The consideration of the effect of observational errors on the problem of manoeuvring ships at sea has, surprisingly enough, only recently become the object of attention by contributors to this Journal, though the basic background was provided many years ago by Sadler. The discussion at the Technical Committee of the Institute emphasized the importance of a statistical treatment of the problem and this led first to a brief note by myself in which systematic and random errors in range and bearing were discussed, and later to a full survey by Hollingdale in which it was shown that the total time to the instant of closest approach should be divided between observing time and manoeuvring time. An article by Proctor, though rather out of the main stream of the ideas pursued by the above mentioned contributors, breaks new ground in the sense that comparisons are made between different types of manœuvre on a probabilistic basis.