Remarks on the Delaunay reduction