Ronald Fisher's The Design of Experiments (link is Wikipedia rather than Amazon since it is long out of print) is interesting for historical context. The book is often credited as founding the whole field, and certainly did a lot to promote things like blocking, randomisation and factorial design, though things have moved on a bit since.
As a period document it's quite fascinating, but it's also maddening. In the absence of a common terminology and notation, a lot of time is spent painstakingly explaining things in what now seems comically-stilted English. If you had to use it as a reference to look up how to calculate something you'd probably gnaw your own leg off. But the terribly polite hatchet job on some of Galton's analysis is entertaining.
(I know, I know -- how the readers of tomorrow will laugh at the archaisms of today's scientific literature...)