Keep in mind on most systems, floating point binary uses 2's compliment and has a sign bit. Also, where the fraction, as I call it, begins and where the exponent end depend on how many bits the floating point numbers use for that system. For instance, on 64bin floating point numbers there ususally a sign bit, 11bits for the exponent and 54bits for the fraction, with a 1024 bias for determining where the decimal is.