B What Makes the Six Colors in Newton's Experiment Special?

[JB321]

I recently read an article written by Newton which outlined the process and results of his "crucial experiment".
From what I understand, Newton says that light can continue to be split until it reaches the basic colors and then it simply stops. What is it about these frequencies of light that makes them so special? Are there six exact frequencies that are the base of all the colors? Or are there ranges of frequencies that Newton simplified to one color? I asked my physics teacher and he responded with, "I don't know..."

Thanks in advance,
JB

 

[blue_leaf77]

Is that "6 colors" part mentioned by Newton?

 

[pixel]

Or are there ranges of frequencies that Newton simplified to one color?

That's the likely explanation. If he used a prism, then he saw bands of color each of which contain a range of wavelengths.

I should mention, however, that the color of each wavelength can be matched by weighting of three suitable light sources. This gives rise to the CIE standard observer color matching functions which are the basis of the CIE L*a*b* color system.

 

[Drakkith]

There are more than 6 colors. Far more. But the real idea to take away from this is that a perfect spectrum is continuous and is measured in either wavelength or frequency, not colors. You can take a slice of the spectrum and expand it until you reach the resolution limit of your system, the spectrum doesn't simply stop at some point.

Remember that color is subjective. My father is red-green colorblind, so his experience of color is very different from mine, and there are supposedly people with more than 3 types of cone cells in their eyes (the cells that allow you to see color), so their experience of color is also far different from mine.

On the other hand, the frequency/wavelength of light is not subjective. Everyone with the proper measurement device will agree on the frequency and wavelength of any sample of light.

 

[JB321]

So why is it that if you split a ray of light several times, you will end up with an irreducible color? Say you isolate the blue section of a spectrum emitted from a prism. Why does the blue frequency not separate into different frequencies of blue? (maybe it does, but Newton seemed to say that it didn't) Is is because the frequencies are so close together that it is difficult to isolate them?

 

[blue_leaf77]

So why is it that if you split a ray of light several times, you will end up with an irreducible color?

Because the solution of the temporal part of EM wave equation is of sinusoidal form $\sin \omega t$ (or equivalently $\cos \omega t$) which has a definite frequency/wavelength called harmonics. Any EM radiation of arbitrary temporal profile can be decomposed into sum of these harmonics. That's why when a light ray is dispersed by prism or diffraction grating you get light rays each with a single indivisible frequency.

 

[pixel]

Why does the blue frequency not separate into different frequencies of blue? (maybe it does, but Newton seemed to say that it didn't) Is is because the frequencies are so close together that it is difficult to isolate them?

Yes, any spectrometer will have a finite resolution.

 

[DaveC426913]

and there are supposedly people with more than 3 types of cone cells in their eyes (the cells that allow you to see color), so their experience of color is also far different from mine.

Tetrachromats have a fourth cone that is slightly different than the usual green ones we all have. They can detect finer gradients of green than the rest of us. They might see subtleties that we don't. But that's about it.

 

[jbriggs444]

Because the solution of the temporal part of EM wave equation is of sinusoidal form $\sin \omega t$ (or equivalently $\cos \omega t$) which has a definite frequency/wavelength called harmonics. Any EM radiation of arbitrary temporal profile can be decomposed into sum of these harmonics. That's why when a light ray is dispersed by prism or diffraction grating you get light rays each with a single indivisible frequency.

But, in fact, you do not.

 

[blue_leaf77]

But, in fact, you do not.

Do not what?

 

[jbriggs444]

Do not what?

You do not get a set of discrete indivisible rays.

 

[Khashishi]

There are infinite shades of blue.

 

[blue_leaf77]

You do not get a set of discrete indivisible rays.

Whether it is discrete or continuous after being dispersed depends on the spectrum of the light. Some vapour lamps have discrete spectrum, neglecting the effect of line broadening.

 

[Simon Peach]

There are infinite shades of blue.

How can you have infinite shades of one colour? The blue light, for instance, has only a given range of wavelengths 450 - 494 nm so that range is not infinite.

 

[pixel]

How can you have infinite shades of one colour? The blue light, for instance, has only a given range of wavelengths 450 - 494 nm so that range is not infinite.

450.001, 450.002, 450.003...or however fine you want to make it.

Are you only thinking in terms of integers?

 

[Simon Peach]

450.001, 450.002, 450.003...or however fine you want to make it.

Are you only thinking in terms of integers?

Yes I agree that the numbers could be infinite, but the colours can't be. But thus is only nit picking really

 

[Drakkith]

Yes I agree that the numbers could be infinite, but the colours can't be. But thus is only nit picking really

I think Khashishi was just making the point that you can divide the portion of the spectrum corresponding to blue light into an infinite number of pieces.

 

[pixel]

Yes I agree that the numbers could be infinite, but the colours can't be. But thus is only nit picking really

You are correct when talking about human perception of color as it takes a certain amount of color difference to be just noticeable. Instrumentally, we can resolve spectral differences finer than that.

 

[PeroK]

How can you have infinite shades of one colour? The blue light, for instance, has only a given range of wavelengths 450 - 494 nm so that range is not infinite.

How many possible frequencies are there in this range?

 

[DaveC426913]

How many possible frequencies are there in this range?

Just as there are infinite points between 0 and 1, so there are infinite frequencies between 450 and 494.

 

[sophiecentaur]

How many possible frequencies are there in this range?

The number of 'identifiably' different frequencies depends entirely on the bandwidth / signal to noise ratio, of your measurement system. The eye is not all that good as a measuring instrument. Only under stringent conditions do you actually need the 'millions of colours' that high quality colour displays use. Those 'millions' refer to the number of points in two dimensional CIE colour space and not just spectral component frequencies and implies something like 0.1% discrimination over the spectral range. As a bandwidth for measurement of frequency, that is pretty rubbishy. Frequency measurement can be done easily to 10 significant figures in some ranges of EM waves.
We specify colours in our language with much cruder accuracy. This link has a picture of the Cie Chromaticity chart, showing a very crude categorisation of areas with named colours. We can distinguish between large areas of colour (patches / contours are just detectable) when the colour space is divided into around a million smaller areas but we don't 'remember' colours with anything like that level of discrimination. (You can't take a new tie home from the shop and be sure that it will match the shirt that you left in your wardrobe but the difference will scream at you when you lay them side by side).

 

[Daanh]

And how about those fifty shades of grey, eh?

 

[Algr]

"Color" can have totally different and even conflicting meanings if you don't define what you are measuring.

If you mix light from a red laser and a green laser, and adjust them to the correct brightnesses, you could produce light that a human would see as identical to a yellow laser. But a prisim would reveal that this light had nothing in common with an actual yellow laser. The idea that red plus green makes yellow is pure biology, not physics.

 

[Auto-Didact]

I recently read an article written by Newton which outlined the process and results of his "crucial experiment".
From what I understand, Newton says that light can continue to be split until it reaches the basic colors and then it simply stops. What is it about these frequencies of light that makes them so special? Are there six exact frequencies that are the base of all the colors? Or are there ranges of frequencies that Newton simplified to one color? I asked my physics teacher and he responded with, "I don't know..."

Thanks in advance,
JB

From what I remember, Newton posited 7 not 6 colors. He chose 7 mostly for numerological reasons, i.e. purely because 7 is a holy number in Christianity.

 

[swampwiz]

The first thing that must be understood is that "color" is a psycho-physical phenomenon, and its *perception* by humans is dependent on 2 systems: the electromagnetic wave/photon (i.e., light) receptors of the retina, and the processing of the output from those receptors by the brain. Normally, humans have these receptors (called "cones" by physiologists) that are tuned to have a maximum transfer function (i.e., how much output gets generated per input) at 3 specific wavelengths of light, that are recognized as being "red", "green" & "blue". Photons of light have a specific wavelength associated with them (i.e., in whatever relativistic reference frame an observer is in), and each photon of a specific frequency generates a specific output for each of these 3 tunings of receptors, and thus, there is a 3-D vector space for the receptors' output. The brain processes this triple coordinate as being a total "color". If the coordinates all the have same value, the brain processes it as "white" of some type (including "grey"); if the coordinates are very high in only one of those coordinates, the brain processes it as the respective color (including a darker shade).

Unless one is looking at the output from a laser device, the light that any set of receptors observe is in the form of many individual wavelengths that can be approximated as a continuous spectrum. This spectrum is in essence the "true color" of any light, with the cones & brain mapping that out to some "perceived color". To a certain level of accuracy, there are infinitely many combinations of individual wavelengths, each at some intensity, that is perceived by any particular brain as the same exact color; this is the reason why television works - a real spectrum is observed by an camera, which then produces a 3-D vector space signal that can be displayed by a display device, which a viewer would perceive in 3-D vector space as being the same as what would be perceived if viewing that original spectrum. As one might expect, this 3-D signal is best matched to human color perception by matching up with the wavelengths that correspond to the cones' maximum transfer functions, and is the reason why color is regarded as being a RGB (red-green-blue) coordinate.

There is a class of spectra called pure color (maybe it is called something else) in which the spectrum is only a single wavelength; such spectra is perceived by the brain as having a certain 3-D color coordinate value, and typically a display cannot reproduce such a spectra (except for those spectra that exactly match the output spectra of the individual display). However, these pure colors basically map to a 1-D curve within the 3-D vector space (or 2-D if intensity is normalized); there are plenty of other colors that the human color perception perceives, but these are artificial colors. A lot of these colors are close to a true color, but there is one particular section of color that is totally artificial - the colors from purple to red, which is due to the fact that the mixture of pure colors must be between these 2 colors, but yet not be along the pure color spectrum. A prism separates out light because the index of refraction is slightly different for different wavelengths; the glorious natural phenomenon of a rainbow has the same mechanism, although the particulars of geometric optics that makes it so is quite an interesting topic in its own right.

Now, as for the OP's original question of there being "6 colors", typically being in order along the pure color spectrum as violet, blue, green, yellow, orange, red, that is just perception as the brain has a hard time picking out any more of a fine gradation of colors. Now this might be my opinion, but when I look at a rainbow, I tend to notice a very large section between blue & green, that is typically know as cyan, so I would say that there really are 7 colors, not 6.

Hope this helps.

 

[sophiecentaur]

when I look at a rainbow, I tend to notice a very large section between blue & green, that is typically know as cyan, so I would say that there really are 7 colors, not 6.

Actually, a rainbow is not a good source for observing spectral colours because the background sky desaturates them grossly, even for the most striking rainbow. If you look at a well spread out spectrum, you will be very aware of the gradation from one wavelength colour to the next and that there are distance and identifiable colours between any two of the 'well known' ones. It's just that we don't bother to use any more than seven colours to describe the spectrum and it isn't worth making a finer scale. That is, unless you are interested getting colorimetric fidelity and good matching of copied colours. In which case you find hundreds of different and distinguishable by A-B comparison (but not memorable) colours along the spectrum.

 

[TheLegendOfCars101]

Just as there are infinite points between 0 and 1, so there are infinite frequencies between 450 and 494.

But are do the difference between the infinitesimal frequencies make a substantial difference ?

 

[Algr]

In video and digital art, an alternative to the Red, Green, Blue color space is HSB, or Hue, Saturation, Brightness. For both HSB and RGB, digital systems today are assumed to need at least 8 bits per channel (24 bits total) to produce photographic quality images.

So, ignoring saturation and brightness, those systems devote 8 bits to defining just the hue. That means 256 colors circling the rainbow from red, through yellow, green, blue, and back to red. (Excluding white, black, pink, brown ext. ).

Early digital systems with less then 8 bits per channel had to inject noise into the signal or other tricks to avoid obvious banding or problems due to insufficient numbers of colors.

Thus we can reasonably presume that typical humans can perceive around 200-300 colors, assuming high brightness and saturation, and not counting such variations as colors.

 

[davenn]

But are do the difference between the infinitesimal frequencies make a substantial difference ?

that's irrelevant ... the point is there IS a difference

 

[FactChecker]

But are do the difference between the infinitesimal frequencies make a substantial difference ?

"substantial" is a poorly defined term. It depends on what you are trying to detect and use color for. The eye may be very limited, but a prism could be used to detect very small frequency differences.

 

[Lish Lash]

"So why is it that if you split a ray of light several times, you will end up with an irreducible color? Say you isolate the blue section of a spectrum emitted from a prism. Why does the blue frequency not separate into different frequencies of blue?"

The three types of cones in the human eye are bandpass filters that respond to broad, overlapping ranges of colors. The "green cone", for example, doesn't just detect green wavelengths, it responds to a wide range of colors from orange to yellow to green to aqua. The "red cone" also responds to yellow and orange, as well as red wavelengths, but in different proportions than the green cone. When both red and green cones respond with equal intensity (and there is little blue cone response) the brain interprets this combination as yellow.

The important point to understand is that an individual cone cannot distinguish among the different colors it responds to. For example, a green cones cannot tell the difference between blue-green and yellow-orange colors, and responds in exactly the same way to both. Likewise, a blue cone would respond in exactly the same way to two shades of blue that were close to the same wavelength. You would only be able to distinguish between them if one of the blue colors stimulated the green cone more than the other.

 

[sophiecentaur]

"So why is it that if you split a ray of light several times, you will end up with an irreducible color? Say you isolate the blue section of a spectrum emitted from a prism. Why does the blue frequency not separate into different frequencies of blue?"

The three types of cones in the human eye are bandpass filters that respond to broad, overlapping ranges of colors. The "green cone", for example, doesn't just detect green wavelengths, it responds to a wide range of colors from orange to yellow to green to aqua. The "red cone" also responds to yellow and orange, as well as red wavelengths, but in different proportions than the green cone. When both red and green cones respond with equal intensity (and there is little blue cone response) the brain interprets this combination as yellow.

The important point to understand is that an individual cone cannot distinguish among the different colors it responds to. For example, a green cones cannot tell the difference between blue-green and yellow-orange colors, and responds in exactly the same way to both. Likewise, a blue cone would respond in exactly the same way to two shades of blue that were close to the same wavelength. You would only be able to distinguish between them if one of the blue colors stimulated the green cone more than the other.

You have to draw a huge distinction between colour vision and spectroscopy. Human colour vision is very approximate and uses just three wide and sensors to analyse and classify approximately, the whole of the visible spectrum of light entering the eye. It treats single spectral lines and light with complicated spectra in exactly the same way. The brain uses just three signals to categorise all the light from an object. That sensation is what we call colour. A spectrometer, otoh, splits the light into regions of FINITE bandwidth. It's the same as a radio receiver or RF spectrum analyser. The notion of a 'single' frequency is a bit nonsensical in this context. It's just a bit of convenient Maths. Every measurement method has a Resolution Bandwidth that blurs the resulting analysis. Zero bandwidth means no energy admitted, meaning you couldn't measure it. The resolution of a "prism" is limited by the width of the slit. Too thin a slit would mean not enough light gets through. Catch 22.

 

[Simon Peach]

The initial question was about colours. As I said before there cannot be infinite colours, the range of the visible electromagnet spectrum is "Visible light is usually defined as having wavelengths in the range of 400–700 nanometres (nm), or 4.00 × 10−7 to 7.00 × 10−7 m, between the infrared (with longer wavelengths) and the ultraviolet (with shorter wavelengths)". So unless you're going to split a nanometre there cannot be infinite range of colours.

 

[jbriggs444]

The initial question was about colours. As I said before there cannot be infinite colours, the range of the visible electromagnet spectrum is "Visible light is usually defined as having wavelengths in the range of 400–700 nanometres (nm), or 4.00 × 10−7 to 7.00 × 10−7 m, between the infrared (with longer wavelengths) and the ultraviolet (with shorter wavelengths)". So unless you're going to split a nanometre there cannot be infinite range of colours.

What's the problem with splitting a nanometre?

 

[Drakkith]

So unless you're going to split a nanometre there cannot be infinite range of colours.

You can easily split a spectrum up into sub-nanometer wavelength. This is routinely done in spectroscopy.

 

[OmCheeto]

Why are there only 6 colors?
I recently read an article written by Newton which outlined the process and results of his "crucial experiment".
From what I understand, Newton says that light can continue to be split until it reaches the basic colors and then it simply stops. What is it about these frequencies of light that makes them so special? Are there six exact frequencies that are the base of all the colors? Or are there ranges of frequencies that Newton simplified to one color? I asked my physics teacher and he responded with, "I don't know..."

Thanks in advance,
JB

Being a brilliant mathematician, Newton probably figured out pretty quickly, that to name an infinite number of colors would take him an extraordinarily excessive amount of time out of his finite time here one earth, which would interfere with his further experiments.

I'm sure Newton, were he alive, would tell you; "There are more than 6 colors."

 

[FactChecker]

So unless you're going to split a nanometre there cannot be infinite range of colours.

But of course there can be infinitely many colors in a finite range. (In fact uncountably many colors.) There is no conceptual problem with splitting a nanometer. Any two frequencies with the slightest difference represent different colors.

 

[pixel]

So unless you're going to split a nanometre there cannot be infinite range of colours.

The D-lines in the spectrum of sodium are at 588.9950 and 589.5924 nanometers. These are two instrumentally resolvable lines. Spectral lines don't just fall on integer values of nanometers.

You could say these represent two different colors. But as others have alluded to, if we mean by "color" the human visual response to light, then these are not visually distinguishable.

 

[rootone]

The human visual sensors (light sensitive neurons in the retina), come in three distinct types.
These are distinguished by response to what we see as red green and blue wavelengths.
However the response is not to a precise wavelength, just are most sensitive to wavelengths in a general range.
That gives six possible combinations that our visual cortex can process, rg, rb, gr, gb, bg, br, to associate a colour to an object.
We also have sensors which don't particularly recognise wavelengths, but just total amount of light.

 

[jbriggs444]

The human visual sensors (light sensitive neurons in the retina), come in three distinct types.
These are distinguished by response to what we see as red green and blue wavelengths.
However the response is not to a precise wavelength, just most generally they detect wavelengths in a general range.
That gives six possible combinations that out visual cortex can process, rg, rb, gr, gb, bg, br, to associate a colour to an object.
We also have sensors which don't particularly recognise wavelengths, but just total amount of light.

Even if our cones provided only binary color recognition, two raised to the third power is eight, not six. In addition, I fail to see the distinction between gr and rg.

 

[rootone]

Even if our cones provided only binary color recognition, two raised to the third power is eight, not six. In addition, I fail to see the distinction between gr and rg.

That is mathematically correct of course, but I'm not going to argue with nature itself, or evolution.
My personal experience is that I see six identifiable colours in a rainbow.
Not seven, I can't tell the difference between violet and indigo using that scheme

 

[sophiecentaur]

That is mathematically correct of course, but I'm not going to argue with nature itself, or evolution.
My personal experience is that I see six identifiable colours in a rainbow.
Not seven, I can't tell the difference between violet and indigo using that scheme

That is a totally personal, subjective assessment of what you see. Under other conditions, you (despite what you claim) would almost certainly be able to distinguish between two sources of spectral colour, both of which would fall within the 'yellow' band. If what you say is true, why would they need 'millions of colours' for good quality digital colour TV?

 

[pixel]

That gives six possible combinations that our visual cortex can process, rg, rb, gr, gb, bg, br, to associate a colour to an object.

Not sure what you are getting at here. That there are only 6 colors?

 

[Algr]

Here we go:

I originally made this to taunt people on an artist forum who didn't have a proper understanding of color theory. But I think it's a perfectly valid approach to examining the psychological perception of colors. I honestly do see all these colors as nearly equidistant. It's hard to explain that given a trinary understanding of color. My understand is that language, and how you were taught color names as a child, can effect your ability to perceive colors. So we can blame the evil Crayola corperation for all this 6 color talk. ;)

But as for this talk about not being able to split colors further - who actually performed this experiment? Ref please? It seems to me that if you kept trying to expand a spectrum further, you would simply end up with something too dim to see the subtle changes of color.

 

[sophiecentaur]

The term 'primaries' has a specific meaning. It is the least number of colours that can be mixed to produce all the others. So, with Additive Mixing (Using Lights, as in TV), you only need RG and B. There is no need for Yellow.
When mixing paints and pigments, the three primaries are Yellow, Cyan and Magenta. Other pigments are often used, to allow for brighter or more saturated colour display.
I suggest you read up about the existing theories on colorimetry. They are well based and have stood the test of time. Do that before trying to come up with an alternative theory of your own.

 

[sophiecentaur]

I honestly do see all these colors as nearly equidistant

I'm not sure how you are concluding that the colours are 'equidistant'. The CIE Colour space diagram is the standard way of plotting the chrominance values of colours and the distance between to 'just perceptibly' different colours depends on where they are in colour space. The 'Macadam Diagram' in this link (In the Tolerance section) shows how our resolution between adjacent colours varies over colour space. Those Ellipses (scaled by a factor of 10, to make it more obvious) show that, in the greens area, we are much less fussy than in the reds and pinks area. We are particularly sensitive to differences in skin tones as we use that information to judge emotions. I guess that there are few examples of highly saturated green colours in nature so we do not need to be able to distinguish there. Leaves of different shades of green will all have desaturated colours, where our colour vision is more discriminating. The details must all be down to evolutionary advantages and cost.

if you kept trying to expand a spectrum further, you would simply end up with something too dim to see

Yes. That is absolutely correct. The narrower the bandwidth, the less energy is admitted (that applies to RF signals as well) and the resulting signal can become too low to 'measure' or detect. This is 'why' our light sensitive cells use very wide band analysis filtering. If you use a spectrometer to look at fine details of the spectrum of light from stars, you will be limited by the actual amount of light power that arrives in a very narrow range of wavelengths that fall on your detector (or film).

 

[Algr]

The chart is my response to the "six colors" in the title of this thread. Take a look at the triangle at the bottom of the chart. Compare it to the square above and to the left. These are easily more different than red is from orange, and yet most people would have difficulty assigning names to them. They are both "Blue". We see one as darker than than other because our eyes aren't very sensitive to blue light. I'm well aware that yellow isn't really a primary (for additive color) but it seems to have a psychological identity so unlike its components. There is good reason why traffic lights are red-yellow-green, and not red-green-blue.

Apparently Isaac Newton would have called the primary color in the square "indigo" and the secondary in the triangle "blue". In modern language, the blue triangle is "Cyan".

 

[sophiecentaur]

The chart is my response to the "six colors" in the title of this thread. Take a look at the triangle at the bottom of the chart. Compare it to the square above and to the left. These are easily more different than red is from orange, and yet most people would have difficulty assigning names to them. They are both "Blue". We see one as darker than than other because our eyes aren't very sensitive to blue light. I'm well aware that yellow isn't really a primary (for additive color) but it seems to have a psychological identity so unlike its components. There is good reason why traffic lights are red-yellow-green, and not red-green-blue.

Apparently Isaac Newton would have called the primary color in the square "indigo" and the secondary in the triangle "blue". In modern language, the blue triangle is "Cyan".

Like I already said, you should try to read some stuff on modern colorimetry (1930's actually) and on the tristimulus theory of colour vision. (Google and learn) There is loads of information about this theory which is the basis of colour TV and film. TV colour fidelity is pretty damn good these days so your theory will need to be a bit more than one diagram with coloured squares and triangles. You need to account for the response of the eye to pale brown, khaki and the colour of facial skin as well. It is not a trivial subject.

 

[Algr]

I don't really have a "theory" these are just observations.

 

[sophiecentaur]

I don't really have a "theory" these are just observations.

The diagram you presented implies that you have tried to systematise what you have observed. The CIE colour diagram has a place for all the colours you (we) experience and shows how they fit into a universal system that works well. You are right to say that many nontechnical but highly creative users of colour use their own naive (I use the term in a non-judgemental way) models. I know that approach can produce excellent results, despite its technical flaws and they have been doing it for thousands of years.
I still recommend that you look into the present state of knowledge of Colorimetry. It could help you with reconciling your subjective observations with the accepted system. It may involve some brain ache, though!