Decision-making procedure based on fuzzy relational composition may require that the involved connectives are not idempotent to handle signi cant information properly. Yet, non-idempotence may result in meaningless inference when the universes of discourse are dense. Here we propose a novel kind of connective and a formal de nition of relational composition for dense universes of discourse which allow the aggregation of di erent path of reasoning con rming each other without incurring in total loss of information.