The orthometric height and the holonomity problem

When height networks are being adjusted, many geodesists advocate the approach where the adjustment should be done by using geopotential numbers rather than the orthometric or normal heights used in practice. This is based on a conviction that neither orthometric nor normal heights can be used for the adjustment because these height systems are not holonomic, meaning–among other things–that height increments (orthometric or normal) when summed around a closed loop do not sum up to zero. If this was the case, then the two height systems could not be used in the adjustment; the non-zero loop closure would violate the basic, usually unspoken, assumption behind the adjustment, namely that the model claiming that height differences are observable is correct. In this paper, we prove in several different ways that orthometric and normal heights are theoretically just as holonomic as the geopotential numbers are, when they are obtained from levelled height differences using actual gravity values. This disposes of the argument that geopotential numbers should be used in the adjustment. Both orthometric and normal heights are equally qualified to be used in the adjustment directly.