It is always reassuring to find a kindred spirit thinking along the same lines as oneself. R. Williams's proposal for a table of latitude parts (34, 247) is a reiteration of my paper of twelve years ago (23, 233).

The equations of a rhumb-line track were established by Sadler (9, 371) and it always seemed to me to be a more elegant solution to solve the equations themselves rather than to apply corrections to inaccurate equations, hence the approach suggested in my paper.

Lenart derives boundaries of a Collision Threat Parameters Area (CTPA), which defines the range of velocity and heading which own vessels must not adopt if it wishes to maintain a specified minimum spacing from a given threat. Lenart's treatment of the problem is algebraic, and a geometric demonstration of the validity of the CTPA boundaries may be of interest.

In Fig. 1, O is the position of own vessel. OX and OY are Cartesian axes, fixed in direction as own vessel manoeuvres. T is the position of a threat, whose velocity is represented on the diagram by a line TT´ whose length gives the distance the threat will travel, relative to own vessel's present position, in some time τ, 0·2 hours, say.

When computers were first used for navigation — mostly air navigation — there were debates as to whether all the computing should be done centrally or implemented by a series of smaller computers, each doing certain limited tasks. The debate was largely academic, since the technology for distributed processing did not really exist. Consequently, the development of navigation systems was centred on single processors and special-purpose input and out-put devices.

In Gerald Forty's paper published in Vol. 36, September 1983, line 36 on p. 325 should read: ‘When using the lower limb with a shade…’

A technical session on the above subject was held in London on 13 April 1983. The topic was introduced by Mr M. St Erme Cardew, who first stressed the distinction between navigational data and positional data. The former, unlike the latter, are referenced by time so that there is a sense of urgency in processing navigational data. Inherently these are always out of date, possibly only slightly, due to the finite processing time required. Some extrapolation is always required. He recognized the importance of errors, possibly gross ones, and distinguished these as due either to the assumptions made by the navigator being incorrect or to errors in the data. He advised checking the calculational methods, possibly by undertaking, via a computer, two or more independent analyses. Checking that incoming data were both stable and within specification was imperative. In an emergency it was necessary that human intervention into computer-produced information be accepted.