As far as I've found so far, they achieved 546.1 nm for green and 435.8 nm for blue wavelengths with the use of mercury vapaour lamps and filters either of gelatine on glass, or chemicals dissolved in water held in glass. The 700 nm red source is not readily found in secondary sources, so I'm looking at the papers describing the work. So far, I figure they relied on a tungsten lamp branded as Ediswan pointolite that was either incandescent or arc light, and a filter.
boof (OP) double-posted this 1 year ago, 1 hour later[^][v]#1,288,376
OK I think I see what's been going on. The coloar matching experiments can be done with any suitably different set of wavelengths to serve as the preferred red, green, and blue triple. The apparatus can use any sort of broad-spectrum lamp with suitable filters that allow specific measure and reporting of three such wavelengths. The wavelengths as defined in the previous post are for standardized reporting of proportion values -- mathematically, any results of any three specific red, green, and blue wavelengths can be translated with respect to those stated wavelengths. The standards for green and blue were chosen because any lab can generate them for reference with a mercury vapaor lamp. The red is less important as any wavelengths about 650 and longer aren't perceived as distinguishable, and so any light source and recommended red filter will do.
boof (OP) replied with this 1 year ago, 9 minutes later, 1 hour after the original post[^][v]#1,288,378
oh here's a good quoate:
"The point about them is, however, that they have for some time been
adopted as standards for colorimetric purposes by the National Physical Laboratory,
because of the practical convenience in standardising colorimeters without
the aid of spectroscopic apparatus."
another:
"It is
unfortunate that the mercury lamp does not possess any bright red line, as in that
case an ideal colorimeter could be designed using the lamp as the source. As it is,
with ordinary sources, .436 μ is rather far down the spectrum to give sufficient
intensity for its satisfactory use as a primary in a colorimeter and, further, .5465 μ
being somewhat towards the yellow-green makes the area of the colour field lying
outside the colour triangle rather bigger than could be desired. It is a difficult
point to decide which is the better compromise; personally I should like to see
standardised three primaries which could, if desired, be actually used as primaries
in a colorimeter, in which case no arithmetical transformations would be necessary."
boof (OP) replied with this 1 year ago, 5 days later, 5 days after the original post[^][v]#1,289,323
It should be noted that any so-called primary coloars can not, as often stated, generate all perceivable coloars. The model is mathematical: such primaries can only be said related to all others if it is permitted to set one of the three to a negative value. When using the coloar-matching apparatus, a negative value corresponds to achieving a match by adding one of the primaries to the test coloar so that the two mixed will match the two remaining primaries mixed.
boof (OP) replied with this 1 year ago, 30 minutes later, 5 days after the original post[^][v]#1,289,340
hmmm OK, some time ago, a mathematical standard way to describe the coloars that we can see was invented. They based it on the use of three wavelengths of light mixed at whatever levels are needed to match single test wavelengths. We can thus assign three numbers to any perceived coloar. The weird thing is, you can't really make all the coloars with any real perceivable primaries. The work-around is to instead add one of the three primaries to the test wavelength and those together can match the two remaining primaries. Mathematically, that means you can describe any single wavelength coloar as three numbers if you permit a negative value for one of the primaries. The model also works to describe any coloar regardless being from a single wavelength or not. Most non-single wavelength coloars do not need the negative value.
Now, people found the idea of negative numbers in the model to be weird, so they later transformed the model into an equivalent one that uses all positives. Geometrically, that's like shifting and skewing points on the graph to get out of the negative areas.
boof (OP) replied with this 1 year ago, 4 minutes later, 5 days after the original post[^][v]#1,289,344
@previous (D)
It can be called a fundamental RGB coloaur. Most people do not encounter it however, and instead use standard sRGB not to define all perceivable coloar, but rather all possible coloaur output for a given system as for a monitor. https://en.wikipedia.org/wiki/SRGB