Many years ago a very nice, very experienced dyer on the CanSpin List took pity on me, a new dyer with vague if any colour sense. Annette explained how to choose dyes for mixing and referred me to a book titled Blue and Yellow Don't Make Green by Micheal Wilcox. It was written for painters but the concepts apply to dye work as well. My MIL, who likes very specific Christmas and birthday lists, was happy to get me a copy and I learned a lot from it. In brief, it explains that primary colours are only pure in theory. In the real world blues aren't really pure, they either lean toward green or purple. The same goes for yellow, leaning to orange or green, and red, leaning to purple and orange. Most basic colour theory texts do talk about 2 sets of primaries: "Painter's" or "warm" primaries are red, blue and warm yellow; "Printer's" or "cool" primaries are cyan, magenta and cool yellow and go on to note that the painter's primaries make clearer oranges and the printer's primaries make clearer greens and neither makes a clear purple.
Wilcox suggests working with all 6 primaries as a base set and understanding what happens when they mix. To make the clearest, brightest secondary colours you need to mix the primaries that lean towards them. Painter blue and printer magenta make the clearest, brightest purples, mixing painter blue and painter red, or magenta and cyan make duller purples. In the same way, cyan and printer yellow make clearer greens and red and warm yellow make clearer oranges. Once you understand how the primaries interact it's much easier to get the mixed colour you are looking for.
Thus, I work with 6 basic dye stock colours from Ciba: cyan, magenta, tartrazine, eriosin red, erionyl yellow and royal blue. I can get a really wide range of secondary and tertiary colours using just these 6 dyes. It's really amazing what a difference changing one dye colour can make. As an example, I tried to duplicate China Rose - honestly, I do take notes, but sometimes I either don't believe them or I write them down wrong... (or both which was the case here).
Clockwise from the top: The original China Rose, made when I was trying to recreate "Antique", retry 1 made with Erionyl Red, Royal Blue and Tartrazine, retry 2 made with Erionyl Red, Royal Blue and Eriosin Yellow and the final (correct I think) retry made with Magenta, Royal Blue and Eriosin Yellow. The final retry is a little darker than the original but that's probably measurement error.
The takeaway (besides "never trust E's notes") is the incredible difference between the 3 batches. Changing one primary at a time makes a big difference.
Here is another example. This time, I didn't have notes on which primaries I used, I just made an assumption based on Wilcox's theory. I was trying to make this:
When I made it I was using dyes from ProChem - I had painter's blue, turquoise, warm yellow and cool yellow to choose from to make the green. Now I'm trying to match those colours using the Ciba dyes. I assumed that I had used turquoise and cool yellow. So I used cyan and tartrazine (because at that point I was still trying to figure out which yellow was which) and got this:
Hmm. Not quite what I expected. Brighter and, well, greener. Back to the dye table I used royal blue and tartrazine, as the royal blue should quiet the green a bit:
The picture is a rather blue shifted (battles with cameras and photo editing are another post entirely), but you can still see that it's a lot closer to the original. In person it's much closer to the original.
As a final example, I present yesterday's dye room adventure:
As I said, I haven't quite figured out which Ciba yellow is which. On top of that, the Ciba dyes are not "leveled" like the ProChem ones. That means that when you mix equal quantities of 2 1% stock solutions of the levelled ProChem dyes you get a secondary colour that is exactly halfway between the two primaries. If you make 1% solutions of the Ciba dyes and mix them 50/50 the resulting colour won't likely be halfway between them because they are not equal strength. In particular, tartrazine is very strong and eriosin yellow is very weak. When I got the Ciba dyes the supplier suggested mixing strengths that do seem to work most of the time. Most of the time... It doesn't help that e. yellow and e. red are the hardest dyes to get into solution and e. yellow really, really likes to separate.
I was aiming for this:
I started with e. yellow and e. red. which I think are the painter's primaries and should make the best oranges. I got this:
Uncooked it looked awfully red. The e. yellow just got eaten by the e. red. Recall that dyes are not WYSIWYG, so that didn't mean it wouldn't turn out, however, the steamer holds two batches, so right away I tried e. red and tartrazine:
Ah, that's it. It's darker probably because I used the medium DOS range and I think the original is the light DOS range.
1. E. Yellow is a pain to work with, it's weak and it doesn't stay mixed long enough to measure it accurately, but now I'm pretty sure it's a cool yellow. Theoretically the clearest greens should come from mixing it with cyan.
2. I do not always want the clearest, brightest secondary. I much prefer the blue green colourway made with the "wrong" primaries (tartrazine and royal blue).
2. Tartrazine is really strong and makes really nice oranges. It's on the warm side of yellow but only slightly and it has an acid cast which is why I thought it was the cool one at first. It's been making decent greens but I might get closer to my original colours if I used e. yellow instead.
3. Switching dye suppliers requires a certain amount of adjustment to one's recipes. Really, I count myself lucky I've been able to match things up as well as I have.
4. Wilcox is right, even primaries that appear very similar, as the two yellows do, can create very different mixed colours. It pays to have a range. It's probably worth exploring using more than one of each primary in a mix to get even more possibilities. Or not, as it is there are already 4 combinations for each secondary colour and 8 combinations for each tertiary colour using just one of each at a time. Should be enough to keep me occupied for quite some time yet.