The convert links process, and maybe recursive retrieval process are touching embedded javascript code and mangling the code up a bit. Historically, I believe wget stayed away from touching javascript.

I attempt to grab the following page using:

wget -rkE -l inf -P wget_test -D dimensionality.com http://www.dimensionality.com/freebeeexamples/freebieexample1-javascript.html

The original source code includes the following line:

document.write("<img src=\"rw1");

After wget converts links, the line reads as:

document.write("<img src="http://www.dimensionality.com/freebeeexamples/\&quot;rw1&quot;);"

Ideally this line should be left as is, untouched.

Tested on wget v1.16 custom compiled from savanna source, on GNU/Linux Debian 7, and package install of wget v1.18 on Cygwin.