If you make a simple model where the rim is width W, the tire's "flattened-out" width is L (ie distance from bead to bead), its inflated/installed width is D, and you assume that the tire takes the shape of a circle who's bottom is cutoff by the rim... then there are two interesting regimes:
Fat rim (where D ~ W.... tire looks like a half circle with rim width roughly the diameter):
* dD / dW ~= 1
Narrow rim (where D >> W):
* Rate of change of width of tire, D, with rim width, W, is:
dD / dW ~= 1 / Pi < 1
With your road bike rim I think you're pretty close to the first regime: the widths of the rims are fairly comparable to the inflated width of the tire, and for every 1mm increase of rim width you therefore get an increase of roughly 1mm of inflated tire width all else being equal.
Unfortunately, with the mountain bike, I think you're really part way between these two regimes, and so adding 1mm in rim width will get you somewhere between 1/Pi ~ 0.3mm and 1mm of increased tire width. Given that 0.2 inches is 5mm, getting 5mm from the tire change costs you 50g, whereas with the rim change for 70g the 2mm of rim width increase will probably only get you something like 1mm in tire width. So there's really no comparison here at all in terms of the two.
Of course, as you said, that's not to say that there aren't other variables that might make the wider rim worthwhile. But weight per tire width is definitely not one of them.
Enjoying my Funk
Last edited by dwaharvey on Mon Feb 25, 2013 9:17 pm, edited 1 time in total.