CRI is technically, according to the American National Standards Institute, the measure of how similar colors appear under illumination by a test source, compared to under a reference source that has the same correlated color temperature.

In practice, it’s often known as the color accuracy, a numerical rating of the color quality produced by a light source. Incandescent bulbs are the benchmark for the test (scoring 100) while a good LED bulb will be rated at 80 or above (this number will almost always be printed on the back of the box). Incandescents—and the sun, for that matter—are the baseline because they emit light at all points along the visible spectrum, while LEDs and CFLs have a spiked, intermittent pattern. Higher CRI value bulbs are available for graphic designers, art galleries, artists, and so forth. But they increase the cost and most consumers won’t notice a considerable difference above 80. When using a bulb with a score below 80, the results will vary based on the qualities of the particular bulb, but viewers will typically observe the color of the surrounding objects to be “dingy.” Red and blue tones will appear to be off and skin tones can look unhealthy.

CRI is determined by measuring the color of eight samples and then taking the average of those ratings. While that number is representative of overall color accuracy, this testing method is not perfect. By using an average it means that a particular light source might be very strong in some areas and very weak in others. A particular bulb might be very accurate with Dark Greyish Yellow (known as TCS02), yet work quite poorly on Light Violet (TCS07). This could leave the bulb with a relatively high CRI rating, despite its low-quality light under particular circumstances. For this reason and others—including the fact that the test is 40 years old—CRI is seen as a less-than-perfect tool, but it remains the only one on the Lighting Facts label on bulbs.

There are other, more saturated colors that can be used in order to augment the testing, but these are generally not included in the normal CRI rating. The most notable of these is Strong Red (known as TCS09 or R9), which serves as a stand-in for certain skin tones, food types, and other colors we often encounter. When speaking about color accuracy, lighting companies will often relay their CRI score and then call out a high R9 CRI value if their bulb is capable of it.

When it comes to measuring the quality of light, the CRI is the test used most often. However, it’s not the only assessment available. Color Quality Scale (CQS) is a newer testing method under development by the National Institute of Standards and Technology (NIST) and has a number of proponents. Its methods are not unlike those of CRI, but it’s a newer model, developed over the last few years specifically with LED lighting in mind. The CQS uses 15 color samples, many of which are more saturated than the eight CRI shades, also operates on a 100 point scale, and addresses many of the problems with CRI testing. For example, CRI treats all deviations from the correct value as being the same, when it’s generally accepted that humans prefer overly saturated color to undersaturation, thus making shifts in this direction arguably preferable. CQS factors this in, thus focusing somewhat less on color fidelity and more on perception.

You’re probably asking yourself, why does this matter so much? It’s not just that great light is important. A bad test is easy to game (or is it that easy gaming makes for a bad test?). This means that manufacturers could design around the test, making for a cheap lamp that does well in testing but is not able to deliver the light quality that buyers expect. This would be bad for the buyers and bad for the industry.

It’s also been noted that, as in other industries, outdated testing procedures can hold back progress, as manufacturers are forced to comply with old standards as opposed to advancing at the rate of technology’s progress.

The shade of white light is measured on a temperature scale using degrees Kelvin (K). This is important because we expect to see a certain shade of color from indoor lighting, about 2500K–3000K. Warmer colors are more reddish and have a lower color temperature, while cooler colors are bluish and have higher Kelvin ratings. This, as you’ve surely noted, is the opposite of what you might expect, with cooler colors actually having a high Kelvin count.

Older LEDs and CFLs typically operated at “cooler,” bluer temperatures, and consumers didn’t like the hospital-like lighting they provided. New LED bulbs have been able to lower their color temperature to approach that of incandescents, primarily by using improved phosphors. A phosphor, which will be covered more extensively later, is a material that radiates light without having to be heated. They are often used in LED lamps, in tandem with the LEDs, in order to dial in characteristics like the color temperature.

As far as LED bulbs go, a more accurate term than color temperature is the correlated color temperature (CCT). Incandescents have a “true” color temperature, but color temperature is approximated for LED bulbs so their value is understood as being correlated (that is to say that the shades of white LEDs produce don’t land perfectly on the Planckian locus like incandescents, but they are close enough that the color temperature scale still works). The CCT is sometimes defined as “the absolute temperature of a blackbody radiator whose chromaticity resembles that of the light source,” but for our purposes we’ll be happy with “color temperature.”

The general rule is that warmer tones will result in a less efficient lamp. This is because more red phosphor is needed to achieve the lower CCT and this is less efficient than yellow phosphor. This is the case not because of the phosphor, but because the energy difference between the blue photons emitted by the LEDs and the resulting red photons is greater than between blue and yellow.

Computer geeks may think of Haitz’s Law as basically Moore’s Law for LED lighting. The law states that every 10 years the cost-per-lumen from light sources will drop by 10x while the light produced from an LED will increase by 20x. Put more simply, LEDs will get better and cheaper over time.

The contrast to Moore’s Law is that LEDs only need to reach a certain point when they are cheap enough and bright enough for the necessary applications, while we will always have a demand for exponential growth in computing. This is why we know LED bulbs will replace CFLs: in just a matter of years they won’t cost $30, they will cost $15, or $10, or possibly $5. The LEDs sometimes account for up to 60% of the manufacturer’s costs (though this varies based on the bulb in question and who you ask), so reducing their price is an important part of increasing sales.

The key is that this is a steady trend, one that’s been observed since the 1970s. It’s not a scientific law in the strict sense, but it means that we can likely count on LED prices dropping in the future at a predictable rate, and we can plan accordingly.

These terms are essentially interchangeable. A lamp is simply something that produces light, just like a bulb. However, not all things that produce light are what we would describe as “bulbous,” so the term “lamp” is the catch-all.

The typical incandescent bulb is known as the A19. It carries this name because it has an “A” shape and is 19 multiplied by 1/8-inches in diameter (2 3/8-inches) at its wide point. So your normal, everyday bulb will be an A19 with an Edison screw base. And any trip to the hardware store will tell you that’s not the only type of bulb—you might have a light fixture that uses the GU (bayonet-style) base and a parabolic aluminized reflector light (PAR) shape, or any number of other options, but the A19 is the most popular.

Within the industry, the term “lamp” is popular because not all lights are bulbs. In fact, when it comes to solid-state lighting, lots of the lights come in non-bulb varieties, such as overhead LED arrays like Lighting Science Group’s Flat LowBay. In the future, an increasing number of lights will be fixtures, which means they will be moving away from the traditional bulb shape because it’s no longer necessary.

Lighting is a highly regulated field. Agencies like the Department of Energy and companies like United Laboratories track products closely and make sure they adhere to certain standards, so that bulbs don’t suck down more power than they say they will or cause pesky fires. LM-79-08 is a testing procedure for solid-state lighting that looks at criteria like efficiency (lumens-per-watt), power usage, total flux, and chromacity. LM-80 is the test for the maintenance, or longevity, of a solid-state light source. It tests bulbs for thousands of hours (6,000 at the minimum) over a range of different temperatures. It only offers recorded test data; no data is extrapolated based on the observed results.

The useful lifetime of LED bulbs cannot be fully tested, not just because of their extremely long lives but because by the time a thorough test was even close to being finished, technology would have passed that model over and replaced it with another. As such, tests are designed to work around this issue. A valid sample size is determined and those bulbs are tested for a set amount of time. After a thousand, or so, hour “seasoning period”—where the bulbs can actually brighten—bulbs are tested for another 5,000 hours. During this time, light levels and bulb failures are recorded. The resulting chart can be extrapolated to determine the point at which the life of the bulb will be over. This is typically at the time when they produce a total of 70% (known as L70) or 50% (L50) of their initial lumens (the group behind the test says either “lumen maintenance life” an be used). The life is determined by the Illuminating Engineering Society of North America’s (IESNA) LM-80-08 testing procedure.

When rating the lifetime of a lamp, certain conditions are maintained. There is a set ambient temperature (25° Celsius), bulb running temperature (usually under 90° Celsius), and amperage (generally 350 milliamps) so that conditions are the same from one test to another, and real-world conditions are being reflected.

Tests are devised and accepted by groups like the Illuminating Engineering Society of North America. Hence the full name of a test might be: “IESNA LM-79-07, Illuminating Engineering Society of North America, Approved Method for the Electrical and Photometric Measurements of Solid-State Lighting Products.”

Incandescent bulbs were rated by their wattage consumption number. This worked because they were the only lighting option, so we could tell that a 60W bulb was brighter than a 40W one. This (admittedly poor) rating system doesn’t work for LED bulbs because they can produce the same amount of lumens using less wattage. So, in order to sell bulbs to consumers who don’t know the rating system, LED bulbs are often sold by their wattage equivalency. For example, a Samsung A19 LED bulb runs at 10W but produces 550 lumens. That lumen number means it has a 40W equivalency.