A couple of recent weblog posts, at Microdoc News and Virtuelvis, have echoed something I've been meaning to write about for a while — that Google makes such heavy use of a page's <title> contents when indexing a site.
For a long time this site had a PageRank of zero, although for a few days now, it's been four. What I noticed was that, despite have a PR of zero, and being linked from a vanishingly small number of other places, my articles were often turning up in the first page of Google results for two-word queries containing words from my article titles. Since my posts appear on pages of their own, the post titles are part of the page titles (along with my name). Only rarely do I get visitors that haven't been referred by searches containing words from my page titles.
Google is also pretty accurate about which page is returned. Even though my index page contains the full text of most of my posts, the post titles are only <h2> on the main page, but <h2> and in the <title> on the archive pages. Google always picks the archive pages for a given query. The only exception being in a search for Martin Kenny, which appears in all page tiles — it dutifully returns the site's main page, and the About page.
Since I've discovered this, I've become a little self-conscious about what goes in my post titles :-)