On the Imperfection of the Geological Record