Efforts to determine the mean size of ship ‘domains’ from measurements of the distribution of ships in the vicinity of an individual ship have been made by several workers, particularly by Goodwin, who made extensive observations in the Sunk Lightship area of the North Sea. The work involved plotting the positions of vessels at time intervals of a few minutes to estimate the number of ships in an annulus about a central ship, or alternatively the number per unit area at various distances from a central ship. These figures were used to obtain distribution curves from which the average domain could be estimated. The work of Curtis and Barratt on the validation of simulator studies of passing manoeuvres in fog, in which the nearest point of approach of ship's tracks was determined, serves to illustrate an alternative method based on the distribution of the density of tracks. In both procedures use is made of the point at which the distribution function first returns to the value which would obtain if the distribution were uniform, that is, if ship domains were absent. Some doubts have been expressed as to the validity of using the position of the cross-over point as a measure of the average domain size, and also whether it would be more appropriate to employ the position of the maximum value of the functions.