diff --git a/src/float-precision.c b/src/float-precision.c
new file mode 100644
index 00000000..264eb4eb
--- /dev/null
+++ b/src/float-precision.c
@@ -0,0 +1,26 @@
+/*
+ * Compile and run the program, observe what happens.
+ * Then recompile it with the -DADDTWO and observe once again.
+ * Why is that ?
+ */
+
+#include <stdio.h>
+
+int
+main(void)
+{
+	float f = 1 << 23;
+	size_t cnt = 0;
+
+	printf("%zu\n", sizeof (f));
+
+	while (1) {
+		printf("%16.2f %zu\n", f, cnt++);
+#if ADDTWO
+		if (f >= 16777216.0)
+			f += 2;
+		else
+#endif
+			f += 1;
+	}
+}