--- a/src/sysdep.cc
+++ b/src/sysdep.cc
@@ -52,6 +52,7 @@
 #include "error.h"
 #include "sysdep.h"
 #include "defun.h"
+#include "f77-uscore.h"
 
 extern "C"
 {
@@ -69,6 +70,8 @@
 #else
 LOSE! LOSE!
 #endif
+
+extern double F77_FCN (d1mach) (const int&);
 }
 
 #ifndef STDIN_FILENO
@@ -81,6 +84,9 @@
 // Octave's idea of not a number.
 double octave_NaN;
 
+// The floating point format on this system.
+floating_point_format native_float_format = OCTAVE_UNKNOWN_FLT_FMT;
+
 #if defined (HAVE_FLOATINGPOINT_H)
 #include <floatingpoint.h>
 #endif
@@ -110,6 +116,39 @@
 }
 #endif
 
+union equiv
+{
+  double d;
+  int i[2];
+};
+
+struct float_params
+{
+  floating_point_format fp_fmt;
+  equiv fp_par[4];
+};
+
+#define INIT_FLT_PAR(fp, fmt, sm1, sm2, lrg1, lrg2, rt1, rt2, dv1, dv2) \
+  do \
+    { \
+      fp.fp_fmt = (fmt); \
+      fp.fp_par[0].i[0] = (sm1);  fp.fp_par[0].i[1] = (sm2); \
+      fp.fp_par[1].i[0] = (lrg1); fp.fp_par[1].i[1] = (lrg2); \
+      fp.fp_par[2].i[0] = (rt1);  fp.fp_par[2].i[1] = (rt2); \
+      fp.fp_par[3].i[0] = (dv1);  fp.fp_par[3].i[1] = (dv2); \
+    } \
+  while (0)
+
+static int
+equiv_compare (const equiv *std, const equiv *v, int len)
+{
+  int i;
+  for (i = 0; i < len; i++)
+    if (v[i].i[0] != std[i].i[0] || v[i].i[1] != std[i].i[1])
+      return 0;
+  return 1;
+}
+
 static void
 octave_ieee_init (void)
 {
@@ -157,6 +196,61 @@
   octave_NaN = DBL_MAX;
 
 #endif
+
+  float_params fp[5];
+
+  INIT_FLT_PAR (fp[0], OCTAVE_IEEE_BIG,
+		   1048576,  0,
+		2146435071, -1,
+		1017118720,  0,
+		1018167296,  0);
+
+  INIT_FLT_PAR (fp[1], OCTAVE_IEEE_LITTLE,
+		 0,    1048576,
+		-1, 2146435071,
+		 0, 1017118720,
+		 0, 1018167296);
+
+  INIT_FLT_PAR (fp[2], OCTAVE_VAX_D,
+		   128,  0,
+		-32769, -1,
+		  9344,  0,
+		  9344,  0);
+
+  INIT_FLT_PAR (fp[3], OCTAVE_VAX_G,
+		    16,  0,
+		-32769, -1,
+		 15552,  0,
+		 15552,  0);
+
+  INIT_FLT_PAR (fp[4], OCTAVE_UNKNOWN_FLT_FMT,
+		0, 0,
+		0, 0,
+		0, 0,
+		0, 0);
+
+  equiv mach_fp_par[4];
+
+  mach_fp_par[0].d = F77_FCN (d1mach) (1);
+  mach_fp_par[1].d = F77_FCN (d1mach) (2);
+  mach_fp_par[2].d = F77_FCN (d1mach) (3);
+  mach_fp_par[3].d = F77_FCN (d1mach) (4);
+
+  int i = 0;
+  do
+    {
+      if (equiv_compare (fp[i].fp_par, mach_fp_par, 4))
+	{
+	  native_float_format = fp[i].fp_fmt;
+	  break;
+	}
+    }
+  while (fp[++i].fp_fmt != OCTAVE_UNKNOWN_FLT_FMT);
+
+  if (native_float_format == OCTAVE_UNKNOWN_FLT_FMT)
+    panic ("unrecognized floating point format!");
+     
+  cerr << native_float_format << "\n";
 }
 
 #if defined (EXCEPTION_IN_MATH)
@@ -515,15 +609,8 @@
 DEFUN ("isieee", Fisieee, Sisieee, 1, 1,
   "isieee (): return 1 if host uses IEEE floating point")
 {
-  Octave_object retval;
-
-#if defined (IEEE_BIG_ENDIAN) || defined (IEEE_LITTLE_ENDIAN)
-  retval = 1.0;
-#else
-  retval = 0.0;
-#endif
-
-  return retval;
+  return (double) (native_float_format == OCTAVE_IEEE_LITTLE
+		   || native_float_format == OCTAVE_IEEE_BIG);
 }
 
 #if !defined (HAVE_GETHOSTNAME) && defined (HAVE_SYS_UTSNAME_H)