International audienceThe main theoretical claim of the paper is that a slightly revised version of the analysis of mass quantifiers proposed in Roeper 1983, Lønning 1987 and Higginbotham 1994 extends to collective quantifiers: such quantifiers denote relations between sums of entities (type e), rather than relations between sets of sums (type ). Against this background I will explain a puzzle observed by Dowty (1986) for all and generalized to all quantifiers by Winter 2002: plural quantification is not allowed with all the predicates that are traditionally classified as " collective ". The Homogeneity Constraint will be shown to be too strong and it will be replaced with the weaker requirement of Divisiveness (for both collective and mass...