In order to ascertain the force and direction of the wind, I made an extremely simple and effective apparatus which is fully shown (see Fig. 38). Whilst conducting these experiments it occurred to me, when a large aeroplane was used, that after it had been travelling for a considerable time, it would impart to the air in the path of its travel, a downward motion, and that the lifting effect would be greatly reduced on this account. In order to test this, I provided four light brass screws and mounted them, as shown at _x_, on a hardened polished steel point much above their centre of gravity, so that they balanced themselves. On account of the absence of friction, they were easily rotated, and responded to the least breath of air that might be moving. One morning when there was a dead calm, I placed four of these screws equidistant around the whole circle. Some of them rotated very slowly in one direction and some in another; some alternated, but all their motions were extremely slow. However, upon setting the machine going with a large aeroplane and a powerful screw, I found after a few turns that the air was moving downwards around the whole circle at a velocity of about 2 miles an hour, but as the screw was a considerable distance below the aeroplane, I estimated that the actual downward velocity of the air in which the aeroplane was travelling was about 4 miles an hour. The result of my experiments are clearly shown in an unpublished paper which I wrote at the time, and as it is of considerable historical interest, I have placed it in the appendix, notwithstanding that there may be certain repetitions. [Illustration: Fig. 36.--The machine attached to the end of the rotating shaft--_a_, _a_, the body of the machine; _b_, _b_, four-legged spider secured to _a_, _a_; _c_, _c_, parallel bars; _d_, _d_, vertical member to which the aeroplane _g_, _g_ is attached; _h_, _h_, the screw; _f_, _f_, wires for preventing distortion of the aeroplane.] [Illustration: Fig. 37.--Marking off the dynamometer. In order to ascertain the actual amount of power consumed in driving the propeller, a brake was put on in place of the screw, a weight applied, and the engine run at full speed. In this way all the uncertain and unknowable factors were eliminated.]