Gravar-mail: Research groups: How big should they be?