Automation has proceeded much further in connection with air navigation and air traffic control than it has in comparable aspects of the maritime service. There are many factors to account for this. By its nature, air travel involves risks which are greater and more urgent than those of sea travel and which, in focal areas, are high in all conditions of weather; its passenger/cargo ratio is vastly greater and this inevitably stresses the need for every possible aid to safety. These characteristics and others more obvious lead to ground control, control leads to centralization and uniformity and thus to the ready introduction of advanced developments and their imposition where appropriate on the airborne user.

English sailors came more than half a century late into the world of transatlantic voyages and new discoveries. Nor should this surprise us when we remember the exasperating delays they had to expect when getting out of our long river estuaries into the narrow seas, and out of the narrow seas into the Atlantic Ocean. Once there they still had to encounter prevailing head-winds, where a Columbus had only to launch himself on the steady trade winds to discover America.

Most of the Ordinary Meetings of the Institute provide an opportunity for specialists to report on work that has been carried out, or experience that has been gained. From time to time, however, we have a meeting designed specifically for the purpose of discussion. The main aim of such meetings is to enable interested persons, with quite different backgrounds of knowledge and experience, to come together and consider collectively a subject of common interest from their various viewpoints. The present discussion is on the significance of gross human errors and blunders in navigation.

Future transport systems are difficult to predict, but one can perceive three main possibilities. The first is that transport will become entirely automatic, so that where, for example, one now has an automatic pilot during flight, one will also have automatic take-off and automatic landing. Flights will be pre-programmed, and will require no supervision in the vehicle itself or even from a remote-control room. Whether or not such a system is practicable is not within the scope of this paper; but there does not appear to be any purely mechanical problem connected with it that could not be overcome. However the production of equipment of the necessary reliability raises formidable human problems. problems. Not the least among these is the difficulty of maintaining rigorous standards of inspection; another is the rationalization and speeding up of faultfinding and repair procedures. It looks as though if we were to dispense with the human element in one sphere, its importance in another would be enhanced.

Captain Majendie would restrict the use of the word blunders to those aberrations of human behaviour which cannot be attributed to any avoidable cause. For my purpose this definition is unduly narrow and I use the concept in a slightly wider sense, namely: ‘errors, or aberrations, without apparent cause’. An understanding of the underlying causes of blunders is essential to their proper study, and to the reduction of blunder rates. Assuming that these causes are factors of human behaviour, there should be some connection between the occurrence and causes of blunders in different fields of human activity. In the hope that this may be so, I summarize the conclusions drawn from investigating blunders over a period of 25 years with a staff of 20–30 engaged on numerical computation and proof-reading.

Certainly genuinely unavoidable aberrations in human behaviour exist. In surface navigation, however, these momentary blunders seldom lead to fatalities such as collisions or groundings, although in shiphandling they can cause collisions when coming alongside or when manoeuvring in a confined space, but usually only of a minor nature.

The examination in the R.N. Navigation School of some sixty collisions and groundings has revealed that in only two have aberrations played any significant part in the subsequent disaster. This is quite easily explained for two reasons. First of all ships at sea usually close one another or approach a danger at a comparatively low speed. Furthermore the prudent mariner reduces speed when danger exists and when the situation becomes tense. There is therefore usually time for the nature of a blunder to be recognized and for the necessary corrective action to be taken, which, if not resulting in immunity, will reduce the gravity of the incident. Sixty knots is an exceptionally high relative speed on the surface, yet this provides minutes for human reaction as opposed to seconds or even split seconds normally available in the air.

In a recent paper I endeavoured to give a statistical solution of the problem of the separation of aircraft tracks with safety. In it I assumed that the distribution of errors of navigation made by the aircraft was a normal gaussian one; but I added a proviso that the effects of blunders might modify the result. From enquiries of various navigators as to the frequency with which they considered they were likely to make blunders I gathered that with a good navigator an undetected blunder might occur on 1 per cent of occasions and further it seemed reasonable to assume that a blunder might be defined as an error greater than three times the standard error of normal navigation. These assumptions are of course quite arbitrary but they seemed to be suitable to give examples of the effect on track separation; this was done in an unpublished paper. In these examples it was supposed that aircraft flying in the same direction were using two parallel tracks distant L apart and on each track aircraft were flying at fixed intervals but quite independently of the aircraft on the other track.

The problem considered in this note is that of determining an upper limit for the chance that an error will not exceed some given quantity. Where the pattern (or frequency distribution) of the errors is either known or assumed, the solution to the problem is merely a matter of calculation. This note starts by making the minimum assumptions about the pattern of errors, and then examines the effect of gradually increasing the restrictive assumptions.

In the course of the development of the practice of astronomical navigation, a straightforward approach to the subject could never be adopted mainly because time (that is, Greenwich Mean Time) either could not be obtained or could not be relied upon to adequate precision. Methods of observation and techniques of reduction were introduced to give the best results from the equipment then available; and these traditional methods, which have continued long after the need for them has vanished, have tended to obscure and to complicate the fundamental simplicity and generality of astronomical navigation.

However much opinions may vary as to the proper way to treat the problems raised by the use of radar, there is no doubt that something must be done to help navigators resolve them. The International Conference on Safety of Life at Sea held in London in June 1948 saw fit to declare in its Recommendation No. 19 that the possession of radar in no way relieves a vessel from the obligation of strictly observing the requirements laid down in the Rules for Preventing Collisions, and in particular with Rules 15 and 16 (sound signals in fog, moderate speed in fog, &c). It is common knowledge that navigators have not found constructive guidance in this text, and accidents have multiplied. It has been argued that the increase in accidents has not been proportionately as great as the increase in tonnage, but the argument is not very weighty. In the first place, accidents have become particularly serious because of the size and crossing speed of ships; and in any case it is a poor consolation to masters anxious to observe the Rules, but not quite certain how to do so, to be told that there is no cause for anxiety as there are not as many accidents as all that. This argument, even if it does not actually discourage navigators anxious to observe the Rules still runs the risk of making hesitant navigators err on the wrong side. It is therefore necessary to do something, and the first thing seems to be to state the problem concisely.

The flight instrument systems used in large modern transport aircraft represent a considerable development from the standard instrument panel in use at the beginning of this decade. Radio and navigational information is coordinated in a single display and flight direction is provided by means of a separate instrument in which it is normally composited with attitude information. The system contains sufficient redundant information to provide positive warning of a malfunction, and is normally used in conjunction with an automatic pilot which is capable of carrying out the approach accurately and reliably, monitoring control being exercised by the pilot by means of the flight system. Synthetic damping and automatic throttle control of airspeed may also be provided.

Statistics based on autopilots in current airline service suggest that an acceptable standard of safety for scheduled automatic landing is not attainable by single-channel autopilots. Adequate safety during the automatic landing manoeuvre can, however, be attained by a multiplex autopilot now under development. Provided that the necessary ground aids are installed, it appears feasible to introduce scheduled automatic landing by 1970.

The accuracy of winds measured in flight using a doppler navigation system has been investigated by comparison with simultaneous measurements from a Meteorological Office radar wind-finding station and also by flying different tracks using a modified multiple-drift technique. The errors of wind measurement arise from errors in the ground-speed vector as measured by the doppler navigator and also from errors in the true airspeed vector as obtained from the readings of the airspeed indicator, altimeter, outside air thermometer and compass. The trials, which were made at various altitudes to 48,000 ft. in a Canberra aircraft of the Meteorological Research Flight, showed that, providing suitable in-flight calibrations are carried out, the error in the vector wind can be reduced to about 5 kt. but that without these calibrations it may be 10 kt. or more.

In this paper, a group of eight directional stones on the coral island of Arorae in the Gilbert Islands are described. The stones seem to have been used to direct canoe voyages to distant islands with the help of horizon stars en route, the stars' movements and bearings being studied and memorized in relation to the correct stone on the night before departure. They perhaps mark the site of a Polynesian navigation school which existed somewhere between A.D. 1000 and 1500. The directions given by the stones appear to have been corrected for drift due to the Equatorial Current. So far as is known these stones are unique, and have not previously been reported in print.

On the island of Arorae in the Gilbert Islands there is a group of eight or nine stones which are said by the natives to have been used by early canoe voyagers to set their courses on departure to distant islands. No other such stones for navigation have been reported from the Pacific, which makes the coral stones of Arorae of unique value in the history of navigation. The stones are some of the few pieces of real evidence in the long-standing controversy on the nature and range of Polynesian navigation. The stones are called ‘Atibu-ni-Borau’ by the natives, which means ‘stones for voyages’. They have no legends or traditions which shed any light on the origin or actual use of the stones, but there are several clues available from elsewhere in the Pacific which enable the use of the stones to be determined with a high degree of probability.

A new type of parallel ruler on rollers, devised and patented by Mr. Ian A. McIntosh of Sydney, deserves the attention of marine navigators. The new ruler is apparently similar to the brass-roller type used in the Navy, but it has the great advantage of omni-directional mobility. It also has much easier manœuvrability than the parallel rules used in the Merchant Service, known as Captain Field's or the ‘walk-about’ type.

In the Brest to Las Palmas de Gran Canaria race held last August by the Sail Training Association I was skipper and navigator of the 7-ton yawl Artica II of the Italian Navy (which incidentally obtained the best corrected time of all three classes). From the navigational point of view the race offered no particular difficulty because although long—1400 miles—we were on a reach most of the time, a condition that makes the navigator's job much easier. It might in any case be interesting to those who are keen on ocean racing, to hear about some of the dodges (certainly not new to them) used on board Artica II:

(a) A gnomon pin (about two inches long) had been welded in the centre of the compass rose: when the Sun was low and the boat sufficiently steady the shadow of this needle enabled us to have very good compass checks.

I was particularly interested in Commander C. E. N. Frankcom's note on the Beaufort wind scale at the end of the article by the Hydrographer of the Navy on Admiral Sir Francis Beaufort (this Journal, 11, 266).

Commander Frankcom states that from time to time attempts have been made to abandon the Beaufort scale and adopt some other form of scale but that efforts to do this have failed. Efforts to improve on or change the form of the Beaufort scale are, in my opinion, likely to continue to fail for the very good reason that the existing scale is almost flawless mathematically.