Flight before the Wright brothers demonstrated their powered man–carrying flying machine was based on the philosophy of vehicles with high inherent stability combined with pilots whose only duty was to steer. This combination resulted in low controllability and poor manoeuvrability, with a strong response to the disturbances of rough air. The Wright brothers broke with the high–inherent–stability concept and combined inherently unstable aircraft with three–axis control operation by human pilots to achieve stable flight systems with good controllability and good manoeuvrability. This change from inherently stable to inherently unstable aircraft was the basic contribution made by the Wright brothers to start the age of flight.

In this paper, a picture of the state of flight control before the Wright brothers is built up from references to the literature. A similar picture is drawn for flight control developments after the Wright brothers, and the effect of their efforts on the progress of events is discussed. Plots with time as the independent variable and non–dimensional ordinates are used to illustrate the overall pattern of past developments in flight control and to indicate the course of future progress.

I intend in this lecture to discuss tlie ways in which I think analogue computing can be of assistance to aeronautics in the future. I shall therefore be less concerned with how they are built, than with how they are used, and in fact, what circumstances justify using them at all. Up to a few years ago the only large computors were analogue ones, so the user had no choice, but when the automatic digital computor became established as a practical machine (and even before) there were many people to suggest that the analogue was obsolete. This view is now less widely held, even by makers and designers of digital machines, and reviews of aeronautical computing aids have usually found some place for the analogue computor, without establishing any general principles on which the choice has been made, except, perhaps, to suggest that the accuracy required is the criterion.

I shall start by describing three of the types of computor most commonly used in aeronautics, the “ response simulator,” the “ electrolytic tank” and the “network analyser” and use these to illustrate the particular properties of analogue computing.

Wartime experience with aircraft steels emphasised the desirability of a complete revision of specifications for aircraft steels from both the technical and inspectional points of view. This was undertaken in association with the steel makers and the Society of British Aircraft Constructors under the aegis of the British Standards Institution. The aims were

1. (i) To reduce the number of basic types of steels to a minimum, consistent with satisfying the demands of aircraft engineers.

2. (ii) To transfer as many materials from the D.T.D. series of specifications to the B.S. Aircraft series.

3. (iii) To prepare an Inspection and Testing Schedule, which would exist separately from the Materials Specifications, but applicable to all steels in whatever form they may be required.

4. (iv) To achieve a degree of economy in alloying elements, again without detriment to the satisfying of designer's requirements.

The Inspection and testing schedule was issued as British Standard S100 in January 1949, but significant difficulties in its full application appeared in practice and it was not until the second half of 1952 that agreement was reached to put into practice the current standard 2 S100.

The work described was undertaken under a Ministry of Supply Contract to investigate the fatigue strength of certain materials where the expected endurance in use was less than the range of values usually examined in fatigue testing.

The material was supplied in the form of a 6 ft. x 3 ft. x 1/4 in. sheet of aluminium alloy MG5.

Tensile test pieces having a parallel length one inch wide by 4 1/2 in. long were taken in both directions from two diagonally opposite corners of the sheet. Later, similar longitudinal test pieces were taken from areas near the other two corners.

The plane bending fatigue test pieces were of the form shown in Fig. 1 and were all cut in the longitudinal direction. The waisted test portion was polished on the edges, the original sheet surface being retained on the faces. Steel plates were clamped to the ends of the test pieces by bolts passing through the holes in order to localise the strain in the test portion as much as possible.

The Classical method of solution of the stability of an axially–loaded continuous beam is by means of the three moments equation, using the Berry Functions, which are functions of the axial load. As the axial load approaches a value equal to the critical value for a pin–jointed beam, the Berry Functions tend to infinity, and the use of the three moments equations —(i. e. treating the end fixing moments as the independent variables)—leads to certain difficulties in the complete solution of the problem.

The major difficulty lies in the question of stability. The critical value is determined by the vanishing of the determinant of the coefficients of the fixing moments in the three moments equations. This value could be found by plotting the determinant against end load (c. f. Pippard and Pritchard). However, in a problem involving a large number of bays, the calculation necessary to do this is likely to be considerable, for there may be many branches to the curve.

As pointed out in previous work, in the method of collocation the functions used usually need not be orthogonal or integrable in closed form. Therefore function of different types may be mixed together in representing unknown quantities. Such function may be aptly called hybrid functions. Also, it has been found that usually it is better to represent unknown quantities by similarly looking functions instead of by a series involving higher harmonics. The idea is that by using similar looking functions it is possible to avoid poor regions for collocation, such as near nodal points, and so on, and closer fit in certain regions is assured. Thus if one knows approximately what the function sought looks like, one should try to find a set of function all of which resemble this desired shape, instead of taking an arbitrary set of harmonics when using the method of collocation. Simple examples below illustrate this approach and indicate the good results that can be expected in general.