsell array. We have to rely on found evidence.

This does not mean looking at colour terms in translations of ancient texts, or even in a dictionary. See point 1 above. Translation is not one-to-one.

Dictionaries can help. But they aren’t evidence, they’re tools. Take for example the LSJ entry for kyaneos (1889 New York edition, with some modernisation):

properly, dark-blue, glossy-blue, of a serpent’s iridescent hues ... (Iliad 11.26, 38, Hesiodic Shield 167); of the swallow, Simonides 21; of the halcyon, Aristotle HA 9.14.1; of the skin of the porpoise, Aristotle HA 6.12.3; of the deep sea, Simonides 18, Euripides IT 7; ... 2. generally, dark, black, of the mourning veil of Thetis ...; of clouds ...; of the brows of Zeus ... of the hair of Hector ...

And so on. The entry gives the impression that there’s no single correct translation, and rightly so, and it lists a bunch of ancient sources as evidence (I’ve omitted most of them here). From there, we can do a lexical survey and look at the kinds of objects and surfaces that are called kyaneos; we can actually look at the sources, and expand beyond the ones that LSJ list; we can look at the context in the sources — metaphors, connotations, implied lighting conditions, and so on.

Even then, a dictionary is still just a tool, not an authority. Lexicographers may have their own biases about how colour works. Plato, Timaeus 68b-c, treats lampros as a colour, but if you look at the LSJ entry you’ll see no sign of that.

Yes, that’s partly because lampros usually denotes brightness. But it’s also partly because we don’t have a single word for ‘metallic silver with a blue tint’ in English. We really ought to, though, considering how often we see that colour. Like, for example, in the sky.

(Yes, I went there. An ancient Greek might well complain that modern English has no word for sky colour!)

A lampros car

Different languages, different boundaries

Translation is not one-to-one. Let’s illustrate: take ‘ground’ and ‘floor’. English-speakers are comfortable with the distinction between them. But in German there’s one word for both, Boden. If you’re translating German-to-English and you come across Boden, you cannot give the correct English equivalent unless you know the context: you have to know whether the Boden is indoors or outdoors.

Similarly with colour terms. German Lila and Purpur have no exact equivalents in English; English ‘crimson’ and ‘chartreuse’ have no exact equivalents in German. That doesn’t mean they can’t be translated! If you know the context, you can come up with a workaround. English ‘lilac’ and ‘violet’, are light shades of Lila: Lila is more general. Similarly Purpur covers English ‘magenta’, but it also includes royal purple and the red of a toadstool. Similar things apply to English ‘crimson’ and ‘chartreuse’. Hellgrün (‘light green’) is the dictionary rendering of ‘chartreuse’, but the English word is normally about halfway between ‘yellow’ and ‘green’.

What about ancient Greek? Well, think of your favourite image editor, and its colour picker:

Colour palette in the Windows programme Paint.NET, with HSV selectors in the red box

Outlined are the controls for three parameters which define the physical parameters of any colour uniquely. ‘Hue’ for the part of the spectrum the colour falls on; ‘saturation’ for the range from grey to vivid; and ‘value’ for lightness-darkness. (When Albert Munsell devised this system in the early 1900s he used ‘chroma’, not ‘saturation’.) In the colour circle at the left, the direction from the centre represents hue, and distance from the centre is saturation. It leaves out value: that would be a third dimension, extending from white to black.

Each colour term refers to a region of the palette. But the boundary of that region is arbitrary — to an extent, at least. There are non-linguistic constraints: the typical human eye has receptors for three colours, and that biases our colour perception ability; evolved cognitive traits may well give a higher priority to red, as a critical colour. But outside those biases, different languages may well assign colour terms to regions of different shapes and with different boundaries.

Most English speakers would be comfortable using ‘blue’ to refer to all of the top left quarter of the circle. But we wouldn’t feel nearly as comfortable grouping all of the bottom right quarter under a single term.

In ancient Greek, by contrast, colour term usage suggests that porphyreos could cover at least a full quarter of the circle, not just ‘purple’. If you take non-vivid hues on the left and lower-left of the circle, and extend it to darker values as well, that’s glaukos. The top left, in a narrower band than English ‘blue’, is kyaneos, again weighted towards darker values. All of the bottom part of the circle would be chlōros.

Some terms do have tidy equivalents in English: erythros is ‘red’, leukos is ‘white’. But others aren’t nearly as easy. For them, to translate them correctly, you have to know the context.

A selection of ancient Greek colour terms, plotted on the colour palette based on lexical usage and a healthy dose of guesstimation. Note that this wheel is only one slice of the available range: the full palette would have a third dimension extending from white (maximum value) to black. Kyaneos extends to lower values all the way down to black (it gets used of hair and Ethiopian skin), and glaukos is also frequently a bit darker than this (the colour of olives or vine leaves). At their darkest shades, kyaneos and melas represent black approached from opposite sides of the wheel: in some passages of ancient texts they appear as synonyms. Some caveats: I’ve left out plenty of terms here (prasinos, ōchros, etc.). Plus, this is at best an approximation, at worst guesstimation, so allow plenty of room for corrections — not to mention disagreement between ancient sources.

There are other parameters too. Hue, saturation, and value represent only the physical characteristics of coloured light. English terms like ‘navy’ and ‘pastel’, and qualifiers like ‘vivid’ and ‘violent’, carry connotations of a colour being vivid or washed out in comparison with its context. Maria Michela Sassi, a scholar of ancient philosophy, identifies three other parameters as significant in Greek colour terms (2017):

• Saliency — related to how we as humans are programmed to perceive colours. For example, if we are hardwired to detect redness as a matter of urgency, then red will be much more universal than other colours.
• Colour event — the subjective experience of colour, including the context in which it is seen (relative vividness, lighting, etc.) and its cultural meaning.
• Glitter effect and material — scattering and textural effects resulting from the type of surface being observed. She cites porphyreos as a key example, in reference to things like the shimmering of pigeon neck-feathers. I’d suggest aithōps as another.

Sassi is absolutely correct that these are all important. Lampros, for example, has a specular quality that can’t be conveyed by a single point on the Munsell spectrum. There are languages where parameters like these are even more important. But today, I think, we can afford to take a simplified approach: we can still convey the problems with how ancient colour terms are represented, while sticking to Munsell’s parameters.

The main point to hammer home (and the Munsell spectrum is enough to make that point) is that English colour terms are neither more nor less arbitrary than ancient Greek. There’s no reason to treat ‘blue’ as an objectively defined region on the palette, any more than there is with Greek glaukos.

Gladstone, Newton, and others

William Gladstone, the 19th century British politician and Prime Minister, is often credited as the source of the idea that Homer had no word for ‘blue’. Sometimes he’s even said to have claimed that ancient Greek as a whole didn’t have ‘blue’. As we’ve seen, that’s false. But Gladstone isn’t totally guilty of creating the myth. (I’ve also seen it credited to Goethe: he’s completely innocent.)

Gladstone does make a starkly racist declaration that ancient colour systems are ‘less mature’ than contemporary English. He refers to the ‘paucity of [Homer’s] colours’ (1858: 457-458), at the same time as producing long lists of them. And he points out (correctly) that Homer never applies a colour term to the sky (483). But he never says, ‘There is no word for blue.’ (In one place he writes of three English colour terms that don’t have exact counterparts in Homer, and mistakenly writes ‘violet’ for ‘blue’: 459, line 6 from bottom. Even if he had written ‘blue’, as he obviously meant to, he’d still be wrong.)

Be that as it may, Gladstone’s assumptions are terrible. He’s an essentialist through and through. He assumes in advance that there are seven ‘primary’ colours — the seven of Newton’s canon — and that there’s something universal about them. He lists eight colour terms that appear in Homer, then carries on (1858: 459):

Now we must at once be struck with the poverty of the list which has just been given, upon comparing it with our own list of primary colours, which has been determined for us by Nature, and which is as follows:

Red. Orange. Yellow.

Green. Blue. Indigo.

Violet.

He adds ‘white’ and ‘black’ to these, then asserts that four of the Greek ones are equivalent to four of the English ones. On the next page he grudgingly adds another 13 Greek terms, and proclaims that they ‘have very slight claims indeed to be treated as adjectives of definite colour’. He doesn’t give any reasoning, but it’s clear enough. Many of them he takes as synonyms for ‘glittering, shiny’ or ‘gloomy’; several are comparisons, like ‘rose-coloured’ or ‘marbled’; and two, chlōros and glaukos, absolutely are colour terms, but Gladstone excludes them simply because they don’t line up with Newton’s seven.

Newton’s colour wheel. Left: Newton 1704, fig. 11. Right: a corrected version that actually follows Newton’s specifications (indigo is supposed to be the only segment that’s much narrower than the others; 1704: 114).

Gladstone’s overconfidence in the objectivity of English terms comes partly from Isaac Newton’s materialist approach. Newton (1704) studies the splitting of white light into component colours, the relationship between different coloured light and different refractive properties. The quantifiable nature of refraction gives the impression that everything he says is objective. And for the bits about refraction, that’s fine. But when we start adding on linguistic boundaries, as if they’re as real as the refractive indices, then there are going to be problems.

Johann Wolfgang von Goethe (1810) criticised Newton too, but his critique wasn’t linguistic: it was more about the idea that refractive indices exhaust the nature of colour. We could say that Goethe’s understanding of colour was phenomenological: he preferred to understand colour in terms of qualia — irreduceable atoms of experientiality. Even today, qualia still pose problems for philosophers of mind. Personally, I think the more insidious problem is translation between languages. (Not that I subscribe to the Sapir-Whorf hypothesis — don’t go thinking that!)

Goethe is the most influential authority for the notion that ancient Greek colour terms are mainly about brightness. Gladstone certainly inherited that. So does Eleanor Irwin, in her study of colour terms in Greek poetry (1974). But the notion is much too reductionist. Early Greek philosophers are partly to blame too. Some of them tried to reduce everything in the cosmos to a single element, and hand in hand with that, thinkers like Theophrastus and Aristotle tried to reduce all colours to a simplistic dualism (Theophr. De sens. 59; Arist. De sens. 439a-440b; see Irwin 1974: 22-27).

But that kind of dualism is just what happens if you’re too reductionist. If you’re a 19th century scholar and you use ‘black, dark’ to translate all of kyaneos, melas, ioeis, and ēeroeidēs; ‘bright, shining’ for lampros, aithōn, aithōps, sigaloeis, charopos, argennos, and argos; and ‘grey’ for glaukos, phaios, and polios — well, don’t be surprised if you come away thinking there isn’t much variety in Greek colour terms.

Irwin’s study is an improvement, with an awareness of Munsell coordinates. She gives a review of scholarship from the 1700s to her own time. But she still succumbs to an awful lot of the older essentialism. On one level, she’s aware that Greek colour terms have a range of potential translations. But she still persists in pinning Greek words to a single English word. And, I’m afraid, she inherits a lot of Gladstone’s ethnocentrism.

The Homeric Greeks had not yet learned to think in abstract terms. ‘What is colour?’ is a question they would never have formulated, let alone been able to answer. (p. 22)

... ‘bright’ [λαμπρόν], not strictly a colour term at all ... (p. 25)

... if ξανθόν is ‘yellow’, then [Aristotle] lacked a particular term for orange. (p. 26)

I don’t know of any general treatments of Greek colour terminology in the last 40 years. The most recent good treatment, according to a 1982 review, is a 1977 Erlangen dissertation written by Helmut Dürbeck. Unfortunately it’s somewhat difficult to get hold of. I haven’t read it, and there are no copies in my country. We could do with a major update, published by a major publishing house.

Edit, several hours later: Professor Melissa Funke of the University of Winnipeg has very kindly alerted me to her book-chapter on the use of Greek colour terminology in 19th-20th century classical scholarship, Funke 2018. I haven’t got access to a copy yet, but I’m looking forward to reading it!

Metaphor

Irwin does at least show some willingness to allow that colour terms are sometimes metaphorical ... sometimes. ‘We find λειριόεις “lily-white” used of sound in Homer and Hesiod, and if we refuse to call it a “metaphor” ...’ (pp. 27-28). Why would we refuse to call it a metaphor, though? I’m guessing because Irwin was trained not to apply a modern concept to ancient poetry, on the grounds that that would be an anachronism. But just because ‘metaphor’ wasn’t widespread as a literary term, that doesn’t mean it didn’t exist — any more than ‘blue’ didn’t exist. Nowadays, in the 21st century, it’s harder to imagine why anyone would ever refuse to admit the possibility of metaphor in ancient poetry.

Some of the most troubling uses of colour terms in Greek — troubling to those who conclude that the ancients must have been physiologically different, or something like that — can easily be explained as metaphor. Take the ‘green’ blood in Euripides, Hekabe 126-127:

γνώμῃ δὲ μιᾷ συνεχωρείτην
τὸν Ἀχίλλειον τύμβον στεφανοῦν
αἵματι χλωρῷ

Unanimously you must concede
to adorn Achilles’ grave
with chlōros blood

Gladstone admits that this can’t be literally green blood, but ‘green’ in the metaphorical sense of ‘fresh, new’ — though, like Irwin, he too avoids the word ‘metaphor’. But Gladstone’s words betray a bias. He doesn’t credit Euripides for an ingenious oxymoron. Instead, he treats the line as an infelicity, and blames it on a deficiency in the ancient Greek colour sense (1858: 492: ‘When the epithet [chlōros] could be thus used, colour could be only very carelessly and faintly expressed in [ancient Greek] minds’).

Similar things apply to Homer’s ‘bronze sky’ and Pindar’s ‘blue earth’. ‘Bronze’ was never a colour term. ‘Bronze sky, iron earth’ is standard 7th-century-BCE imagery, with connotations of being harsh and unyielding: the same image appears in Assyrian and Hebrew texts of the same period (see my piece from 2016). And Pindar (Hymns fr. 33e.3-6) —

χθονὸς εὐρεί-
ας ἀκίνητον τέρας, ἄν τε βροτοί
Δᾶλον κικλῄσκουσιν, μάκαρες δ’ ἐν Ὀλύμπῳ
τηλέφαντον κυανέας χθονὸς ἄστρον.

(Delos,) the broad earth’s
immoveable wonder. To mortals, it’s called
Delos; to the blessed ones on Olympus,
‘the far-visible star of the kyaneos earth’.

On one level kyaneos is used here as a synonym for melas, in the familiar formula ‘black earth’. Apparently that’s enough to justify using kyaneos in a metaphorical sense. At the same time, we don’t know what the flavour of Pindar’s metaphor is: maybe it’s related to the idea that Delos is born from the dim-blue sea, maybe it’s something to do with religion, we just don’t know. Plenty of metaphors are lost on us now. Homer’s ‘wine-looking sea’ is one of them. There are plenty of theories as to what metaphors like that mean, but often there’s no clear winner.

The Sapir-Whorf hypothesis

Let’s end with a mention of Sapir-Whorf. The Sapir-Whorf hypothesis is that linguistic categories have an effect on cognition.

In connection with colours, the idea would be that if the ancient Greeks didn’t have a word for ‘blue’ — which, as we’ve seen, isn’t true in any sensible way — then that would mean they weren’t even able to conceive of the colour blue. In some popular accounts, this might even mean that they were unable to perceive the colour blue!

This is, of course, bollocks. It would make about as much sense if someone were to say: Greek has a word, glaukos, which denotes the colour of a clear sky and vine leaves, and modern English doesn’t, so that must mean that English-speakers can’t even perceive the colour of the sky or of vine leaves!

Strong Sapir-Whorf is nonsense, and all cognitive scientists know that.

Much, much weaker forms of the hypothesis are still being explored, though. For example, one recent study of the effects of colour terminology on speakers of Mandarin and Mongolian (He et al. 2019) suggests that, while different linguistic boundaries between colour terms have no noticeable effect on people’s ability to recognise and categorise colours, they do have an effect on the speed at which people sort colours. And, moreover, the study finds that this effect is linked to verbal working memory: that supports the idea that language is involved in some parts of cognitive processing.

But that doesn’t mean that ‘the way you see colour depends on what language you speak’, as a 2018 article in The Conversation put it. That title was so misleading that the authors had to step into the comments and try to explain what they meant. But their explanations didn’t exactly clear things up —

This does not mean that we can´t physically perceive the full gamut of colours, but that we do perceive them differently depending on the words that we hold to describe them.

‘Perceive them differently’ is very, very vague. It doesn’t need to be as vague as that. Language has an effect on cognitive processing of colour: that’s clear, and it isn’t difficult to explain. But ‘we perceive them differently’ is a gigantic overstatement. It implies that there’s something incommunicable about colour terms in different languages. It begs the question, in other words. It takes qualia for granted. And it does so before even starting to explore whether we really ought to be talking about ineffable qualia. It’s much more precise to say only what is meant: that we’re talking about how fast people can sort colours, and how this is affected by language.

References

• Dürbeck, H. 1977. Zur Charakteristik der griechischen Farbenbezeichnungen. Habelts Dissertationsdrücke, kl. Phil. 27 (Bonn).
• Funke, M. 2018. ‘Colourblind: the use of Greek colour terminology in cultural linguistics in the late nineteenth and early twentieth centuries.’ In: Varto, E. (ed.) Brill’s companion to classics and early anthropology. Brill. 255-276.
• Gladstone, W. E. 1858. ‘Homer’s perceptions and use of colour.’ In: Studies on Homer and the Homeric age, vol. 3 of 3. The University Press (Oxford). 457-499.
• Goethe, J. W. von 1810. ‘Erste Abtheilung. Griechen.’ In: Zur Farbenlehre, vol. 2 of 2. J. G. Cotta’schen Buchhandlung (Tübingen). 1-59. (= 1879. Goethe’s Werke, vol. 36. Gustav Hempel (Berlin). 10-47; = html text version).
• He, H., et al. 2019. ‘Language and color perception: evidence from Mongolian and Chinese speakers.’ Frontiers in psychology 14 Mar. 2019, 10:551.
• Irwin, E. 1974. Colour terms in Greek poetry. Hakkert (Toronto).
• Newton, I. 1704. Opticks: or, a treatise of the reflexions, refractions, inflexions and colours of light. Sam. Smith and Benj. Walford (London). (Archive.org copy)
• Sassi, M. M. 2017. ‘The sea was never blue.’ Aeon.co.