The approximation property implies that convolvers are pseudo-measures

This paper (not for formal publication) grew out of the authors' attempts to understand Cowling's argument that for a locally compact group $G$ with the approximation property, we have that $PM_p(G)=CV_p(G)$ ("all convolvers are pseudo-measures".) We have ended up giving a somewhat self-contained survey of Cowling's construction of a predual for $CV_p(G)$, together with a survey of old ideas of Herz relating to Herz-Schur multipliers. Thus none of the results are new, but we make some claim to originality of presentation. We hope this account may help other researchers, and in particular, that this might spur others to study this problem.