I was kind of curious about the HTML documents that contain more than one million (and more) characters of inline CSS, and was even more curious after checking what the first two urls in Figure 5-2 looked like. Those websites should not rely on that much CSS at all - on the contrary, it would probably be quite a challenge to write their stylesheets in such a way that you actually use one million characters.

So I CTRL+U'd them and saw the following.

http://www.macuisinechezvous.com - Line 338 is where the HTML code suddenly stops with (linebreaks inserted for readability):

<a href="http://www.tournemain.com/" target="_blank">
<img src="http://ng38.free.fr/tournemain/banniere/reftournemain.gif" width="88" height="36" border="0" alt="Tournemain - recettes en image"></a>
<!-- Début du marqueur leguide34 http://www.leguide34.com -->

And then, on line 524630 (!), there's a closing </style> tag (where was it opened?). By the way, this is also a page that contains a dozen consecutive lines like this:

<p style="line-height:100%; margin-top:0; margin-bottom:0;" align="center"><span style="font-size:14pt;">&nbsp ;</span></p>

http://www.moundsviewschools.org/belair/home-BA.html - On line 19, a 3000-line inline <script> begins, which is completely empty as expected. Then, for the good looks of the page, they add 786-thousand lines of inline CSS that consist of only whitespaces. Another interesting part of their code (linebreaks removed for readability):

<br><br><br><br><br><br><br><br><br>
<br><br><br><br><br><br><br><br><br>
<br><br><br><br><br><br><br><br><br>
<br><br><br><br><br><br><br><br><br><br>
</body>
</html>

My point is, shouldn't these pages (and possibly more with ridiculously large (and suspect) inline CSS) have been ignored in the statistics?