Empirical estimates of cumulative refraction errors associated with procedurally constrained levelings based on the Gaithersburg- Tucson refraction tests of the National Geodetic Survey

Analyses of results of the National Geodetic Survey's leveling refraction tests indicate that the standard deviation about the mean (σ) for high-scale minus low-scale rod readings closely correlates with measured refraction error. Use of this relation in conjunction with values for σ obtained from routinely constrained surveys provides a basis for estimating the refraction error associated with levelings of stipulated order and class.