diff --git a/examples/readme_examples.cpp b/examples/readme_examples.cpp
index 92853c5..db891e3 100644
--- a/examples/readme_examples.cpp
+++ b/examples/readme_examples.cpp
@@ -213,6 +213,11 @@ bool basis_operation() {
     | std::ranges::to<std::u8string>()};
   return to_utf8_1 == to_utf8_2;
 }
+
+static_assert(
+  !std::ranges::equal(u8"foo" | to_utf32, std::array{U'f', U'o', U'o', U'\0'}));
+static_assert(
+  std::ranges::equal(u8"foo" | to_utf32, std::array{U'f', U'o', U'o'}));
 #endif
 
 bool readme_examples() {
diff --git a/paper/P2728.md b/paper/P2728.md
index 80c9f55..2ac491e 100644
--- a/paper/P2728.md
+++ b/paper/P2728.md
@@ -286,6 +286,27 @@ The table below compares the error handling behavior of the `to_utf16` and
 
 ![](../../paper/images/error_handling.png){ width=750px }
 
+### Special CPO Logic for String Literals
+
+The `to_utfN` CPOs use a heuristic to detect null-terminated ranges and omit
+the null terminator so that it doesn't appear in the output. If the input
+range satisfies `is_bounded_array_v`, is nonempty, and its last element is 0,
+then the last element is omitted.
+
+Without this logic, you'd have:
+
+```c++
+static_assert(
+  std::ranges::equal(u8"foo" | to_utf32, std::array{U'f', U'o', U'o', U'\0'}));
+```
+
+Instead, with the heuristic in place, the null terminator is eliminated:
+
+```c++
+static_assert(
+  std::ranges::equal(u8"foo" | to_utf32, std::array{U'f', U'o', U'o'}));
+```
+
 ### The `_or_error` Views Are Basis Operations for Other Error Handling Behaviors
 
 You can use the `_or_error` view to implement the same behavior that the