Summary At Florence, in 1955, it was accepted that, in the problems of levelling, the notion ofpotential was scientifically clear, and that the altitude could derive from it only through a conventional process. In the same manner, when we want to have a geometric reference of the points at the earth surface, we use the coordinates (X Y Z) in thegeneral cartesian trihedron as fundamental unknowns, the geodetic coordinates (λϕH) deriving from (X Y Z) through a conventional process. Practically, in order to set up the observation equations, it is necessary to define local trihedrons (laplacian trihedrons), deriving from the cartesian general system through a fixed transformation, and to refer all the unknowns in these local trihedrons. All the observations used in Geodesy can be expressed simply and without any singularity in these local trihedrons. The links between classical geodetic nets, the astrogeodesy, the combination between classical and spatial geodesy, become easier. In astronomical controls, “longitude, latitude, azimut” must be replaced by: W-E deflection, N-S deflection and Laplace azimuth. Thus all the observation equations can be set, even in polar regions. A practical application of the new formulae was done successfully by L.F. Gregerson (Geodetic Survey of Canada).