On the possibility of a second kind of mantle plume