The errors in phase-difference measurement which arise when using the Decca Navigator system, operating at frequencies near 100 kc/s, in the immediate neighbourhood of obstacles such as trees and telegraph wires, have been examined; the results of the measurements are described and discussed. Vertical obstacles show some uniformity of behaviour, but long horizontal conductors produce widely varying effects. A tentative theory of the phase errors due to a single tree is given in an Appendix, and fairly good agreement with observations is obtained.In Section 2 of the paper, errors obtained at good sites on the base line extensions of the red and green lattices are shown graphically. The effects of the near-field components and of the finite conductivity of the ground are apparent. Good agreement is obtained between the mean of the observations and a composite theoretical curve, based on the use of Norton's plane-earth theory and Bremmer's curved-earth theory of ground-wave propagation. Rapid fluctuations of error about the mean curve are observed, indicating that the phase velocity of the ground wave may show variations from point to point of up to 1% of the mean value.