Books used to contain original research and theories. (Think Stanley Milgram's book Obedience.) It's only in the last 30 years that science books tend to just be collections of other people's papers rewritten for the lay audience. Because modern academia rewards people for churning out large amounts of derivative work but not for spending time coming up with groundbreaking theories, pretty much the only time scientists write books is when they get 600k advances (like Gilbert).