This book is filled with superb graphics and insights into the study of variability and the graphical methods that can be used to summarize data. It is written in many short chapters, each of which is based on one of the authors famous papers, that can be easily digested in about 5 or 10 minutes. While a few places Wainer goes off on brief tangents about the math (that distract from the flow and are hard to follow because the lack of detail) the book flows remarkably well especially given its fractured origins.
The graphics (especially the color plates) are a well rendered collection of great historical value. Graphics by Playfair, Florence Nightingale, Galton, Minard and people tracking deaths in Nazi occupied Lithuania are all talked about in detail. Beyond the art, some of the strong points of the book include: insights into what goes wrong when people do not study variability, the dissection of good and bad graphics to point out what makes a good graphic good, and a great discussion about the graphics of changes over time and how easy it is to misinterpret them.
The down side of the book is that many of the graphics can be rendered fairly easily with modern analytic software (like SAS, S-Plus or R) but the author does not give links for the novice. If you want to combine this with a book that discusses how to make good scientific graphics (like the examples in this book) take a look at the masterpieces by Cleveland Visualizing Data and/or The Elements of Graphing Data.