VMS Help FORTRAN, /G_FLOATING  *Conan The Librarian (sorry for the slow response - running on an old VAX)

  /[NO]G_FLOATING                 D=/NOG_FLOATING

  Controls how the compiler implements REAL*8, COMPLEX*16, DOUBLE
  PRECISION, and DOUBLE COMPLEX declarations and constants.

  This qualifier can also be specified on the OPTIONS statement.  The
  default is /NOG_FLOATING, which directs the compiler to implement
  double-precision quantities using the VAX D_floating data type.
  The /G_FLOATING qualifier directs the compiler to implement such
  quantities using the VAX G_floating data type.

  If your program requires the G_floating form of double precision
  for its correct operation (that is, it uses a range larger than
  10**38), you should use the /G_FLOATING qualifier in an OPTIONS
  statement in your source program.  The implementation of REAL*8 in
  Compaq Fortran 77 is further discussed in your language reference manual.

                                 NOTE

          You should not mix the  D_floating  and  G_floating
          data  types  in routines that pass double-precision
          quantities between themselves.