Floating point data is stored as sign, mantissa and exponent. It is base10 arithmetic, although the method of storing the actual numbers is base2 (binary). Binary storage doesn't change the essence of the floating point number being represented--base10.
(The mantissa can get you into...
A good quick book, if still in print, is "A Fortran Coloring Book". It teaches with humor and is very thorough. I'm just not sure it is still in print.
