One of the conclusions reached by recent studies of Barnes et al. and latr Hide was ''that atmospheric excitation alone was sufficient to account for the observed polar motion over (the studied) period, that there is apparently no need to invoke substantial excitation either by the fluid core or... earthquakes.'' The purpose of the present paper is to point out that their argument that led to the above conclusion is unjustifiable (hence whether the conclusion is in reality true or not is still an open question). I demonstrate this through a physical ''thought'' experiment and a numerical simulation. In essence, they show that if we want to compare a geophysically observed excitation function &psgr;a(t) with the excitation function deduced (via deconvolution) from the polar motion observation m(t), we should do so directly (the ''direct approach'') To compare m(t) with the polar motion computed (via convolution) from &psgr;a(t) (the ''integration approach''), as Barnes et al. and Hide did, is misleading.