Gould does an excellent job balancing readability and technical details in this account of the discovery and classification of fossils found at the Burgess Shale in Canada. Some of the fossils reveal truly bizarre creatures of the Cambrian period, which were shoehorned into existing phyla when originally classified, but upon recent, closer study were shown to belong to completely new phyla of their own. The fascinating creatures are thoroughly described.
Gould convincingly argues that these examples of wild early diversity undermine the idea of evolutionary 'progress' from a few basic forms 'up' towards ever-increasing complexity and diversity. Instead, the Burgess Shale points to huge diversity early-on, which was decimated by outside chance, its survivors (not obviously 'fitter' than those that didn't survive) determining life as we know it. Gould argues that this 'decimation by lottery' means that evolution could have very easily taken many entirely different paths than it did (the power of contingency), including no humans and no conscious creatures of any kind. Fascinating stuff.
Conway Morris, one of the paleontologists responsible for the re-classification of the Burgess Shale 'weird wonders' and who is featured prominently in this book, has written a book (Crucible of Creation) opposing Gould's interpretation of the Burgess Shale, and I'll probably give that book a go at some point.