This paper first gives a brief account of the deviation theory applicable to magnetic compasses and a unified theoretical treatment of the principal heeling-error and pitching-error effects. Consequences of symmetry in soft-iron correctors are then considered, and a system of six equal soft-iron spheres at equal distances along three mutually perpendicular axes is shown to have no effect at the compass position, any diametral pair thus having a complete negative counterpart in the remaining four.

This leads to the deduction of the possibility of a simple form of corrector system in which soft-iron correctors remove those parts of the principal inclination errors which are due to induced magnetism. If the remaining part due to permanent magnetism is assumed to be removed by vertical magnets in the usual manner, the correctors when once adjusted would then require no readjustment on change of locality, in distinction from all known correctors at present in use or in course of manufacture, which require continual readjustment in these conditions.

The calculation and publication of the errors in measured altitudes due to tilt of the sextant serve a useful purpose in impressing users more strongly with the need for care than does the unbacked advice to rock the sextant always given in the text-books, and also in indicating that carelessness carries greater penalties the greater the altitude. The work once done is done for ever, so we may as well get the sum right.

More than 1200 American radar equipments have been fitted to ships and this paper gave same American views on its operation; the views put forward were not necessarily different from those held in this country but certain experiences of inland navigation and blind pilotage which would be new to British operators were described. It was stated that when commercial production of marine radar started in the U.S., opinion was divided on the relative merits of 3 cm. and 10 cm. radar, some manufacturers concentrating on one and some on the other. Although 10 cm. equipment is still manufactured, all manufacturers now have a 3 cm. radar in production.

Certain advantages of radar assisted navigation were mentioned and these included the ability to determine another vessel's course and speed by plotting and the considerable saving in fuel and time that could be achieved by the accurate knowledge of the ship's position in coastal waters no matter what the weather.

Anomalous propagation in the Great Lakes region was described as being so common that shipmasters often regard the increased range of radar in the area as being the normal range of the set.

The use of radar by tugs on the Mississippi and its tributary the Ohio was described. Here up to fifteen barges of 100 ft. are ‘shoved’ by the tugs into narrow locks in weather which sometimes conceals half the tow. The small bump in the trace, representing the lock, is recognized by the pilots. One of the outstanding problems of this form of radar navigation was stated to be that the bridges (of which there are 1700 over the Mississippi) were such good reflectors that discrimination between them was not possible at a mile, at which distance the tow gets into position to pass between the piers.

It may be useful at the end of this symposium to ask broadly what has been achieved so far in the field of navigational radar and what are the outstanding problems.

In this country alone, in the four years since the end of the war, rather more than 300 commercial radars have been manufactured to an exacting specification and fitted to merchant vessels, and a much larger number has been produced in the United States; during 1949 the British radar industry states that it hopes to fit more than a ship a day.

I wish to thank Dr. Clemence for pointing out1 the flaw in the interpretation drawn in my paper that the error in observed altitude due to tilt of the sextant is a maximum when the altitudes are in the neighbourhood of 45°. The purpose of a correction table is either to supply the correction or to indicate the most favourable or unfavourable circumstances. I agree that my table does not fulfil this purpose; the argument β is unknown. Moreover the table gives corrections for all combinations of altitude and β this is incorrect, as some values of β are impossible for some altitudes, and a square form of table is therefore not permissible.

A method of obtaining aircraft altitude to within 50 feet at altitudes of 10,000 to 20,000 feet is described. This accuracy is necessary in a few special cases. The method consists of determining the height of a pressure level at some convenient point by radar or other means, and then calculating from a knowledge of the wind vector the change of height of this pressure level between the check point and the point at which an accurate altitude is required. The limitations of alternative methods of height finding, such as the use of an aneroid altimeter corrected for temperature at the upper level, are discussed.

For most purposes the aircraft type aneroid altimeter gives a sufficiently accurate indication of height, provided corrections are made for the temperature at flight level. In the application of radar to topographical surveying, however, it is necessary to know the altitude of the aircraft used in the measurement to within 50 feet when flying at heights of 10,000 to 20,000 feet. This can be achieved only by special techniques. The method here outlined has been used extensively during investigations into the use of radar for surveying. It is simple and appears to give satisfactory results.

In a paper under the above title (Journal, Vol. 1, p. 266) J. P. G. Worlledge has made the useful suggestion that sextant certificates should give a critical table for the corrections instead of the present practice of giving the corrections at a few equally spaced readings. Unfortunately the method which he gives of deriving the proposed form from the present form is unsound. The centring error is correctly stated as

where Z is the sextant reading (the difference between Z and the true angle in this formula is negligible). But part of the centring error is included in the index error, so that to obtain the part of the centring error shown on the certificate the value of C for Z= 0 must be subtracted. This yields

for the formula of the correction.

This paper describes how it was found possible to estimate the accuracy, under operational conditions, of navigators' wind-determinations by the 3-course drift method. This was done by studying the self-consistency of the set of three drifts logged by the navigator; the technique which was developed, using the size of the circle inscribed in the ‘triangle of error,’ can be used expeditiously by persons with no special training. This technique is applicable to other position line problems. Of the total inaccuracy of the winds found by the navigators, three quarters was contributed by the errors of observation of the drifts, and one quarter by the process of deriving the wind from the set of drifts, using the Dalton Computor.

The validity of the method was checked by comparing wind measurements by the 3-course drift method with the value obtained from Air Position Indicator readings taken in the course of practice bombing exercises; the accuracy of the latter value was also estimated from internal consistency. The check was satisfactory; and in particular, contrary to widely held belief, showed that the navigators did not systematically underestimate drift angles.

In planning a flight over polar regions, celestial navigation is of paramount importance: Loran and Consol coverage is doubtful and Gee chains non-existent; interpretation of the P.P.I. screen is hampered by lack of accurate survey and the camouflaging of coastlines by ice packs. Thus navigational accuracy and the amount of activity needed to maintain it depend largely on the foresight with which the flight is planned to render celestial information available.