Tests have been carried out on round bar specimens of various strong aluminium alloys (unclad) in the fully heat-treated condition in both tensile and compressive loading. A careful study of the results establishes beyond reasonable doubt that there is a progressive change in the tangent elastic modulus with stress. The variation is appreciably greater than that predicted by the consideration of second-order effects in the conventional stress and strain representation. It is found that the modulus decreases with tension and increases with compression by about 5 per cent at 30 tons/in.2, the form of the variation being a continuous function. Tests on two alloys of aluminium other than those of the normal zinc-bearing and the copper-bearing types were also carried out. The first of these was a simple fully heat-treated binary alloy of 4 per cent copper which was very similar in behaviour to D.T.D.364; the second was a 5 per cent alloy (N.S.6-½H) which, after its initial loading, showed only very slight elastic modulus variation in the subsequent loading and unloading lines. An additional test to compare naturally and artificially aged D.T.D.364 did not reveal any difference in the degree of modulus variation.

In view of the present experimental findings, the use of the “second” modulus line in the proof stress determination of unclad strong alloys should be adopted generally, if practicable. A high level of internal stress can cause slight errors in this method because the elastic limit is then prematurely reached. As an occasional check on the method when applied to routine mechanical testing, the elastic component of strain can be found for the last point on the loading line by returning to the zeroing load.

In the Appendix the difference between true and nominal stress-strain relationships is calculated. The meaning and theory of modulus variation is also discussed and aluminium alloys are compared with a high-tensile steel.

A simple expression for the form of the variation is proposed.

The drag in two-dimensional flow is expressed in terms of the distributions of static and stagnation pressures along a traverse line downstream of the aerofoil. Sources of error are discussed with regard to their effect on the accuracy of drag measurement in small tunnels at medium Mach numbers.

Several plate buckling problems are solved, using a reduction method. By this method the solution of an orthotropic plate can be reduced to the solution of an isotropic plate and the solution of a plate with bi-axial loading can be reduced to the solution of a plate with uni-axial loading and so on. Plates with simply-supported ends and various boundary conditions at the sides with uni-axial and bi-axial loading are considered and the necessary reduction equations are given.

A systematic discussion of partitioning as a tool for matrix inversion is presented, together with various methods and applications which have been of help in actual computations. New concepts are introduced, among them those of super-matrix and of square partitioning. The most usual type of partitioning, that into 2x2 sub-matrices, is discussed in detail, showing the orderly arrangement of the calculations in an auxiliary matrix. Further sections deal with matrices of the continuant type, and with special types of symmetry in the arrangement of the sub-matrices. The greatest advantage of the method of partitioning for the inversion of these types (as compared with the elimination method) lies in a considerable reduction in the number of arithmetical operations.

A method is given for the solution of the flutter determinant on a general purpose electronic digital computer. The method has been programmed for the N.R.D.C. 401 Computer and details of this programme are given. A quicker programme, applicable when only structural damping is present and there are two degrees of freedom, is also discussed. Appendix I discusses in detail the types of errors which occur when the method is applied and Appendix II considers the anticipated loss of accuracy which will be encountered when using the method in cases where the number of degrees of fredom is large.