<para>On the other hand, there are some features of MAD which make one
truly wonder what the designers were thinking — the handling of
-array indexing, for example. If 1962 weren't a couple of years too
-early to attribute that to an LSD overdose it would make a tempting
-hypothesis.</para>
+array indexing, for example, with two different addressing systems one
+of which could mutate at runtime. If 1962 weren't a couple of years
+too early to attribute that to an LSD overdose it would make a
+tempting hypothesis.</para>
<para>In most of the respects we've described, however, MAD is very
representative of its time. This will become apparent from a look at
<para>could have been used as tthe second statement of this program
instead of the BOOLEAN declaration.</para>
+</sect2>
+<sect2><title>Example 3</title>
+
+<para><emphasis>Problem:</emphasis> To approximate
+<subscript>a</subscript>∫<superscript>b</superscript> f(x) by Simpson's
+Rule, for an arbitrary interval [a, b] using N equal subintervals
+(where N is an arbitrary even integer).</para>
+
+<para><emphasis>Analysis:</emphasis>By Simpson's Rule</para>
+
+<!--
+This is not a displayed equation in the original, but sort of inlined.
+Also the a attached to the integral sign is before it, not after,
+but eqn doesn't seem to be able to do that.
+-->
+<mediaobject>
+ <imageobject>
+ <imagedata fileref='equation5.png'/>
+ </imageobject>
+</mediaobject>
+
+<para>where y<subscript>i</subscript> =
+f(x<subscript>i</subscript>), and a = x<subscript>0</subscript>,
+x<subscript>1</subscript>, ... x<subscript>N</subscript> = b are the
+partition points of the interval [a,b].</para>
+
</sect2>
</sect1>
</chapter>