exponent bias
The fixed number added to the real exponent before it is stored - 127 for float, 1023 for double - so the stored field is never negative.
The exponent bias is 2^(w-1) - 1 for a w-bit exponent field (127 for binary32, 1023 for binary64); the stored field is E = e + bias, which lets the field be compared as an unsigned integer.
Examples
binary32's bias is 127, so a stored exponent of 130 means 2³.
See also