Correct typography in correction

content.html

@@ -421,7 +421,7 @@ <h2>Handling math with varying precision</h2>
     The obvious solution to that is rounding: after each mathematical operation, if the result does not have a bit pattern in our system,
     it snaps to the closest available value.
     (The default rounding mode is supposed to be <a href="https://en.wikipedia.org/wiki/Rounding#Rounding_half_to_even">rounding half to even</a>,
-    as it minimizes error accumulation - but most CPUs use round-to-nearest by default anyway.)
+    as it minimizes error accumulation &mdash; but most CPUs use round&hyphen;to&hyphen;nearest by default anyway.)
 </p>
 <p>
     Notice that now, each bit pattern actually represents a range of values &mdash;