Curtis N wrote in post #4621625
The more brightly exposed your reference area is, the more precise the data will be (as long as you don't hit the upper limit of the camera's dynamic range, as Leo indicated).
Here's the best analogy I can come up with: Suppose you have a digital scale that gives you readings in pounds. If you weigh something and it says 5 pounds, it could be anywhere from 5.00 pounds to 5.99 pounds. So two objects that vary in weight by nearly 20% will appear to be equal. But if you weigh something that weighs 500 pounds, then anything 2% greater than that (501 pounds) will be shown as a different weight. In relative terms, the scale is more precise when heavier objects are weighed.
Obviously a scale with a 600 pound capacity won't accurately weigh anything heavier than that. An object weighing 601 pounds and an object weighting 1,000 pounds would both appear to weigh 600. This is why an overexposed image with one or more RGB values at 255 will not be a suitable white balance reference.
Now keep in mind that an overexposed black reference could be theoretically just as precise as an underexposed white object, if the RBG values are the same. But using a light grey color allows a very precise reference without the need to adjust the camera's exposure relative to the overall scene, so the Whibal card is made that way. It's a reference that gives you precise data without the risk of having any clipped channels, provided the reference shot is more or less properly exposed.
Applying your analogy, Curtis, then I would tend to think that the closer to mid-gray the color, the more accurate it will be. I'm thinking this because an image sensor is linear while image tones are not. So, to convert a raw linear image to, say, a JPEG image, the highlight end of the data would get compressed and the shadow end would get stretched.
But all that is academic. For all intents and purposes, I'm sure it'll be safe to say that using any neutral gray scale tone will yield a good white balance, no?