experiments were being tried, we will say at 50 miles an hour, it was always possible to turn on steam until the red liquid mounted to 50. This device was very simple and effective, and saved a great deal of time. In order to prevent the twisting of the radial arm, a piece of stiff oval steel tube 12 feet long was secured between the arms at _j_, and on each end of this tube were attached the wires _u_, _u_. This not only effectually supported the end of the arm, but at the same time prevented twisting and made everything extremely stiff. Of course, while the machine was running at a high velocity, centrifugal force had to be dealt with, and in order to prevent this from causing friction in the articulated joints of the weighing apparatus (Fig. 36), thin steel wires _k_, _k_ were provided. As this apparatus was in the open, it was found that the slightest movement of the air greatly interfered with its action. On one occasion when a fabric covered aeroplane, 4 feet long by 3 feet wide, was placed in position, the four corners being held down by the wires _v_, _v_, and the apparatus driven at a high velocity, a sudden gust of wind snapped two of the wires, broke the aeroplane, and the flying fragments smashed the screw, and this notwithstanding that each of the four wires was supposed to be strong enough to resist at least four times any possible lifting that the whole aeroplane might be subjected to. [Illustration: Fig. 35.--The little steam engine used by me in my rotating arm experiments; the tachometer and dynamometer are distinctly shown.]