Tufte shows you why it's important to do graphs well. Cleveland shows you _how_. The last quarter of the book details experiments in human visual perception that rank how well we detect certain things: relative angles not on a common baseline (i.e. pie charts) justly come out at the bottom of the list. One of a only handful of books I've labelled "JXH ONLY". If I loan you my copy, know that you are special.