Each pixel in the CCD chip is represented by 16 bits, which can be 0 or 1. This means it can encode a range of 65536 (2^16) different numbers. Since the intensity of any object will never be negative, the natural range to choose would be 0 to 65535.
However, read noise introduced when amplifing the tiny signal stored in the CCD chip to something big enough to be a readable signal can be positive or negative. The amplifier automatically adds a small bias level to each pixel, so it doesn't lose the negative values. To correct for this, we simply subtract the bias value from all of the images, pixel by pixel.
But why can we have negative numbers now? The CCD chip must be as fast as possible to be able to read out quickly and accurately, and extra bits (extra information) slow it down. We care more about accuracy than speed in processing, so now we can afford more bits at the cost of more time.
Questions? Comments? Ideas? Email all the