I meant that rationals are a subset of the reals, and while floats are also a subset of exact reals a "typical" fixed bitwidth implementation will have a signif

I meant that rationals are a subset of the reals, and while floats are also a subset of exact reals a "typical" fixed bitwidth implementation will have a significantly greater range in floats than rationals. I don't know that it's necessarily as significant a difference as I'd first considered it to be, though.