On the orbit structure of the logarithmic potential