Tim,
Be aware that for sites with a large proportion of single-visit sessions (like blogs, where a user reads a single article), the time-on-site value can mislead. Google Analytics fires a single call each page load, and the time-on-page calculation is the difference between the page load timestamp and the subsequent page load. From this it is clear that the last page in a visit does not get a time-on-page assigned - and for single-page visits there is no accumlated time.
However, Google Analytics calculates time-on-site by dividing the total accumulated time by the number of visits (including bounce visits for which no time is accumulated), so the reported average time on site is lower than the actual time-on-site. If this is a problem, then there are a number of documented methods to track time-on-page by firing GA events after an elapsed time (I'd only suggest doing this for the first page on a session), or alternate analytics packages which automatically log while the page is open (check out ClickTale).