The colors of objects that you see are actually those colors being reflected back at you. If you start with white light shining on “red” object, the blue and green parts of the white light are absorbed by the object and the red is reflected back for you to see. If instead of a “white” light that has a good balance of red, blue, and green, you have a “white” light that is weak in the red part, the blue and green still get absorbed by the object, but there is less red to be reflected back at your eye. The same object is going to look less “red” under the second “white” light.

Since an incandescent source is like a black-body radiator, the color spectrum of of the light generated by the filament has a decent mix of all the visible colors. However, since the colors present in the “white” light of a fluorescent is made up by phosphors which each do just a part, there can be a poor mix of colors in the light. The better the phosphors do at creating that mix of colors the better the CRI rating.

The more important color is, the higher the CRI of your lamp should be. Manufacturers will make similar lamps (in terms of size, power, etc.) with different CRI ratings since the lower CRI lamps can be made more cheaply. People will either select the lower CRI lamps because color doesn’t matter as much as cost or because they don’t know that there is a difference. For example, for linear tube fluorescents, like you find in offices, the cheaper lamps will have CRI ratings in the low to mid 70s and the higher quality lamps will have CRI rating in the low to mid 80s.

For your home or office I recommend getting the lamp with the highest CRI you can find. For standard fluorescents I only specify lamps with a minium CRI of 85. For compact fluorescents I have to drop my minimum to a CRI of 82, since higher CRI rated lamps aren’t readily available.

Note that both the CRI and the CCT of a fluorescent lamp will be printed on the lamp itself, so if you need to find out what something is you can just look at it. For a linear tube style, the information will be printed on the glass at one of the ends. For a compact fluorescent lamp the information will be printed on the base. Most of the time the information is just printed as “82 CRI” and “3500K.” Sometimes only the manufacturer’s coding will be printed, but typically it will show the CRI by either a “7″ or “8″ somewhere in the lamp code (for low to mid 70′s or low to mid 80′s as appropriate) and the CCT as the first two digits of the CCT (such as 27 for 2700K or 35 for 3500K).

To see an example of what happens when you use a lamp with a low CRI rating most people can probably just find a nearby streetlight. Many streetlights are still high-pressure sodium because HPS lamps are very energy efficient. However, they also have a CRI around 22 and a CCT of around 1800K. That’s why the “orange glow” of a streetlight makes everything look orange or gray.

If you have streetlight around you that look more “white” than orange they are either very old mercury lamps (which are now banned from new installation but utility companies who stockpiled them can use up their inventory until they replace the fixture) or metal halide lamps. They probably have a CCT of around 4000K and a CRI ranging from about 50 (mercury vapor) to 65 to 75 (metal halide). There are also new ceramic metal halide lamps that can have a CRI as high as the low to mid 90′s.

Using streetlights doesn’t provide a perfect example since the darkness of night also affects how we perceive colors, but it’s a good shorthand.

• Color temperature usually only matters for non-filament sources: fluorescents, compact fluorescents, metal-halide, induction, etc. (and it’s really the correlated color temperature, so it will be labeled CCT)
• There are many options for color temperatures of fluorescent lamps. Don’t just pick something up because it’s on the shelf. Think about the space and select a color temperature that is going to help. You may need to go somewhere and see some samples of different color temperatures before you decide.
• Higher color temperatures give a greater perception of brightness. There are complex optical reasons for this that you don’t really need to understand, just know that a CCT of 4000K will appear brighter than a CCT of 3000K even if the measurable quantity of light is equal.
• Color temperatures higher than 4200K tend to be perceived as too cold for comfort. They are useful for very visually-intensive tasks, so you find them in manufacturing facilities where there are lots of small parts being used, but not homes or professional offices.
• Color temperatures lower than 3000K will seem to be too amber during the daytime. Things may appear to be visually “mushy.” However, at night 3000K can be very nice.
• Typically, places used and lit during the daytime benefit from higher color temperatures and places used and lit during the nighttime will benefit from lower color temperatures.
• Dimming fluorescents will not shift the color like it does with incandescents and halogens. This is actually a negative, since during the daytime you need higher lighting levels and a higher color temperature to be comfortable, but in the evening you’ll want to turn down the intensity and color temperature.
• I have found the best middle of the road CCT for fluorescents is 3500K. No single color is going to be the best for every purpose, but 3500K is a pretty decent compromise.
• If you have fixtures will multiple lamps you can mix a warm and a cool to try control the balance of “white” light better. However, don’t mix colors from a single-lamp fixture to the next since it will be start to look just plain weird.

The reason CCT is important is because there are so many different phosphor mixes that can be used. Unlike incandescent lamps, there is a wide range of mixes and each mix, relating to a CCT, establishes what we see as “white.” It is this establishment of “white” that makes any and all of this matter.

By using a different phosphor mix, we make white objects appear either warm or cool, based on the (correlated) color temperature of the light source. All other objects, from artwork to skin color, reflects the same “bias” toward the color we have established as white light. This is why early adoption of fluorescent office lighting was so painful. The phosphor mix and color temperature of the light was shifted so far toward the higher, cool end of the lighting range people didn’t like the way things looked. There were other problems, but the main thing most people took away from the experience was an association of fluorescent lighting being “cold.”

However, we could just as easily make our fluorescent lighting “warm” by using lamps with a CCT with a lower number. For example, you could use lamps with a CCT of 2700 to try to match the incandescent lamps you are used to having in the home. In my experience, most people don’t like a fluorescent lamp with a CCT of 2700K either, since it doesn’t look like a regular incandescent lamp with a color temperature of 2700K. That’s part of the “correlated” aspect, and there is also some color rendering issues which I’ll discuss in following posts.

The trick to good fluorescent and compact fluorescent lighting is finding the CCT you like for any particular use. In more residential settings, a 3000K lamp may be preferred, whereas in an office or kitchen 3500K may be better. You can’t just pick one and use it for everything, although that would be preferable to not picking at all and just installing whatever shows up. Selecting the right color temperature is a part of the lighting design process.

When a black body radiator is heated it begins to glow. There’s a cool java applet on the Olympus Microscopy Resource Center website that demonstrates this idea with a picture of a horseshoe. At the lower range it begins by glowing red. Then as it gets hotter it goes through yellow and white phases, eventually shifting into the very bright blue-white range. At any particular color, we refer to the temperature of the blackbody as the “color temperature.” (A note on that java tutorial, the temperature indicated on the scale refers to the tips of the horseshoe. The other colors move down the horseshoe to show that the tips are being heated and the bottom is cooler than the tip. The temperature does not relate to all the colors shown in the image.) The temperature is on the Kelvin scale, which is the same as celsius plus 273. That is, -273˚C is 0 K, 0˚C is 273 K, and 100˚C is 373 K.

For lighting, it is easier to refer to the temperature generating a color than trying to describe the color itself, which changes slowly and is not really part of our language of color. If you are looking at that java applet from Olympus, try to describe the difference between the color of the horseshoe tips at 2700K, 3000K, and 3200K. You’ll quickly realize why referring to a color temperature is easier.

A bit of confusing terminology, however, is that we refer to red the red side as “warm” and the blue side as “cool.” This is based on our traditional color associations of red and blue. It can be confusing because the bluer tint comes as you increase the temperature. Hence, you may hear people refer to “raising” the color temperature to “cool” the light source.

The filament of a regular incandescent lamp is not a true black-body, but it does go through the same color process as it is heated. That’s why when you put a regular lightbulb on a dimmer it shifts toward the red as you dim it down: it’s the horseshoe in reverse. Therefore, if you’re looking at incandescent lightbulbs color temperature isn’t a very useful metric. Regular and halogen lamps at full power usually end up around 2700K to 3050K, and that can be changed by adjusting the voltage. The real purpose of discussing color temperature is for fluorescent lamps, which I’ll get to tomorrow.

One of the biggest problems with artificial lighting is that our eye and brain determine a lot of color information by comparison instead of some sort of mental color wheel. I’ve stuck an image file into this post to provide an example. The image to the right is made up of exactly 3 colors, but the purple color appears to be either a lighter or darker purple depending upon whether the adjacent color is the blue or the pinkish color. Also, the purple bar that runs all across the middle seems to change color from one side to the other.

The issue for artificial lighting is that as we look around us, our eye spots the brightest source of light and our brain “sets” that as white light. We are mentally doing the opposite of the what my sample image shows: instead of using relativity of colors to “see” either bright or darker versions of purple when the purple is actually the same, we are using relativity of colors to see different versions of “white” and then perceiving that as always the same “white.”

For the common user, a lightbulb (unless a specialty colored lamp) is going to be perceived as white. For the professional, we know that different sources are giving us different whites and we need a way to document the differences. We do that by specifying the “color temperature” of the source.

For the next post, I’ll explain what color temperature actually is and how we come up with the numbers.

Common claims:

• It is closer to “natural” daylight, which is its own benefit: This claim is more or less meaningless. Daylight changes throughout the day due to atmospheric conditions (is it cloudy or clear) and time of day (mid-day sun or morning sunrise). All artificial lighting sources are static in their output, what you get at 3:00 PM on a cloudy day is that same as 2:00 AM at night. Some LED sources are collections of LEDs that can be programmed to change color, but that has nothing to do with mimicking daylight. Trying to hash out the meaning of this claim gets you mired in spectral distribution curves which is a huge topic, so for now just keep in mind that “full-spectrum” sources are just modifications of existing fluorescents. Except GE Reveal and the like, which are modified incandescents.
• You get better color from full-spectrum sources: This is sort-of true. Full-spectrum fluorescents use a different phosphor mix that can sometimes have a higher CRI (color rendering index) than a typical fluorescent. The increase in CRI is often accompanied by an decrease in efficiency. You can also get non-full-spectrum fluorescents with higher CRI ratings than standard for slightly more cost than a standard lamp but still less than a full-spectrum lamp. The stand-out exception to this is the incandescent full-spectrum sources, which decrease the CRI of their bulbs.
• You get increased worker productivity with full-spectrum: There are two parts to this: the first is just based on the ability to see tasks well. In most circumstances productivity for visual tasks is linked to monochromatic tasks, such as reading black ink on white paper. For these tasks it is the amount of light that determines good visibility, not the color or quality of the light. In this regard full-spectrum offers no benefits. There is some research suggesting that since the rods–which are blue-green sensitive–control the size of the pupil, light to the cool, bluish side causes the pupil to constrict and thereby increase acuity and you can get equal acuity with less power by using bluish light. I think it is unreasonable to get into this kind of detail when you just want to buy a lightbulb, so the simple answer is “no.” However, if your work is very color sensitive, such as fine art or clothing production, you want to maximize the CRI of your sources to affect productivity.
• It has psychological benefits: This is the second part of the “increased productivity” claim. By definition, a psychological benefit is “all in your head,” so in terms of the benefit claim it is true. If you like full-spectrum lighting and don’t mind the disadvantages then you are getting a psychological benefit. Does everyone feel better equally, there’s no way to say, which is why it makes a good marketing claim. I have never found any research that shows a physiological link that leads to a psychological benefit. This is also often linked to the “natural” daylight claim. It is true that most people feel better working and living under daylight than artificial light. However, daylight changes throughout the day and with the weather, and I think the change over time is one of the main reasons people like daylight.

This post is getting longer than I like for a daily reading, so I’ll deal with the health concerns tomorrow. That will give you a chance to rest up and stay with me.

1. I hate the color. It’s not so much the color as the lack of color. Incandescent, like sunlight, has a wide “spectral distribution,” which simply means it looks white because it has all the colors of the rainbow mixed together. Because the light in a fluorescent lamp is generated by the phosphor coating on the inside of the lamp (see yesterday’s post), rather than incandescence, the spectral range is reduced. Think of a rainbow with parts missing. Objects and people that you see will looked dingy if the colors on them are not present in the fluorescent lamp’s “rainbow.” Although most people can’t quite put their finger on it, that’s why they think that things just look bad under fluorescent lights.
2. I hate the flickering. The “flicker” that some people see is a real problem and can lead to headaches as well as annoyance. It is caused by the ballast. Old ballasts operate at 60 cycles per second, the same as the cycle supplied from the power company. Each end of the lamp is actually firing off 60 times per second, which is slow enough some people can see it, usually in their peripheral vision.

There are good fixes for these complaints, which is tomorrow’s posting topic.