Directional statistics deals with probabilities of directions (unit vectors), axes (lines through the origin) and rotations. For example, one can use directional statistics to model the distribution of bond angles in molecules, wind directions or observations on the celestial sphere. The book by Mardia & Jupp provides a unique overview of this field of statistics, and contains a wealth of information on distributions and techniques that deal with directional data. If you find yourself dealing with directional data by using conventional distributions (ie. on the wrong manifold) or by discretizing your data (and thereby making the problem intractable), buy this book and find out about the right way to deal with your data. This 'right way' is moreover often surprisingly straightforward and tractable!