(In)accessible comonads and (non)Grothendieck toposes

Hi all, I am just wondering where it was first stated (for both directions) that the category of coalgebras for a comonad on a Grothendieck topos E is again Grothendieck if and only if the underlying endofunctor of E is accessible. A modern argument might go as: the topos of coalgebras is Grothendieck if and only if it is locally presentable if and only if the endofunctor is accessible, the original probably just mentioned preservation of filtered colimits. Many thanks, David Roberts [For admin and other information see: http://www.mta.ca/~cat-dist/ ]