The differences in moments of inertia of a homogeneous globe of viscous liquid arising from steady convection in its interior are evaluated on the basis of a linearized theory of such motion as developed by Chandrasekhar in 1952. It is shown that if the physical characteristics of such a globe are made to approximate the well-known properties of the Moon, with the same difference C - A of momenta about its principal axes of inertia, Chandrasekhar's theory requires the velocity of convective motion to be of the order of 10 −8 cm/sec (too small for the establishment of steady flow for a globe of lunar dimensions in 10 9 years); and the observed Rayleigh number is several hundred times as large as that required theoretically for the stability of the respective flow. As the physical basis of Chandrasekhar's theory should be closely applicable to the lunar problem, it is concluded that internal convection conforming to it—if it exists—does not lend itself for the actual explanation of the difference in momenta C - A of the lunar globe as deduced from its librations.