A FORTRAN 77 compiler must support both single- and double-precision data, but is not required to support quadruple-precision. If a FORTRAN 77 compiler supports quad-precision data, the name of the conversion function is implementation dependent, so you'll have to check the compiler manual. In Fortran 90 or later, the compiler is also not required to support quad-precision data, but if it does support it, the standard way to convert data is to call the real() function with a kind type parameter