MR. M. GREGORY commented that the slow convergence or divergence of the moment distribution process at loads near the critical load had tended to make the calculation of critical loads for elastic rigidly jointed frames by this method rather tedious. He quoted the reason Dr. Bolton had given: “Basically this difficulty arises because the testing distortion used is not the critical mode, but merely the rotation of one joint. If the critical mode were to be used as the testing distortion, one cycle would be sufficient to decide whether the calculations were converging or not.” Mr. Gregory thought this had become clear to most people who had used the Hoff moment distribution convergence criterion. It took many distributions for the effect of a single disturbance to be felt throughout the whole structure, and many more for the carry-over to reflect back to the originally disturbed joint, and the first step towards reducing the length of the calculation was obviously to apply disturbing moments at several joints rather than at one joint, particularly if the disturbances could be given the correct sign, which was often the case if the desired buckling mode could be pictured. This method had previously been used in the University of Tasmania,
but it was interesting to see it carefully worked out in detail in Dr. Bolton’s paper, particularly with regard to the information which could be gained by keeping running totals of the disturbing moments after each distribution, and he complimented Dr.
Bolton on this advance.