Comment by ses1984

Or you could approach it another way, store the data using the full range of the system float type and normalize it to [0,1] through accessor methods.

Can float types even support 0, 1 inclusive? You just can't represent natural numbers like that with floats...