diff --git a/doc/fp_utilities/float_next.qbk b/doc/fp_utilities/float_next.qbk
index 60d9df9f08..1e299f3882 100644
--- a/doc/fp_utilities/float_next.qbk
+++ b/doc/fp_utilities/float_next.qbk
@@ -261,7 +261,7 @@ next floating-point value, but `x - u` is not necessarily the previous value.  S
 value.  The corner cases occur at power of 2 boundaries.
 * When the argument becomes very small, it may be that there is no floating-point value that
 represents one ULP.  Whether this is the case or not depends not only on whether the hardware
-may ['sometimes] support denormals (as signalled by `std::numeric_limits<FPT>::has_denorm`), but also whether these are
+may ['sometimes] support denormals (as signalled by `boost::math::detail::has_denorm_now<FPT>()`), but also whether these are
 currently enabled at runtime (for example on SSE hardware, the DAZ or FTZ flags will disable denormal support).
 In this situation, the `ulp` function may return a value that is many orders of magnitude too large.
 
diff --git a/doc/sf/lambert_w.qbk b/doc/sf/lambert_w.qbk
index 22b13b6ac0..27591d9fee 100644
--- a/doc/sf/lambert_w.qbk
+++ b/doc/sf/lambert_w.qbk
@@ -289,7 +289,7 @@ The domain of /W/[sub -1] is \[-/e/[super -1], 0\). Numerically,
 For example, for `double`: lambert_wm1(-2.2250738585072014e-308) = -714.96865723796634 [br]
 and for `float`: lambert_wm1(-1.17549435e-38) = -91.8567734 [br]
 
-* `z < -std::numeric_limits<T>::min()`, means that z is zero or denormalized (if `std::numeric_limits<T>::has_denorm_min == true`),
+* `z < -std::numeric_limits<T>::min()`, means that z is zero or denormalized (if `boost::math::detail::has_denorm_now<T>() == true`),
 for example: `r = lambert_wm1(-std::numeric_limits<double>::denorm_min());` and an overflow_error exception is thrown,
 and will give a message like:
 
diff --git a/example/inspect_fp.cpp b/example/inspect_fp.cpp
index 71ea686f41..5de25f2ac7 100644
--- a/example/inspect_fp.cpp
+++ b/example/inspect_fp.cpp
@@ -20,6 +20,8 @@
 #include <boost/math/tools/config.hpp>
 #endif
 
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
+
 //------------------------------------------------------------------------------
 
 bool is_big_endian()
@@ -111,9 +113,9 @@ template<class T> void print_table()
 {
     print_row("0", (T)0);
     print_row("sn.min", std::numeric_limits<T>::denorm_min(),
-          std::numeric_limits<T>::has_denorm);
+          boost::math::detail::has_denorm_now<T>());
     print_row("-sn.min", -std::numeric_limits<T>::denorm_min(),
-          std::numeric_limits<T>::has_denorm);
+          boost::math::detail::has_denorm_now<T>());
     print_row("n.min/256", (std::numeric_limits<T>::min)()/256);
     print_row("n.min/2", (std::numeric_limits<T>::min)()/2);
     print_row("-n.min/2", -(std::numeric_limits<T>::min)()/2);
diff --git a/include/boost/math/ccmath/next.hpp b/include/boost/math/ccmath/next.hpp
index 30045a0a36..602c3bc9f9 100644
--- a/include/boost/math/ccmath/next.hpp
+++ b/include/boost/math/ccmath/next.hpp
@@ -101,7 +101,7 @@ constexpr T get_smallest_value(const std::false_type&)
 template <typename T>
 constexpr T get_smallest_value()
 {
-    return get_smallest_value<T>(std::integral_constant<bool, std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present)>());
+    return get_smallest_value<T>(std::integral_constant<bool, std::numeric_limits<T>::is_specialized>());
 }
 
 template <typename T>
diff --git a/include/boost/math/special_functions/lambert_w.hpp b/include/boost/math/special_functions/lambert_w.hpp
index c57ce1d5bd..014282e0cf 100644
--- a/include/boost/math/special_functions/lambert_w.hpp
+++ b/include/boost/math/special_functions/lambert_w.hpp
@@ -55,6 +55,7 @@ BOOST_MATH_INSTRUMENT_LAMBERT_W_SMALL_Z_SERIES_ITERATIONS  // Show evaluation of
 #include <boost/math/special_functions/fpclassify.hpp>
 #include <boost/math/special_functions/log1p.hpp> // for log (1 + x)
 #include <boost/math/constants/constants.hpp> // For exp_minus_one == 3.67879441171442321595523770161460867e-01.
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
 #include <boost/math/special_functions/pow.hpp> // powers with compile time exponent, used in arbitrary precision code.
 #include <boost/math/tools/series.hpp> // series functor.
 //#include <boost/math/tools/polynomial.hpp>  // polynomial.
@@ -1797,7 +1798,7 @@ T lambert_wm1_imp(const T z, const Policy&  pol)
       return -tools::max_value<T>();
     }
   }
-  if (std::numeric_limits<T>::has_denorm)
+  if (boost::math::detail::has_denorm_now<T>())
   { // All real types except arbitrary precision.
     if (!(boost::math::isnormal)(z))
     { // Almost zero - might also just return infinity like z == 0 or max_value?
diff --git a/include/boost/math/special_functions/next.hpp b/include/boost/math/special_functions/next.hpp
index 07c55e8d4a..1b56932423 100644
--- a/include/boost/math/special_functions/next.hpp
+++ b/include/boost/math/special_functions/next.hpp
@@ -82,8 +82,8 @@ inline T normalize_value(const T& val, const std::true_type&)
 }
 
 template <class T>
-inline T get_smallest_value(std::true_type const&)
-{
+inline T get_smallest_value(std::true_type const&) {
+   static_assert(std::numeric_limits<T>::is_specialized, "Type T must be specialized.");
    //
    // numeric_limits lies about denorms being present - particularly
    // when this can be turned on or off at runtime, as is the case
@@ -106,11 +106,12 @@ inline T get_smallest_value(std::false_type const&)
 template <class T>
 inline T get_smallest_value()
 {
-#if defined(BOOST_MSVC) && (BOOST_MSVC <= 1310)
-   return get_smallest_value<T>(std::integral_constant<bool, std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == 1)>());
-#else
-   return get_smallest_value<T>(std::integral_constant<bool, std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present)>());
-#endif
+   return get_smallest_value<T>(std::integral_constant<bool, std::numeric_limits<T>::is_specialized>());
+}
+
+template <class T>
+inline bool has_denorm_now() {
+   return get_smallest_value<T>() < tools::min_value<T>();
 }
 
 //
diff --git a/test/test_classify.cpp b/test/test_classify.cpp
index e6f51a0c77..609a355b04 100644
--- a/test/test_classify.cpp
+++ b/test/test_classify.cpp
@@ -11,6 +11,7 @@
 #include <boost/limits.hpp>
 #include <boost/math/concepts/real_concept.hpp>
 #include <boost/math/special_functions/fpclassify.hpp>
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
 #define BOOST_TEST_MAIN
 #include <boost/test/unit_test.hpp>
 #include <iostream>
@@ -105,7 +106,7 @@ void test_classify(T t, const char* type)
          }
       }
    }
-   if(std::numeric_limits<T>::has_denorm)
+   if(boost::math::detail::has_denorm_now<T>())
    {
       t = (std::numeric_limits<T>::min)();
       t /= 2;
diff --git a/test/test_difference.cpp b/test/test_difference.cpp
index b64f2835c4..7f4ff4f8f2 100644
--- a/test/test_difference.cpp
+++ b/test/test_difference.cpp
@@ -90,7 +90,7 @@ void test_values(const T& val, const char* name)
    test_value(-boost::math::tools::epsilon<T>(), name);
    test_value(boost::math::tools::min_value<T>(), name);
    test_value(-boost::math::tools::min_value<T>(), name);
-   if (std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present) && ((std::numeric_limits<T>::min)() / 2 != 0))
+   if (std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>() && ((std::numeric_limits<T>::min)() / 2 != 0))
    {
       test_value(z, name);
       test_value(-z, name);
diff --git a/test/test_gamma.hpp b/test/test_gamma.hpp
index 5b8c2724e7..c21573dac6 100644
--- a/test/test_gamma.hpp
+++ b/test/test_gamma.hpp
@@ -11,6 +11,7 @@
 #define BOOST_TEST_MAIN
 #include <boost/test/unit_test.hpp>
 #include <boost/test/tools/floating_point_comparison.hpp>
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
 #include <boost/math/tools/stats.hpp>
 #include <boost/math/tools/test.hpp>
 #include <boost/math/constants/constants.hpp>
@@ -319,7 +320,7 @@ void test_spots(T, const char* name)
       BOOST_CHECK(sign == -1);
    }
 
-   if(std::numeric_limits<T>::has_denorm && std::numeric_limits<T>::has_infinity && (boost::math::isinf)(1 / std::numeric_limits<T>::denorm_min()))
+   if(boost::math::detail::has_denorm_now<T>() && std::numeric_limits<T>::has_infinity && (boost::math::isinf)(1 / std::numeric_limits<T>::denorm_min()))
    {
       BOOST_CHECK_EQUAL(boost::math::tgamma(-std::numeric_limits<T>::denorm_min()), -std::numeric_limits<T>::infinity());
       BOOST_CHECK_EQUAL(boost::math::tgamma(std::numeric_limits<T>::denorm_min()), std::numeric_limits<T>::infinity());
@@ -336,7 +337,7 @@ void test_spots(T, const char* name)
    //
    // Super small values may cause spurious overflow:
    //
-   if (std::numeric_limits<T>::is_specialized && std::numeric_limits<T>::has_denorm)
+   if (std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>())
    {
       T value = (std::numeric_limits<T>::min)();
       while (value != 0)
diff --git a/test/test_ibeta_inv.hpp b/test/test_ibeta_inv.hpp
index 9e303e9235..ba98901773 100644
--- a/test/test_ibeta_inv.hpp
+++ b/test/test_ibeta_inv.hpp
@@ -8,6 +8,7 @@
 #define BOOST_TEST_MAIN
 #include <boost/test/unit_test.hpp>
 #include <boost/test/tools/floating_point_comparison.hpp>
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
 #include <boost/math/special_functions/math_fwd.hpp>
 #include <boost/math/tools/stats.hpp>
 #include <boost/math/tools/test.hpp>
@@ -305,7 +306,7 @@ void test_spots(T)
       BOOST_MATH_CHECK_THROW(::boost::math::ibeta_inv(static_cast<T>(2.125), -n, static_cast<T>(0.125)), std::domain_error);
       BOOST_MATH_CHECK_THROW(::boost::math::ibeta_inv(static_cast<T>(2.125), static_cast<T>(1.125), -n), std::domain_error);
    }
-   if (std::numeric_limits<T>::has_denorm)
+   if (boost::math::detail::has_denorm_now<T>())
    {
       T m = std::numeric_limits<T>::denorm_min();
       T small = 2 * (std::numeric_limits<T>::min)();
diff --git a/test/test_lambert_w.cpp b/test/test_lambert_w.cpp
index 6841f8d6fa..33269e68dd 100644
--- a/test/test_lambert_w.cpp
+++ b/test/test_lambert_w.cpp
@@ -491,7 +491,7 @@ void test_spots(RealType)
   }
 
   // denorm - but might be == min or zero?
-  if (std::numeric_limits<RealType>::has_denorm == true)
+  if (boost::math::detail::has_denorm_now<RealType>())
   { // Might also return infinity like z == 0?
     BOOST_CHECK_THROW(lambert_wm1(std::numeric_limits<RealType>::denorm_min()), std::overflow_error);
   }
diff --git a/test/test_next.cpp b/test/test_next.cpp
index 305dd8fee2..b4f05b4379 100644
--- a/test/test_next.cpp
+++ b/test/test_next.cpp
@@ -56,7 +56,7 @@ void test_value(const T& val, const char* name)
 
    BOOST_CHECK_EQUAL(float_distance(float_advance(val, 4), val), -4);
    BOOST_CHECK_EQUAL(float_distance(float_advance(val, -4), val), 4);
-   if(std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present))
+   if(std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>())
    {
       BOOST_CHECK_EQUAL(float_distance(float_advance(float_next(float_next(val)), 4), float_next(float_next(val))), -4);
       BOOST_CHECK_EQUAL(float_distance(float_advance(float_next(float_next(val)), -4), float_next(float_next(val))), 4);
@@ -122,7 +122,7 @@ void test_values(const T& val, const char* name)
    test_value(-boost::math::tools::epsilon<T>(), name);
    test_value(boost::math::tools::min_value<T>(), name);
    test_value(-boost::math::tools::min_value<T>(), name);
-   if (std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present) && ((std::numeric_limits<T>::min)() / 2 != 0))
+   if (std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>() && ((std::numeric_limits<T>::min)() / 2 != 0))
    {
       test_value(z, name);
       test_value(-z, name);
@@ -135,7 +135,7 @@ void test_values(const T& val, const char* name)
    if((_mm_getcsr() & (_MM_FLUSH_ZERO_ON | 0x40)) == 0)
    {
 #endif
-      if(std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present) && ((std::numeric_limits<T>::min)() / 2 != 0))
+      if(std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>() && ((std::numeric_limits<T>::min)() / 2 != 0))
       {
          test_value(std::numeric_limits<T>::denorm_min(), name);
          test_value(-std::numeric_limits<T>::denorm_min(), name);
diff --git a/test/test_next_decimal.cpp b/test/test_next_decimal.cpp
index 3d532335bc..07d59aa877 100644
--- a/test/test_next_decimal.cpp
+++ b/test/test_next_decimal.cpp
@@ -61,7 +61,7 @@ void test_value(const T& val, const char* name)
    }
    BOOST_CHECK_EQUAL(float_distance(float_advance(val, 4), val), -4);
    BOOST_CHECK_EQUAL(float_distance(float_advance(val, -4), val), 4);
-   if(std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present))
+   if(std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>())
    {
       BOOST_CHECK_EQUAL(float_distance(float_advance(float_next(float_next(val)), 4), float_next(float_next(val))), -4);
       BOOST_CHECK_EQUAL(float_distance(float_advance(float_next(float_next(val)), -4), float_next(float_next(val))), 4);
@@ -132,7 +132,7 @@ void test_values(const T& val, const char* name)
    test_value(T(-boost::math::tools::epsilon<T>()), name);
    test_value(boost::math::tools::min_value<T>(), name);
    test_value(T(-boost::math::tools::min_value<T>()), name);
-   if (std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present) && ((std::numeric_limits<T>::min)() / 2 != 0))
+   if (std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>() && ((std::numeric_limits<T>::min)() / 2 != 0))
    {
       test_value(z, name);
       test_value(T(-z), name);
@@ -142,7 +142,7 @@ void test_values(const T& val, const char* name)
    test_value(radix, name);
    test_value(T(-radix), name);
 
-   if(std::numeric_limits<T>::is_specialized && (std::numeric_limits<T>::has_denorm == std::denorm_present) && ((std::numeric_limits<T>::min)() / 2 != 0))
+   if(std::numeric_limits<T>::is_specialized && boost::math::detail::has_denorm_now<T>() && ((std::numeric_limits<T>::min)() / 2 != 0))
    {
       test_value(std::numeric_limits<T>::denorm_min(), name);
       test_value(T(-std::numeric_limits<T>::denorm_min()), name);
diff --git a/test/test_students_t.cpp b/test/test_students_t.cpp
index 59e3fc1860..ad4dc4187b 100644
--- a/test/test_students_t.cpp
+++ b/test/test_students_t.cpp
@@ -21,6 +21,7 @@
 #define BOOST_TEST_MAIN
 #include <boost/test/unit_test.hpp> // Boost.Test
 #include <boost/test/tools/floating_point_comparison.hpp>
+#include <boost/math/special_functions/next.hpp>  // for has_denorm_now
 
 #include <boost/math/concepts/real_concept.hpp> // for real_concept
 #include <boost/math/tools/test.hpp> // for real_concept
@@ -390,7 +391,7 @@ void test_spots(RealType)
          //
          // Bug cases:
          //
-         if (std::numeric_limits<RealType>::is_specialized && std::numeric_limits<RealType>::has_denorm)
+         if (std::numeric_limits<RealType>::is_specialized && boost::math::detail::has_denorm_now<RealType>())
          {
             BOOST_CHECK_THROW(boost::math::quantile(students_t_distribution<RealType>((std::numeric_limits<RealType>::min)() / 2), static_cast<RealType>(0.0025f)), std::overflow_error);
          }