Realising that my previous test may have been somewhat 'unfair', in that it hit the CPU frequently, I came up with a revised test to show battery life under 'light browsing'.
Input conditions were simple; WiFi on, minimum brightness, and SeaMonkey pointing at a website that auto updates the page. This should simulate - to some extent - light surfing.
Now these are curious results... especially when compared to the last set. While it shows that overclocking will shave runtime of your battery, it also shows that underclocking to 360MHz will make you loose runtime too.
Perhaps there is a sweet spot one can underclock to to improve battery life under Pupeee, perhaps there isn't. But 360MHz most assuredly is not that sweet spot... perhaps 450MHz would be better? Perhaps it will be just as bad? Could it be that Aus got it right and the lines cross at 630MHz?
More than three hours of runtime on a 5200mAh battery is pretty good, no matter how you look at it. Makes me almost afraid to test with my aftermarked 6600mAh battery... logic dictates that I should get a bit more than four hours out of it.