--- a/configure.ac	Fri Jun 16 22:30:00 2017 -0700
+++ b/configure.ac	Sat Jun 17 10:26:38 2017 -0700
@@ -2455,7 +2455,7 @@
 ## Check for nonstandard, but common math functions, that we need.
 
 dnl Use multiple AC_CHECKs to avoid line continuations '\' in list
-AC_CHECK_FUNCS([erf erff erfc erfcf exp2f])
+AC_CHECK_FUNCS([exp2f])
 
 ## Check for math defines such as M_LN2 in math.h
 AC_CACHE_CHECK([for MATH DEFINES in math.h],

--- a/liboctave/numeric/lo-specfun.cc	Fri Jun 16 22:30:00 2017 -0700
+++ b/liboctave/numeric/lo-specfun.cc	Sat Jun 17 10:26:38 2017 -0700
@@ -133,29 +133,9 @@
 #endif
     }
 
-    double
-    erf (double x)
-    {
-#if defined (HAVE_ERF)
-      return ::erf (x);
-#else
-      double retval;
-      F77_XFCN (xderf, XDERF, (x, retval));
-      return retval;
-#endif
-    }
-
-    float
-    erf (float x)
-    {
-#if defined (HAVE_ERFF)
-      return erff (x);
-#else
-      float retval;
-      F77_XFCN (xerf, XERF, (x, retval));
-      return retval;
-#endif
-    }
+    double erf (double x) { return std::erf (x); }
+
+    float erf (float x) { return std::erff (x); }
 
     // Complex error function from the Faddeeva package
     Complex
@@ -172,29 +152,9 @@
       return FloatComplex (ret.real (), ret.imag ());
     }
 
-    double
-    erfc (double x)
-    {
-#if defined (HAVE_ERFC)
-      return ::erfc (x);
-#else
-      double retval;
-      F77_XFCN (xderfc, XDERFC, (x, retval));
-      return retval;
-#endif
-    }
-
-    float
-    erfc (float x)
-    {
-#if defined (HAVE_ERFCF)
-      return erfcf (x);
-#else
-      float retval;
-      F77_XFCN (xerfc, XERFC, (x, retval));
-      return retval;
-#endif
-    }
+    double erfc (double x) { return std::erfc (x); }
+
+    float erfc (float x) { return std::erfcf (x); }
 
     // Complex complementary error function from the Faddeeva package
     Complex
@@ -211,7 +171,7 @@
       return FloatComplex (ret.real (), ret.imag ());
     }
 
-    // Real and complex scaled complementary error function from Faddeeva package
+    // Real and complex scaled complementary error function from Faddeeva pkg.
     float erfcx (float x) { return Faddeeva::erfcx(x); }
     double erfcx (double x) { return Faddeeva::erfcx(x); }
 
@@ -225,7 +185,8 @@
     erfcx (const FloatComplex& x)
     {
       Complex xd (x.real (), x.imag ());
-      Complex ret = Faddeeva::erfcx (xd, std::numeric_limits<float>::epsilon ());
+      Complex ret;
+      ret = Faddeeva::erfcx (xd, std::numeric_limits<float>::epsilon ());
       return FloatComplex (ret.real (), ret.imag ());
     }
 
@@ -261,7 +222,8 @@
     dawson (const FloatComplex& x)
     {
       Complex xd (x.real (), x.imag ());
-      Complex ret = Faddeeva::Dawson (xd, std::numeric_limits<float>::epsilon ());
+      Complex ret;
+      ret = Faddeeva::Dawson (xd, std::numeric_limits<float>::epsilon ());
       return FloatComplex (ret.real (), ret.imag ());
     }