The meaning of the fact that the angular momenta of protons and neutrons are oriented in opposite directions, shown by Otsuka [Phys. Rev. Lett. 71, 1804(1993] for the angular-momentum-projected Nilsson wave functions, is reexamined using a two-rotor model. It is shown that this fact does not necessarily mean an unphysical situation that proton and neutron ellipsoids are rotating freely in opposite directions. On the contrary, it originates in a close binding of the two ellipsoids, which is accompanied by a large spreading of the relative angular momentum due to the uncertainty principle. It is also demonstrated that the elimination of the spurious center-of-mass motion does not substantially change the situation.