Maximal floating point exponent for double precision; greater exponents are principally possible (up to 16383), but not supported in all math functions.