After a comment-to-comment talk with Branco Collin, I realised that the reason why the stripped-down front page that is fed to my various websites shows up with such a large font (the default) in Firefox is that the page is broken somehow. A few passes through the W3C CSS and HTML validators helped me find the error that caused CSS to be ignored. However, both validators would be a lot more useful if one or more if the following things were true:
1. if it was possible to validate CSS if the HTML page it is on contains errors; or
2. if it was possible to set the validator to ignore ampersands, which cause the bulk of the error messages in the validator and which in a blog with multiple entries take a non-trivial amount of time to fix (meaning they're still not fixed - I checked in Firefox if what I thought was the cause of the problem had gone away, and it had, so I stopped).
I understand the need for strictness in XHTML. But the point of the validator is to help weed out errors, not to distract a developer with technically correct but ultimately useless information.
