“…However we look at curvature, it must deform the "symmetric" object (G, M ) into a "lumpy" object. All the existing approaches to the concept of curvature (see, for instance, [14], [29], [6], [8]) circle around this fundamental idea and this note and [2] are no exceptions. It is commonly accepted today (at least in Riemannian geometry, see [15] and the recent work [8] for parabolic geometries) that this lumpy object is a principal bundle P → N with structure group H ≃ H p , dim N = dim M, together with some extra structure on P → N , like a torsionfree connection.…”