One of our "founding members", Louis Loynes, ran the Byraz Colour Bureau in Monmouth Street, London.

He created a system of numbering colours using dozenals and published a manual about the system.

I've started to scan it for publication here on the web and you can find the first (test) pages at
BYRAZ.

The BRYAZ bipyramid seems to resemble an HSV color cylinder, with theta=hue, r=saturation, and z=brightness. Unimpressive today, but it must have seemed more revolutionary back in 1959.

P.S. You don't have the only copy of the book. It's for sale at amazon.co.jp.

.jp?
Japan? How did a copy get out there? !!

Dan is quite right Shaun as I found the same page but my PC crashed as I was replying. However, what seems odd is that the book they list appears to be a paperback and it also has an ISBN. Now as far as I am aware the SBN system was developed in the UK in 1965 and the ISBN system was standardized in 1972. Since you say this book was published in 1959, this predates ISBNs and therefore if a version exists with an ISBN it must be a re-print.

The rest of the data listed is,

Byraz Colour Coordinating
Louis Loynes
095016271X

The page in question is
Amazon Japan

There's no ISBN on my copy. It's dated 1959 and I got it from Louis himself soon after it appeared.
There's the number 535.6 above the address on the first page but this will be the Dewey-decimal Libraray reference.
Can't read anything on the Jap Amazon page.

Cover is green and rather grubby after all these years ...

Louis was also very keen on heraldry; he had a huge card index.
Try "Louis Loynes" in Google for other leads.

Probably so: 535 is the Dewey Decimal code for "light & paraphotic phenomena", which presumably would include color.

Well, his approach is completely arbitrary from the very beginning. His choice of five "prime colours" (the six elementary percepts except, arbitrarily, green) is not any more "natural" than the seven arbitrarily chosen colours of Munsell (who chose the six elementary percepts plus, arbitrarily, one of the four binary hues). Either you choose:

1) A colour space based on retinal stimuli (e.g., RGB, CYMK and the like), with three so-called "primary colours" (actually, "primary light frequencies"; probably the most serious and unfortunately widespread misunderstanding regarding colour perception is the equation of light frequencies with color perceptions). To which some more may be added in order to extend the range of representable colours (compensating for the fact that the absorption curves of the cones overlap); e.g., as done in Hexachrome. This type of colour space is useful in order to deal with the reproduction of colours by means of pigment mixes or on TV screens, but other than that it is quite unintuitive to use (you need a table to look up the appropriate colour code, which in most instances is not directly inferrable from its appearance, nor viceversa; e.g., white is perceived as a completely pure and achromatic colour, not at all as a mixture of the chromatic hues red, green and blue as the RGB notation would seem to imply: if such a combination of hues were possible —which is not, under normal viewing conditions: red and green do not mix, as neither yellow and blue—, it would look as some extremely strange colour that one might describe as "greenish magenta", nothing even remotely similar to what white actually looks like).

2) Or you choose one based on the perceptual elementaries: white, black, red, yellow, green and blue (the only colours whose perceptual appearance, i.e. what they "look like", cannot be described in terms of others; all other colours can be described in terms of those six: e.g., "orange" = "red + yellow", "purple" = "red + blue", "brown" = "red + yellow + black (+ white)", "pink" = "red + white (+ blue)", etc.). For example, as done in the NCS, which is very intuitive to use (you can easily estimate the colour code from the colour appearance, and viceversa), but is not as useful if your primary concern is the reproduction of colours on print or on screen.

He's just using the traditional idea that red, yellow, and blue are the primary colors and green is the mixture of yellow and blue. You might call it "outdated", but I wouldn't call it "completely arbitrary".

I'll carry on scanning the book in tol put the rest on site.
There's also a section on Newton v. Goethe colour theory.

I think it would be helpful if the book weren't in black and white.

I regret only the cover was in colour; because it was printed on coloured paper.
I think Louis said there was a full-colour copy in the British Museum Library.

Just to liven up the forum, I'll comment on a colour ordering system I am currently working in that has many similarities to this one. Although, in fact, I have noticed this similarity now when I've had another more careful look at BYRAZ. Mine was not inspired by it, actually; I worked it out independently from my own long-cherished idea of representing colours as simple proportions between elementary percepts (particularly, as dozenal-compatible proportions), and just happened to arrive at a very similar system (though clearly distinct in some important aspects). I hadn't paid any attention to BYRAZ when this thread appeared here, because I saw it was based on three "prime" hues, instead of the four unique ones, and to be sincere, the last thing I was (and still am) interested in is yet another colour system that, like RGB, CYMK, Munsell, etc., etc., is not based on the six elementary colour percepts.

This system of mine (I doubt between calling it "WKYBRG" or "RYGBWK") is not based on the painter's mixing primaries (which are arbitrary), nor on retinal stimuli (which are completely unintuitive to use), but on the elementary colour sensations of human trichromatic vision as described by Hering's opponency (W white, K black, Y yellow, B blue, R red, G green, and these are not an arbitrary choice, but the elementary components of how our brains "think" colour; the six completely pure, unique colours). So, in a way, it is intermediate between the BYRAZ and the NCS systems. The colour space of this system can be represented as an octahedron (rather than the double triangular pyramid of BYRAZ, or the double cone of NCS). I'm still working on pictures to illustrate it, but for the moment I'll describe some of the niceties of this system (which are not featured in BYRAZ nor in NCS -- nor in Munsell, OSA-UCS, Coloroid, RAL, Pantone, CIELab, RGB, CYMK or Hexachrome, for that matter):

- The colour perceptions of normal trichromatic human vision can be classified into the following (perceptual, not linguistic) categories:

- Achromatic type 1 (pure): white, black
- Achromatic type 2 (mixed): grey
- Saturated type 1 (unary hue): yellow, blue, red, green
- Saturated type 2 (binary hue): orange, yellowgreen, bluegreen, purple
- Insaturated type 1 (unary hue + pure achromatic): light or dark + yellow, blue, red, green
- Insaturated type 2 (unary hue + mixed achromatic): greyish + yellow, blue, red, green
- Insaturated type 3 (binary hue + pure achromatic): light or dark + orange, yellowgreen, bluegreen, purple
- Insaturated type 4 (binary hue + mixed achromatic): greyish + orange, yellowgreen, bluegreen, purple

For example, most browns (which are dark or greyish oranges) belong to the insaturated type 3 or 4 categories. Furthermore, these categories can be classified according to their "purity":

- 1 elementary percept (purest colours): categories "pure achromatic" and "unary hue"
- 2 elementary percepts: categories "mixed achromatic", "binary hue" and "insaturated type 1"
- 3 elementary percepts: categories "insaturated type 2" and "insaturated type 3"
- 4 elementary percepts (least pure colours): category "insaturated type 4"

This categorization is reflected in the geometrical organization of the octahedral colour space: If you divide the octahedron into four equal tetrahedra, each defined by four percepts (W, K, and a pair of non-opponent hues), the purest colours in each tetrahedron are located at its vertices, those composed of 2 percepts are located along the edges, the ones composed of 3 percepts are located on the faces, and the least pure colours (those composed of 4 percepts) are located in the space inside. That is, the more "prominent" its location, the purer the colour.

- Going further with the above categorization, one can describe 35 "elemental categories" of colours, according to their composition in terms of combinations of perceptual elementaries. These are:

Pure achromatic: W, K
Mixed achromatic: W+K
Saturated unary hue: Y, B, R, G
Saturated binary hue: Y+R, Y+G, B+R, B+G
Insaturated type 1: W+Y, W+B, W+R, W+G, K+Y, K+B, K+R, K+G
Insaturated type 2: W+K+Y, W+K+B, W+K+R, W+K+G
Insaturated type 3: W+Y+R, W+Y+G, W+B+R, W+B+G, K+Y+R, K+Y+G, K+B+R, K+B+G
Insaturated type 4: W+K+Y+R, W+K+Y+G, W+K+B+R, W+K+B+G

or

1 percept: W, K, Y, B, R, G
2 percepts: W+K, W+Y, W+B, W+R, W+G, K+Y, K+B, K+R, K+G, Y+R, Y+G, B+R, B+G
3 percepts: W+K+Y, W+K+B, W+K+R, W+K+G, W+Y+R, W+Y+G, W+B+R, W+B+G, K+Y+R, K+Y+G, K+B+R, K+B+G
4 percepts: W+K+Y+R, W+K+Y+G, W+K+B+R, W+K+B+G

Now, the colour system I'm proposing is not made to be dozenal by an a priori arbitrary decision; rather, its dozenality arises naturally from the fact that elementary percepts can combine in mixtures of 1, 2, 3 or 4 (of course, under normal viewing conditions, leaving aside the exceptional possibility of overriding opponency using the convoluted Crane & Piantanida technique). In this system colours are named in terms of proportions between its elementary percept components (this is very similar to the BYRAZ approach, and dissimilar to the NCS approach, because in NCS the amount of white in a colour is only referenced indirectly); for example, a medium grey can be notated as "w1k1" (the numerical coefficients meaning 1 part of white and 1 part of black, from a total of 1+1 = 2 parts). So a medium dark purple could be described as"k1b1r1" and a medium greyish purple as "w1k1b1r1", but the first notation references thirds (1+1+1 = 3) and the second references fourths (1+1+1+1 = 4), so they belong to different subdivision schemes of the colour space, while the notations "k4b4r4" and "w3k3b3r3" both reference twelfths: 4+4+4 = 3+3+3+3 = 12 (= *10), so these belong to the same subdivision scheme. This means that the dozenal subdivision of the octahedral colour space, which I will call the "WKYBRG-12 gamut" (by "gamut" here I mean the set of all colours belonging to a regular subdivision of the continous colour space, that is, the set of all colours that can be described with combinations of coefficients that sum up to a fixed number), is the minimum gamut that contains at least one representative for each of the above 35 "elemental categories". Thus, in other words, the system is "naturally" dozenal, although not necessarily dozenal (the same dark purple can be described in terms of thirds as "k1b1r1", in terms of sixths as "k2b2r2", in terms of ninths as "k3b3r3", in terms of twelfths as "k4b4r4", in terms of fifteenths as "k5b5r5", etc., although there is a subtlety regarding their choice which I'll explain later). So the users can choose whatever "gamut" fits them better for their purpose, not necessarily the WKYBRG-12 one, but this one would be the most "standard". The WKYBRG-12 gamut contains 1496 or *A48 colours, which are more than enough to choose from for everyday practical purposes.

- Another feature of this system is that it not only allows for the choice of different "gamuts", it also allows for the choice of different colour "spaces". The octahedral space is the one corresponding to trichromatic vision, but spaces corresponding to dichromatic and monochromatic vision (built around 4 and 2 elementary percepts, respectively) can be described easily as subsets of it. Moreover, theoretically it can also be expanded to notate colours in larger dimensions, by adding more pairs of elementary percepts. Seeing that dichromats add one pair of opponent hues to the colour space of monochromats, and that trichromats add another pair of opponent hues to the colour space of dichromats, one may hypothesize that the perceptual colour experience of tetrachromats adds another pair of opponent hues to the colour space of trichromats. Since these two novel unique hues are unknown to trichromats, they lack names in human languages, so for the sake of argument let's call them "septarine" and "octarine" (names inspired by "the colour of magic" and the fact that they would be the seventh and eighth elementary percepts after white, black, yellow, blue, red and green). I'll notate "septarine" as S and "octarine" as E (from "eighth", because O could be confused with "orange"). So this system not only encompasses trichromatic gamuts (among them, the "standard" WKYBRG-12), but also monochromatic gamuts (among them, the "standard" WK-2), dichromatic gamuts (among them, the "standard" WKYB-6), tetrachromatic gamuts (among them, the "standard" WKYBRGSE-60), etc. The WKYBRG-60 gamut would be the trichromatic subset of the "standard" WKYBRGSE-60 tetrachromatic gamut (the one containing at least one occurrence of each of the 107 tetrachromatic "elemental categories", among its over four and a half million tetrachromatic colours), and it adds fifths and tenths to the list of compatible subgamuts, so this would be a natural choice for an "extended" or "rich" gamut (the WKYBRG-60 gamut offers 151,341 or *73,6B9 colours). The WKYBRG-4 (or "square") and WKYBRG-6 (or "hexagonal") gamuts, if taken together also include colours from all 35 "elemental categories", although this union set does not form a "gamut" but instead a "palette" or selection from the WKYBRG-12 gamut. This "palette" of "square"+"hexagonal" colours offers a selection of 297 colours (85 square and 231 hexagonal, with 19 of them both square and hexagonal), allowing gradations of 2, 3, 4 and 6 steps between elementary percepts, and makes for a good "basic palette" for quick reference and selection.

Tomorrow I'll comment further on other aspects of this system, and possibly already offer some illustrative pictures. I'll address, for example, the question of the perceptual distance between colours, since this is not designed to be a perceptually uniform space the way Munsell, OSA-UCS or Coloroid are, but offers some advantages over these ones while retaining the possibility of finding colours that are perceptually equidistant. The advocates of those systems seem to think that perceptual equidistance, and particularly equal lightness, is the central parameter around which colour spaces must be organized, grossly disregarding the fact that, for example, pure colours like red and yellow are fundamentally different from mixed colours like orange. While, on the other hand, the fact that yellow is a lighter colour than red is an inherent feature of that elementary percept (yellow is lighter than red per se), so we actually gain little by simply placing yellow closer to white than red in the colour space: the knowledge that yellow is closer to white is inherent in the characterization of those percepts, so lightness needs not be a fundamental parameter to organize the colour space, as it can be derived from other parameters. If the lightness levels inherent in the pure hues are compared with lightness levels in the grey scale and this way they are assigned a relative inherent level of lightness, then if you know the composition of a colour in terms of its component elementary percepts, you can easily compute the lightness of that colour (e.g., a medium orange has a lightness level halfway between those inherent in yellow and red); while from a notation where colours are referenced primarily in terms of lightness, you cannot extract their elementary components or tell how pure or mixed the colour is by computing with that lightness parameter. That is, computations can be performed with colour notations based on elementary components so as to "deform" the octahedral colour space into a perceptually uniform one, while the reverse is at least difficult with colour notations based on a perceptually uniform space, which tell us little if anything about the composition of a colour in terms of elementary percepts.

I will admit to being totally lost on this! I hope someone else will be able to be able to comment on your ideas.

What is it that you don't understand? I think the notation is fairly easy to use, although I find it much harder to try to explain it in words than to simply show it through illustrative images.

Look at the following illustrations for the WKYB-1, WKYBRG-1, WKYB-2 and WKYBRG-2 gamuts:




(Note: the actual RGB colours shown in these illustrations are merely approximations to the ideal colours represented by the notations)

WKYB-1 and WKYBRG-1 are unary gamuts, that is, they contain only pure colours. While WKYB-2 and WKYBRG-2 are binary gamuts, that is, they contain colours that can be defined in terms of halves (e.g., one half of white plus one half of blue = "w1b1", two halves of red = "r2", etc.).

The leading 2 in the notations of the WKYB-1 and WKYB-2 gamuts indicates that the notation refers to a colour within a 2-dimensional colour space, i.e. a colour space with just white-to-black (WK) and yellow-vs-blue (YB) dimensions. Such is the kind of subjective colour space experienced by red-green colour blinds (you can check how the world looks like in "2D colour" by means of colourblind filters; e.g. here you can see the Google logo as it appears to red-green colourblinds).

The leading 3 in the notations of the WKYBRG-1 and WKYBRG-2 gamuts indicates a colour within a 3-dimensional colour space, i.e. a colour space with white-to-black, yellow-vs-blue and red-vs-green dimensions, which is the one of normal human colour vision.

As you can see, the same colour can have different notations; e.g. "2b1", "3b1", "2b2", "3b2", etc., all of them indicate a pure blue. The choice of notation depends on what colour space and what gamut you want to contextualize that colour in, which is relevant for example to define what are its "nearest" colours; e.g., the nearest colours to "2b1" (pure blue in the context of 2D colour space and unary combinations, that is, in the context of the WKYB-1 gamut) are "2w1" and "2k1" (i.e., pure white and pure black), while the nearest colours to "3b2" (pure blue but in the different context of 3D colour space and binary combinations, that is, in the context of the WYKBRG-2 gamut) are "3w1b1", "3k1b1", "3b1r1" and "3b1g1" (i.e., azure, navy, purple and turquoise). The choice of gamut also defines the tolerance; e.g., the "nearest" colours to "2b60" (pure blue in the 2D sexagesimal gamut) are "2w1b59" (an almost pure blue with a faint white tint) and "2k1b59" (an almost pure blue with a faint black tint), so even a tiny amount of white or black added would be enough for the resulting colour to correspond to a different notation in the WKYB-60 gamut, while the "nearest" colours to "2b1" (pure blue in the 2D binary gamut) are "2w1" and "2b1" (pure white and pure black), so it doesn't matter that much if the shade of blue you use for "2b1" is not the purest one available, because the WKYB-2 gamut doesn't include any notation other than "2b1" for a colour that is predominantly blue---so the notation "2b1" encompasses a broader range of acceptable blue shades than the notation "2b60", and the colours corresponding to the notations "2b60", "2w1b59" and "2k1b59" would all be valid as correspondences to the notation "2b1".

This other illustration shows one representative sample for each of the 35 "elementary categories" of colour components I talked about:



As I explained, the WKYBRG-12 gamut is the minumum gamut containing at least one colour shade from each of those 35 colour categories. These categories correspond to the geometric parts of the WKYR, WKYG, WKBR and WKBG tetrahedra that subdivide the colour octahedron:
- the six unique colours (w, k, y, b, r, g) are located at the vertices of the tetrahedra
- the shades of binary colours (wk, yr, yg, br, bg) are located along their edges
- the shades of ternary colours (wky, wkb, wkr, wkg, wyr, wyg, wbr, wbg, kyr, kyg, kbr, kbg) are located on their faces
- the shades of quaternary colours (wkyr, wkyg, wkbr, wkbg) are located in their internal spaces
As you can see, the more elementary components a colour is composed of, the less pure it looks. Also, if you take only the top third of the image, you get the 11 "elementary categories" of 2D colour space (w, k, y, b, wk, wy, wb, ky, kb, wky, wkb).

Any comments? Questions? Suggestions?