g1smd - 11:06 pm on Sep 25, 2006 (gmt 0)

If a site is fully indexed as www and all links inside the site always point to www, and all non-www requests are 301 redirected to www, then you can't get any of it indexed at non-www, ever. It is bomb proof to that particular effect.

If a site links internally to www, but does not have the redirect in place, then if you link from outside to their non-www version, then the non-www page you linked to gets indexed. However, because all internal links are hard-wired to www, there is no back up vote for that non-www URL from within the site, and so the non-www URL soon gets dropped out of the index again.

If a site points internally using relative links, and does not have a redirect installed, if you now link to their www URLs then those www URLs get indexed. All those www URLs point to yet more www URLs within the site, and they get indexed too.

If you now also link to them using the non-www then all the links out of those pages also point to non-www URLs (it's relative addressing, right), and so all of those non-www URLs get indexed too.

Where internally the site can use either format, they will now have a very big Duplicate Content mess to fix.

.

I won't elaborate on some of the other finer points of this, but I have tested it to see what happens and to find ways of bomb-proofing sites against such measures.

Those initially started at just www and non-www issues, and have extended into the handling of multiple domains, varying parameters, URL CaPiTaLiSaTiOn issues, http vs. https, and so on, too.