I wonder what the words say.....

As I said, absolutely not required.
In theory, the purpose of separating the channels is to get a greyscale version of each of the three colour channels. Working in greyscale in 8 bits means having 256 tones levels only. On the other hand, to be able to read the text with very low contrast means that the edit process will have to heavily 'stretch' tones to make the writing visible. Very subtle grey tone differences will be stretched possibly to black to white. Increasing the local contrast suggests that the 8-bits colour depth will produce banding and artifacts.
Before adding personnal comments about this in our case, I'd like to suggest that this is a perfect test image to give the opportunity for everybody to test the difference, given that Elements can easily work in 16 bits mode except for layers, local tools and some filters.

Now, my personal comments:
- for our purpose, banding and artifacts are very secondary to the ability to read the writing.
- There has been hot discussions about the advantage of opening an 8-bits image and converting it to 16 bits for editing. That's what happens when you edit a jpeg in ACR. I have been challenged by Pica to provide some more tests results and theoretical explanation. My results today do confirm that there is really a slight advantage. I already tried to show that advantage without success a few years ago, now I have a test protocol using the Equalize filter which confirms where 16-bits editing from an 8-bits picture is useful.
However, I must say that the results have given surprising results:
- With the equalizer filer, you can't show banding... but the 'equalization' effect is far better (flatter histogram).
- The explanation of 'rounding errors' seems logical... but does not hold.
- The explanation of 'cumulative' edits (similar to cumulative saves to jpeg) seems to be totally irrealistic in normal workflows. In ACR, there is no cumulative effects, they are all applied in a single editing session.
- Those tests have been run with careful examination of histograms, to detect 'combing' and posterization risks. Now I can easily tell what the histograms really mean. There are 'combed' histograms with no visual degradation of the image... and there are fine, smooth histograms for very visible posterization!