This method applies to the observation of angles, usually obtained by repetitions; and instead of equating directions, as in the first method, its object is to furnish the most probable values for the several observed angles from a supernumerary number of angles.