Robert_Charlton - 8:21 pm on Jun 20, 2011 (gmt 0)
I was reflecting that this is almost as if Google were parsing .htaccess files. I was also noticing that I couldn't find any "large" prominent sites (large enough to be OK to use as an example) that were having this problem. Then a little bell went off.
Large sites generally have dedicated servers, and they most likely use httpd.conf rather than .htaccess. Could that be one of the distinctions involved here? I'm also noticing that sites I know that are hosted on VPS, even though they are using .htaccess for canonicalization, don't show these results.
I am seeing problems from sites hosted on IIS as well, but that's another can of worms.
This isn't a complete answer, but it's a start, I think, and maybe someone can build on it.