Dynamic range - 72db ??

This mystery number I can explain, having dug through the footnotes. It is simply "twenty times the log base ten of the ratio of the maximum electron count that a photosite can hold to the RMS variation in the electron count recorded by taking a photo in darkness"! Quite transparent, I would think.

In more familiar photographic terms, you can divide this number by 6 to approximate the S/N ratio in "stops", or factors of two. So the maximum measurable signal level is twelve stops brighter than the noise.

Note that the noise measure is variation in signal, not average signal: the average value can be corrected for (like printing through film base plus fog level?); the variation is the visible "confetti" of noise.

Now, if you accept Kodak's suggestion that things need to be about ten to fourty times above noise to look good, which is roughly three to five stops, you get a usable subject brightness range of something like six to nine stops. Vague, but I do not think the theory justifies any more precise statement.