"Wow! Three times as good as a Treo 750!"

Except it's not, of course. Even if it's too much to ask for error bars on this sort of graph (the Iphone's lead over the Samsung appears to be about three minutes in five-and-a-half hours: is their testing really accurate to the nearest 1%?), surely it's not rocket science to start the graph at the origin?

I think statisticians are like Empegs: most of the people who direly need one, don't even know it...

Peter